Written evidenceCONTENTS

Memorandum from the City of London CorporationAnnex A: Comments regarding the Content of the 2011 Census Household Questionnaire:Annex B: National Statistics Small Area Geography Policy Review and ConsultationMemorandum from Local Government Data Unit—WalesMemorandum from the South West of England Regional Development AgencyMemorandum from the Demographics User GroupMemorandum from the Statistics User ForumMemorandum from Professor David Martin, Royal Statistical Society Study GroupMemorandum from the Joint Industry Committee for Population Standards (JICPOPS)Memorandum from the Census and Geodemographics Group of The Market Research SocietyMemorandum from The Market Research Society (MRS)Memorandum from the Statistics CommissionMemorandum from Suffolk County CouncilMemorandum from Mr A R Thatcher CBMemorandum from CACI Ltd.Memorandum from the Greater London AuthorityMemorandum from Newcastle upon Tyne City CouncilMemorandum from the South West ObservatoryMemorandum from CLIP Population Sub-group, LA-side, Nottinghamshire County CouncilMemorandum from East Sussex County CouncilMemorandum from the Association of Census DistributorsMemorandum from the London CouncilsMemorandum from South Tyneside CouncilMemorandum from Milton Keynes CouncilMemorandum from the London Borough of BrentMemorandum from Slough Borough CouncilAPPENDIX AMemorandum from Islington CouncilMemorandum from Wandsworth CouncilMemorandum from Hertfordshire Census Users' GroupMemorandum from the Local Government AssociationAnnex AAnnex BMemorandum from the LGA High Ethnicity Authorities Special Interest GroupMemorandum from the London Borough of MertonMemorandum from the London Borough of NewhamMemorandum from Luton Borough CouncilMemorandum from the Metropolitan Police ServiceMemorandum from the Equality and Human Rights CommissionMemorandum from the Economic and Social Research CouncilMemorandum from Westminster City CouncilMemorandum from the Local Intelligence Network Cornwall (LINC) Business and Economy ModuleMemorandum from Barnet CouncilMemorandum from the London Borough of CamdenMemorandum from the Association of Regional ObservatoriesMemorandum from the Medical Research CouncilMemorandum from Mr Philip RedfernMemorandum from Tyne and Wear Local Authority Districts'APPENDIX 1: ESTIMATING INTERNATIONAL IN-MIGRATION—HOW THE NEW ONS METHODOLOGY DISADVANTAGES THE NORTH EASTMemorandum from Cornwall County CouncilMemorandum from Worcestershire County CouncilMemorandum from Manchester City CouncilMemorandum from David Coleman, Professor of Demography at Oxford UniversityAPPENDIX 1: SOME COMMENTS ON CURRENT UK SOURCES OF POPULATION AND MIGRATION DATAAPPENDIX 2: LIST OF VARIABLES FOR WHICH DATA IS MAINTAINED IN THE BELGIAN MUNICIPAL REGISTERS

Memorandum from London Borough of HackneyMemorandum from the Office for National StatisticsAnnex A: RECOMMENDATIONS OF THE INTER-DEPARTMENTAL TASK FORCE ON MIGRATION STATISTICS

Memorandum from the City of London Corporation

INTRODUCTION1. The City of London Corporation welcomes the opportunity to contribute to Sub-Committee's

inquiry. The City of London works closely with Greater London Authority through the Demographics Liaison Group and has been supportive of the constructive dialogue the City of Westminster has initiated on the robustness of population and international migration statistics at a local level. The City Corporation supports much of what has been said by the GLA and the Central and Local Information Partnership subgroup. The paragraphs below set the particular views of the City Corporation.

USES AND DEFINITIONS2. Information regarding the estimation of population and the projection of population is used to feed

into planning policy approaches, departmental business plans and ultimately service delivery. Inaccuracies in the population counts can, therefore, have an effect at the end user point of delivery of services with increased pressure placed on services if the number of residents is underestimated and local authorities consequently will be under funded. This leads to an inability to cope with an increased residential population in terms of environmental services, community and children's services as well many other service functions the local authority is expected to deliver.

3. The usual resident population is defined by ONS as1

“someone who spends the majority of their time residing at that address. It includes:

@UL@— people who usually live at that address but who are temporarily away from home (on holiday, visiting friends or relatives, or temporarily in a hospital or similar establishment) on Census Day;@/UL@

@UL@— people who work away from home for part of the time, or who are members of the Armed Forces;@/UL@

@UL@— a baby born before 30 April 2001 even if it was still in hospital; and@/UL@

@UL@— people present on Census Day, even if temporarily, who have no other usual address."@/UL@

The definition of usual resident used in the Census is a clear definition. It is only sufficient, however, if a further defining question is asked on second residences to clarify the number of usual residents who work away from home for part of the week. As is stated in the City Census 2001 document published by the City2 :

“In Central London visitors contribute significantly to the total number of people requiring local authority services, normally during the working week. Therefore, incorporating visitor numbers is integral to the planning of services in areas where many persons occupy residential premises on short term lets or use the premises as second homes. Unfortunately no visitor information was collected during the 2001 Census. Therefore visitor numbers will have to be estimated from Census dwelling counts."

No information on short term migrants or visitors is included in the resident estimates. This is a significant shortfall of the estimates, as these figures are also required to accurately estimate service needs, as both short term migrants and visitors require local authority services.

ROLE OF THE CENSUS4. The Census of Population is the only complete source of demographic information provided on a

consistent basis throughout the country and is vital for establishing need and demographic profiling of communities. The Census is currently too infrequent and using 2001 data for over 10 years does not show an ideally accurate quantification of need within communities. Further surveys and local needs assessments must be completed in the interim period. Central London boroughs have high levels of migration and high population turnover due to a highly transient population. Measuring this every 10 years therefore is not sufficient as much can change in this time.

5. A consistent interim Census that provides basic demographic information half way through the 10 year period would be useful in ensuring the accuracy of estimates and projections and providing more detailed

1 Census 2001: Definitions, ONS, 20042 City of London Resident Population: Census 2001: An introduction, June 2005

information that cannot be compiled from other sources. This does not have to be a complete Census, but one with a number of basic demographic questions.

6. The City of London's Census 2001 response rate was third lowest in London. The Census test run in Camden in May 2007 provided some valuable insights into improving the Census response rate. Many of the reasons for low response rates given by Camden are similar to those in the City of London. These reasons include gated estates, mobile 24 hour population, logistical problems with the posting back of forms and a high proportion of visitors. The City of London fully supports the constructive comments submitted by the London Borough of Camden. The recommendations given by Camden can be found in the document “2007 Census Test: London Borough of Camden Evaluation Report" (August 2007), a number of which are as follows:

@UL@— ONS should work closely with local authorities;@/UL@

@UL@— a local address list should be used (where it can be provided); and@/UL@

@UL@— embark on a publicity campaign for the Census.@/UL@

7. Related to this, it is important that the public have a clear understanding of the importance of the Census to service delivery. An explanation of this should be included in Census forms to ensure the public has access to this information.

8. The City of London believes the Census 2011 would be more beneficial if it included questions on second residences (most importantly) and industry. The reasons for this are given in the consultation document to ONS “Comments regarding the Content of the 2011 Census Household Questionnaire" submitted to ONS on the 12th of February 2007, this document is attached (Annex A).

9. In order to ensure good quality, effective consultation responses and statistical information it is important to have effective co-ordination at all levels. At a London wide level constructive dialogue between boroughs and other authorities is coordinated by the GLA. At a national level the work done by ONS is welcomed, however, mistakes were made during the 2001 Census compared to the results of the 1991 Census. It is therefore important to learn the lessons from previous years and take into consideration the changing nature of a more transitional population base.

10. The format the Census data is made available in is, on the whole, a useful way to distribute the data. The geography used for the distribution of the data has been consulted on by ONS and the City of London has commented in the paper “Census 2011 Response" submitted 16 February 2007 (attached at Annex B). The City makes comments regarding a business or employment geography as well as the need for more robust data at lower output levels.

11. Integration with other datasets is important, as the Census is used in many different ways by many different authorities, however it is difficult to deliver integration in practice. The most important aspect of this is to ensure clarity and focus of the Census, effective good quality results are key.

MID YEAR POPULATION ESTIMATES12. The mid year population estimates, as previously mentioned, are generally soundly measured for the

City, however the exclusion of short term migrants and visitors provides an underestimate of the amount of people the City is required to provide services for. Being an international business City, the City of London has a large proportion of short let accommodation which is used by business visitors for whom services need to be provided year round. The City has a high proportion of migrants and a highly mobile population, from the 2001 Census 23% of residents were migrants, 4% of these were migrants from outside the UK. Similar issues to these are faced by neighbouring boroughs.

13. The User need in terms of mid year estimates for the City therefore relates to the inclusion of short term migrants and visitors, as these groups are currently not counted despite their use of services. It is essential that the City Corporation is aware of the number of people using short term accommodation. The fundamental Census base data is therefore essential for all future population counts and it must be ensured that this is of a high quality.

SURVEY AND ADMINISTRATIVE DATA14. The monitoring of Development activity is important for local authorities as this tracks the number

of residential properties and tenure, which is a useful source to be able to validate population data but is by no means

a complete source for population estimates. Data from General Practitioners and Health Services do not provide robust data for a central business district such as the City due to high migration and high population turnover. The City also has a high proportion of young 20-35 year old men who are well known to rarely register with a doctor.

CO-OPERATION WITH STAKEHOLDERS15. Due to the importance of these figures to all local authorities in England, accuracy is essential as

these figures feed into the delivery of services at every level, including setting policy requirements. The ONS has made every attempt to consult with users at the borough level as can be noted from this document, where ONS consultation responses are included. In order for ONS to provide valuable statistics it requires the necessary investment to ensure the data it delivers is the data that is required by local authorities, regions and England as a whole. ONS needs to be resourced and supported to maintain a constructive dialogue with other bodies such as local authorities and should use this channel as a useful source of information and collaboration to deliver high quality statistical data fundamental to planning local authority services.

November 2007

@ANAME@Annex A

Comments regarding the Content of the 2011 Census Household Questionnaire:

INTRODUCTION:The City of London Corporation, and this response, is in support of comments given by the Greater London

Authority, the CLIP subgroup, Local Government Association and London Councils. A key issue for the City is the need to quantify occupancy of second residences.

THE CASE FOR SECOND RESIDENCES:Introduction:

At present there are a small number of sources where data on second homes is available such as the Survey of English Housing, the Omnibus Survey and the Census. However each of these have draw backs, the SEH does not collect information at Local Authority level, the Omnibus Survey is available at a cost and there are problems with each of the above using a different definition of a second home. For these reasons estimates of second homes in the City range from 10% (Census, 2001) to 27% (ODPM; Savills Research).

It is necessary to have a consistent well-defined source of data on second residences which can provide accurate base data for analysis, as the number as second residences impact highly on the delivery of services within the working week in the City of London.

Census 2001:

In the Census of Population 2001, information was collected on residences that were not considered to be the respondents' main residence. However this question was not clearly defined as to who this includes. This results in some ambiguity with the term “not main residence".

According to the Census 2001 Glossary a “Usual Resident" is defined as:

@UL@”someone who spends the majority of their time residing at that address. It includes:@/UL@

@UL@— people who usually live at that address but who are temporarily away from home (on holiday, visiting friends or relatives, or temporarily in a hospital or similar establishment) on Census Day;@/UL@

@UL@— people who work away from home for part of the time, or who are members of the Armed Forces;@/UL@

@UL@— a baby born before 30 April 2001 even if it was still in hospital; and@/UL@

@UL@— people present on Census Day, even if temporarily, who have no other usual address".@/UL@

This definition means that people who work away from home for part of the week are enumerated at their family home and not at their work away from home. For the City this results in an undercount of the number of occupants of residential premises during the working week resulting in undercounting of persons in occupancy of residential accommodation and utilising local authority services.

However, with a question on second residences there is a risk of over-counting respondents at both their addresses. In order to reduce the risk of over-counting a clear definition of second residences is required with regard to which household constitutes the second residence.

The data format released in accordance with question 17 in the Census 2007 Test would be useful in the format total number of people occupying second residences during the week in a tabular format. As this could then be compared with the number of residents living permanently in the City.

Questionnaire Design and Layout:

The City of London also supports the need for a four page questionnaire and therefore supports Option C and Option D detailed in the document “Taking Forward the Content of 2011 Census Household Questionnaire".

Support for 4 Page Questionnaire:

Industry:The Annual Business Inquiry provides a source for industry of workers in the City of London classified according to SIC codes, however this does not provide information on City residents and their industry of employment.

Industry within the context of the Census is essential for providing in depth analysis of worker profiles. The origin-destination statistics allows cross tabulation analysis of the entire City of London workforce. Knowing the number of workers within an industry and being able to cross tabulate where they are travelling from, their age structure and sex for example provides a valuable overview to the structure of the City employment. This feeds into policy decisions and service provision.

Industry information also provides valuable base data for comparison with the ABI to track changes to the workforce structure.

Carers:

Information regarding the number of carers is required to compile:

@UL@— Information strategies around identifying and supporting carers to encompass the fact that very substantial numbers of carers start caring every year@/UL@

@UL@— Costs associated with caring, such as charging for non-residential community care services, to lower the long term impact of caring.@/UL@

@UL@— All health and social care strategies to ensure that carers needs for support are identified early and the appropriate support provided early to mitigate the negative effects of being a carer@/UL@

This information would benefit form being collected by a consistent and reliable source such as the Census of Population.

It is important to note that the key aspect for the City of London with regard to Census data is the quality of the results and the usability of the data. Data that is consistent and that can be reliably used at a ward level is essential.

Date: 16 February 2007

@ANAME@Annex B

National Statistics Small Area Geography Policy Review and Consultation

COMMENTS FROM THE CITY OF LONDONThis document comments on the Census geography used for 2001 and how it will be used in 2011. The City

of London is in support of the GLA submitted documents and responses and those submitted by the CLIP subgroup for this consultation.

For the City of London it is most crucial that a new view of statistical disclosure control alongside the flexibility of boundaries is implemented to obtain both confidentiality and usability of data.

THE 1991 CENSUS RESULTS:The 1991 Census geography divided wards into enumeration districts, some of these districts were

suppressed due to confidentiality but the majority of the areas with large residential populations had data available that could be built up to form all the City's residential areas including the Barbican, Golden Lane, Mansell Street and Middlesex Street. Therefore analysis could be undertaken for areas that had meaning to local authority staff and partners.

The Map 1 below illustrates the residential areas (shaded in green) that are defined in the City's UDP and were analysed in the 1991 Census. The 2001 Census was unable to analyse at this level. Analysis by these neighbourhood definitions is highly desired by the City of London.

REPORTING AND ANALYSING 2001 CENSUS RESULTS:The City has had a number of problems in analysing 2001 Census data at a local level. Due to the disclosure

control measures the City has no reliable Census data for 2001 at output area and ward level. Therefore the City cannot use Output Areas as a building block for other geographies as used in 1991. Lower layer Super Output area data is available, however the randomness of the boundaries prevents any real useable data analysis (this will be dealt with later in this document).

The new measures for output areas and disclosure control prevent any meaningful comparison with previous years Census except at borough level.

RANDOMNESS OF BOUNDARIES:ONS have attempted to constrain the output areas to conform to similar population homogeneity

characteristics; however, for the City with areas of distinctive residential characteristics this has not been achieved. Map 1 illustrates the residential areas and the residential distribution of the City as compared to the Lower level Super Output Areas.

Map 1

SUPER OUTPUT AREAS IN RELATION TO CITY OF LONDON RESIDENTIAL AREAS AND DISTRIBUTION

/home/PKB/DATA/386539/FOLIO8T.EPS> SPECIFIC EXAMPLES OF SUPER OUTPUT AREA PROBLEMS:It can specifically be seen in Map 2 that parts of the Barbican development, which has residents of similar

demographic characteristics, has been included with the Golden Lane Estate which has distinctly different characteristics. Therefore data analysis of the area cannot be done accurately.

Map 2

GOLDEN LANE ESTATE AND SUPER OUTPUT AREA

/home/PKB/DATA/386539/FOLIO8B.EPS>Map 3 shows the Barbican Residential Area has been split into three very different LSOA's with half of the

Barbican incorporating part of Smithfield and both halves losing residents to the SOA incorporating Golden Lane. Therefore data analysis of the Barbican as a whole is difficult.

The Barbican has a large number of residents and the minimum population thresholds can be reached for the area.

Map 3

THE BARBICAN RESIDENTIAL AREA AND LSOA

/home/PKB/DATA/386539/FOLIO9T.EPS>The four smaller areas shown in Map 4 below (the Temples; Smithfield; Carter Lane and Queenhithe) would

ideally be able to be analysed individually, however as population thresholds are likely to be breached it is possible to group these four areas together, as long as boundaries of the areas are completely within the LSOA.

Map 4

SMITHFIELD, TEMPLES, CARTER LANE AND QUEENHITHE RESIDENTIAL AREA AND LSOA

/home/PKB/DATA/386539/FOLIO9B.EPS>The LSOA shown below incorporates 2 residential areas (Petticoat Square and Mansell Street). Where the

City's residential areas do not conform to minimum thresholds for LSOA's it makes sense to group similar areas together such as is done with the Super Output Area below, E01000005. Map 5 shows a good example of grouping areas as the SOA encompasses the entire area without breaking up any residential areas boundaries.

Map 5

PETTICOAT SQUARE AND MANSELL STREET LSOA

/home/PKB/DATA/386539/FOLIO10.EPS>At present the boundaries of the Super Output Areas are illogical and don't tie in with any local concepts

including current ward boundaries. Ward geography will always be needed by local authorities and a set of output areas that can be built up to create wards would be useful.

The City of London does not currently have the capability with output areas to build them into meaningful areas as all output areas are subject to disclosure control and using building blocks that have been through various disclosure processes does not allow for anywhere near accurate information. The LSOA area are also seen as too big to be used as small area building blocks for wards. The City recognises the need for confidentiality as well as the need for high quality data and would like the ONS to consider the idea of a “licensed" user by permitting controlled access to raw data according to restrictions on the outputs. This would make it possible for authorities to create areas of their own choice (and would possibly make the City able to use OA to aggregate to residential areas.)

The ward boundaries will constantly change but it is necessary to have information on these areas for electoral processes. Local Authorities are essentially electoral authorities and thus require information at electoral geographies. Output Areas should therefore relate to wards. Shown in the maps below are the current wards and the pre-1994 wards and how they relate to the LSOA's. The LSOA's make sense according to the previous wards and at that level they could have provided at least some useful information, the fact that boundaries are currently completely unrelated makes analysis difficult. Commonality of boundaries is seen to be useful in data analysis.

However it is essential that the areas that various departments are working with make sense to them locally and are areas they can identify with.

City of London current wards and Super Output Areas

/home/PKB/DATA/386539/FOLIO11T.EPS>City of London pre-1994 wards and Super Output Areas

/home/PKB/DATA/386539/FOLIO11B.EPS> WORKFORCE OUTPUT AREAS:Output Areas are based on residential populations and household size ranges. However in the City there is a

large workforce (about 300,000) that far outweighs the number of residents (about 9,000). The workforce is distributed more evenly than the residents across the entire City. The link attached provides valuable information on the employment distribution and the location of the City's LSOA's. [http://213.86.34.248/NR/rdonlyres/C9D54DCA-B0F5-429C-AD2E-88B13D7C08D4/0/DP_PL_EmploymentDistribution.pdf].

For this reason the Super Output Areas used are not representative of the workforce for example Super Output Area E01000004: City of London 001D encompasses the majority of the City's workers. The Map below illustrates the current LSOA's and the 2003 workforce distribution. It can be seen that the majority of the workforce fall in one large LSOA.

The ONS release employment data such as the Annual Business Inquiry at SOA level but as the SOA's are not based on workforce there is little analysis that can be done with this.

It would be very useful if these output areas could be split further into smaller business Super Output Areas as this would allow for further analysis of the ABI data. Alternatively for an area as small as the City and with a large workforce, ABI and employment data would be very useful at OA level as this should not breach disclosure control policies.

/MILES/PKB/DATA/386539/PageEv11.eps,s(35p6)>The number of workers in the City also plays an important part in the electoral process as the City is the only

area in the country in which the number of workers significantly outnumbers the residents, and therefore, to be truly representative of its population, offers a vote to City organisations so they can have their say on the way the City is run. Accurate data on the workforce is then useful in determining franchise in the City.

@HR25@

Memorandum from Local Government Data Unit—Wales

BACKGROUNDThe Local Government Data Unit—Wales (Data Unit) is an organisation created to bridge between the

public and private sectors. It was established in April 2002 with the aims of

@UL@— Improving the collection local government data@/UL@

@UL@— Encouraging the development of additional sources of information, and@/UL@

@UL@— Improving the capacity of government to take evidence based decisions on policy and service delivery.@/UL@

The Data Unit is wholly owned by the Welsh Local Government Association (WLGA) and limited by guarantee. The funding comes from local government and Welsh Assembly Government Statistical Directorate.

The Data Unit has worked most closely with the Welsh Assembly Government and Welsh Local Government, but also with other public sector organisations, to provide a service valued by our clients. As a part of this work the Data Unit has become involved with local and central government on population in relation to both the financial and service issues involved.

Staff of the Data Unit sit on the Central and Local Information Partnership (CLIP) census group, Office of National Statistics (ONS) local authority census liaison steering group, Census Advisory Group for Wales and 4Nations liaison committee. We have also been in discussions with other government departments about administrative data which supplements our knowledge on population numbers and attributes. The Data Unit is a joint chair of the Welsh Statistical Liaison Committee.

In responding to the request for evidence we will consider the following population data sets offered by H.M. Government

@UL@— Mid Year Estimate of Population (MYE) issued 14 to 15 months after the base date,@/UL@

@UL@— Internal and international migration estimates,@/UL@

@UL@— National population projections,@/UL@

@UL@— Small Area Population Estimates (SAPE) issued 18 to 20 months after the base date,@/UL@

@UL@— Census of Population taken every 10 years.@/UL@SUMMARY

The Data Unit welcomes this opportunity to address the Select Committee on the concerns we have identified from Welsh local government and other persons in Wales who are involved with serving the public interest.

A useful starting point for the discussions is the article “A Demographic Statistical Service for the 21 Century" by Peter Fullerton3 which considers many of the questions asked by the Committee.

The essential elements of counting the population are for resource allocation, both financial and to the client groups identified from the count. The census provides the only reliable small geography information source with attribute data for the population.

As society becomes more complex in its living arrangements and dynamic in terms of relocations, it is more important to have reliable, timely benchmarks of the population. The purposes to which the population estimates and census data are put are becoming more complex and as such require different interpretations of the bases for counting that population, be it residents, day time or seasonal.

3 Demographic Statistical Service for the 21 Century, Peter Fullerton, http://www.statistics.gov.uk/about/Methodology_by_theme/downloads/Demographic_Statistics_Service.pdf

The main concern relating to the issue of population and migration is that of uncertainty and pressure on services. It is recognised that there is uncertainty about the population in Britain at any point in time and the need to improve that information.4 The financial impact of migration was highlighted by the report jointly between Home Office and DWP.5

The cost to local authorities to provide services for this difference between official figures and locally based delivery to those present has been recognised. Emergency funding of around £250 million (Local Government Association and Guardian article6 ) has been requested.

To undertake a reliable count of the population there needs to be either an universal mandatory population register, or reliable information on the locations where the people to be counted can be found. In Wales (and possibly Scotland and Ireland) this requires the recognition and equality of use of languages other than English.

Every ten years when the results of the Census of Population are released, there has been a need to re-base the population estimates for the following ten years and correct those made for the last ten year period. The need to reduce the gap found at the end of each ten year period has been one of the reasons for the formation of the ONS Improving Population and Migration Statistics project.7

It is recognised that administrative and survey data is used to compile population estimates. It is also likely that more administrative sources could be used to test and quality assure the outcomes. There will need to be quality checks made on the administrative sources to ensure they are themselves of known reliability.

The uncertainty introduced by the gap between official estimates and observations, either every ten years or by those working with the population day to day, leads to tensions between the expectations of the population, the service providers and central government. This does not help mutual trust, relations between organisations or community cohesion.

Many are concerned that the uncertainty is a symptom of the Treasury restrictions on funding the ONS with regard to survey and research initiatives, particularly relating to the International Passenger Survey and other surveys of the population present.

Another concern is for the continuing improvement of communication between governmental organisations, particularly central government departments. Better communication will enable departmental initiatives to be less myopic and enable cost savings through improved sources of administrative data. In recent years ONS have been engaging more with users of and experts in population related data, this is welcomed. The extension of this trend is supported by the local government community. Perhaps the Statistics and Registration Services Act will help through the activity of the scrutiny department on methodology used and to enable the release of more statistical information to support population estimation.

Lessons on the balance between reliability of administrative data and the need for national censuses can be learned from other EU and international statistical services.8 However, without a population register and national spatial addressing infrastructure, it is difficult to see how administrative sources can replace a census.

RESPONSE TO THE REQUEST FOR EVIDENCE

1. USES AND DEFINITIONS

4 Paper to the Home Office Migration Impacts Forum “Improving International Migration Statistics", Peter Goldblatt (ONS) 21 June 2007.5 Cross Department submission to the House of Lords Select Committee on Economic Affairs (Cm 7237, October 2007).6 Local Government Association press release 1 November 2007 http://www.lga.gov.uk/PressRelease.asp?lSection=0&id=SX7C5E-A78498C8, and report “Estimating the Scale and Impact of Migration at the Local Level, Institute of Community Cohesion “Guardian newspaper, Tania Brannigan—Economics Correspondent, 1 November 2007).7 Aims of the project to reduce the problem of large differences between census and population estimates http://www.statistics.gov.uk/about/data/methodology/specific/population/future/imps/default.asp8 Census strategic review Alternatives to a Census: Review of international approaches, ONS, October 2003 www.statistics.gov.uk/downloads/theme_population/international_approaches.pdf

What are the uses of population estimates, and how far do current population estimates meet the needs associated with those uses? What are the effects of inaccuracies or inadequacies in such estimates?Population estimates are used in a local government context for a myriad of purposes. The initial concern is that in Wales grant funding9 is largely based upon population statistics and the need of those counted. It is evident that inaccuracies, in the absolute number or the numbers of persons with specific needs which feature under the funding formula, will lead to under-funding of the services provided by the authorities affected. A rule of thumb for non-specific groups is around £1,000 per person, thus an uncertainty of 1,000 persons would amount to about £1 million of an authority's budget, this uncertainty will affect a combination of staff numbers and services offered.

Local government requires the population estimates to analyse need, identify policies to tackle need, plan services, monitor performance and to evaluate the outcomes of any interventions provided. This requires not only absolute numbers but also the attributes of the population.

There are a considerable number of Acts of Parliament and guidance from government sources that require local government to evaluate their services against their population, specifically in terms of gender, ethnicity, disability and sexual orientation. These all require a detailed knowledge of the base populations.

The current population estimates are issued over a year after the date to which they relate. There has been some recent discussion as to the adequacy and reliability of the sources of migration data, especially for those travelling from outside the country. Efforts have been made by ONS to address the problems identified by work on the Improving Migration and Population Statistics project10 (IMPS). Reports from the IMPS project have identified a number of specific weaknesses where further research is required, a particular concern to a number of Welsh authorities is the enumeration of foreign students, many not in private households.11

The largest element of change in population statistics is usually the migration element. The birth and death information is usually reasonably reliable, unless applied to an inaccurate base population, where the estimates will thus themselves be unreliable. Returning to the migration element, internal migration is based on census information or health registrations and migrant surveys.

The census is taken only once each 10 years and asks only about moves within the 12 month period prior to the enumeration date. It does give reliable figures for small geographic areas, but is subject to disclosure control which makes small numbers unreliable (and potentially also when a series of small numbers are added together for a larger area). The other essential element of the census of population is that it provides the only reliable source of small area attribute data for the population. For example small geography data on ethnicity, family structure and housing tenure are available nowhere else in a nationally comparable form.

Administrative and survey data is used for inter-censal estimates of population. The National Health Service Central Register (NHSCR) is used to record movements of residence of those registered for health care. This register in England is due to be replaced shortly(the position is uncertain in Wales), so if it is to continue to be used, cooperation between Department of Health and ONS is essential for data fit for purpose.12 The international migration element of change is extracted from the International Passenger Survey13 (IPS), asylum seeker data and Ireland data. The survey is too small by itself to provide local authority level information. The IMPS project have devised new methodology which utilises the Labour Force Survey (LFS) and 2001 census to supplement the IPS, but there is uncertainty among local authorities relating to the small sample sizes and the risk of “outliers" to the new methodology, as it has introduced considerable “churn" in some authority area population estimates. These concerns have been transmitted to the IMPS team by Welsh Assembly Government colleagues.

ONS have recently published estimates of “short term migration".14 This element of population change can have a large impact upon the demand for local government services, but has not been a part of the traditional population

9 Information on funding in Wales may be found at; http://new.wales.gov.uk/topics/localgovernment/finandfunding/?lang=en10 Information on the IMPS project is found at http://www.statistics.gov.uk/about/data/methodology/specific/population/future/imps/default.asp11 Welwyn and Hatfield typify local authorities with high proportions of students in their population and the uncertainty this produces http://www.statistics.gov.uk/about/data/methodology/specific/population/future/imps/updates/downloads/WHreportV3.0.pdf12 Administrative data sources and their potential to inform population studies can be found at http://www.statistics.gov.uk/about/data/methodology/specific/population/future/imps/updates/downloads/admin.pdf13 IPS information http://www.statistics.gov.uk/ssd/surveys/international_passenger_survey.asp

estimates, which concentrate on the long term migration intentions. However, seasonal and short visit migration can have a significant effect on the population present at a local level.

How appropriate is a definition of the population based on the usually resident population in the context of the needs of the users?

The basic definition is adequate for many applications, but does not cover needs arising from day time populations, seasonal variations and residents with more than one address (holiday homes, second homes, shared parenting arrangements). A useful article about this subject was published in Population Trends 124. 15

In Wales it is contended that certain areas are particularly affected by holiday homes (including caravans) and second homes used on an irregular basis. To provide appropriate services and ensure sustainability (as required by the Government of Wales Act) these need to be known.

Day time populations relate to movement for work, retail and leisure activities, which require specific service provision, not only by local government but also emergency services for public protection.

Seasonal variation may relate to the tourism industry, the requirement for agricultural labour or the provision of goods and services which may be seasonal.

Thus it can be seen that where possible alternative population estimates would be used to better target and provide services for the population.

How does the quality and range of population statistics in England and Wales compare with that available in Scotland, Northern Ireland and internationally?

After the last census the small area population statistics in Scotland were more usable than for England and Wales, due to disclosure controls added in England and Wales. This led to the migration and travel to work small area statistics being virtually unusable. It is noted that a protocol exists between the home countries to harmonise definitions for the census in 2011, which must help comparability. It is not proposed to comment on international comparisons.

2. THE ROLE OF THE CENSUSHow does the Census contribute to the creation of population statistics? Is the current frequency of Censuses appropriate and, if not, how frequent should Censuses be?

The census is the basis of population statistics in the absence of a universally applied population register. More frequent adjustments to the population can be made from administrative and survey sources, but for reliability and coverage the census is currently essential.

From a local government perspective, a census would be useful every five years. However, what form the census might take and what it would cost will be fundamental issues. It is also noted that many other countries have gone more to administrative data sources for updating population counts, but still find it necessary periodically to undertake a census to check the base numbers.16

What lessons have been learnt from the Test Census carried out on 13 May 2007? What methods should be employed for the distribution and completion of Census forms, including the use of the Internet?

One lesson from Wales is a proper appreciation of the need for a fully bilingual property gazetteer which is current and complete. Omissions and errors arose in Carmarthenshire due to difficulties in reconciling English and Welsh place names (and alphabets).

14 Research Report on Short-term Migration http://www.statistics.gov.uk/about/data/methodology/specific/population/future/imps/updates/downloads/STM_Research_Report.pdf15 Population Bases, Chris Smith & Julie Jeffries, Population Trends 124 pages 18-2616 Information on international approaches to population counts can be found in Census strategic development review Alternatives to a Census: Review of international approaches, ONS 2003 http://www.statistics.gov.uk/downloads/theme_population/international_approaches.pdf

The evaluation report also identified the value of a pre-survey reconnaissance of the area, especially in fast developing locations and those subject to subdivision of properties.

The test covered an investigation on the effect of hand delivery of census forms or postal delivery. The test results showed a marginal difference that could be recovered by targeted follow up by enumerators. It is understood that Scotland are more inclined to favour hand delivery.

With regard reaching the whole population there are two aspects to the problem. Firstly those who do not wish to be enumerated and secondly those that are hard to contact. The more extensive use of post out presupposes a detailed and accurate knowledge of the area to provide an address list. The opportunity to complete returns by phone or internet are welcome, but beg the issue of how to contact those persons to make the request and to provide & secure the identification code for them to use to identify themselves.

What steps should be taken to increase the rate of responses to the Census, particularly in hard-to-reach areas? How effective are plans to measure the extent of non-response to the 2011 Census?

Response rates are measured against contacts made. Coverage is whether or not the whole population is included. Coverage will require good cooperation between ONS and local government to identify all possible locations for enumeration. To persuade those contacted to reply requires them to “buy-in" to the process or outcomes. This will require careful and well targeted publicity showing the need for the information being collected and disclosure controls applied.

Engagement with local populations will help to show them it is a joint operation requiring their help and providing them with appropriate services from use of the results. There will always be those who do not wish to participate. Their attitude will often be shaped from a fear of the data being shared with other government departments or just an adverse reaction to authority. Publicity relating to the security and confidentiality of census data will help if that is the fear. If an aversion to authority is the reason it will require identification and pursuit to be employed.

The proposal for a logging system for each census form issued seems to be a sensible approach. The results of the checking for completed forms will need to be fed back efficiently to the field staff to enable them to follow up. This could be supplemented by cross checking various counts against proven local intelligence, such as numbers of registered students.

To what extent is there a trade-off between the length of the Census form and its role in providing population information? What questions should be included in the 2011 Census?

There is evidence from other countries census forms that it is not necessarily the length that matters but how well people relate to the census and what force of authority it has behind it. The value of the census information must be put against the cost of collection and response rate.

The value of a census is the provision of small area information about the population and their attributes, not the population count alone. There are a number of items not available elsewhere, especially from sample surveys in terms of geographical accuracy. If local data were not to be available the accumulated cost of each local authority doing their own surveys would far exceed a national census and risk not being able to compare the data. The census provides data required by central government from local government and not available elsewhere (sub-local authority data on ethnicity, use of Welsh language, journey to work, tenure of property, family structure, overcrowded dwellings).

Questions to be asked should cover the topics of:

@UL@— Who is present and who resident, by age, gender, ethnicity, national identity, religion, relationships@/UL@

@UL@— What is the household structure (including shared responsibilities for children)@/UL@

@UL@— The tenure of the dwelling (including usage as second home and for holidays)@/UL@

@UL@— The number of bedrooms available@/UL@

@UL@— Activity including industry / occupation information@/UL@

@UL@— Inactivity information (eg retirement, withdrawal, disability, study)@/UL@

@UL@— Movement of residence and to work (ideally also movement to school)@/UL@

@UL@— Education, skills and training (and lack of these)@/UL@

@UL@— General indication of health of the individual@/UL@

@UL@— Caring (voluntary) for others in the household or family@/UL@

@UL@— Ability to communicate through the official languages of the home nations (English and Welsh)@/UL@

To what extent should Census questions and statistical information be coordinated across the United Kingdom and how effective are plans to ensure such coordination?

The coordination is essential, but should include the ability of specific national additions, like use of the Welsh language or Gaelic. The protocol agreed by the relevant registrars and chief statisticians should provide that coordination.

What forms should the outputs of the 2011 Census take, how and when should they be made available and how should they be integrated and coordinated with other information on population?

The outputs should provide for general use, which might include general statistics on paper. Where the level of detail and complexity increases it is more appropriate to disseminate the data by electronic means. This should include mass data storage and internet as vehicles for transmission. An additional facility should empower the users to compile their own custom inquiries, subject to them not being disclosive. This could be enabled by a web based engine to combine base statistics on request. This implies the data used would be pre-prepared to be non-disclosive.

During the last census, when significant errors were identified (Westminster, Manchester, Cardiff)17, the population figures were adjusted to correct the error, but the census information was not reworked to be consistent. This can produce discontinuities in data sets.

The data should be available as soon as possible after the event without increasing the risk of the need for revisions, which cause problems and confusion. Whether simple counts could be released earlier than more complex cross tabulations would depend on the methods of processing used on the forms.

As the census is used to rebase the estimates of population produced by ONS there will necessarily be delays to the rebasing of those historic series to the census standard. This will also be true of survey data, where the census is used to produce population estimates from samples and to check the stratification of responses.

Whilst putting the information collected into the public domain is important there is an obligation not to make it possible for personal information to be disclosed. ONS have investigated a number of methods to statistically disclosure control (SDC) the data. It is earnestly hoped that whatever method of SDC is chosen, it will not produce the problems observed in 2001 in England and Wales, most notably the non-additive nature within tables and the variation of counts relating to identical data items eg the total number of households in various tables were reported with different totals within the same geographic area. It resulted in problems identifying an unique base figure and very little use being made of the origin—destination tables for migration and journey to work, which was a waste of a valuable resource. Current investigations by ONS are focusing on three techniques, users hope the one chosen will provide adequate disclosure controls whilst retaining additivity within tables and identical totals for data items between tables.

3. MID-YEAR POPULATION ESTIMATESHow accurate and useful are the mid-year population estimates for England and Wales, including information available for local authorities and Strategic Health Authorities?

17 Results of 2001 Census based Local Authority Population Studies. Reference: http://www.statistics.gov.uk/pdfdir/census0704.pdf

The accuracy of the statistics appears to vary by the local authority, specifically by the characteristics. There have been a number of characteristic clusters of local authorities identified by the IMPS project. It is recognised that areas with high proportions of students are more difficult to count. In Wales Cardiff, Ceredigion, Carmarthenshire, Swansea, Newport and Rhondda Cynon Taf have issues with the treatment of students in population estimates. This may relate to the low sample rate for the International Passenger Survey which identifies foreign nationals entering Britain. It is asserted by local authorities that a high proportion of foreign students live in non-private accommodation, so may not be picked up by the Labour Force Survey which is used to apportion the internationals to sub-national intermediate geographies. The 2001 census is then used to allocate internationals from intermediate geographies to local authorities, but there were issues, particularly in Cardiff, with student and international areas being undercounted (Gabalfa, Adamsdown, Butetown) with that census.18

How appropriate is the methodology by which mid-year population estimates are reached?

The revised methodology adopted this year (August 2007) has caused some churn in the series of population estimates. It is believed that the revised method is generally an improvement compared to the previous approach, but there are exceptions which need to be investigated.

What are the most appropriate ways to ensure a common level of accuracy in mid-year population estimates for different areas and what steps should be taken to improve the accuracy in areas where population is hard to measure?

The different characteristics of areas lead to different problems when estimating populations. The cluster case studies being undertaken by the IMPS team should be progressed and consultation with the local authorities in hard to count areas should be used to exchange intelligence and quality assure the outcomes. For independence and consistency, ONS must always make any decisions about controversial issues. There are a number of administrative sources available locally with which to test estimates when they are produced. It is recognised that there is a balance between more extensive quality checks, consultation and timeliness of the estimates for the financial rounds.

What is the right process for making revisions to mid-year population estimates so as to meet the needs of users?

The main impact of revisions is upon the allocation of finances. This affects the Barnett Formula and WAG Local Government Finance allocation of resources to local authorities (as with DCLG in England). Changes may only affect the amount of change between the pre and post revision years, but they impact on revenue budgets and hence jobs and service provision. Any revisions to methodology need to come with an assessment of the likely reasons for the churn induced in the population estimates. This assessment would enable finance distributors to assess whether transition measures need to be put in place to bridge changes in methodology (hence revisions to the population estimate series).

A recently published Population Revisions Policy has undertaken not to revise the estimates for a further two years. Stability is welcome, but when future changes could potentially reverse past changes which affect funding of local government, the “churn" effect can impact upon jobs and services that can be delivered. It is necessary for some form of agreement with the funding bodies to smooth the changes generated by revised population estimates from improved methodology.

4. THE ROLE OF SURVEY AND ADMINISTRATIVE DATAWhat role can and should be played by survey data in the compilation of population statistics?

Without a mandatory population register, it is inevitable that survey data will play some part in estimating the population. A more reliable International Passenger Survey for immigrants and emigrants would improve reliability and geographical definition possible. Interaction between surveys (like the Integrated Population Survey) can provide more reliable data sets and more extensive topic information that could be used to support population estimates.

What role can and should be played by data from General Practitioners and other health service data in the compilation of population statistics?

18 The IMPS report on the case study into Welwyn & Hatfield noted Cardiff had a correction of +3,050 population to 3 electoral divisions (see page 64 of that report).

The National Health Service Central Register (NHSCR) provides internal migration information, but looses the (flag 4) indicator of a recent entrant to the UK should they move within. As the NHSCR in England is to be replaced with a new system in the near future, it is imperative that the Department of Health and ONS work together to ensure the opportunities offered by a new system are not lost. The retention of an international migrant registration flag in the register for a fixed period of time would prove very useful. There is still uncertainty as to whether this new system will be taken up in Wales. There will always be the issue of those who do not register or de-register, however providing this becomes a known error it can be taken into account.

What other data sources can and should be used in the compilation of population statistics?

Other material can be gleaned from the Pupil Level Annual School Census (PLASC, pupils registered in schools), Higher Education Statistics Agency (HESA, students in education or training institutions), Department of Work and Pensions (registration for employment, receipt of benefits or allowances), Her Majesty's revenue and Customs (people paying taxes and duties) and the Workers Registration Scheme in the Home Office.

The PLASC and HESA data is likely to be most reliable in locating the residence of elements of the population. As pensions and benefits are not universal, they would only serve to check the order of magnitude. HMRC data is not currently considered reliable, especially for residential addresses or to construct family groups.

5. COOPERATION WITH STAKEHOLDERSHow effectively has the Office for National Statistics cooperated with stakeholders with an interest in and information relating to population statistics and how can cooperation be improved by the Statistics Board and by its Executive Office?

The ONS has been much more open in recent years and has consulted and listened to interested parties. This is an improvement, but in some cases does not reach those who need to be involved.

The opportunity exists through the scrutiny function of the Board to oversee the approaches taken. In addition it is recommended that more “interest groups" be formed to consider the actions and implications of ONS work. The Central and Local Information Partnership (CLIP) groups, particularly on the census and population are very useful. There needs to be more thought given to the most effective way to provide information requested by central government from local service providers. This could reduce duplicate or piecemeal surveys and ensure the answer was to the underlying problem, not superficial question.

In addition to the above the IMPS team have engaged with users to explain the methodological changes applied to population estimation. In Wales the Welsh Statistical Liaison Committee meets at least four times each year to discuss matters of interest to both Welsh Assembly Government and other public sector organisations. This is valued by the participants.

The Statistics and Registration Services Act has provision for ONS to collect data sets and make statistics available. This in itself has the potential to unlock some useful data sets that are encased in Acts where the provisions for using the data have become out of date. An example of this would be housing data and employment data (for daytime populations). The controlled access to data by registered researchers under section 39 of the Act is welcomed and will hopefully be implemented to suit researchers, ONS and data security considerations.

November 2007

@HR25@

Memorandum from the South West of England Regional Development Agency

The South West of England Regional Development Agency (South West RDA) welcomes this opportunity to contribute to the Treasury Sub-Committee's inquiry into Counting the Population.

The South West RDA uses Office for National Statistics (ONS) population statistics—Census counts and mid-year population estimates—and sub-national population projections. We are not sufficiently familiar with the methodology to comment upon methods and accuracy. As such, our response is largely limited to the importance of population estimates to our remit of regional economic development.

USES OF POPULATION ESTIMATESThe South West RDA leads on the development and implementation of the Regional Economic Strategy

(RES) for the South West of England. The RES needs to be built upon a robust evidence base, comprising statistical analysis and research across the wide range of areas that directly and indirectly influence economic growth across the SW region.

An understanding of the size and nature of the population is fundamental to understanding the economy of the region, how it is performing, relative to the rest of the United Kingdom, and what the prospects are for the future.

Our specific uses of population estimates can be considered under a number of headings:

 (a) Understanding the nature of the SW population

Population estimates are used to understand and monitor the demographics of the SW population. Of particular interest in recent years, is the nature and speed of change in the SW population. The South West has one of the fastest growing populations in the United Kingdom. Importantly, net growth (of around 33,000 people per year over the last decade) is entirely due to inward migration. The region also has the oldest population in the United Kingdom—in 2006, around a quarter of the SW population was over 60 and this is projected to increase to a third by 2029. These issues have clear economic implications, both now and in the future.

As with the rest of the United Kingdom, the SW region has acquired a relatively high number of European migrant workers in recent years—42,000 from 2004 to 2006. International migrants now account for more than one third of the net growth in the SW population, substantially higher numbers than in the years leading up to the 2001 Census. The size and nature of this growing international community, including basic demographics such as age, gender, ethnicity and disability, as well as skills (including but not exclusively language) is increasingly important in understanding the region's population and its role in the regional economy. Currently, this information is largely unavailable.

 (b) Regional disparities

The South West is a very diverse region which varies in demographic make up and economic characteristics. At least seven functional economic zones have been identified. It also has the largest rural population of the English Regions—more than a third of the population at the 2001 Census. This diversity presents a challenge for delivery of economic development and regeneration as different parts of the region require different interventions. Understanding the various local issues and needs requires robust information on population sizes and composition at low geographic levels.

 (c) Planning for growth

The SW population is one of the fastest growing in the United Kingdom—with a 7% increase in population between 1996 and 2006. Based on the recent new UK population projections, we estimate that the SW population, about 5.1 million in 2006, could reach six million by around 2021, ten years earlier than the current sub-national projections suggest.

Population projections are of necessity based on historical population statistics. In order to make reasonable assumptions about growth, robust and accurate historical population figures are required. In order to plan for population growth, these figures need to be as robust, at least for the near future, as sensibly possible.

The Regional Spatial Strategy (RSS) for the South West, produced by the SW Assembly, sets out the planning strategy for managing population growth, in terms of housing and infrastructure, at regional and local levels.

National population projections feed into the regional projections and consequently sub-regional projections—these then feed directly into projections of household numbers, which are then used to determine household allocations at the local authority district level. As housing allocations have the potential to be controversial and politically sensitive in some areas, it is essential that the household projections are based on accurate district level population data.

The South West RDA works closely with the Regional Assembly to ensure alignment of the Regional Economic Strategy and the Regional Spatial Strategy. As a result of the Treasury's recent Sub-National Review of Economic Development and Regeneration, the RES and RSS will be combined in the near future to form a Single Regional Strategy. In addition, there will be a move to increasing the amount of economic development delivered at a local level, through local authorities and led by the RDA. This move to increased local delivery provides additional pressure for provision of statistics that give a better understanding of population and population growth at the local level.

 (d) Economic development

The mid-year population estimates are routinely used as denominators in calculation of a wide range of (economic) indicators and for relative comparisons across demographic groups, different parts of the region, and between urban and rural. For example, economic productivity, average household income, labour market performance (economic activity, employment and unemployment rates) and deprivation (the Indices of Deprivation, etc.).

As such, population figures underpin analysis used throughout the RES evidence base that is used to drive economic policy and deliver economic development and regeneration at local and regional levels.

 (e) Inclusion for all

As a public sector organisation, the RDA has a duty to ensure that equality and diversity are at the forefront of its economic policy development and delivery. This requires robust and consistent statistics on the different “strands" of diversity at a sub-regional level. In particular, there are a lack of robust population data on the ethnic minority and disabled population in the South West.

 (f) Monitoring RDA performance

The nine English RDAs report to the Department for Business Enterprise and Regulatory Reform. All RDAs have been tasked to raise regional economic performance by working to promote regional regeneration, investment, skills, training, employment, efficiency and competitiveness. In particular, the RDAs remit is to reduce the gap in economic growth rates between the nine English Regions.

As part of this, RDAs have been set regional economic performance PSA targets specifically on growth in GVA per head in each region. Robust population measures are required both to set realistic targets and to monitor progress towards those targets over time.

Do current estimates meet these needs?

Broadly, the current ONS mid-year, regional population estimates, at least in terms of total population counts, appear to meet our needs. However, at a regional level it is difficult to assess the accuracy of the SW population. Colleagues across the SW local authority areas are in a better position to comment upon accuracy at the local level.

The SW population has been growing and changing relatively quickly in recent years. The mid-year estimates provide little understanding of the nature of this change. In particular, statistics on the levels of international migration into the region are poor and are over dependent on the International Passenger Survey. Better information is needed on where migrant workers (and their families) are in the region, from which countries they are from, age, gender, disabilities, skills and needs.

Clearly, this is particularly important for determining realistic local authority budgets and for appropriate local service provision. For the RDAs, it is important to understand the nature of change and how it affects regional and sub-regional labour market dynamics, and of course underpins RDA delivery around the growing sustainable communities and economic inclusion agendas.

THE ROLE OF THE CENSUS (a) Importance and frequency of Census

The 10-year population Census is fundamental to accurately assessing not just the size but the nature of the population, and for provision of detailed data on the population—both population attributes and distribution at a low level geography.

Since 2004 and expansion of the European Union, the United Kingdom, including the South West, has seen a relatively large influx of workers from the EU accession countries. In addition to the impact that this has on local population sizes, migrants are likely to have slightly different issues and needs.

As a result of rapid change in the SW population, the 2001 Census figures are now largely redundant. We estimate that the South West population may have grown by as much as 230,000 people since the 2001 Census. ONS modelled estimates of the region's ethnic minority population suggest that, although a relatively small component of the SW population in 2001, this is a fast growing sector of the population. Again, the ethnic minority population in different areas may have different issues and needs, and it is likely that service provision must be adapted to account for this and other changes in the population.

In light of the large changes in the SW population in recent years, an earlier Census than 2011 is clearly desirable if not practicable.

 (b) Length and content of 2011 Census

Naturally there is a trade off between Census form length and its role in providing accurate population information—the longer the form, the poorer the response rate and quality of response. Of course, there are also cost considerations. However, given the fundamental importance of the Census to a wide range of stakeholders, including appropriate service level provision in the right places, it is concerning to stakeholders that it may indeed be the cost of producing and analysing the 2011 Census that is driving what will and will not be included in the 2011 Census questionnaire.

The South West RDA has contributed to a number of consultations around the 2011 Census. The primary purpose of the Census is to provide an accurate count of the UK population, including key demographic information on age, gender, disability and ethnicity. While for diversity monitoring, it is important to monitor religion and sexual orientation, it is unclear whether the Census is the correct vehicle for assessing these issues.

A key concern is that where possible the 2011 Census should allow easy and consistent comparison with previous Censuses or at least the 2001 Census. As an economic development agency, we would also argue for collection of information on contribution to the work force, qualifications/skills, travel to work patterns and migration, and importantly on languages.

COOPERATION WITH STAKEHOLDERSIn terms of the 2011 Census, ONS has carried out a number of consultations with stakeholders on the content

of the Census, on the geography, and on various question topics. We recognise that it is very difficult to balance the needs of the diverse needs of the various stakeholders for the Census and believe that ONS have indeed succeeded (to date) in managing this task sensitively.

November 2007

@HR25@

Memorandum from the Demographics User Group

EXECUTIVE SUMMARYThe Demographics User Group (DUG) represents the views of 15 commercial companies—Abbey, Argos,

Boots, Co-operative Group, Marks & Spencer, M&S Money, Nationwide, O2, Powergen, Sainsbury's, Tesco, The Children's Mutual, Whitbread, Woolworths, and Yell—which make extensive use of population statistics produced by government.

Commerce, like government, needs reliable population statistics when making investment decisions which total many billions of pounds each year. Members of the Demographics User Group wish to see:

@UL@— More use made now of government administrative files for statistical purposes.@/UL@

@UL@— Government in England & Wales following Scotland's lead in creating a definitive address file for the 2011 Census.@/UL@

@UL@— ONS targeting difficult areas in England and Wales for hand delivery of 2011 Census forms, as will be done in Scotland.@/UL@

@UL@— Progress towards the creation of a population register for statistical purposes, validated by a final traditional Census in 2016 (rather than waiting until 2021).@/UL@1. COMMERCIAL COMPANIES' NEED FOR POPULATION STATISTICS

1.1 When deciding how best to invest many £ billions to meet the needs of local consumers, companies seek current and accurate statistics for small areas for both resident populations (including short-term migrants), and workplace populations. Government is in a uniquely strong position—when compared with even the biggest retailer—to assemble comprehensive information, due to its complete national coverage, and powers of compulsion.

1.2 DUG members fully appreciate that many changes in society—more migration, daily commuting, weekly commuting, and more entry phones—make it increasingly difficult to assemble reliable and current statistics about the population using traditional methods such as voluntary sample surveys and the decennial Census, and believe that government should also pursue potential new sources.

2. TO WHAT EXTENT ARE USERS' NEEDS CURRENTLY BEING MET?2.1 Resident population

@UL@— 2001 Census. This still forms the bedrock for decision-making, providing detailed statistics for very small areas. It is recognised that the Census coverage was inadequate in a small number of local authorities, but this has not prevented heavy use of the data as the best available source. However, the results are now more than 6 years old, and some parts of the country have experienced rapid change. Inaccurate statistics inevitably result in bad investment decisions.@/UL@

@UL@— Mid Year Estimates. Whilst these have the advantage that they are more recent, those at Local Authority level are of limited use to commercial companies. The ONS's recent development of updated estimates for smaller areas is welcome, but they do not provide sufficient geographical or subject detail.@/UL@

@UL@— Short-Term Migrants. DUG members appreciate that the Mid Year Estimate population is of usual residents, and excludes people who fall below the radar by being here for less than 12 months. The ONS's recent publication of its first estimate of STMs is welcome, but the estimate itself seems implausibly low (see below). The absence of reliable statistics results in the sub-optimal targeting of investment.@/UL@

2.2 Workplace population

@UL@— Workplace populations—such as the number of people working within a kilometre of Boots on Victoria Street—are also of great interest to retailers. However, 2001 Census statistics still have to be used, as no new information has been produced since.@/UL@3. WHAT IMPROVEMENTS SHOULD BE MADE NOW, BEFORE THE 2011 CENSUS?Make more extensive use of government administrative files

3.1 We believe that the statistical landscape is changing, and that ONS should react accordingly. Response rates to voluntary sample surveys have been declining, and for some topics (such as migration) there are real difficulties in getting truly representative samples, and accurate answers. On the other hand, various arms of government are creating huge files of administrative records, which are part of the functioning of modern societies. These will not provide a complete alternative, but they are current, and have postcoded addresses, enabling them to be used to create aggregate statistics for small areas.

3.2 Administrative files are already used to produce some of ONS's Neighbourhood Statistics, and the new Statistics Act increases the scope for data sharing: its potential benefits should be explored as a top priority. Valuable lessons can be learned from countries such as Finland. It is good to see that ONS's Draft Work Programme for 2008-2012 highlights this as a new work package.

3.3 Examples of potential sources include:

@UL@— Pupil-Level Records (which include pupils' home addresses).@/UL@

@UL@— National Insurance records. The NINo Registrations of people arriving to work from overseas are of particular interest, and could be analysed longitudinally.@/UL@

@UL@— Department of Work and Pensions benefits records, such as pensioners.@/UL@

@UL@— National Health Service Central Register (which is already used for migration estimates).@/UL@

@UL@— Inter Departmental Business Register, using HMRC and NI records, which might be developed to provide statistics about workplace populations as well as businesses.@/UL@

Publish more statistics as “experimental"

3.4 Whilst none of these sources will immediately provide perfect sources of new aggregate statistics, they can provide valuable pointers to recent trends, and a yardstick to assess the plausibility of existing sources. ONS should publish more such statistics, together with explanations and caveats, to help users build a broad picture.

3.5 Taking the ONS's recent work on Short Term Migrants as an example, there are estimated to have been only 43,000 Short Term Migrants in England and Wales in June 2005, with just 16,000 of these in Greater London. These estimates are based on the International Passenger Survey, and appear to many users to be implausibly low when contrasted with administrative data such as National Insurance Numbers, which in 2005-6 recorded 235,640 new NINo registrations in Greater London alone. Many of the latter no doubt returned to their countries of origin within 12 months, but having the NINo counts available provides a valuable additional insight.

4. 2011 CENSUS: WHAT MUST BE DONE TO MAXIMISE COVERAGE/AVOID FAILUREGovernment must create a definitive address file now for England & Wales

4.1 The ONS needs a reliable address register as the basis for 100% Census coverage. Past efforts to produce a definitive national file have foundered on issues of the ownership of intellectual property by different parts of government (Ordnance Survey, Local Government, and the Post Office). To the outside world this looks like a scandalous waste of public assets. ONS is now planning to evaluate the rival sources, and potentially stitch them together to produce an ad hoc file for the 2011 Census alone. This contrasts with Scotland, which already has a definitive national file. Government should bang heads together now to produce a definitive national address file.

Difficult areas should be targeted for hand delivery of Census forms

4.2 The ONS is planning to post out Census forms, rather than deliver them by hand, as has always been done in the past. Even if a definitive address file existed, in some areas (such as Camden, which was selected for the Test Census), many addresses have several separate households within them. Information about individual residents might be obtained from government administrative files—ONS and DUG are also exploring the scope for using commercial companies' customer files to indicate multiple households—but can only be confirmed, and Census forms handed over, on a personal visit. It is significant that in Scotland it is planned to continue with traditional hand delivery of Census forms.

5. A LONG-TERM VISION

5.1 This note has made a case for greater tactical use of administrative files to increase our understanding of recent population changes. It is encouraging to note that a feasibility study is currently underway which is looking at the better use of administrative data as an analytical base to drive improved policy research, analysis, evaluation and decision making across HM Government.

5.2 Looking further ahead, the ONS's earlier plans for an Integrated Population Statistics System should be refreshed in the light of experience gained by those countries which have already built population registers, notably Finland.

5.3 A move towards population registers will require validation by a final traditional Census, which should be held in 2016.

POSTSCRIPT: ONS'S CO-OPERATION WITH STAKEHOLDERSP1. The ONS has been making significant efforts to understand users' needs for better population

information through consultations, its specialist Census Advisory Groups, and meetings with users including the Statistics User Forum.

P2. In responding to users' needs, it is essential that the ONS is seen to be independent. DUG members do not feel that it has succumbed to pressure from particular Local Authorities to change the population estimates for their areas. We do, however, have concerns that the plans for the 2011 Census have recently given higher priority to the needs of Government Departments, to the detriment of users more generally—”the public good" as enshrined in the new Act.

P3. When compared with commercial companies, the ONS is often cautious, being slow to take on new priorities and practices. We think that this is partly cultural—to staff, the costs of failure are greater than the benefits of success; and partly financial—the ONS's job is becoming more difficult due to social changes, and it is difficult to fund new initiatives at the expense of established datasets.

We are looking forward to dialogue with the new Statistics Board to address these issues.

November 2007

@HR25@

Memorandum from the Statistics User Forum

The Statistics User Forum (SUF) provides an umbrella organisation for 21 groups of statistics users. Detailed information about SUF is available at: http://www.rss.org.uk/main.asp?page=1607

SUF very much welcomes the Treasury Sub Committee's enquiry, which gives users of population statistics a special opportunity to express their views to policymakers.

Three of SUF's User Groups with particular interests in population statistics—the Association of Census Distributors, the Demographics User Group, and the Market Research Society's Census and Geodemographics Group—are making their own submissions to the committee. They also wish to highlight some priorities which are held in common, and which are shared by many other members of the Forum.

@UL@— Make more use of existing administrative files to create new population and migration statistics cheaply. For example, National Insurance Number records, which have recently been used to create local statistics of new registrations of overseas workers, could be analysed longitudinally to see how long such workers stay.@/UL@

@UL@— Create a national address register for England and Wales. This is a matter of great urgency as the basis of good coverage for the 2011 Census. Such a register is already being created in Scotland@/UL@

@UL@— Deliver 2011 Census forms by hand in difficult areas in England and Wales, as will be done in Scotland.@/UL@

@UL@— Maximise the value of the 2011 Census by including a fourth page which includes all the questions on deprivation/affluence that ONS are currently planning to exclude. It seems that central government departments are currently exerting undue influence in promoting questions on migration which are unproven, and also of doubtful value.@/UL@

@UL@— Continue to develop population estimates for areas below the local authority level to help understanding of very varied local conditions and change.@/UL@

@UL@— Develop ONS's vision for an Integrated Population Statistics System which would link results from the 2011 Census to administrative records and to its Continuous Population Survey. Such a system would solve two drawbacks of the “traditional" Census, viz. the production of data only every 10 years, with a two-year lag in obtaining results.@/UL@

November 2007

@HR25@

Memorandum from Professor David Martin, Royal Statistical Society Study GroupI am writing as Chair of the Royal Statistical Society's Census Study Group in response to the call for written evidence for the new inquiry into Counting the Population. The Census Study Group engages in a range of cross-sectoral activities including the promotion of events specifically to inform users and provoke debate concerning census development in the UK. The views expressed here have been collated from contributions by members of the study group committee which represents expert census users in central and local government, academic and commercial settings with a common interest in maximising the value and utility of UK census data. The structure of this response follows the numbered topics and a selection of the more detailed questions identified in the call for evidence. Our response is necessarily weighted in favour of the census-specific aspects of the inquiry.

1. USES AND DEFINITIONSWhat are the uses of population estimates, and how far do current population estimates meet the needs associated with those uses? What are the effects of inaccuracies or inadequacies in such estimates?

Population estimates have at least three critical uses which are applicable across sectors. The first concerns resource allocation and the determination of financial settlements, of particular concern to central and local government. A significant strand of academic investigation has centred on the sensitivity of these figures but they are of greatest impact for local government, where errors in baseline population figures can have significant financial consequences, particularly where funding from central government lags behind growth or important changes in population composition.

A second area of use is as the basis for denominator rates for numerous applications which include deaths, accidents, health, crime and conceptions. More broadly, central and local government, business and academia are all concerned to understand changing population characteristics. Specific studies or administrative sources may often provide the counts of interest but there is a general reliance on official population estimates, broken down into appropriate age groups, as the basis for determining rates. Uncertainty in the base population figures can often obscure or may at times even reverse apparent changes in the phenomena of interest, which will often have local or national policy consequences.

This indicates a third area of use which is the planning of local services—or in the commercial context, business decisions—on the basis of estimated population need and demand. The census provides key elements of client and customer need, while the various sources of population counts help determine its size and trajectory. The modelling of these factors is fundamentally influenced by the quality of the underlying demographic count data.

In growth areas, current annual population estimates fail to anticipate and keep pace with the additional demands on resources resulting from a growing population. There are particular weaknesses associated with all types of migrant populations. Inaccuracies in base population figures are especially sensitive when interest relates to small geographical divisions and to population subgroups, due to the uneven spatial and statistical distribution of most sources of error. Specific local authorities, most notably Manchester and Westminster were directly and adversely affected by enumeration problems in the 2001 Census. Although total population figures were revised, the detailed census counts remain unchanged with the result that analysis of subpopulations is known to be unreliable but no alternative sources are available. Unreliable or demonstrably inaccurate basic population counts will call into question the entire edifice of resource allocation decisions, target-setting, prevalence rates and area profiles which are essentially reliant on this core information.

How appropriate is a definition of the population based on the usually resident population in the context of the needs of the users?

The usually resident base provides the best available information but the increasing complexity of household composition and multiple residences makes it necessary to capture some additional information on the numbers of visitors and there is a strong need for more information to be available on a workplace basis for the many departments and organizations which are concerned with servicing daytime populations rather than populations at residential addresses. This information can only grow in importance as more information is required to support the generation of sustainable development and transportation systems.

How does the quality and range of population statistics in England and Wales compare with that available in Scotland, Northern Ireland and internationally?

One of the most obvious differences between Scotland and the rest of the UK in the 2001 census was the less restrictive approach to statistical disclosure control adopted in Scotland. This is demonstrated both in the smaller size

of output areas and the somewhat less damaging statistical disclosure control methods applied to the published datasets. There is strong user demand both for consistency and for moves towards the Scottish rather than the ONS 2001 positions in this respect. We note that there are other differences in practice which cause challenges to those undertaking whole-UK analysis such as the different timing of publication of mid year population estimates in Scotland compared to England and Wales. Although there are headline agreements between the Registrars General to work together on the principles of statistical disclosure control, we are concerned at other emerging differences between GROS and ONS such as strategies for address checking and the delivery of census forms in 2011: these types of divergence will have differential and unpredictable impacts on data quality for users.

2. THE ROLE OF THE CENSUSHow does the Census contribute to the creation of population statistics? Is the current frequency of Censuses appropriate and, if not, how frequent should Censuses be?

The census continues to be essential to the creation of baseline population statistics and forms a foundation and reference point for alternative sources of population estimates. Due to its high population coverage, it is however unrivalled as a source of socio-economic population detail at small area level. While some of these topics are covered in greater detail by survey datasets, none offers any small area geography. An increased frequency of census would be useful to researchers and planners, especially with regard to population characteristics which are known to change quite rapidly, such as the ethnic composition of the population. Researchers would generally like more frequent information but are prepared to accept a decennial census providing it is sufficiently resourced to produce outputs of the absolutely highest quality, allowing it to continue to be used as the key reference dataset. Group members acknowledge the potential benefits of an “integrated population statistics system" (IPSS) such as that proposed by ONS in 2003, or partial censuses on a five-yearly basis, but emphasize the importance of demonstrating the utility of any such approach before retreating from the proven model of the current decennial census.

What lessons have been learnt from the Test Census carried out on 13 May 2007? What methods should be employed for the distribution and completion of Census forms, including the use of the Internet?

What steps should be taken to increase the rate of responses to the Census, particularly in hard-to-reach areas? How effective are plans to measure the extent of non-response to the 2011 Census?

One of the single most important underpinning strategies for increasing response rates is to have an address list of the highest possible quality. The 2007 test showed that the best results were achieved by hand delivery of census forms and this approach must be retained for the hardest to count areas. Strong liaison with local government will be necessary to assist ONS in the creation and checking of local address lists for mail out. In particular, it will be important to identify and count the number of dwelling spaces in each area and to take full advantage of the address referencing system to track census forms. We observe that the current situation in which the National Land and Property Gazetteer and Ordnance Survey address products effectively compete, each having different strengths and weaknesses and no strategy for integration, is a major threat to the census operation. ONS currently propose a comprehensive national address check in the run-up to the census, yet this entire expensive process would be unnecessary if a single definitive national address list were maintained, in which case much of the address-checking resource could be devoted to other aspects of census data collection and production.

Much better census publicity will be required in 2011, both locally directed, pointing out the importance to local communities of providing essential information for the provision of local services, and nationally, to create a supportive culture for this major data collection exercise by central government.

To what extent is there a trade-off between the length of the Census form and its role in providing population information? What questions should be included in the 2011 Census?

Most census study group members suggest that a four-page census form would not be too long—providing it is well-designed, although the well-recognised trade-off between form length and completion rates is acknowledged. Many groups express a strong desire for a question on income to be included in the 2011 form although this is less pressing among local government users. It is felt that too many new questions on national identity and aspects of migration have been added whose value is unproven. There is a widespread sense that the 2001 topic set should largely be retained.

To what extent should Census questions and statistical information be coordinated across the United Kingdom and how effective are plans to ensure such coordination?

There are clear differences in the positions articulated by different sectors with regard to the demand for UK-wide coordination, with strong requirements for consistency from the business and academic communities who are required to operate on a UK-wide basis and little demand from health, local government and central government departments who operate on a national or local basis, for which UK-wide statistics are of minor importance.

What forms should the outputs of the 2011 Census take, how and when should they be made available and how should they be integrated and coordinated with other information on population?

All group members articulate a strong desire for outputs to be delivered as fast as possible and in the most flexible possible formats. The output production process of the 2001 census is generally considered to have been too slow and the derailing effect of late changes in statistical disclosure control methodology are particularly identified. The lesson from these experiences is to consult and agree on methodological issues well in advance of data production and then to focus on the most efficient generation of outputs using contemporary data and metadata standards.

There continues to be a broad spread of incompatible requirements with regard to census output geographies, with local government articulating most strongly the need for statistics for wards, which are at the same time some of the least important output units for business and some academic users. It does not seem possible to identify any single set of small geographical units which meets all needs, perhaps due to the fact that central government is itself inconsistent in the geographical bases on which it operates, thus requiring different sectors and organizations to operate within inconsistent geographical frameworks. It is only possible to draw the conclusion that output aggregation systems must incorporate the greatest possible levels of flexibility with regard to user specification of geographies and counts.

3. MID-YEAR POPULATION ESTIMATESHow accurate and useful are the mid-year population estimates for England and Wales, including information available for local authorities and Strategic Health Authorities?

How appropriate is the methodology by which mid-year population estimates are reached?

What are the most appropriate ways to ensure a common level of accuracy in mid-year population estimates for different areas and what steps should be taken to improve the accuracy in areas where population is hard to measure?

What is the right process for making revisions to mid-year population estimates so as to meet the needs of users?

Mid-year population estimates are important as they determine the level of funding received by local government. Errors in the estimates can thus lead directly to problems in funding local services. They also form valuable constraints to local population forecasts and the detailed and supporting information can be valuable to feed into these. The methodology is generally appropriate but failures occur in areas with a more transient population, a large armed forces population, a large student population, or some other “hard to count" group, or one which is not included in the population estimates. There is broad confidence in the mid-year estimates for larger geographical units, but concern is expressed in various ways with regard to the methodology applied to smaller areas. In particular migration, which plays a very significant element in population change, is currently poorly reflected in the estimates and this will have major impacts in specific areas.

4. THE ROLE OF SURVEY AND ADMINISTRATIVE DATAWhat role can and should be played by survey data in the compilation of population statistics?

What role can and should be played by data from General Practitioners and other health service data in the compilation of population statistics?

What other data sources can and should be used in the compilation of population statistics?

Survey data can provide extremely valuable insights into detailed topics but cannot provide an alternative source of baseline demographic data such as that currently provided by the census. They may be most useful when seeking to infer other aspects of the population, such as service access and acceptability. Survey data can also play a valuable role in updating age profiles of, for example, migration propensity. Health service data derived from GP registrations are the nearest available to whole population administrative data, and an important source of internal migration data. Other data sources, such as national insurance number allocations and worker registration scheme

data can be indicative in ascertaining some data about migrants, but need augmentation to be able to provide any information about the stock of migrants, rather than the in-flows. The electoral roll and schools census data can further extend population coverage. Council tax records can also be another valuable additional tool, which can provide additional information, for example about numbers of second homes and also play a role in validating census dwelling counts. Essentially, the power of these sources can only be realised by integration as each offers only a partial view. The principal difficulties identified in relation to all administrative databases are the conflict between operational and statistical requirements and the fact that the demographic information obtainable is usually very basic. This means that sources such as the GP registration lists do not currently hold any of the more useful applied (and sensitive) data which might help to address the deficiencies of existing methods. There is a clear window of opportunity surrounding the 2011 Census for cleaning and testing administrative sources and investigating their exact relationship with the census methodology for obtaining small area demographic information.

5. COOPERATION WITH STAKEHOLDERSHow effectively has the Office for National Statistics cooperated with stakeholders with an interest in and information relating to population statistics and how can cooperation be improved by the Statistics Board and by its Executive Office?

The organizations represented by the RSS Census Study Group generally report positively on the relationship between ONS and its stakeholders, and particularly that there have been improvements since 2001. Most sectors report patchy involvement in ONS consultations due to the difficulty of understanding the full consultation timetable and the absence of any central resource to identify relevant meetings and advisory groups. The structure of sectoral census advisory committees and the CLIP liaison group on population estimates currently plays a valuable role. While the academic sector is represented by a single identifiable ESRC-funded census programme, other sectors are variably represented by self-organized special interest groups or isolated expert users, making effective consultation challenging from both ONS and user perspectives. It is important that the new Statistics Board should seek to clarify and improve communications between ONS and users rather than adding additional layers of consultation whose relationship to operational decisions is unclear to users.

November 2007

@HR25@

Memorandum from the Joint Industry Committee for Population Standards (JICPOPS)

JICPOPS is a non-profit making organisation providing population and household data for advertising, marketing and market research needs and particularly for the major advertising research currencies (BARB, the National Readership Survey and so on). So we are interested in your endeavour and would like to submit the following for consideration:

JICPOPS (with the help of CACI and Experian) provides industry-standard population statistics down to postcode sector level for the whole of the marketing, advertising and market research industries. Without reliable population data, advertisement exposure data would become impossible because it is necessary to know how many people there are before one can decide the best way to reach them. JICPOPS is owned by all the media bodies: TV, Radio National Press, Regional Press, Outdoor, etc and the advertising industry bodies (IPA & ISBA).

JICPOPS needs (a) households and (b) population estimates by postcode sector (9,000+), by social grade within age group, within gender.

For households, JICPOPS believes that there should be an accurate and definitive (and single) national address list, with the postcode and grid reference being provided for each property.

In addition:

@UL@— Gender: Male / Female.@/UL@

@UL@— Age: 7 groups: 15-24, 25-34, 35-44, 45-54, 55-64, 65-74, 75+.@/UL@

@UL@— Social grade is established from other data, ie qualifications, occupation, job title, industry, size of industrial or commercial establishment. All data to be provided for previous occupation or employment, to be able to establish the social grade of retired people.@/UL@

@UL@— JICPOPS also needs to know whether people can understand, speak, read or write English.@/UL@

November 2007

@HR25@

Memorandum from the Census and Geodemographics Group of The Market Research Society

EXECUTIVE SUMMARYS1. This evidence comes from market researchers and geodemographers, professionals who collect and

analyse statistical information.

S2. Good quality population figures are a key part of running many organisations and businesses in the UK, both in the private and public sectors, showing the number and nature of customers, and helping to serve them efficiently from national down to local level.

S3. The Census provides the bedrock which underpins much of the analysis, thus a successful 2011 Census is paramount. We have been generally well served by official statistics, but, with changes in society and users' rising expectations, traditional sources and approaches are now strained, and the inquiry is apt and welcome.

S4. We have addressed each of the Sub-Committee's areas of concern, and our top priorities are—the creation of a national address register as a matter of great urgency as the basis for the 2011 Census and population estimates, open and accessible for consistent application in all data collection, analysis and output

@UL@— provision of a modest addition to the 2011 Census budget to provide space on the form for essential but well tried questions, to give the best return on the overall investment@/UL@

@UL@— the continued production of official population estimates for areas below the local authority level to help understanding of very varied local conditions and change, although clearly getting the figures right for local authorities is the highest priority@/UL@

@UL@— a greater emphasis on plans for output throughout statistical operations, with the simple aim of maximizing value through accessibility in the public domain—taking opportunities to push forward use of new sources such as administrative data@/UL@

@UL@— action by the new ONS Board to ensure that the wider user community plays a full part in shaping the statistical programme.@/UL@

S5. We have been considering our evidence in conjunction with that from other organisations and user groups through the Royal Statistical Society's Statistics Users Forum, and we share and support many of the views of the Association of Census Distributors and the Demographics User Group, as well as views expressed in the recently published Statistics Commission report on managing risks in the 2011 Census.

DETAILED SUBMISSIONThe Market Research Society and its Census and Geodemographics Group

1. The Market Research Society (MRS) is the world's largest association representing providers and users of market, social, and opinion research. The MRS 2007 annual survey of the UK market research industry showed that total industry revenue increased by 2.4 per cent during 2006 reaching a value of £1.353 billion. The UK is the second largest market and social research market in the world.

Full details on MRS and its activities are available via: http://www.mrs.org.uk

2. Geodemographics can be defined broadly as “the analysis of people by where they live", and involves analysing demographic data within small geographical areas to serve a wide range of clients in the public and private sectors. The annual turnover of geodemographic activities in the UK has been estimated at £200 millions alone. The Census and Geodemographics Group (CGG) is an MRS committee and was founded in 1989 to represent the interests of this important activity, and to advise the MRS. The CGG has specialists in market research, retail site location, market and database analysis, as well as census distributors and academic researchers.

3. The CGG is involved with Census developments through representation on the ONS Business and Professional Interests Advisory Group, and with wider matters through membership of the Statistics User Forum as well as through an extensive network of contact in the market research industry.

Official statistics and non Government use

4. Good quality population figures are vital to geodemographic analysis now and in future. It is founded on the availability of robust small area statistics—especially from censuses—and, whilst much “in house" information now complements the official statistics, there is no foreseeable alternative source for the counts and estimates of the whole population and its characteristics.

5. Geodemographers have been generally well served by the official statistics. However, with changes in society coupled with statistics users' rising expectations putting traditional methods of information collection under strain, a wide ranging inquiry is appropriate and welcome.

6. The CGG is firmly of the view that Government is in a uniquely advantageous position to collect and improve statistical information for all users. It is also in the position to benefit the whole information economy by making statistical output freely available at marginal cost and with minimal restrictions on its use. The general moves since 2000 to this approach should be given every encouragement.

7. It is, however, very evident to those like the CGG involved in the plans for official statistics that the rate and extent of recent changes in UK society are challenging the traditional methods of collecting statistical information suited to a more stable society. Risks from use of dated or inaccurate information are growing. It is, for example, more than twenty years since the census organisations began to examine alternatives to decennial censuses, and it is arguable that progress has not matched changing circumstances. The Inquiry is therefore a timely opportunity to push forward improvement in traditional methods, new sources such as administrative data, and in delivery of output.

TOPICS RAISED BY THE SUB-COMMITTEEUses and definitions

Uses of population estimates

8. The geodemographics community needs not only good quality periodic benchmarks of population numbers and characteristics at levels from the UK to the neighbourhood, but also regularly updated estimates for the same areas, as a basis for operating businesses and services effectively, particularly to help anticipate and meet the differing requirements of specific local populations.

9. Serious inadequacies in benchmarks or estimates, particularly if selective in terms of area or type of people, undermine the whole geodemographic approach. Indeed, in the absence of official estimates, the media research industry developed its own annual population estimates at postcode sector level (through the Joint Industry Council for Population Standards). The introduction of official sub-local authority estimates from 2001 onwards, albeit on an experimental basis, is therefore welcome.

Definitions

10. Counts of the usually resident population are generally the most appropriate base for geodemography. Changes to definitions, variation in and between sources, and revisions to figures over recent decades have not always been helpful. The objective of the “One Number Census" set for 2001, but not in the end achieved, should be paramount.

UK consistency

11. The basic population counts are sufficiently consistent throughout the UK for geodemographic purposes, and compare reasonably well with the statistics available in similar countries. But unhelpful variation, going beyond the differing needs of the devolved UK administrations, persists in content and dissemination, for example there is no single official source for UK-wide local statistics on the Web or in other electronic form.

THE ROLE OF THE CENSUSContribution of the Census

12. The main value of the Census is as a benchmark, still uniquely so at small area level. It is best when well tried methods are used to deliver straightforward results giving a snapshot of the country and showing relationships between key characteristics of the population. A specific value of the Census to market and other survey

based research is for organising fieldwork and weighting results to make samples representative of populations as a whole.

13. In current circumstances more frequent full Censuses might not be cost effective, and could deflect investment from development of alternative sources. Moreover, methods for intermediary sample or rolling censuses are undeveloped, although more frequent full counts would be welcome by users if there were no detrimental consequences to other sources.

14. The principal objective for 2011 must therefore be as complete as possible coverage of the population with avoidance of bias, with the means provided to detect and correct any under-coverage and bias.

The 2007 test census

15. In simple terms the test has shown that even well tried methods will be stretched to the limit by the nature of contemporary society.

16. The prospects of a good Census in 2011 remain in some doubt. Response rates in the 2007 test were at a historic low, quite possibly making it insufficiently representative of populations as a whole for a test of more contentious aspects. Moreover, major question marks remain over the availability of an adequate address framework for the proposed mail out of forms. Such concerns are set out in more detail a recent report by the Statistics Commission on managing the risks of the 2011 Census, and we are sure that this will be considered in the Inquiry.

17. In these circumstances use of the internet, with further unresolved issues of coverage and of security of personal data, and with no report of any large scale trial, would be a risk too far, despite the attractions of instant data capture.

A fit for purpose address framework

18. The first step towards ensuring a good Census in 2011, and an absolute priority, is a better address framework than the one-off compromise currently planned for the Census alone.

19. Such a framework must be accurate and properly updated as a record of physical places where people live, and must be openly available as a consistent base for all applications in the Census and elsewhere. None of this basic and essential information should be part of trading regimes.

20. For the greatest benefit there must be no restrictions on the use of output derived from the processing of address locations, likely for example to be a very large part of the output from the 2011 Census where coordinate (geographical) references which are the intellectual property of Ordnance Survey—a trading fund—will be used to allocate data to areas for the presentation of statistics. Charges should simply recover marginal additional costs.

21. The long running and persistent failure of Ordnance Survey, the Royal Mail, and the local authorities to work together for the creation of a national address framework is well known, and needs to be resolved at an appropriately high level where matters of funding and intellectual property can be decided for the benefit of the country as a whole.

Response rates and question content

22. The public and media must not perceive the Census as an unreasonable burden. Long term PR from ONS and Census users about the value of the Census can help, but primarily the Census must not be a vehicle for over elaborate or modish methods or questions. In particular the Census does not offer a prime solution to the challenge of measuring rapidly evolving patterns of migration and immigration.

23. There are well established guidelines which should be used to design successful self completed Census forms. Questions should be succinct and objective, with purpose clear to respondents, and contention minimized. The Census is not an “omnibus" survey for single short term issues, questions must have value over the long term for all user sectors. Topics should relate to one another to show relationships between characteristics. Significant value is added if further key characteristics such as social groupings can be derived by combining data from existing topics without placing a direct burden on the public.

24. Our view is that, provided that the questions are clear and non-contentious, then increasing the length of the 2011 Census form should not significantly impact on the response rate. ONS have already illustrated this through their testing of four page against three page questionnaires.

Loss of valuable question topics

25. The risk of losing valuable topics is well illustrated by the fact that the number of new question topics under consideration means that established labour market topics, including industry of occupation, hours worked and qualifications, are under serious threat of being excluded unless additional space is provided. These topics are the basis for widely used social classifications such as NS SEC and Social Grade, and add significantly to area classifications for geodemographic analysis. They also help in organising sample surveys.

26. We suggest that ONS should also consider including “proxy questions" for topics that would otherwise take up disproportionate space on the Census form, for example “age at end of full-time education" could be an acceptable surrogate topic for “qualifications", being easy to answer and taking up much less space on the form.

27. The cost of increasing the form length from three to four pages per person, which we understand to be some £22 million, is relatively small in comparison with the gain in benefit from the additional topics and is an investment which we feel Government departments would find well worthwhile and be able to support. This would be a small addition to the 2011 Census budget to permit a combination of questions giving the best return on the overall investment.

UK consistency

28. Geodemographic analyses are generally undertaken on a UK-wide basis, and benefit substantially from the availability of a consistent core of census questions and local statistics. The Census Acts (Section 4.2) allow users to commission output, and in fact the vast majority of 2001 Census output was produced under this arrangement. So consistency is in part a reflection of common needs among users, provided that there is sufficient consistency in questions and coded input, and provided that arrangements made by the three census organisations for commissioning and disseminating output cover UK wide needs.

Output

29. Our strongly held view as users is that there should be a greater emphasis on plans for accessible output throughout Census preparations, and indeed in all statistical operations, with the simple notion that the Census is only worthwhile if results are disseminated effectively, and with the clear aim of maximizing value through availability.

30. The statistical results from the 2001 Census are effectively in the public domain for all to use. This is in line with Treasury policy on boosting the UK knowledge economy by giving access to data where costs have already been met for Government purposes, and was implemented through the Treasury Invest to save programme.

31. Census output must continue to be freely available for use without restriction. In addition, boundaries in digital form. which are integral to the use and presentation of the output, must also be in the public domain.

MID-YEAR POPULATION ESTIMATESMethodology and accuracy

32. The CGG does not have particular expertise or evidence in this area, but it does have a major interest in up-to-date population estimates and projections for small areas, and we strongly support the continued production of official estimates for a range of small areas below local authority level. Our understanding that this is to be part of the next four year ONS programme of work, and this is welcome.

33. Although the prime concern is getting accurate estimates at local authority level, many need up to date information on the widely varying circumstances within the authorities. Any methodological improvements should therefore be for a coherent as possible package at all geographical levels. In addition, the sub local authority estimates made by ONS, which currently have the status of experimental statistics, should be established as permanent series of National Statistics without delay.

Improvement

34. We suggest that a better address framework, discussed in paragraphs 18-21 above, might offer a way of improving estimates of population numbers and change. As the effective tracking of the most mobile and elusive people may well remain problematic, it may be productive to see what improvements could be gained for surveys, administrative databases and estimates by the use of single source comprehensive and up to date “hard" information on all places where people live. For example, changes in “stock of dwellings" would indicate population changes and would also identify targets for additional surveys. An opportunity to do this arises from preparations for the 2011 Census.

35. There should be an urgent investigation of whether the address framework which the Census requires could also offer a basis for surveys and estimates of migration and population in the reasonably near future, within the timetable of the 2011 Census.

Revisions

36. Typically, revisions do not significantly change geodemographic analyses, but the impact of change on information systems can be unwelcome and an ideal position would be minimal revision.

THE ROLE OF SURVEY AND ADMINISTRATIVE DATA37. The CGG has limited information on the suitability and availability of administrative data sets and

suspects that coverage of the more mobile and elusive people, which is the most challenging aspect of estimating population, may be deficient in many such sources. Nevertheless, the bases are current and comprehensive, and we see considerable scope for increasing statistical output from records such as national insurance, benefits and health.

38. However, sample surveys have the advantage of being purpose built, and we feel that they should play an important part in the estimation of the number and characteristics of local populations through the application of analytics and modelling techniques. Cuts in funds for surveys should be reversed.

39. The potential offered by a better address framework is covered in paragraphs 18-21 above.

COOPERATION WITH STAKEHOLDERS40. The geodemographics community has reasonable relations through annual meetings with the

National Statistician and through the Census advisory machinery, although we feel that engagement through the latter has been detrimentally diminished recently. The ONS workload has meant that different sectors, with differing requirements from the Census, have shared advisory group meetings. We also feel that the way in which relations with stakeholders is managed tends to be divisive, for example the needs of central government departments are perceived as being handled separately (and given higher priority). This may put at risk across-the-board user support for statistical operations such as the 2011 Census.

41. We therefore suggest that there should be action by the Statistics Board and Executive Office to ensure that the wider user community plays a full part in shaping statistical programmes. It must recognise the economic benefit and importance of good official population statistics particularly for the private sector. Many information needs are in fact common between sectors, and the Board must ensure that relationships between producers and users are not divisive.

November 2007

@HR25@

Memorandum from The Market Research Society (MRS)

With members in more than 70 countries, MRS is the world's largest association representing providers and users of market, social, and opinion research, and business intelligence. All individual MRS members and MRS Company Partners agree to self-regulatory compliance with the MRS Code of Conduct. Full details regarding MRS and its activities can be found via: www.mrs.org.uk

MRS welcomes this opportunity to respond to the Treasury Sub-Committee inquiry and confirms that no part of this submission needs to be treated as confidential.

MRS has an advisory committee, the MRS Census and Geodemographics Group (CGG), which recommends MRS policy in relation to geographic, demographic and census information. CGG is submitting a detailed response to the inquiry, which MRS fully endorses.

MRS would also like to identify the following additional points on the priorities listed in the CGG submission:

@UL@— The national address register, identified as a requirement for the 2011 Census, would be an invaluable data resource for the market research industry and would strengthen all consumer research undertaken in this country. @/UL@

@UL@— Pressure of space on the 2011 Census form means that labour market questions, that are essential for market research purposes, will be dropped from the 2011 Census, unless a fourth page of individual questions can be funded.@/UL@

@UL@— The research industry requires updated population estimates for sub-local authority areas, for a host of purposes ranging from market research through to geodemographics.@/UL@

November 2007

@HR25@

Memorandum from the Statistics Commission

SUMMARY@UL@— Good population statistics are fundamental for policy development and for planning and allocating funds for public services. For many years migration, rather than births and deaths, has been the dominant influence on population change. The Statistics Commission has pressed for improvements to migration estimates since 2003, arguing that there are potentially large economic costs to not knowing with sufficient precision the size and geographic distribution of the population. @/UL@

@UL@— We welcome various recent initiatives by the Office for National Statistics to improve the estimates, but we now want to see other government departments take the necessary steps to provide ONS with better raw data from administrative systems. @/UL@

@UL@— Drawing on the Commission's recent report Counting on Success: The 2011 Census—Managing the Risks, we summarise our key recommendations to the incoming Statistics Board regarding preparations for, and risks facing, the 2011 Census.@/UL@

@UL@— The Commission supports the new data-sharing provisions in the Statistics and Registration Service Act 2007 and urges Whitehall departments to cooperate with ONS to develop and share data for statistical purposes, subject to proper safeguards. Recent and serious concerns about the physical transfer of administrative records between government agencies are not a counter-argument. There are adequate technical solutions available if implemented and monitored properly to ensure both individual confidentiality and benefit to the whole populace in terms of reduced cost and better official statistics. @/UL@

@UL@— The Commission would like to see the Statistics Board give high priority to developing cross-government planning of official statistics within a broader umbrella of user and stakeholder engagement. At present no coherent mechanism exists to identify and address the big issues, many of which cut across responsibilities of multiple departments. @/UL@

@UL@— In the context of migration, it is also important to work jointly between countries to establish reciprocal measurement; every immigrant is an emigrant somewhere else. In some respects this will also be true between England and the three devolved administrations—to the extent that they have distinct administrative systems.@/UL@THE USES OF POPULATION ESTIMATES

1. Population estimates—by area and age—are the bedrock of every national system of official statistics. As well as measuring the most fundamental characteristics of society, they underpin other key statistics from macro-economic indicators, such as GDP per capita, to small area statistics such as unemployment rates or crime rates. They are used as the core component in statistical formulae that allocate very large sums of public money to the devolved administrations, to local government, the health service and public services. They are also the basis for projections such as forecasts of household numbers and life expectancy, which influence assumptions about housing demand, social care and pensions. To the extent that population estimates are inaccurate or misunderstood (see below), they have the potential to undermine administrative processes, invalidate research findings and policy decisions, and reduce the value for money of public services.

2. Statistics of population, as routinely published by the Office for National Statistics19, are residence-based. That means they relate to the assumed “night time" population. If they are used as an indicator of day-time demand, say in relation to the expected need for street cleaning, they may be misleading. The extreme case is the City of London where the day time population is well over 30 times the night time one, but similar if smaller discrepancies exist in other inner urban areas. So the concept of “accuracy" is not straightforward and it may be more useful to ask whether the statistics are fit for the purposes for which government and society needs them. The answer to this appears to be that often they are; but sometimes they are not. Examples of cases where population figures fall short of being fully fit for purpose include when they are used in relation to inner city areas; when they are used at a very small area level; in the final few years before a decennial census (when they are likely to have “drifted", at least for some areas); and when they are used as specific indicators of need to spend on public services that are not closely related to resident population numbers (eg crime against visitors).

19 See for example UK population set to increase to 65 million over the next ten years, ONS News Release, 23 October 2007. http://www.statistics.gov.uk/pdfdir/pproj1007.pdf

3. There are two main components of change in the population, natural change (births and deaths) and migration. Over at least the last ten years, migration has been the dominant influence on population change. At the level of local authority population figures or for smaller geography, the migration component will be more dominant still—because, at these geographical levels, migration has both an international component and a between-areas component. And the dominance of the migration component will be bigger again in the most “hard to count" areas as identified for census purposes. So, for those areas which present the greatest challenge statistically, much of the challenge relates to migration (international and internal).

4. In 2006, we set out in a letter20 to several government ministers the consequences of not having adequate population and migration data:

@UL@— Weak data can lead to inefficiency in the allocation of grants to local authorities, the NHS and other public services. We are not in a position to estimate the cost to the public purse of such inefficiency but we believe it could be very substantial.@/UL@

@UL@— Mis-measurement of migration could contribute to failure to predict accurately the demand for services from the NHS, in education, social services etc. Over-provision for such services is wasteful of public money; under-provision is liable to lead to unnecessary suffering. There could also be longer-term consequences in relation to the planning of social housing and other infrastructure.@/UL@

@UL@— The development of policy on immigration, the monitoring of the impact of current policies, and the provision of services targeted at immigrant communities may all be adversely affected by weak data.@/UL@

@UL@— The 2011 Census may be left vulnerable if the migration estimates in the years prior to 2011 are not robust. Essentially, the danger is that if the Census appears to contradict earlier estimates, as happened in 2001 in some areas, there will be a loss of confidence in the quality of census data.@/UL@

5. The response to that and subsequent letters has been essentially that the Government recognises the need to improve migration statistics but no specific commitments have been made to our knowledge. Ministers mention their support for the Inter-departmental Task Force on Migration Statistics set up by the Office for National Statistics in 2006 and the Improving Migration and Population Statistics (IMPS) research programme.

6. The Commission believes that ONS's report of the Inter-Department Task Force on Migration Statistics21 (published in December 2006) presents broadly the right path ahead but it is one that requires substantial commitments from other government departments. In January 2007, ONS also published a report looking into the feasibility of estimating short-term migration,22 and published feedback23 on this in April and “experimental estimates" in October.24 Short-term migration is important as it is not normally included in the official figures for migrants (which, following international practice, only counts as “migrants" those intending to stay for more than a year). But, perhaps particularly following EU enlargement, short-term migration has the potential—and already the reality in some areas—to have significant implications for local services.

7. In April 2007, ONS published information on changes in the way migration will be estimated for the mid year estimates from 2007 onwards. However, this largely applies new methods to existing sources which have known weaknesses. A number of local authorities have expressed concerns about the new approach because it essentially re-distributes existing figures rather than improving the underlying estimates.

20 Statistics of International and Internal Migration, letter to the Minister of State at the Home Office and others, 8 May 2006, http://www.statscom.org.uk/uploads/files/correspondence/Letter0300.pdf21 Report of the Inter-departmental Task Force on Migration Statistics, ONS, 15 December 2006 http://www.statistics.gov.uk/about/data/methodology/specific/population/future/imps/updates/downloads/TaskForceReport151206.pdf22 Short-term migration feasibility report, ONS, 25 January 2007 http://www.statistics.gov.uk/about/data/methodology/specific/population/future/imps/updates/downloads/Short-termMigrationFeasibilityReport.pdf23 Summary of Feedback received on Short-term migration feasibility report, ONS, April 2007 http://www.statistics.gov.uk/about/data/methodology/specific/population/future/imps/updates/downloads/STM_feedback.pdf24 Research Report on Short-term Migration, ONS, 25 October 2007 http://www.statistics.gov.uk/about/data/methodology/specific/population/future/imps/updates/downloads/STM_Research_Report.pdf

8. We note that ONS, together with DWP and Home Office, are taking the initial steps recommended in the IMPS research programme—improving the information that exists about migration across government, reviewing administrative sources and carrying out research, through tracking and linking data sources. 25

9. These various initiatives, welcome as they are, will only deliver better migration estimates if other government departments take the necessary steps to provide ONS with better raw data from administrative systems. Statistical surveys, of the kind ONS can undertake, may well have a role in the future to fill out the detail on the population in areas of high turnover and to look at specific issues (such as the experience of immigrants in securing jobs). However, sample surveys are not in general a satisfactory basis for measuring the movement of population into, out of, and around the country. National sample sizes cannot be large enough to give reliable local area information.

10. The 2011 Census is of course an exception in terms of survey size and will be a very important source of population and migration data, but it will be a one-off event and will not solve the problems for the longer term. And censuses are not necessarily the ideal tool in this context. As was seen in 2001, the “hard to count" areas in the Census tend to be both where migration is a major influence on the population and where the Census results are most likely to be called into question.

ROLE OF THE CENSUS11. The decennial Census has the potential to offer relatively reliable and detailed information on the

population at a local level; and it is currently the only source in the UK that can do so. The Census acts as a baseline for subsequent year's population estimates, which accordingly become less reliable in the latter years of the decade. The Statistics Commission has been considering census-related matters for some years, most recently in our report Counting on Success: The 2011 Census—Managing the Risks, 26published earlier this month. The title of that report refers to the fact that we are all counting on the success of the Census, but success is far from being a certainty. We also need to keep in mind that the Census will be by far the biggest statistical project ever undertaken in the country, and as such, may of necessity draw resources and managerial priority away from other initiatives, such as the better exploitation for statistical purposes of the administrative records held by government departments.

12. In the report, we conclude that high-level discussions about what will replace, or at least supplement, censuses after 2011 need to be taken forward in parallel to work relating to 2011. We have been impressed by the cost savings, and quality improvements, that have been made in several Scandinavian countries that now rely mainly on registers of population, households and businesses. And also by the advantages, but lesser cost savings, of moving to a short census form supported by a large continuous survey, as in the United States. Whatever future path is determined by Government for the UK, we believe 2011 should be the final Census of its traditional kind and planning for the longer term, at the top level of government, should start now. It is not sufficient for this to be a debate amongst experts and academics, the major political and financial considerations of adopting a new approach need to be central to the planning.

13. Recognising that censuses are by their nature problematic and vulnerable to disruption, we highlight in the report some aspects of the preparations for 2011 on which more could be done by government to assure success. We have made recommendations to the incoming Statistics Board, including the following:

@UL@— The Statistics Board should engage all interested parties, including Parliament and the devolved administrations, in building a consensus on how success will be judged for the 2011 Census and publicising the outcome. We make some suggestions in the report but our main concern is that there should be a consensus building process and a clear outcome early enough to be helpful to the census offices.@/UL@

@UL@— Judgements about the quality of the Census will be heavily influenced by whether results are approximately consistent with expectations prior to the Census—a problem in 2001 for some areas. It is therefore important that expectations are well informed and for this it will be necessary to have robust information on population movements—that is both international and internal migration—ahead of 2011. The Statistics Board may have to accept that there is already a credibility problem facing the census estimates in view of the current position on migration estimates. We suggest in the report that the best way to handle this may be to pinpoint those geographical

25 Statistical Evidence on the Economic Impact of Immigration Statistical Submission to the House of Lords Select Committee on Economic Affairs, ONS, DWP and Home Office, October 2007. http://www.parliament.uk/documents/upload/EA235%20ONS.doc26 Counting on Success: the 2011 Census—Managing the Risks, Statistics Commission Report No. 36. http://www.statscom.org.uk/uploads/files/reports/Report%2036%20Census%202011.pdf

areas in which the Census results may prove controversial and engage local authorities and other stakeholders on what further steps might be taken locally.@/UL@

@UL@— Government has not been successful in establishing a single good quality national address register to reconcile the Address Point product from Ordnance Survey with the Land and Property Gazetteer managed by the Local Government Information House. This will hamper Census enumeration and the Statistics Board will have to give some priority and resources to assessing the scale of the problem and finding ways to mitigate the weaknesses in the available information.@/UL@

@UL@— The Statistics Board should actively seek to promote understanding of the risks facing the 2011 Census; and do so across all levels of government and public administration in the UK. It should identify and emphasise the scope for many organisations inside and outside government to help contain those risks. As part of this, special interest groups which may use the Census as a vehicle to gain publicity need to be identified and engaged in constructive discussion. The time for this is now, not just in the final months before the Census. @/UL@THE ROLE OF ADMINISTRATIVE DATA

14. The Statistics Commission is aware that ONS has carried out research directly with some local authorities to evaluate local population estimates using various local and national sources of information—and has recently published a review of the potential use of some administrative sources in relation to making population estimates.27

15. However, we think that more local area research, particularly to evaluate official estimates against several administrative sources of information in the most problematic areas, should be carried out and published by government. ONS cannot, of course, “negotiate" local population figures; such estimates must all be produced in the same systematic way. The aim would instead be to get a better understanding of the scale, geographical variability and distribution of the problems with population and migration estimates.

16. There is a question of mind-set here. Just because population estimates for all areas have to be produced in a uniform way does not mean that ONS or others in government should feel obliged to defend every local estimate as being the best population estimate for that particular area. We would like to see ONS publishing its own critique of specific local estimates, perhaps on occasion saying that the local evidence suggests that an estimate produced by uniform methods is too low or too high. This might create problems about the use of the data in matters such as local authority funding formulae, but in reality those problems already exist and are mostly well known to exist. Transparency about this would minimise misunderstandings.

17. Much more work is going to be needed to develop new sources of data to achieve better estimates. We have argued that to produce better estimates, ONS will need to receive raw information on people moving into and around the country from the big Whitehall departments: the Home Office, the Department for Work and Pensions, Communities and Local Government, HM Revenue and Customs, and the Department of Health amongst others. We have observed some reluctance on the part of departments to develop and share information for such purposes and we believe this will need to be resolved at ministerial level. The data-sharing provisions of the Statistics and Registration Service Act offer a new tool that should now be used to the full.

18. There are of course important technical and data protection implications in bringing administrative information together; any data on individuals would have to be kept totally confidential and the information used only for aggregate statistical purposes. The recent debacle over loss of personal data by HMRC indicates the need for much enhanced processes to protect (and demonstrate the protection of) personal data; it does not invalidate the sizeable benefits that can arise from such use. We believe these technical and data protectin implications need to be tackled openly, rather than having the effect of keeping the debate behind government doors for fear of political reaction.

19. The Audit Commission recently recommended more joint working between local and national government, such as sharing data on National Insurance numbers.28 We would support any initiative of this type.

20. There are no quick or cost-free answers. Government needs to decide whether improving estimates of population and migration in the long term is of sufficient importance to justify the commitment of substantial resources to address the impediments. Other countries have solved the problem through the use of integrated

27 A Review of the Potential Use of Administrative Sources in the Estimation of Population Statistics , ONS, http://www.statistics.gov.uk/about/data/methodology/specific/population/future/imps/updates/downloads/admin.pdf28 Crossing borders—Responding to the local challenges of migrant workers, Audit Commission, January 2007

population, household and business registers. This is one possible solution for the UK but there are perhaps others of merit given the culture and infrastructure in the UK.

21. One model that has been suggested to us would be that the core information on immigration/emigration and internal migration would be obtained from new or existing administrative systems (managed by the immigration service at ports in the case of immigration/emigration and from several sources for internal migration) and brought together with existing administrative information on births and deaths to produce a reasonably robust population register which would contain minimal information about individuals. That would then be supplemented by sample surveys that use the core population register as a “spine" whilst enriching the statistical picture. This is only one approach but it illustrates the sort of option that needs to be considered seriously.

22. In the shorter term, there is an opportunity to carry out a major experiment on how well existing administrative data held by various government departments (notably the Department of Work and Pensions) could be used to build not only an “administrative census" but also to deliver better estimates of migration. If this were done and the results compared with those of the 2011 Census, we would have a much better idea of whether we can get reliable, frequent and up-to-date population and migration figures from administrative sources.

ONS COOPERATION WITH STAKEHOLDERS23. The Statistics Commission recognises that the Office for National Statistics has made efforts to

understand users' needs for population and migration statistics, notably in its Census Advisory Groups, by attending meetings with users, including the Statistics User Forum, by consulting on future work priorities and through publishing a lot of information on its website to inform users of its thinking. However, we believe that devoting some ongoing resource to better understanding the uses made of these (and other) key statistics would be well worthwhile and would point to ways to improve the service provided. The Commission's own report, The Use Made of Official Statistics, 29illustrates the richness of the use of this kind of statistical information across society and discusses the meaning of “public value" in that context.

24. We would like to see the new Statistics Board giving high priority to the cross-government planning of official statistics; and between governments in some respects. At present, no coherent mechanism exists to identify and address the big issues—as is witnessed by the case of migration statistics. The Statistics Board will, itself, be part of the solution here but there are many largely autonomous players, including the devolved administrations. We need new mechanisms that will engage all the relevant parts of government in finding ways of achieving the benefits of comparable data drawn from robust sources.

The Statistics Commission was set up in 2000 to “help ensure that official statistics are trustworthy and responsive to public needs", to “give independent, reliable and relevant advice" and by so doing to “provide an additional safeguard on the quality and integrity" of official statistics. It operates openly and independently, with all its papers normally available publicly.

November 2007

@HR25@

29 The Use Made of Official Statistics, Statistics Commission Report No. 33 http://www.statscom.org.uk/uploads/files/reports/Report%2033%20Use%20Made%20March%2007.pdf

Memorandum from Suffolk County Council

SUMMARY1. During the last six years ONS population estimates for one of the Districts in Suffolk have varied by

more than 27% (15K); the Council on the other hand believes the population is growing slowly. Any performance indicators, rates or proportions prepared using the ONS estimates will also exhibit a considerable degree of variation which could influence decision making. This is an exceptional example but serves to show the degree of variability that can occur.

2. Since there is increasing evidence that newcomers from abroad are living in rural areas, ONS may need to reconsider its Census enumeration procedures to cope with this.

3. There seems to be some scope for Local Authorities to evaluate the ONS population estimates prior to publication, for example by making simple comparisons with earlier estimates or with other variables.

4. There may be some merit in ONS enhancing its surveys in areas with a poor demographic record.

INTRODUCTIONThis response uses the estimates for Forest Heath to enhance the response from the Local Government

Association by illustrating the points that the LGA has made. Forest Heath is a district in Suffolk for which ONS has difficulty preparing annual population estimates and projections, largely because the district has a high proportion of people whose demographic characteristics are unusual. About a fifth of the population are employed or associated with the two US military bases there; these people tend to be younger than average, all the military are employed and many personnel and their families arrive and leave the two bases each year, usually to go abroad. The time spent in this Country varies considerably, but some are effectively visitors. Their moves may constitute international migration, migration with the rest of the UK or be measured by counts of the military and their families.

1. USES AND DEFINITIONSWhat are the uses of population estimates, and how far do current population estimates meet the needs associated with those uses? What are the effects of inaccuracies or inadequacies in such estimates?

The table below sets out all the estimates prepared by ONS for this district for mid-2000, mid-2001, mid-2005 and mid-2006, plus estimates prepared by Suffolk County Council for mid-2000 and mid-2005. All the Council's estimates are marked in bold print, including the ONS estimate for mid-2001 which the Council considered acceptable. The Council prepares its own estimates for two reasons:

@UL@— It needs estimates of total population for areas smaller than a Local Authority District for service delivery and policy formulation; these figures represent the sum of parish estimates.@/UL@

@UL@— Experience since 1974 has shown that the Council's estimates have proved a more accurate predictor of the Census results than those prepared by ONS. Although the Council uses various indicators available from ONS to calculate its estimates, it has the added advantage of local knowledge.@/UL@

TOTAL POPULATION OF FOREST HEATH

Source Date to which figure refers

Estimate Date of publication

SCC mid-2000 56,175 01/04/2002ONS mid-2000 70,772 17/12/2001ONS mid-2000 62,100 01/09/2003ONS mid-2000 62,300 01/10/2004ONS Held on 29/04/2001 55,510 2001 CensusONS mid-2001 55,603 01/09/2002ONS mid-2001 56,223 01/11/2003ONS mid-2001 56,145 01/09/2004ONS mid-2005 63,117 06/07/2006ONS mid-2005 61,057 22/08/2007SCC mid-2005 56,936 25/10/2006

ONS mid-2006 62,129 22/08/2007

Over this six year period, the ONS estimate varied from 55,600 (first draft mid- 2001) to 70,800 (first draft mid-2000) whereas the Council's estimates varied from 56,200 (mid-2000) to 56,900 (mid-2005). The Census result was however 55,510, somewhat smaller than both suites of data, but considerably nearer to the Council's estimates than those prepared by ONS.

It is usual—and often mandatory—to use ONS estimates in preference to any other population estimates to calculate any performance indicators since these estimates are prepared consistently across the Country enabling comparisons to be made. Indicators are also calculated as soon as the latest estimate has been released ie they utilise the first draft rather than any later version. Although many of the estimates have been updated this is not usually carried out until some time has elapsed, by which time the performance indicator has been calculated, utilised and is considered superseded by one using an estimate for a later date.

In the case of Forest Heath any performance indicators that utilise population estimates in the denominator will fluctuate considerably, particularly since the smallest ONS estimate in the table is only about three quarters the value of the largest estimate. It is not always possible to tease out whether the fluctuations in the indicator are genuine and therefore indicative or simply caused by the denominator. If the wrong interpretation is made then this could have serious consequences.

Forest Heath is an unusual district in that it has an exceptional population with characteristics that make population figures difficult to calculate and the time period has been specifically chosen to span the least accurate population estimates (since these are developed successively from the previous population Census.) Most of the time, most districts will not exhibit such variability. However this exercise demonstrates the degree of variability that can occur in the population estimates and thus any other derived variables. During this period, performance indicators, percentages or rates would be expected to be fairly uniform or changing slowly over time whereas the variability in the denominator means that such derived variables are easily distorted.

2. THE ROLE OF THE CENSUSWhat steps should be taken to increase the rate of responses to the Census, particularly in hard-to-reach areas? How effective are plans to measure the extent of non-response to the 2011 Census?

The conventional hard-to-reach groups generally live in urban areas, but there is increasing evidence that many migrant workers live in rural areas. For example, Suffolk County Council has noted the large number of foreign-born children starting school this term, many of whom live in rural areas. There will obviously be many adults without children coming here as well, who cannot be so readily tracked. During its Census test in Carmarthenshire, ONS admitted that it did not find it easy to work in rural areas owing to the distances and thus time (and costs) involved in travelling. An allowance need to be made in rural areas to ensure this new hard-to-reach group is properly enumerated and included in the results, especially as we do not seem to have any other adequate source of hard data down to local level.

3. MID-YEAR POPULATION ESTIMATESHow accurate and useful are the mid-year population estimates for England and Wales, including information available for local authorities and Strategic Health Authorities?

There is an additional facet to the inadequacy of estimates of international migration that is specific to Forest Heath. There seems to be an overlap between internal and international migration with changes in the numbers of foreign armed forces and their dependants, as analysis of change in this District since 2001 shows that most of the change occurred in the young adult age groups—the age of the armed forces. The level of change suggests the military increased by as much as 60%, an increase which is not substantiated by the US military. This means either visitors are being included in the population estimates or that newcomers are being double counted in some way. ONS says it will investigate this.

What are the most appropriate ways to ensure a common level of accuracy in mid-year population estimates for different areas and what steps should be taken to improve the accuracy in areas where population is hard to measure?

In view of the comment above, an appropriate organisation to contact would be the US military which can corroborate the increase in numbers and holds records of staff and dependants for whom they are responsible for comparison with ONS' figures.

Many Local Authorities conduct a “reality check" on the population estimates or on the year-on-year change in either the total population or population structured by age. The latter approach enabled identification of the root cause of the population change in Forest Heath since 2001 (an apparent increase in the number of young adults that has not actually occurred). A simple check on population change involves comparing this with changes to the housing stock. Whilst it is acknowledged that housing stock increases do not necessarily imply an increase in the population (owing to declining household size, increasing number of second/weekday homes and changes in the number of vacant properties), the reverse is not true. Population increases must coincide with increases in the housing stock; otherwise household size will have to increase and more homes will be the main home. During April 2001 to April 2005, the housing stock of Forest Heath increased by 575, whereas the population increased by 4,912 (using final ONS estimates), implying the average household size increased from 2.28 to 2.43. Council figures suggest the average household size fell from 2.28 to 2.26, a more likely scenario. If this simple check had been carried out sooner, ONS may not have published population estimates that seem counter-intuitive.

4. THE ROLE OF SURVEY AND ADMINISTRATIVE DATAWhat role can and should be played by survey data in the compilation of population statistics?

It is ironic that the sample size of the Annual Population Survey is so small in Forest Heath, a District with a difficult demographic profile. In districts with such a poor demographic record, perhaps the sample size should be inflated to assist with improvements to any estimates?

November 2007

@HR25@

Memorandum from Mr A R Thatcher CB

This evidence is concerned solely with the definition of the population. It describes the effects which the choice of definition had on the estimated population of England and Wales in 1991 and 2001, and the effect on the difficulty of counting the population.

The great problem in counting the population is that people move about This was well-recognised from the beginning. The problem was solved in the traditional censuses by asking every householder to give the required census details for every person who slept at the address on census night. This was a clear and unambiguous instruction. Every person would appear on only one return, so there would be no duplications or omissions. The enumerators identified the households, gave out the forms and then collected them again, checking that they had been completed. The total number of named persons on the returns was the “population present". The population present was obtained in this way at each census from 1841 to 1981.

The population present included persons from abroad, who were identified on the returns, and these could be subtracted for the purpose of estimating the national resident population of England and Wales. In 1981 an estimated addition was also made for residents who were temporarily abroad on census night, either visiting or on holiday. However, these were additions or subtractions which started from a figure which was well-defined and known, namely the population present.

From 1961 onwards there was a growing demand from users for population estimates to show the “usual residents" in local areas. In 1971 and 1981 both the “usual residents" and the “population present" were collected and published. In 1991 both were collected, but only the “usual residents" were processed and published. In 2001 only the “usual residents" were collected.

The effect of these decisions can be seen in retrospect to have been considerable. Apart from visitors from abroad, who can easily be excluded from the count, the difference between the “population present" and the “usual residents" at national level consisted of many of the people who were recorded as being present as visitors at wherever they were staying on census night. If they gave their name and census details, and also gave a usual address in England and Wales, then they were counted as part of the national resident population derived from the population present. However, they were only counted as part of the “usually resident" population if the h d been declared by least ore household to be a usual resident at that address who was absent on census night

A serious problem was caused by the fact that a large number of these visitors were not recorded as being “usual residents" at any other household. In many cases this may have been because a whole household was absent from its usual address, but it is not difficult to think of other reasons. If people move about frequently, nobody is quite sure what is their usual address. Also a householder, faced with the task of completing a long census return for someone who is not present to supply the details, and who is not quite sure whether he is a genuine “usual resident", may find it easier to leave him off the form.

Whatever the reasons, it was discovered long after the event (from the longitudinal study) that in 1991 the number of visitors in this category was about 500,000. If the visitor data collected in the 1991 census had been processed in the same way as in 1981, it would have been seen that these 500,000 named individuals existed and had not been included in the original count. Some of their numbers were later “imputed", but that is another story. If the “population present" had been calculated in the traditional way in 1991, the history of the population estimates would have been different.

In 2001, when there was more mobility, the presumption is that the gap between the “population present" and the “usually resident" population will have been even higher.

The censuses of 1991 and 2001 gave full rein to the desire of users to have just statistics of “usual residents". The difficulties this caused for counting the population were not anticipated. Perhaps the time has come when it might be worth while to reconsider the pros and cons of the choice between the “population present" and the “usual resident" methods as the basis for the census count itself in 2011.

The population present has great simplicity. It is quite clear which persons should be included on each census return. There is no need to fill in forms for people who are not there. The householder simply includes on the return everyone who was present on census night, and also gives the usual addresses of those who were visitors.

In the “usual resident" method, in contrast the householder has to make decisions about who should be shown on the return There are marginal cases of doubt, about who is and who is not a “usual resident", and this is where errors can creep in.

Statistics of the number of usual residents in each local authority area are given directly by the “usual resident" method but they can also be obtained easily by the “population present" method. By reallocating (ie transferring) the visitors to the local authority area which contains their stated usual address, one obtains the number of usual residents in each local authority area, both those who were present and also those who were absent on the census night because they were visiting another address in Great Britain. Presumably the same method can be applied to smaller areas using postcodes.

It should be mentioned, though that there may be a practical problem in transferring all the absent residents back to their individual households, if there are difficulties in identifying the household. If the “usual resident" method worked perfectly, this would provide household compositions including those members of the household who were absent on census night because they were visiting another address. However, we already know that in practice the “usual resident" method missed many of these absent residents. Both methods may therefore have a problem in dealing with the effect of absent residents on household composition.

In comparing the “population present" method with the “usual resident" method for the purpose of counting the population, there can be no question that the “population present" method is much simpler for the form-filler. It also seems to pick up at least 500,000 more people. These are substantial advantages.

November 2007

@HR25@

Memorandum from CACI Ltd.

EXECUTIVE SUMMARY1. CACI, together with other Census distributors, is part of an industry which provides value-added

information products and services based upon population statistics. We address a wide range of applications in the public and private sectors.

2. We believe that ONS works to high quality standards. However the time taken to publish data and output inconsistencies between the various UK statistical agencies can limit the value of the data.

3. New challenges to the measurement of population arise from the increase in hard-to-count population and from increased rates of migration. While various approaches may be used to reduce the resulting problems within a traditional census-based strategy, there may be a case for considering other approaches.

4. The current high level of interest in migration-related questions risks damage to the value of the 2011 census if important labour market questions, such as industry and level of qualifications, are excluded as a result.

5. There may be significant opportunities for making better use of statistics derived from administrative records.

6. The maximum economic and welfare benefits from population statistics can be achieved via the earliest possible publication of unrefined aggregate data.

INTRODUCTION7. CACI Ltd. is one of the largest commercial organisations with a specialism in providing products

and services based upon the population geography of the United Kingdom. A large part of our business makes use population statistics together with other data as the basis of providing services to a wide range of public and private sector end users. We have the longest history of all companies working in this sector, with experience of using censuses from 1971 onwards.

8. Together with the other members of the Association of Census Distributors, we are part of a competitive industry which is responsive to end-user needs and works efficiently to provide value-added information products and services. We have been able to combine the census with other data sources to create innovative and valuable solutions to real-world problems. We believe that over 90% of private sector use of population statistics is delivered via a census distributor, together with an increasing volume of public sector usage.

9. An important factor in the growth of this industry and of its value to the UK economy has been the adoption of the principles outlined in the Treasury “Cross Cutting Review of the Knowledge Economy : Review of Government Information" (2000). This study observed that:

@UL@”It is generally accepted that information derived from the statutory and normal workings of government forms the largest single information resource in a developed economy. In the new knowledge economy, “infomediaries" can aggregate and repackage apparently disparate data sets for end-users. But the importance and potential value of the information collected by and for departments and agencies may not always be fully appreciated within government and there appear to be obstacles which currently prevent its widespread re-use."@/UL@

10. The Cross-Cutting review further gave general support to marginal cost pricing for government data, trading funds excepted.

USES AND DEFINITIONSThe uses of population estimates

11. We make use of statistics on both the number and the key characteristics of the population such as age, affluence, household structure, education, industry, and ethnicity.

12. Our work in the public sector includes the following :

@UL@— CACI provides consultancy for major regeneration projects across the country and an evidence base for many Local Development Frameworks.@/UL@

@UL@— NHS organisations employ our classification tools and advice to target Public Health Improvement initiatives and refine Strategic Needs Assessments.@/UL@

@UL@— Police Forces and Community Safety partnerships work closely with CACI to maximise their understanding of neighbourhoods to develop local policing strategies.@/UL@

@UL@— Education Departments within local authorities and the Department of Children Schools and Families analyse pupil attainment and measure school performance using classifications and indicators developed by CACI.@/UL@

@UL@— Local authorities employ our tools to prioritise their e-government and service transformation projects.@/UL@

@UL@— Non-Departmental Public Bodies such as Visit England, the Arts Council and Welsh Tourist Board employ our tools to measure participation across all social groups.@/UL@

13. The uses of population estimates in the commercial sector include the following :

@UL@— Selecting locations for retail outlets, assessment of available market size and design of the outlets.@/UL@

@UL@— Assessment of shopping centres in terms of present and future projected market sizes@/UL@

@UL@— In market research exercises, to plan interview schedules which are representative of the general population@/UL@

@UL@— Facilities requirements assessment for utility companies@/UL@

Many high value private sector investments depend upon the availability of accurate population statistics.

The usually resident population base

14. We use the following three population bases, listed in decreasing order of importance

@UL@— The usually resident population base@/UL@

@UL@— The daytime population base. How many people work in a neighbourhood?@/UL@

@UL@— Estimates of tourists and visitors. These can be important in planning resources in certain areas of the UK.@/UL@

Estimates of the number and type of households are also very important to our work.

15. Most statistics are currently published on the usually resident base. The census gives a decennial snapshot of neighbourhood working population. Information on tourists and visitors is generally difficult to come by.

International comparisons

16. We have a perspective which is primarily UK-wide, since most of our clients operate throughout the UK, and secondly EU-wide because of the increasing interest of UK firms in expansion into Europe..

17. To date the quality of statistics available in the UK has been very high in comparison with most other European suppliers. ONS works to high quality standards and its data is normally available freely, openly and for small geographical areas.

18. UK-wide consistency is particularly important to us. Inconsistencies between the Scottish census on the one hand and the England and Wales census on the other have sometimes caused us difficulties. We have some

concern about the apparent general movement towards increased independence between the various statistical agencies in the UK, and the resulting likelihood of decreasing consistency between statistics for the various regions in the UK into the future.

THE ROLE OF THE CENSUS19. We believe it is important to view the census as part of a coherent set of methods used to measure

population statistics.

20. At present the census is the only mechanism which approaches a complete count of population and households and their characteristics within local neighbourhoods. It is the only time ONS get out onto the streets to count everybody and no other source gives an effective alternative at present.

21. There seem to be two main underlying long-term challenges to its effectiveness:

@UL@— The conduct of a comprehensive count of this kind is becoming progressively more difficult. The hard-to-count population is growing and becoming progressively harder-to-count.@/UL@

@UL@— The census is infrequent, while population is becoming more mobile. Once published, the results lose accuracy more rapidly than in the past.@/UL@

22. There appear to be two plausible approaches to dealing with these trends.

23. The first is to base the ongoing strategy around the existing framework of regular, but infrequent, “traditional" censuses. If this route is chosen, the options available to reduce the impact of the trends include:

@UL@Further attention to methodological detail, such as more thorough effective post-enumeration assessment and a high quality address register.@/UL@

@UL@Conducting the census more frequently—for example at 5-yearly intervals rather than 10-yearly.@/UL@

@UL@The publication of census data sooner after the census date than is currently achieved. This would provide more up-to-date and hence more accurate data. @/UL@

@UL@Subject to protection of privacy and other legal constraints, the better use of administrative records and other data sources. These could potentially be used both to run a better census and to help track changes in the population between censuses.@/UL@

24. The second approach is to move to a new strategy such as that considered in ONS's studies for an “Integrated Population Statistics System", which is based around a continuously maintained population register.

Census questions and the length of the form

25. We are alarmed that the demand for topics such as country-of-origin, nationality, citizenship and similar subjects in the 2011 census, although understandable, has put pressure on the available space on the census form and may lead to the exclusion of key labour market questions such as industry, travel to work, level of qualifications, size of establishment and hours worked. All of these questions, and in particular the questions on industry and level of qualifications, have proved very valuable in the past.

26. At the time of writing, ONS are proposing to exclude all these questions from the 2011 census if the questionnaire has only three pages per individual.

27. The omission of all of these questions would seriously damage the ability of the census to distinguish between deprived neighbourhoods and affluent neighbourhoods. This is particularly the case in Inner London and other large cities where the main household-level measures of deprivation and affluence—house size, tenure and number of vehicles—are ineffective; both rich and poor may live in small rented accommodation with low levels of car ownership.

28. We hope that the census questionnaire will be extended to four pages per individual, and that the labour market questions listed above can be restored. Under current ONS proposals a fourth individual question page

will restore the two most important variables from this list, namely industry and level of qualifications, but will still omit the others.

Consistency between constituent parts of the UK

29. We believe that it is important to make census questions as consistent as possible between the different UK statistical agencies.

30. For example in the 2001 census, a superficially minor difference in the wording of the religion question between Scotland on the one hand and England/Wales on the other led to significant differences in how people responded to the question. This has created great difficulty in comparison of the statistics for this question between Scotland and the rest of the UK.

31. We cannot assess the effectiveness of the current plans, but are hopeful that they will lead to a greater level of consistency across the UK in 2011 than in 2001.

What forms should output from the 2011 census take?

32. We are interested mainly in unrefined small-area tabular output from the census, delivered by ONS in a simple standard data transfer format.

33. The data loses value over time and the greatest benefit to both the public sector and the wider economy can be achieved if it is published as soon as possible after census date.

34. We believe that a demand-driven approach towards identifying applications and requirements for value-added data will normally be more effective and efficient than a public-sector consultation process. If data is made widely available in straightforward formats as soon as possible, then end users and distributors will be effective in creating software systems and value-added data products which meet users' requirements. This approach will be more closely focussed on end-user needs than any public consultation approach leading to a list of extended requirements to be satisfied by ONS. The latter approach could damage the end-user value of the census if it leads to delays in the publication of the data.

35. It may be worth mentioning here that simple tabular output will not limit the use of census data to a few companies who can afford the expertise and resources needed to process it. There is a reasonable chance that it will be possible to load all the 2011 small area census statistics into an Excel spreadsheet by the time they are published.

36. This type of issue did arise when the 2001 census data was published. Initial ONS plans were to bundle the data with their own software. This software effectively prevented bulk access to the data or its export into alternative computer systems, and if it had remained the only form of supply would have severely limited the benefits to be obtained from the use of the census data.

37. ONS involvement in integrating census data with other data sources should, for similar reasons, usually be limited to those things which cannot be done outside ONS. Integration of published aggregate datasets, within each of which the privacy of individuals has been protected, can usually be achieved more efficiently by end users and distributors. An extended exercise by ONS to integrate other datasets could also damage the end-user value of the census if it leads to delays in the publication of the data.

38. Further difficulties arose in 2001 from the adoption of a single publication date for all small area data. We can see no benefits in such an approach and recommend a return to the rolling publication schedules used in earlier censuses in which data became available county by county. A rolling schedule allows end users and distributors to obtain data earlier, to develop applications more quickly and to identify problems with the data sooner.

39. Further delay to the publication of the 2001 census arose because of confusion, late in the process, about disclosure control and the rules for the protection of privacy. We welcome the processes put in place already to determine the approach to disclosure control for the 2011 census.

MID-YEAR POPULATION ESTIMATES

40. We use the ONS Mid-Year Population Estimates as a control statistic for our own more detailed population estimates and projections. We believe the Mid-Year Estimates are the highest-quality statistics available at this geographical level.

41. It is difficult to assess their absolute quality without something better to compare them against, and this is usually only possible when census statistics become available. Even this comparison has historically been contentious.

42. There is an apparent weakness however in their ability to track mobile populations, and in particular the large migrant populations recently arriving from the A8 countries. It is not clear to us how this population can be tracked over short time periods given currently available data sources.

The role of survey and administrative data

43. ONS conduct an effective and wide-ranging programme of surveys. We make use of ONS facilities which allows access to anonymised microdata and hope that these will continue into the future.

44. We think there may be a great opportunities for more effective use of administrative data, including the possible use of school statistics, HMRC data, DWP data, the Electoral Roll and other sources for measurement of population statistics.

45. There are clearly challenges in overcoming legal constraints on the use of administrative data and in establishing a common approach to the protection of privacy which has public confidence. Aggregate anonymised small-area statistics from the full electoral roll, for example, cannot under current regulations be used for statistical purposes, although there are no convincing privacy reasons for this restriction.

46. We hope that the newly independent ONS may have sufficient public confidence to act as an honest broker in co-ordinating and regulating the public release of anonymised statistics based upon administrative data.

November 2007

@HR25@

Memorandum from the Greater London Authority

EXECUTIVE SUMMARY

USES AND DEFINITIONS1. ONS should make efforts to incorporate improved address list and property gazetteer data from both

Ordnance Survey and IDeA into the small area estimation methodology.

2. The main problem with ONS population estimates as currently constructed is that there is insufficient confidence in the estimates of international migration flows. This is a particular problem for London.

3. ONS projections are critical to local authorities in England as CLG uses the projections in the allocation of the local authority grant settlement. Clearly when estimates of international migration and hence population are wrong, and are compounded when carried into projections, the result can be an unfair allocation of scarce resources between areas.

4. Comparing ONS international migration statistics with registrations of overseas nationals for National Insurance shows a major difference—of the order of 100 thousand a year. Either the international migration figures understate true levels of long-term migrant inflows or the difference indicates a large flow of short-stay temporary workers. Either way, local authorities see their costs rise to meet the demands from new migrants (whether long or short-term) for their services. It is therefore imperative that local estimates of short-term migrants are made alongside the mid-year estimates, which include only long-term migrants.

5. It is essential that local estimates are made of study and work stays in the UK of at least one month duration and that these are set alongside the conventional estimates.

THE ROLE OF THE CENSUS6. The GLA would press strongly for ONS to pursue the option of a fourth page of individual

questions per person on the Census form. This would increase the likelihood of including some very important questions for London, providing vital data on qualifications, industry, carers, language, full-time and part-time working, and a wider question on health. In particular, it is vital that we know about both qualifications and industry.

7. The GLA would strongly encourage the development of disclosure control and license arrangements that allow public sector users access (under strict licence conditions) to data—including ward level data—that meet their needs.

MID-YEAR POPULATION ESTIMATES8. The high degree of errors in London borough population estimates totally destroys confidence in the

ONS data and in the analyses and outcomes based upon them. While the estimates are worse for boroughs in London than for any other region there are several urban areas of England with high migration turnover where similar problems exist.

9. The ONS methodology is basically sound—if only the data were available to support it. ONS is between a rock and a hard place—the data available on international migration flows are inadequate and there are no signs of early improvements. The methodological approaches taken to deal with international and internal migration must be more “joined up" by utilising more fully the available NHS data on new registrants from overseas or by registering international immigrants.

10. The ONS-led Interdepartmental Taskforce offered several good suggestions of how to improve international migration data—mostly relating to inflows—but so far only limited resources have been found to implement the administrative changes and research necessary.

11. Neither ONS nor users wish to see a repeat of the large discrepancies between estimates and census results. However, without further pre-2011 revisions to the methodology and data inputs, ONS will be having to alter estimates for some London boroughs by similar amounts—up to 25%—as had to be done after the 2001 Census results had been fully absorbed into the system.

THE ROLE OF SURVEY AND ADMINISTRATIVE DATA

12. If survey data on international migration are to be improved, the sample in London needs to be significantly increased. The GLA hope this will be considered as part of the move towards the Integrated Household Survey (IHS).

13. Internal migration within the UK is built on the GP re-registrations and recorded changes of patient postcodes. Methods must be found to correct the biases caused by delayed registration to enhance the value of the data to the population estimates.

COOPERATION WITH STAKEHOLDERS14. The Statistics Commission contributed strongly to an informed, professional, and empathetic

critique of ONS activities. It is important that the new Statistics Board maintains this critique, alongside other relevant bodies such as the Royal Statistical Society, the Statistics User Forum, and topic user groups.

DETAILED RESPONSEThe GLA response follows the order of the five sections specified, and treats each point made by the Sub-Committee, listed at the top of each section.

1. USES AND DEFINITIONS@UL@— What are the uses of population estimates, and how far do current population estimates meet the needs associated with those uses? What are the effects of inaccuracies or inadequacies in such estimates?@/UL@

@UL@— How appropriate is a definition of the population based on the usually resident population in the context of the needs of the users?@/UL@

@UL@— How does the quality and range of population statistics in England and Wales compare with that available in Scotland, Northern Ireland and internationally?@/UL@

General

1.1. ONS provides population estimates on a number of geographies; this submission will concentrate on local authority level estimates but, at the outset, make some observations on the estimates for wards and super output areas.

1.2. The sub-local authority estimates made by ONS are constrained to the estimates at LA level, so everything that follows pertaining to the higher level will have influence at the lower levels.

1.3. ONS small area estimates are important for local government as they are used by CLG to underpin the Indices of Deprivation.

1.4. Small area estimates suffer from lack of good datasets below LA level; therefore they are understood to be less accurate. However, the key variable in small area population change is the local housing stock—its volume, type and utilisation. Small area estimates must take on board local data that show this. Most local authorities have good council tax records as local finance depends upon it. Address lists and property gazetteers have been a major problem in the past, but with both Ordnance Survey and the Improvement and Development Agency offering products in this area—with both due to be used in the 2011 Census—data are improving and efforts should be made by ONS to incorporate this data in the small area estimation methodology.

Uses by GLA

1.5. The Greater London Authority makes significant use of ONS estimates, not necessarily because they are thought to be sufficiently accurate but because they provide a consistent set of data across all authorities in England and Wales. Therefore when comparing London, or parts of London, with the rest of the country or other parts of the country there is consistency and the GLA is using data to which everyone has access. The range of uses includes using the estimates (either for total population or parts of the population) as denominators in indicators of unemployment, poverty, deprivation or crime.

1.6. One of the main uses of the estimates is not to do with the estimates themselves but the detailed components of annual changes—births, deaths, UK migration, international migration, etc. Births and deaths are the most reliable as they are sourced from administrative information; internal migration is also based on administrative

data but there are known biases. All three components are used directly in GLA demographic models. However the estimates of international migration flows are carefully scrutinised before being used, as these are the most likely parts of the estimation process to be in error or to carry wide confidence intervals.

1.7. If authorities cannot sign-up to the accuracy of ONS estimates for their areas or parts of their areas, there will be difficulty in convincing government departments of the need to consider their own in-house alternative population data.

1.8. The main problem with the estimates as currently constructed is that there is insufficient confidence in the estimates of international migration flows. This is a particular problem for London as it receives about 40% of UK net international migration and alternate administrative data give different interpretations of where the international immigrants settle.

Uses by others with impacts on the GLA

1.9. There are many uses to which the estimates are put by government departments and agencies that have a direct impact on the GLA. The CLG uses the estimates for super output areas to underpin the Indices of Deprivation. These are used in support of funding priorities. ONS uses the estimates as bases for its own subnational population projections. Any error in the estimates is therefore introduced in the projections doubly—first at the base and second in the migration components that steer the projections. These projections are critical to local authorities in England as CLG uses them in the allocation of the local authority grant settlement. Therefore in the current CLG allocation, errors in the estimates for mid-2004 may have a significant impact on future funding for three financial years ahead, ie 2008-11.

1.10. Clearly when estimates of international migration and hence population are wrong, and are compounded when carried into projections, the result can be an unfair allocation of scarce resources between areas. Some areas will receive insufficient and others too much.

Usual Resident Definition

1.11. Until quite recently nobody was concerned that the definition of a usual resident required migrants from overseas to be intending to remain in the UK for at least a year. The 2001 Census did not make this definition sufficiently clear to the population that was being enumerated and it is likely that many people were included who had not been in the UK for 12 months or who were not intending to stay for more than 12 months but who, nevertheless, had settled temporarily in the country. It is likely that many of these people were in London. However, in the last few years, particularly since the enlargement of the EU in 2004, it is noticeable that many more migrant workers are in the UK than the official ONS population estimates indicate. Comparing ONS international migration statistics with registrations of overseas nationals for National Insurance shows a major difference—of the order of 100 thousand a year. Either the international migration figures are understating the true levels of long-term migrant inflows or the difference indicates a large flow of temporary workers intending to stay for less than a year.

1.12. Whichever the case, local authorities are seeing their costs rise to meet the demands from new migrants (whether long or short-term) for their services. While it is acknowledged that ONS has recently issued some experimental statistics relating to short-term migrants in the period mid-2003 to mid-2005, the estimates look inadequate against what is being seen in London in 2007. It is therefore imperative that local estimates of short-term migrants are made alongside the mid-year estimates, which include only long-term migrants. This will enable appropriate planning and funding of the additional services likely to be required, some of which are extremely expensive to provide. Most short-term visitors to the UK are on family visits or are tourists and are putting resources into the economy by paying for their stays of a few weeks. On the other hand there are many short-term labour migrants who are contributing to the UK economy and to the Exchequer while staying for several months but who are presenting to local and health authorities for housing, medical, legal, translation and other services. There are also short-term study visits to the UK. These students require accommodation, mainly in the private sector, and add pressure to local housing capacity.

1.13. It is essential that local estimates are made of study and work stays in the UK of at least one month duration and that these are set alongside the conventional estimates.

1.14. Another major omission from the definition of the mid-year estimates is any acknowledgement of those persons who spend part of their time at one address and part at another. Nobody can have more than one usual residence, but there are many people who live in two homes in separate parts of the country in the course of the

working week. Some may genuinely own two homes but others may live with friends or in rented accommodation near to work from Monday to Friday and return to the family home at weekends. This again puts a burden on local housing capacity and local services without necessarily putting anything back to the locality of the second residence. The travel to work patterns of these people are not correctly picked up in the 2001 Census data so transport planning accuracy is compromised. The 2011 Census may investigate the two-residence problem but meanwhile there is little administrative data that can indicate the extent of the issue. Second-home council tax discounts cover only part of the population with two residences.

Quality Comparisons

1.15. It is very difficult to comment with authority on comparisons with other parts of the UK and overseas, but a few observations are offered.

1.16. In Scotland the definitions used are broadly consistent with those in England & Wales, and therefore the same broad problems of definition and accuracy pertain. However, international migration is less of an issue in Scotland and therefore one expects greater accuracy. The short-term migrant issue is, however, one that has also reached Scotland if one takes National Insurance and Worker Registration Scheme data as guides. Scotland does have one major advantage over England & Wales in that GROS produces its estimates about three months earlier than does ONS.

1.17. Internationally there is one great advantage that several European countries have over the UK. Population registration has been carried out in many northern European countries for decades. Such systems enable a very accurate track to be made of residents as registration acts as a one-stop shop for most local and central government services. In addition some European countries allow for the population register and administrative sources such as tax and health to be linked. Hence most individuals can be accurately and quickly tracked as they move within the country and, to a certain extent, across its borders. A population register has a second advantage of virtually eliminating the need for regular full censuses and also acts as an accurate sampling frame for large social surveys.

2. THE ROLE OF THE CENSUS@UL@— How does the Census contribute to the creation of population statistics? Is the current frequency of Censuses appropriate and, if not, how frequent should Censuses be?@/UL@

@UL@— What lessons have been learnt from the Test Census carried out on 13 May 2007? What methods should be employed for the distribution and completion of Census forms, including the use of the Internet?@/UL@

@UL@— What steps should be taken to increase the rate of responses to the Census, particularly in hard-to-reach areas? How effective are plans to measure the extent of non-response to the 2011 Census?@/UL@

@UL@— To what extent is there a trade-off between the length of the Census form and its role in providing population information? What questions should be included in the 2011 Census?@/UL@

@UL@— To what extent should Census questions and statistical information be coordinated across the United Kingdom and how effective are plans to ensure such coordination?@/UL@

@UL@— What forms should the outputs of the 2011 Census take, how and when should they be made available and how should they be integrated and coordinated with other information on population?@/UL@

Creating Population Estimates

2.1. The Census is fundamental to the creation of population statistics. It is the only source for statistics for small areas. It has been used as a benchmark every decade to check the estimates. The Census provides the only reliable data we have on international immigration and the most comprehensive analysis of the structure of all migration flows within the UK.

2.2. If the frequency of censuses were increased to every fifth year it would improve the accuracy of population statistics, as they would be brought into line more often. This would avoid potentially crippling financial shocks to local authorities.

Test Census Lessons

2.3. Posting the forms out will provide resources for a better follow up in harder to count areas and is a sensible way forward, as long as the resources are then targeted to the right hard-to-count areas and in the right way. For example, some areas will need enumerators who speak particular languages. Experience of the 2001 Census fieldwork has shown that Muslim enumerators will get better response rates in some areas.

2.4. While completion and submission of forms via the Internet would appear to be a sensible option to offer, it has not yet been tested and experience in the United States Census in 2000 has shown that it is expensive to set up and was only used by a very small proportion of the population. Much better value for money, in terms of the statistics collected, might be gained from spending £22 million on a fourth page of individual questions (see paragraph 2.7 below).

Increase the response

2.5. Improved response will be facilitated by ensuring that there is enough time allowed for follow up of non-responding addresses. In 2001 there was very little, if any, follow up in some areas (especially in parts of London) because of delays in the main fieldwork and the time for the planned follow up was spent continuing with the main fieldwork. It is of concern that targets of 94 per cent coverage (ie no worse than in 2001) appear to have been set already. The target should be to include everyone. The knowledge of a reduced expectation may become a self-fulfilling prophecy. Publicity is also important leading up to the Census and there should be wide coverage in the press/radio and television. The Internet might also be useful. Advertising on Google, Facebook, eBay and, with rapid change in this area, sites unknown in 2007, might have even more effect in reaching younger people, especially young men.

2.6. A further response problem is that in 2001 many forms were returned only part completed and the additional data had to be imputed. Better follow-up procedures and form tracking are necessary to attempt to get more total completion on all forms.

Length of form and question content

2.7. The discussions are currently centred on three or four pages of individual questions. ONS has been working on the assumption of three pages per person and based its bid to the Treasury for funding on this assumption. Testing has since shown that having four pages instead of three has no noticeable effect on response rates. In the light of this we would press strongly for ONS to pursue the option of the fourth page. There are some very important statistics at stake, as the inclusion of questions on qualifications, industry, carers, language, full-time and part-time working, a wider question on health and, possibly, income are all dependent upon there being a fourth page, and they are all very important to London. In particular, it is vital to know about both qualifications and industry. There is no other source of this information for small areas and London's plans and strategies rely on knowledge of the skills and qualification levels of the population and the industrial base in London. Other questions such as year of arrival in the UK are important in gaining a better understanding of the population and likely future population change. These questions are all higher priority than a question on income.

2.8. Some new questions currently planned for the first three pages (and hence squeezing space for other topics), have been chosen specifically to get a better understanding of the size and distribution of the resident and short-term populations. Details of second residences are important and there is value in asking year and month of arrival in the UK. However a question on the intended length of stay appears to be out of place on the Census form and a question on national identity does not appear to contribute to counting the population or understanding population structure and dynamics.

UK Co-ordination

2.9. It is unrealistic to expect complete co-ordination of statistics across the UK when some questions are asked differently and others are only asked in some countries. Apart from some measures, such as population by age and sex required by Eurostat, there will be essential differences. The devolved administrations have been given the power to decide on questions relevant to them and this is a good thing. But the perceived effect of this in the GLA is that we always seem to be under pressure to accept compromises in questions that are relevant to London so that we can fit in with what is being asked in Scotland, Wales and Northern Ireland. There is no equivalent voice supporting the need for questions in England that are relevant for us. We fully support the requirements on questions on Welsh language in Wales and other things relevant to Scotland. But the unique and diverse population in London means that we have very different needs for questions. We have little interest in how many Londoners speak Welsh

and do not like to see the space on the form left blank in England (as it was in 2001 and is planned to be in 2011) when it could be used to ask a question more relevant to languages spoken in England. This is especially important given the hundreds of languages spoken in London. To do our jobs effectively we need different information and would not willingly sacrifice this just to get a UK figure. The national comparisons used are UK when possible but England and Wales is acceptable for characteristics that are not a priority in Scotland and Northern Ireland and, on occasions, just a figure for England is used for comparison, particularly when comparing for purposes of local government finance.

2011 Outputs

2.10. Outputs from the 2001 Census are likely to be many and varied. There is a need for basic sets of tables that are consistent and comparable for all local authority districts and wards in England and, sometimes, England and Wales. This is important for local government because of the competition for resources and grants in which local authorities are involved. As CLG (which is the main government department involved) is only responsible for England, this is not required by local authorities to be a UK dataset. A particular issue here is that of geography, and the decision by ONS to use the output areas designed in 2001 as the main geographical base. This means that accurate ward level data are not planned as part of the standard outputs. Local authorities need data at ward level in addition to output areas and lower layer super output areas. This is not only because local councillors represent electoral wards but also because wards are the only areas that are both subdivisions of local authority districts and are large enough to allow publication of the wide range of tabulations that are needed, for example by ethnic group and country of birth.

2.11. The GLA will strongly encourage the development of disclosure control and license arrangements that allow public sector users access to the data that meets their needs. This includes ward level data as well as a range of datasets such as the origin-destination statistics on migration and travel to work. These datasets for wards and output areas in 2001 were very badly damaged by the disclosure control procedures applied to the data. In fact the output area data was, by general opinion, not fit for purpose and was not widely used because of this. License arrangements, allowing access to data that might be considered disclosive if made publicly available, should be made so that approved public sector and other users can access the data they need under strict license conditions.

3. MID-YEAR POPULATION ESTIMATES@UL@— How accurate and useful are the mid-year population estimates for England and Wales, including information available for local authorities and Strategic Health Authorities?@/UL@

@UL@— How appropriate is the methodology by which mid-year population estimates are reached?@/UL@

@UL@— What are the most appropriate ways to ensure a common level of accuracy in mid-year population estimates for different areas and what steps should be taken to improve the accuracy in areas where population is hard to measure?@/UL@

@UL@— What is the right process for making revisions to mid-year population estimates so as to meet the needs of users?@/UL@

Accuracy and Utility

3.1. For England and Wales the accuracy of the population estimates is good. The inputs on births and deaths are highly accurate. Even the problems of UK migration based on GP re-registrations are minimised as only cross-border moves with Scotland and Northern Ireland need to be considered. The major uncertainty remains international migration. However at the national level the 95% confidence intervals in the inflows and outflows as measured by the International Passenger Survey are plus or minus 4% to 5%. This means that, at the most extreme worst case, the error in the net flow would be of the order of 33 thousand, about 14% of the flow. This translates to an annual error in the total population of less than 0.07% ie well under 1% over a decade even in the unlikely circumstance of all annual errors being accumulated rather than compensatory.

3.2. At the local authority level the problems start to become noticeable and, potentially, critical. Considering just London boroughs and comparing the ONS estimates for mid-2006 with GLA's own estimates based on vital statistics and monitored housing availability, one finds differences ranging between +8 thousand and -25 thousand, equivalent to between +4% and -11%. On the plus side, of the 32 boroughs the differences are less than 2% in nineteen boroughs, and in twelve of these it is less than 1%. However, there are three inner London boroughs in which the ONS estimates appear to be between 6% and 11% too high, and a further four where the estimates are 3%

to 4% too low. Apart from just comparing migration-based ONS estimates with development-led GLA estimates, other data can be used to get strong impressions of the inaccuracy of some ONS estimates. Demographic measures such as the General Fertility Rate (births per thousand female residents aged 15-44) can be used to show how the estimates for some boroughs become gradually more inaccurate since 2001.

3.3. Inaccuracy levels such as those described above are similar to those that were corrected by ONS after the results of the 2001 Census were fully analysed. In that instance the estimate for Westminster had been overstated by about 50 thousand residents (about 25%). The degree of errors in London boroughs described here totally destroys confidence in the ONS data and in the analyses and outcomes based upon them—such as the ONS projections used by CLG for local authority grant settlement. While the estimates are worse for boroughs in London than for any other region there are several urban areas of England with high migration turnover where similar problems exist.

Methodology

3.4. The ONS methodology is basically sound—if only the data were available to support it. Those elements that are based on vital registration are very accurate. Some of the administrative data are good, but the use of NHS GP re-registrations data needs more work to correct for biases.

3.5. The use of survey data as the major input to international migration flows and distribution still leaves much to be desired, even after the methodology overhaul that was introduced in 2007. ONS is between a rock and a hard place—the data available are inadequate and there are no signs of early improvements. In 2005 the International Passenger Survey (IPS) only contacted 3000 migrants coming into the country and 800 emigrants. The distribution to regions is now achieved using three years of data from the Labour Force Survey (LFS) amounting to about 2000 contacts. While confidence intervals for the IPS are published at regional level, so far ONS has not issued confidence intervals for its Total International Migration (TIM) flows at any geographic level. The ONS-led Interdepartmental Taskforce offered several good suggestions of how to improve international migration data—mostly relating to inflows—but so far only limited resources have been found to implement the administrative changes and research necessary.

3.6. ONS is fully aware of the inadequacies of its estimates and projections for areas of high migration, as described in the following quote from a letter sent by the ONS Director Social Reporting and Analysis to the ODPM Director of Local Government Finance in December 2005. “Although we acknowledge that there are issues with the population estimates, the published mid-year estimates remain the best estimate of the population. Until our research has concluded you may wish to consider how the estimates and projections are used and whether there is any scope for recognising the particular uncertainty for those parts of the country that are affected by relatively high levels of migration." ONS has the skills to improve methodology, but only within the constraints of the available data.

3.7. A key methodological improvement would be to “join up" international and internal migration more adequately. At present the first registration of international immigrants into the NHS is not used to ascertain residence within the mid-year estimates methodology. However, the onward flows of international immigrants within the UK are picked up by the present system as internal moves based upon changes in GP registration. This “gap" in the methodology can lead to a situation in which areas “lose" population in the estimates that had never been assigned to them and other areas retain population that had never settled in that area. The new ONS methodology of using the Labour Force Survey to distribute immigrants at regional level partly tackles this problem, but nowhere near adequately enough at local authority level within regions. This problem is basically intractable until new migrants to the UK are in some way registered on entry and that registration is carried forward to registration with the NHS.

Ensuring and Improving Accuracy

3.8. As indicated above more effort must be made across government to improve data on migration, particularly international migration, both directly and indirectly. The Taskforce offered suggestions of what should be done.

3.9. More attention should be paid to data already collected, and not analysed, on landing cards for non-EU citizens and better understanding should be made of the registrations of overseas nationals for employment (National Insurance numbers) and health (NHS Flag 4s). National Pupil Dataset and State Pensions claimants are also useful for the young and the old. None of these data offer a different figure due to definitional issues, but they do offer alternate views of trends.

3.10. While waiting for better data ONS should engage in quality assurance of the trends so far calculated between mid-2001 and mid-2006. As mentioned earlier demographic analyses can play a part. Studies of local trends in General Fertility Rates, Standardised Mortality Rates and gender ratios by age should be undertaken and allowed to feed back into amended estimates. If trends in one area are out of synchronisation with national and regional trends this should be a trigger to further investigation as to why—with the basic premise that the most likely reason is inaccuracy in the estimates.

3.11. Other data that can be used for QA are those collected by local and regional authorities, most of which relate to dwellings and homes.

Revisions Process

3.12. Having reached mid-2006 estimates, with a base of mid-2001, there is scope to make a wholesale revision of estimates back to mid-2001 following extensive QA as outlined above. While users want to have the best estimates available, under whatever definition of population is adopted, it is not desirable to have many sets of revisions. It is recognised that estimates for 2002 to 2005 were revised in 2007 and ONS has said that it would not offer any more revisions for at least two years.

3.13. The most appropriate course of action is to develop a QA strategy for estimates using as much administrative data and as many demographic measures as possible and to implement this with a view to revising all existing estimates back to a fixed point at mid-2001. This is unlikely to be able to be implemented before 2009 by which time ONS has indicated a review of internal migration estimates. After a wholesale review the QA strategy should be employed annually as part of the estimates process prior to release.

3.14. ONS has said that it does not wish to see the large discrepancies at mid-2010 of estimates made without and with the 2011 Census results such as were seen after the 2001 Census was published. Nor do the users. Without further pre-2011 revisions to the methodology and data inputs ONS will be having to alter estimates for some London boroughs by similar amounts as had to be done after the 2001 Census results had been fully absorbed into the system.

4. THE ROLE OF SURVEY AND ADMINISTRATIVE DATA@UL@— What role can and should be played by survey data in the compilation of population statistics?@/UL@

@UL@— What role can and should be played by data from General Practitioners and other health service data in the compilation of population statistics?@/UL@

@UL@— What other data sources can and should be used in the compilation of population statistics?@/UL@

Role of Survey data

4.1. Unfortunately it is survey data that causes the main problems in the current estimates.

4.2. Currently, the Annual Population Survey/Labour Force Survey (APS/LFS), which will become the cornerstone of the Integrated Household Survey (IHS), offers a range of data on migration, country of birth, year of arrival and nationality. Indeed, ONS currently use estimates from the Labour Force Survey, as one of the inputs to the international migration methodology.

4.3. However, the LFS/APS is limited in its application due to problems in relation to coverage of the population. Many residents living in communal establishments are not covered by the sample, and the coverage of short-term or recently arrived migrants is not comprehensive and thought to be poor. Grossing procedures muddy the water further, as while the APS may well pick up short-term migrants during fieldwork, and go on to include some of them in the survey sample, the population weighting methodology does not include them.

4.4. Coverage issues are made generally worse in London as response rates on the APS are lower in London than outside London,30 increasing the likelihood of bias and poor coverage of certain population groups.

30 For example, the response rate for wave 1 of the quarterly Labour Force Survey (a key component of the APS) during April-June 2006 was 61.6% in London and 72.3% in the rest of England.

Further, sample size on the LFS/APS is lower for London boroughs relative to other local authorities, making local area estimates drawn from the survey less reliable. If survey data on international migration are to be improved, the sample in London needs to be significantly increased. The GLA hopes this will be considered as part of the move towards the full implementation of the IHS in 2008-09.

4.5. The quality of the final survey estimates is, of course, ultimately dependent on the accuracy of the population data used for grossing the survey sample, which themselves are highly problematic, particularly in London (as explained in Section 3).

4.6. While survey data will always have limitations, there is undoubtedly scope to improve the potential applications of the Integrated Household Survey by tackling these issues. Further, questions to international migrants about their reasons for migration and immigration status could usefully be included in the IHS. This would improve the ability to triangulate and verify data from administrative and survey sources. Survey data are of use only where there is a lack of administrative data to feed into the estimates. The major improvement would come with setting up a Population Register and having better border control of international migration.

GP/Health Data

4.7. Health service data are of two broad types—stocks of registered patients and flows of persons onto the register and within the register. The stocks data are notoriously error-prone due to problems in getting persons who change their address within the UK to re-register promptly. In 2001 the ONS quality assurance process for the Census populations used the health register of the area but the biases that remain in the register made comparisons very difficult for working age adults.

4.8. Delay in re-registration is a particular problem with students and especially young single adult males. The excessive delays in recording a change of address lead to biases in the age/gender structure of moves and, if the delay straddled the 2001 Census, can lead to the ONS estimates “moving" people from an area that they had not been attached to by the Census enumeration, eg a person registered with a GP in Wales but actually living unregistered in London at the time of the Census would effectively be counted as a London resident in the mid-2001 estimates. When moving and re-registering with a GP in Surrey the next mid-year estimates would move a person from Wales to Surrey but not change London. There would therefore be a “loss" to Wales of a person never actually resident and London would “retain" a person who had moved on. This process is, of course, totally anonymised but the statistical principle pertains.

4.9. Another problem with the data is that people moving overseas are not required to de-register. Hence they remain on the register and inflate it. Pruning the registers is done by health authorities but there will be delays in making adequate corrections.

4.10. Internal migration within the UK is built on the GP re-registrations and recorded changes of patient postcodes. Methods must be found to correct the biases caused by delayed registration to enhance the value of the data to the population estimates.

4.11. One, as yet underutilised, aspect of GP registration that is proving useful is the recording of new registrations of people from outside the UK—referred to as Flag 4s. The total numbers of Flag 4s accord well with international inflows but the geographic distribution around the country is quite different to that used by ONS in the population estimates in many areas, particularly as between city centre boroughs and surrounding boroughs. More research on this data source needs to be carried out to incorporate it within the methodology. When new NHS computer systems are implemented it is to be hoped that complete recording of these “international migrants" will be available. At present the “Flag 4" is lost at a subsequent recorded move and the total numbers of new overseas registrations cannot be determined.

Other data

4.12. Very few administrative data sources actually measure total population or changes in the population, but several are good comparators of change in a locality. Local authorities have good records of council tax properties (including certain estimates of vacancy and second residences), new housing developments, houses in multiple occupation and children in schools. These sources can be compared as time series with ONS estimates to determine potential problems of consistency of estimates of change since 2001. While the electoral register does not cover all adults of overseas origin it may also be used as a comparator to estimated changes in the adult population.

5. COOPERATION WITH STAKEHOLDERS@UL@— How effectively has the Office for National Statistics cooperated with stakeholders with an interest in and information relating to population statistics and how can cooperation be improved by the Statistics Board and by its Executive Office?@/UL@

Cooperation with Stakeholders

5.1. For many years, ONS has had a system of maintaining contact with stakeholders, which—on paper—appeared reasonable and adequate. Those at the receiving end have occasionally found the appearance deceptive. GLA staff have played significant roles in Census, population and labour market groups, often representing broader interests such as local government users, or public sector users.

5.2. Among many problems with this complex structure of advisory groups, a few stand out:

@UL@— The ONS approach can feel like “divide and rule", as it liaises separately with various user sectors—eg academics, local and central government, business, health, etc.@/UL@

@UL@— To the new user, ONS may appear to be “the voice of central government", but it has to deal with the Government Statistical Service, the devolved administrations, and other government departments, not least the Treasury. Thus a frequent response has always been “we can't afford it". As an example, demands for a larger sample, or other improvements to the International Passenger Survey, have been around for decades. In the same vein, ONS recognises the difficulties of, effectively, four separate UK censuses but does not have the power to impose a solution.@/UL@

@UL@— ONS rarely, if ever, uses the data it collects and produces, and therefore does not comprehend or appreciate the genuine concerns raised by users (examples abound in the area of census dissemination).@/UL@

@UL@— Somewhat late in the day, ONS has come to appreciate the importance of population counts and estimates, and their significance to the funding and targeting of services. The political and public concern over the rise in international migration over the last decade, and particularly since 2004, has brought about this change. Government statisticians are now suffering—often unfairly—because the political heat has been turned up under their figures. However, the reasons for this are understandable, and reflect in part the frustration at an apparent earlier lack of response.@/UL@

5.3. The Statistics Commission, particularly under Professor David Rhind, contributed strongly in its relatively brief life, to an informed, professional, and empathetic critique of ONS activities. Significant reports include:

@UL@— A Code of Practise for National Statistics.@/UL@

@UL@— The 2001 Census in Westminster.@/UL@

@UL@— Counting on Success: The 2011 Census—Managing the Risks.@/UL@

@UL@— The Use Made of Official Statistics.@/UL@

5.4. It is important to maintain this critique, alongside other relevant bodies such as the Royal Statistical Society, the Statistics User Forum, and topic user groups. The Statistics Board needs to ensure a coherent overall approach that uses its professional integrity and influence to harness sectoral/sectional interests to maximum effect.

November 2007

Memorandum from Newcastle upon Tyne City Council

1. Newcastle City Council appreciates the opportunity to comment on the quality of the ONS projections. We do this in the context of verifying the estimates and testing to see if the projections seem reasonable.

2. SUMMARYThis submission comments on the 2003 and more recent population projections. This submission notes that:

@UL@(a) whilst the population estimation work is difficult, the population projection work is considerably more difficult, leading to many problems;@/UL@

@UL@(b) this suggests that the population projections may not be reliable enough, on their own, to guide the distribution of resources to local authorities;@/UL@

@UL@(c) in particular, charts 3 and 4 present evidence of internal inconsistencies in the projections that give rise to concerns about their overall robustness, and@/UL@

@UL@(d) ONS should undertake and publish quality assessments of its estimates and projections along the lines of this submission.@/UL@

This submission is in addition to that provided by the five Tyne & Wear Metropolitan Districts together with Tyne & Wear Research & Information Unit (TWRI), submitted by TWRI.

3. INTRODUCTIONTo get estimate work correct, in the current climate of high and variable migration, is difficult.

To get projections correct is an even more difficult task. This is primarily because we know more about the past than we do about the future; ie we have to make certain assumptions about the future. However, as the projections are currently used to determine Central Government allocations to authorities they are now very important.

Those who produce and those who use projections may not always be aware of the underlying assumptions that are being made, so testing to see if the projections are plausible is imperative. For example, the inclusion of the A8 countries into the EU will have had an impact on the estimates and will have a

4. MIGRATION COMPONENTSIn order to simplify this submission, we have deliberately ignored commenting on the migration

components, apart from the following general comments. These considerably complicate the underlying issues. By ignoring them we are not saying that they are unimportant, (we appreciate that they are far from unimportant), but in order to make progress we need to deal with each issue separately. Noting, however, that when in-migrants and the local population are combined there are still further assumptions that are made (eg we may need to assume that new in-migrants into an area have similar characteristics [eg death rates, propensity to have offspring or to leave an area] as those who already live there, or are so small as to make no difference).

However, because the outcome of this analysis includes the impact of all of these we are unable to disentangle the impact of migration within the change in the gender-ratios.

5. EXAMPLES OF DIFFICULTIES THAT ARISE WHEN BASE DATA IS WRONGAs is well known there were problems in a number of areas collecting the 2001 Census data and specific

adjustments were made to the data to compensate for this, generally, undercount.

Not all areas were increased and so there will be some areas where the overall figures are more accurate than others. In what follows we give two simplified examples as to the problems caused by incorrect base data.

5.1 Data requirements to project forward the number of deaths in an area

To all intents and purposes we know how many deaths, in any area, there have been in any past year, and so can use these in the estimates.

However, to project the number of deaths for just one year ahead, in an area, we need to:

@UL@(a) for the past, know how many men/women there were in each age group;@/UL@

@UL@(b) how many deaths to each of these people;@/UL@

@UL@(c) calculate the age-specific male/female death rates;@/UL@

@UL@(d) decide how many years data we wish to include, [the more years the more robust but the less the data reflects the most recent situation];@/UL@

@UL@(e) use the national death rates to project what the local death rates will be;@/UL@

@UL@(f) work out how many deaths there will be to men/women in each age group; and@/UL@

@UL@(g) add the results together to give the projected number of deaths.@/UL@

This shows that projecting the number of deaths has some difficulty; in particular if there is any wrong figure this can lead to significant impacts further down the line.

If, for example, in an area we have too high a figure for the number of men in the “75 and over" age group (in (a) above), we know how many deaths there were to men in this age group (b), so our age-specific death rate (c) is too low. If the death rate (e) is expected to decrease, then the age-specific death rate in the area will decrease to a lower rate than it should have done and the number of deaths, in any particular period, will be projected (f) to be less than it otherwise would have been.

Thus the number of older people will be projected to be more than it should have been, compounding the issue, for the second and subsequent years, still further.

5.2 Data requirement to project forward the number of births in an area

Again, to all intents and purposes we know how many births in any area there have been in any area in a previous year, and so can use these in the estimates.

However, to project the number of births, for the year ahead, in an area, we need to:

@UL@(a) for the past, know how many women there were in the relevant age groups;@/UL@

@UL@(b) how many births to these women;@/UL@

@UL@(c) calculate the age-specific birth rates;@/UL@

@UL@(d) decide how many years data we wish to include, [again the more years the more robust but the less the data reflects the current situation];@/UL@

@UL@(e) know how many women will be in these age groups in the next year;@/UL@

@UL@(f) use the national birth rates to project what the local birth rates will be;@/UL@

@UL@(g) work out how many births there will be to women in each age group; and@/UL@

@UL@(h) add the results together to give the projected number of births.@/UL@

This needs to be done for both boy and girl babies separately, because the latter will be women who can give birth at a later time period.

This shows that projecting the number of births in the future has inherent difficulties; in particular, if there is any wrong figure this can lead to significant impacts further down the line.

If, for example, in an area we have too low a figure for the number of women in a particular age group (in (a) above), we know how many births there were (b), so our age-specific birth rate (c) is too high. If the birth rate (f)

is expected to increase, then the age-specific birth rate, in the area, will increase more than it should have done and the number of births will be projected ((g) & (h)) to be more than they otherwise would have been.

The high variability that these errors can introduce to estimates and projections makes it imperative that ONS can assure us of the integrity and robustness of their projections at local authority level.

6. How do we know if the projections are plausible?

In what follows we present a simple method to determine whether or not a projection for a particular area is plausible. This method will not pick up all incorrect data, but it will certainly pick up some unusual features of the material and/or projection methods.

6.1 Testing the plausibility of the projections

One means of considering the accuracy of population projection is to use the gender-ratio. To do this we calculate the percentage of the population of a particular age-group who are male, at any projection time, chart it and see if there are any oddities. In our exercises we have looked at 15 year age-bandings (ie 0-14 year olds, 15-29 year olds, 30-44 year olds, 45-59 year olds, 60-74 year olds and 75+ year olds, as well as “all ages"). In particular, we look to see if there are any shapes which appear odd.

If a shape is unusual, this may suggest that something is amiss. It might be the basic data or it might be the assumptions. It is possible that the method used is actually producing a higher population for one or more authorities, and a lower population for other authorities.

This exercise can be undertaken at any spatial level (eg Regions, Counties, London boroughs, Unitary authorities, County and Metropolitan districts). From this we can determine any unusual regions/authorities data which need examining. Once the basic idea is set up it is not an onerous job to repeat the exercise on any projection (or estimate) data set.

In the past the projections have not been used, directly, for local authority financial allocations. Now that they are used in this way, it is important that we take a critical look at the figures. Note that in proportionate terms, absolute errors impinge on smaller authorities, financially, more than they do larger authorities.

7. EXAMPLESWe have undertaken this exercise for some of the 2003, the original and revised 2004 projections, for the

English regions, Metropolitan districts and London boroughs, giving rise to a number of apparent anomalies.

In what follows we give two examples from the Metropolitan districts projection data for the base year 2003 and the original and revised 2004 projections.

7.1 Example 1 Metropolitan districts (for age band under 15's)

We did this for the 2003 projections, noting that the Newcastle gender-ratio, for the “under 15's", was very high [rising from 51.1% to 58.1% in the projection period]. (See table 1 and chart 1, below).

As a consequence, we wrote to ONS (December 2004) regarding this, asking for an explanation, receiving a reply (January 2005), which although unsatisfactory, left the feeling that there was little point in making any further representation.

Table 1, below, shows that, for example, in the 2003 projections, the “boy percentage", for Newcastle, was projected to rise from 51.9% (in 2003) to 58.1% (in 2028); almost 3 boys for every 2 girls. Comparing to the number of girls, we would need in 2028, an extra 4,800 boys only coming as in-migrants, almost 30% above the number of boys that would have been expected to have be alive; this is unrealistic.

The associated figures, in brackets, show that Newcastle started as first highest and was still first highest at the end of the projection period. However, this underestimates the true scale of this anomaly, as is shown in the attached chart (Chart 1), where the second highest has a rate, for 2028, just under 54%.

For the original 2004 projections, the percentage boys starts at 51.7% (Newcastle has ranked 3rd out of 36 Metropolitan districts.) By the end of the projection period 25 years later (2029), Newcastle had moved to 30th. A

similar picture holds for the revised data. This shows that the introduction of an extra years data has had a significant impact on the figures, leading to questions of their reliability.

Table 1

NEWCASTLE 2003, 2004 ORIGINAL AND 2004 PROJECTIONS

Base year2003

2004 Age band Year Original Revised

0-15 Base 51.91 (1) 51.70 (3) 52.05 (1)Base+25 58.11 (1) 50.75 (30) 50.85 (30)

75+ Base 35.96 (28) 36.76 (23) 36.76 (22)Base+25 46.30 (1) 43.75 (13) 44.58 (4)

All Base 48.72 (16) 48.76 (18) 48.63 (23)Base+25 52.73 (1) 48.97 (25) 48.56 (28)

[Notes: The table shows for three projections (that using the 2003 figures, the original 2004 figures and revised 2004 figures as bases), for age bands “0-15", “75+" and “all ages", for the base year and the base year+25 years, the percentage of the relevant population who are male, for Newcastle upon Tyne, and, in parentheses, the rank of Newcastle within the 36 Metropolitan districts.]

The table also shows that this is also true for those aged 75 & over, where not only is there an inconsistency between the 2003 projections and the 2004 projections, but that there is an inconsistency between the two 2004 projections.

Finally, in this section, the table shows that for “all ages" combined, there is a significant difference between the 2003 and the 2004 projections. Chart 2 shows the results, for the 2003 projections, showing differences between authorities.

7.3 Example 2 Metropolitan districts 2004 projections (original and revised)

Charts 3 & 4 below show the position for the 2004 results for Newcastle and other Metropolitan authorities. Comparing these two charts suggests that there are two outliers showing unusual demographic patterns. We would expect such patterns to cause ONS to critically assess its data.

8. CONCLUSIONSWe question these projections on the grounds that there are unusual and unexplained inconsistencies in the

data. ONS must undertake quality assessments of both its estimates and projections at local authority level to demonstrate that the projections are robust and able to support resource allocation at this level.

/home/PKB/DATA/386539/FOLIO104.EPS>

/MILES/PKB/DATA/386539/new105.eps,s(35p6)> /home/PKB/DATA/386539/FOLIO105B.EPS> /home/PKB/DATA/386539/FOLIO106B.EPS>November 2007

@HR25@

Memorandum from the South West Observatory

1. USES AND DEFINITIONSThe Population estimates provide the foundation for intelligence both at national and sub-national levels. It

is recognised that, generally, the population data available from National Statistics are of a good standard and is readily available.

Persistent gaps continue within the sub-national data, with over reliance on Census information at the lower geographies. The South West Observatory continually receives requests for more up to date information at SOA and Ward levels. The experimental population estimates at SOA level have been well received and we hope these publications will be given permanent status in time.

Improvements are needed in the temporal and geographic coverage of “Ethnicity" and “Country of birth" population statistics to provide essential evidence to support community cohesion efforts, public service planning and investment. Specifically:

@UL@(1) Updated more frequently (through surveys and sampling)@/UL@

@UL@(2) Available at local (sub district level) geography (eg LSOA, MSOA, LA Ward)@/UL@

Why?

The need for improvements in this area of population data supply was brought up as an issue during operation of the SELD pilot (www.seldsw.org) here in the South West region (2005-06). Not only was this issue raised during several workshops run with practitioners across the region, but also in various discussions such as those which followed the publication of Anita Pilgrim's research review commissioned by the Black South West Network (www.bswn.org.uk) in 2006:

@UL@”Review of Research Data on the Black and Minority Ethnic Population in the South West Region of England" @/UL@

@UL@(http://www.bswn.org.uk/BSWN%20Resource%20Library/researchsouthwest/SW_Research_Review/folder.2006-05-17.3695375486/SW_BME_Research_Review_Final_Report.doc) @/UL@

@UL@PowerPoint Summary@/UL@

@UL@http://www.southwestforum.org.uk/docs/POWERPOINTS_27_Oct_06/Needing%20Numbers%20presentation.pdf@/UL@

The South West Observatory is not able to comment on population statistic in comparison to other countries.

2. THE ROLE OF THE CENSUSThere is an increasing call for the Census to be run every 5 years or a “mini" census around population to

assist in the modelling process. This would avoid the retrospective fitting of data to trend analysis.

Processes AND resources should be in place to collate and produce an output from the Census as soon as possible. Any delays create frustration from analysts requiring the information for policy development and funding bids etc.

3. MID-YEAR POPULATION ESTIMATESIt would be useful to extend the format of the mid year estimates by single year as well groups. There can be

some confusion over the content of particular population estimates releases and consistency is the key to providing long term trend information.

4. THE ROLE OF SURVEY AND ADMINISTRATIVE DATAMany Local Authorities utilise the information on population derived from General Practitioners data to give

a more realistic view of population. This information is available at postcode level. In terms of accuracy, this is deemed better than the mid-year estimates produced by ONS in many cases.

Other sources of data that could be used to inform the estimate process can be derived from un/employment statistics, council tax records along with the health data. A more comprehensive and inclusive approach could be developed that may achieve a better estimate of the population.

November 2007

@HR25@

Memorandum from CLIP Population Sub-group, LA-side, Nottinghamshire County Council

1. SUMMARYUses and definitions

1.1 It is important to establish a consistent source for population estimates data. (3.4)

1.2 Factors which are marginal in National terms for total population are significant when applied in a particular context; different definitions, geographies and categories. The importance and the difficulties in capturing these important “marginal uncertainties" should not be underestimated. (3.1)

1.3 We have a concern over the fitness for use of (various) population counts (including population estimates and projections) by Government departments, in particular, but not solely, CLG, and in local government. These include resource planning and allocation, performance measurement and gathering of evidence for policy making. (3.5)

1.4 Definitions used by ONS need to adapt to the changing social and economic situation. We recognise the value in having a single standard definition which can be consistently applied across the country, giving an unambiguous result. Ideally, though, the population estimates system should also provide flexibility to the various users. (3.14)

1.5 Increased concern over perceived inaccuracies in population estimation has led to considerable work to address the concerns. This has deflected effort from other developments in ONS's programme of modernisation and systems development. (3.3)

1.6 There is ambivalence amongst local authorities on the value of a population register. There is support for a more integrated, administrative data based system as proposed with the ONS Proposals for an Integrated Population Statistics System. This should deliver more robust statistics, but could not guarantee widespread acceptance. We also note that there is increasing public suspicion about collection of data on the individual, so a register system would at least have the advantage of being, in principal clear, transparent and in the open. (3.24)

The role of the Census1.7 The importance of the Census in the creation of population statistics cannot be over-estimated; it is crucial. It is the only way that a complete understanding of the linkage between individuals, households, houses and population at a local level can be achieved. (4.2)

1.8 Despite this the currency and accuracy of “simple counts" between censuses becomes increasingly important with the current dynamics of population change. (4.2)

Mid-year population estimates

1.9 In assessing the quality of population statistics, as with any dataset the effects of errors need to be estimated. This informs the resources devoted to its compilation, the cost of securing improvements, and the uses to which it is put. (5.2)

1.10 The revision of statistics creates difficulties for local authorities because of the discontinuity and transition between significantly different totals. This has in part given rise to the concern mentioned in 1.4 above. (5.3)

The role of survey and administrative data

1.11 We support the work of ONS to develop an Integrated Population Statistics System, which would utilise many of the benefits of administrative data. (6.1)

1.12 There is an importance to “marginal" factors, and “rare populations", in particular the collection of data on the scale and movement of migrants, especially international migrants, is the most critical element to be improved. (6.1)

1.13 There are dangers in relying on administrative sources to compile long-term statistical datasets. Statistical output is a by-product of these processes and is subject changes in procedures or definitions, driven by other needs. (6.2)

Co-operation with stakeholders

1.14 ONS resources need to be taken seriously if the concerns are to be addressed; it is false economy to only await concern over statistics. We have, on several occasions expressed concern in CLIP about work streams having to stop or be suspended. (7.2)

1.15 ONS needs cross departmental support in its efforts to improve population statistics. It is an area where co-operation in and thinking, planning and working will provide huge benefits. We gave our support to this approach presented to us by ONSCD at the last CLIP Population sub-group meeting. (7.3)

1.16 It should not be overlooked that ONS has a primary role in serving the interests of government in its widest sense, which will raise tensions between the interests of individual bodies and the need to consistently deliver satisfactory national statistics. (7.4)

2. INTRODUCTION2.1 CLIP is the Central and Local Information Partnership. It was established in 1998 to enable central

and local government to work together to develop efficient and effective statistical information.

2.2 Much of the work of CLIP is undertaken by subject-specific sub-groups. The aim of the CLIP Population Sub-group is to improve communication between the Office for National Statistics (ONS) and some users31 of population statistics produced by ONS. The sub-groups determine their own membership from local authorities, central government, government offices and agencies. For CLIP Population this includes the Ministry of Defence and NHS, the GLA and London Councils group, and statistical bodies for Scotland, Wales and Northern Ireland. Each sub-group has a central government and local authority lead. This submission is from the local authority side of the population sub-group.

2.3 The Sub-group provides a forum where ONS may be made aware of the changing demands and implications of population statistics, and where users are invited to comment upon all proposals for significant change in content, quality, and timing of ONS population statistics, including proposals for changes in data sources. The group is also concerned with the general methodology and procedures used by ONS when producing national and local population statistics and projections; and advises ONS about dissemination of their population statistics.

2.4 The Local Authority side of CLIP Population is aimed at providing impartial, non-partisan advice, comments and suggestions to ONS on a range of issues relating to Population Estimates and Projections.

2.5 The group has had sight of submissions from its members, and the LGA in particular. Therefore this submission builds on rather than repeats the submission of the LGA and concentrates on aspects that the group consider are particularly relevant to their work with population data, its use, and the relationship with ONS.

3. USES AND DEFINITIONSWhat are the uses of population estimates, and how far do current population estimates meet the needs associated with those uses? What are the effects of inaccuracies or inadequacies in such estimates?

3.1 It is notable that small uncertainties which are marginal in National terms for total population are significant when applied in the context of different definitions (eg age, nationality), geographies and categories. The importance and the difficulties in capturing these “marginal uncertainties" should not be underestimated.

3.2 Most particularly in recent years there is the impact of the most uncertain and volatile aspect of population estimation, migration. This is even more significant owing to public perception due to the visibility of “non-british" residents, and the impact of migrants demand on services being identifiable (eg owing to language).

3.3 This has led (rightly) to an increased concern for some authorities of perceived inaccuracies in population estimation. Considerable work has been done by authorities individually, collectively, and by ONS, to

31 CLIP does not cover the private sector, for which ONS has separate liaison groups.

address the concerns. It should be noted that, in our view, this has deflected effort from other developments in ONS's programme of modernisation and systems development.

3.4 Of significance to us is the means of establishing a consistent source for population estimates data. Much of the information put forward in support of better population estimates cannot, for various reasons, be used consistently across the country (England, nor UK). We are satisfied that, within the resource limitations they have, ONS are pursuing the need for better data, it's proper analysis and use, with the right priorities. They have consulted their users on their programmes in this regard.

3.5 A major concern over population estimates is their use (along with population projections) in resource planning and allocation for Government departments, in particular, but not solely, CLG. Bearing in mind also the time delay in preparing population estimates, and still more, population projections, our view is that a more responsive solution for the concerns that LAs have in obtaining suitable resources could be to use other sources of data, particularly service-based data, for short-term resource planning. This would leave the estimates and projections to provide an “underlying, long-term" basis for resource decisions; these are typically reviewed for periods of 2-5+ years (1-2 years following), with projections having terms of 25 years and more.

3.6 This is a matter of the fitness for use of (various) population counts.

3.7 Notwithstanding the need for accurate population information, CLIP Population sub group has discussed areas where the use of population data is inappropriate in the light of its genesis, assumptions and robustness. Examples that we have identified include the following.

3.8 NHS resource and CLG Financial Settlement decisions: These use trend-based population projections that do not reflect a “likely" scenario, and there is a tendency for CLG to use these in a prescriptive fashion. Their use in this way can raise the kinds of difficulties in financing changing service delivery highlighted recently by some authorities. It is noted that this is in contrast to preparation of Regional Spatial Strategies where the inbuilt and inherent sources of uncertainty in estimates and projections are taken account of as one factor to be weighed in decision making.

3.9 Transport planning: A model used by the Department for Transport (DfT), called TEMPRO (for Trip End Model Presentation Program) has, in the past used as an input the trend-based ONS projection of population to derive households (an input to the model). This is increasingly being rectified by the use of “policy-based" projections, obtained through Regional Assemblies and (theoretically) aligned with Regional Spatial Strategies. There are still issues over currency, spatial disaggregation and the use of the modelling that are been addressed.

3.10 All types of indicators increasingly use rates to illustrate year-on-year (or shorter period) change. In some cases these relate data from current administrative sources to out-dated, and “not fit for purpose" data. This is not to say the data are inaccurate, or the definitions unclear, but they are unsuitable for the purpose they are put to. (eg BV 17 uses as denominator a population that includes people not in work, when the numerator relates only to people in work.)

3.11 Performance measurement increasingly looks for denominators based on current estimates of the population. Local authority Mid Year Estimates are prepared a year in arrears, although National Statistics now prepare quarterly national and regional estimates as “Experimental Statistics". The group recognises the difficulties in speeding the process up without further prejudicing quality and reliability or increasing cost. Those drawing up performance indicators need better awareness of the constraints and limitations of population data.

How appropriate is a definition of the population based on the usually resident population in the context of the needs of the users?

3.12 The usually resident population definition is one amongst many population definitions of potential use to local authorities and the more complex living arrangements of modern life. For example spending three days a week at one residence and four with a family at another demonstrates that one simple view of usual residence is out dated and difficult for the public to interpret.

3.13 Definitions used by ONS need to adapt to the changing situation. Increasingly people are subject to non-traditional definitions, for example students with a term time address, people having multiple residences, sometimes abroad and the much discussed topic of migrants and in particular short term international migrants. The changing frequency and nature of travel to work also has an impact upon definitions.

3.14 Having said this we recognise the value in having a single standard definition which can be consistently applied across the country. If for no other reason, it helps to ensure complete coverage and eliminate double counting of people. Ideally, though, the population estimates system would provide enough flexibility to allow authorities to calculate populations against a range of definitions to better inform service provision, performance indicators and policy development.

3.15 ONS preparatory work for the 2011 Census has identified many ways of defining a person (in excess of 40, inc. 23 for UK nationals), most of which are valid for some purpose. In practice LAs use different definitions; a population count could support these by providing different options.

3.16 The range of alternatives would ideally include population present, day/night populations, term-time and holiday populations, weekday and weekend populations. This of course raises its own issue of the appropriate or incorrect use of any particular definition for any particular use. However, as mentioned above (1.10), the adherence to only one definition does not prevent the misuse of the data.

3.17 We have three more specific points to make on definitions:

3.18 1) The local authority grant allocation process makes some use of the concept of “additional population". This uses out-dated Census data on commuting and 1991 survey data on visitors.

3.19 2) Apart from other concerns about the quality of international migration data there is an inconsistency between the resident population, the basis of the Mid-Year Estimates, and the definition of an international migrant (LGA paper refers).

3.20 3) Much current concern about migration arises not only from effects on net changes in resident population. Net change conceals much greater gross in- and outflows. Population churn may generate demands in itself, particularly when composed of people with special characteristics: eg different languages. Indeed many migrants may be “short-term" and so not reflected in the population estimates.

How does the quality and range of population statistics in England and Wales compare with that available in Scotland, Northern Ireland and internationally?

3.21 We would note that population-related statistics in England and Wales compare very favourably with those in the other home countries, with, as far as we are aware, two exceptions. These are, in Scotland census and other data are not subject to disclosure provisions, and that there are housing-based data sources, derived from administrative data, that prove very valuable to verify population data on a consistent basis.

3.22 While internationally there are official population statistics that are considered more robust and accurate, and command high levels of public trust, it is noteworthy that these are almost exclusively based upon a population register

3.23 Thus, as the LGA submission states, it is sensible to monitor how these things are done elsewhere. ONS' intention to develop a business plan for an Integrated Population Statistics system, which would presumably examine approaches taken elsewhere, is therefore welcome.

3.24 We note that there is ambivalence amongst local authorities on a register, but support for a more integrated, administrative data based system as proposed with the ONS Proposals for an Integrated Population Statistics System. This should deliver more robust statistics, but could not guarantee widespread acceptance. We also note that there is increasing public suspicion about collection of data on the individual, so a register system would at least have the advantage of being, in principal clear, transparent and in the open. It is important that a separation is clear between the administrative purposes of the register, and ONS's survey role.

4. THE ROLE OF THE CENSUSHow does the Census contribute to the creation of population statistics? Is the current frequency of Censuses appropriate and, if not, how frequent should Censuses be?

What lessons have been learnt from the Test Census carried out on 13 May 2007? What methods should be employed for the distribution and completion of Census forms, including the use of the Internet?

What steps should be taken to increase the rate of responses to the Census, particularly in hard-to-reach areas? How effective are plans to measure the extent of non-response to the 2011 Census?

To what extent is there a trade-off between the length of the Census form and its role in providing population information? What questions should be included in the 2011 Census?

 (Other questions deleted)

4.1 We wish only to submit three brief, specific, comments on these questions.

4.2 The importance of the Census in the creation of population statistics cannot be over-estimated; it is crucial. The only feasible alternative to obtaining even superficial information on the population (ie counts, age/sex, personal, not relationship information) at district level or below is a comprehensive, administrative register, and this would only provide part of the census statistics. Having said this, in terms of populations statistics, ie “simple counts", the currency and accuracy of information between censuses becomes increasingly important, such that currently, within 3 years of the release of census information, the dynamics of population change are problematic.

4.3 We would like to draw attention to two unique characteristics of the Census. First, and most importantly it is the only way that a complete understanding of the linkage between individuals, households, houses and population at a local level can be achieved. With a significant proportion of the population having several residences and vacant properties concentrated in particular areas it is important that double counting of the population does not take place. With absent households, short term migrants and persons abroad we are dealing with a complex situation which requires either a census or a register which can identify if double counting is taking place. Secondly, it is a key method of collecting local migration and travel to work data both of which have population definition implications.

4.4 We note that “hard-to-reach" groups often correlate to the “hard to deliver services" areas and rightly demand more attention, which is an approach of ONS census planning. However, evidence that a non-geographic, systematic deficiency in the census enumeration, applying to all areas (for example the lack of response for certain aged males), would require approaches that address being “hard-to-reach" across the whole Census geography. In other words, “hard-to-reach" areas are areas where certain “hard-to-reach" groups are concentrated.

4.5 We would support the view that there is significant value of having 4 Census pages per person rather than 3. ONS have judged that response rates will not be affected, and the only deterrent is financial cost. We understand that this is the main difference between the 3 and 4 page versions. Less information would be gathered, in particular on educational qualifications and on industry/business of employer. These are both very important in getting an understanding of disadvantage.

5. MID-YEAR POPULATION ESTIMATESHow accurate and useful are the mid-year population estimates for England and Wales, including information available for local authorities and Strategic Health Authorities?

How appropriate is the methodology by which mid-year population estimates are reached?

What are the most appropriate ways to ensure a common level of accuracy in mid-year population estimates for different areas and what steps should be taken to improve the accuracy in areas where population is hard to measure?

What is the right process for making revisions to mid-year population estimates so as to meet the needs of users?

5.1 Our comments on these questions are, in the main, adequately covered in the LGA submission and earlier paragraphs of this submission.

5.2 In assessing the quality of population statistics, as with any dataset the effects of errors need to be estimated. This informs the resources devoted to its compilation, the cost of securing improvements, and the uses to which it is put. For example, a small error in the population estimate would have a relatively large effect on a budget allocation where each person attracts £1,000 of funding. Conversely, where population is a denominator the effects on, say, an unemployment rate may be less important than errors in counting the numerator (number unemployed).

5.3 With regard to the last question, the revision of statistics creates difficulties for local authorities because of the discontinuity and transition between significantly different totals. Most noticeably in recent times, this occurred following the 2001 Census, although this was not the first instance, and more recent revisions of estimates and projections have repeated the problem. A solution, but not one we would favour, is that the population estimates are shifted by only limited amounts towards the “new" totals.

5.4 This is an intractable difficulty when thought of simply in terms of population estimates or statistics. It is the effects of the revisions that should be the focus of attention. We would particularly draw your attention to paragraph 3.4, with regard resource planning using consistent data across local authorities.

6. THE ROLE OF SURVEY AND ADMINISTRATIVE DATAWhat role can and should be played by survey data in the compilation of population statistics?

What role can and should be played by data from General Practitioners and other health service data in the compilation of population statistics?

What other data sources can and should be used in the compilation of population statistics?

6.1 Our comments on these questions are in the main well covered in the LGA submission and earlier paragraphs of this submission. We note that the work of ONS to develop an Integrated Population Statistics System would support many of the points made in the LGA submission about the benefits of administrative data. We also note, with reference to our earlier comments about the “marginal" factors, and the LGA's comments on “rare populations", that the collection of data on the scale and movement of migrants, especially international migrants, is the most critical element to be improved. Even so, improving administrative sources may not be able to adequately capture transient populations such as student and migrant workers.

6.2 There are dangers in relying on administrative sources to compile long-term statistical datasets. Statistical output is a by-product of these processes and is subject to the effects of changes in procedures or definitions. The difficulties in obtaining consistent measurement of unemployment over time illustrate this. The electoral register was used as an input in the Mid Year Population Estimates until the late 1990s. Differences in quality and procedure between local authorities was seen a major drawback in this source. More recently the prospective termination of council tax in Scotland is raising difficulties for very valuable work on collecting data on dwellings.

7. COOPERATION WITH STAKEHOLDERSHow effectively has the Office for National Statistics cooperated with stakeholders with an interest in and information relating to population statistics and how can cooperation be improved by the Statistics Board and by its Executive Office?

7.1 Recognition of the two points in paragraphs 5.2 & 5.3 by the Statistics Board and by its Executive Office would be very valuable.

7.2 We support the LGA submission that ONS resources need to be taken seriously if the concerns are to be addressed; we have seen the effects of resource limitations through our work liaising with ONS, and have on several occasions expressed concern about work streams having to stop or be suspended.

7.3 We also reinforce the point that ONS needs cross departmental support in its efforts to improve population statistics. It is an area where co-operation in and thinking, planning and working will provide huge benefits. We gave our support to this approach presented to us by ONSCD at the last CLIP Population sub-group meeting.

7.4 It should not be overlooked that ONS has a primary role in serving the interests of government in its widest sense, and the citizens of this country through that role. This is bound to raise tensions between the interests of certain parties, say individual and the need to consistently deliver satisfactory statistics.

7.5 ONS is looked to by users as “the experts" in establishing the most appropriate methodology leading to figures being accepted as “National Statistics" standard, a role which they generally live up to. Communication to its wide user community is fraught with difficulty, however. Revisions in particular could be more widely advertised, through networks such as the (local government) Statistical Liaison Officers, and through user groups. ONS' proposed update service may also be helpful.

November 2007

@HR25@

Memorandum from East Sussex County Council

This response has been prepared at officer-level, by the Research and Information team of the Transport and Environment Department of East Sussex County Council. Consultation has taken place with a wide variety of officers who use population statistics in their work.

Comments here are confined to the use of counts of current population through Mid Year Estimates and Census. They do not include comments on projections, which I assume are excluded from this enquiry.

1. USES AND DEFINITIONSWhat are the uses of population estimates, and how far do current population estimates meet the needs associated with those uses? What are the effects of inaccuracies or inadequacies in such estimates?

Population estimates (Mid Year Estimates) are one of our key datasets. They provide a count of the population at district and county level, and show the age/gender structure also. Some uses include

@UL@— Providing a better understanding of the demographic characteristics of the county and districts to inform and influence service delivery. For this we need to know current estimate of the population, age structure, how we compare to other local authorities, how this changes over time.@/UL@

@UL@— As denominator in calculating rates or proportions. These are used in preparing performance indicators, or targets.@/UL@

@UL@— As various denominators in Revenue Support Grant calculations, thus affecting current and future funding of the local authority.@/UL@

@UL@— As the best estimate of current population, the Mid Year Estimate is used to standardise or control other calculations against. Our own demographic projections use the 2006 Mid Year Estimate as a baseline figure.@/UL@

Current population estimates meet our needs for trusted population estimates at county and district level. Estimates at a lower level of geography (ward, parish and lower layer super output area) are welcomed provided they fully reflected actual change on the ground. However, in view of East Sussex's dependence on large flows of in- and out-migrants, small area estimates are inevitably less reliable. New statistics coming forward through ONS's “experimental" range are a potentially useful product which give more information on ethnic group and small area geography, while implying caution in their use by their naming.

Recent population estimates (for 2002-05) were altered this summer due to a change in methodology, and the outcome for East Sussex seems to be more in line with expected population than the previous set. The Office for National Statistics (ONS) were open and transparent about these changes and their likely outcomes, as well as their knock-on effect on other population-based datasets, and this process was helpful in enabling us to plan for the change and to understand the reasons for change and explain to users.

The effects of inaccuracies in the estimates include:

@UL@— Underfunding (or overfunding) of local authorities.@/UL@

@UL@— Poorer-planned services if denominators are inaccurate, as indicators and targets influence policy choices.@/UL@

How appropriate is a definition of the population based on the usually resident population in the context of the needs of the users?

The usually resident population is an essential key count of population in planning for current needs of the existing population, and to use in preparing population projections which assist in planning for future service delivery. However, it takes no account of seasonal variation (eg visitors), or daily variation (eg workers), or weekly variation (eg. long-distance commuters or second home owners). Each of these counts presents various problems in compiling, but are requested by statistics users from time to time, and would assist in more targetted service delivery. The 2001 Census produced limited statistics for workplace population and daytime population, which were of interest to users.

How does the quality and range of population statistics in England and Wales compare with that available in Scotland, Northern Ireland and internationally?

I am not able to comment on these comparisons.

2. THE ROLE OF THE CENSUSHow does the Census contribute to the creation of population statistics? Is the current frequency of Censuses appropriate and, if not, how frequent should Censuses be?

The Census, as a full survey of the entire population, provides a benchmark against which population estimates can be assessed and recalibrated after each Census. In addition it provides rich socio-demographic and economic detail of the population at all possible levels of geography which remains useful throughout the entire intercensal period and is not replicated in any form elsewhere.

A more frequent census would be welcomed from a research user point of view, but has huge cost and organisational implications.

What lessons have been learnt from the Test Census carried out on 13 May 2007? What methods should be employed for the distribution and completion of Census forms, including the use of the Internet?

What steps should be taken to increase the rate of responses to the Census, particularly in hard-to-reach areas? How effective are plans to measure the extent of non-response to the 2011 Census?

To what extent is there a trade-off between the length of the Census form and its role in providing population information? What questions should be included in the 2011 Census?

To what extent should Census questions and statistical information be coordinated across the United Kingdom and how effective are plans to ensure such coordination?

I have no specific comments to make on the organisation of the 2011 Census.

What forms should the outputs of the 2011 Census take, how and when should they be made available and how should they be integrated and coordinated with other information on population?

The output for the last (2001) Census used a new software called Supertable, which as users we found hard to master, and quickly stopped using it once other outputs became available. I would therefore suggest that tried and tested formats for output would be preferable, or that users are able to use new software in advance of data release.

Outputs for the 2011 Census should be mainly electronic, with much web-based data. We also have a specialised requirement for data in a format that SASPAC, our census analysis software, can use. After the last census there was a delay in receiving SASPAC-compatible data because ONS were not able to supply it to the SASPAC team ahead of the release date. Steps should be taken to ensure that this does not happen again.

With regard to when the data should be made available, we would prefer a fully-accurate dataset to be published later, rather than to have to cope with revisions as we did during the release of 2001 data. This caused us difficulty in processing and publication of Census data.

As far as integration or co-ordination with other population information goes, the 2011 Census should form the underpinning data for the 2011 Mid Year Estimate, though they will be different owing to the different date to which each relates. The 2011 MYE should then form the basis for recalibrating the MYEs for the period 2002-10, as has previously been done.

In addition, understanding of census data by general users would be improved by ensuring that comparable datasets show the same numbers for the same counts (for example, total population by ethnic group for a ward equals total population by economic status for that ward). This will have an impact on the disclosure methods used by ONS (to preserve confidentiality).

3. MID-YEAR POPULATION ESTIMATES

How accurate and useful are the mid-year population estimates for England and Wales, including information available for local authorities and Strategic Health Authorities?

My answers to question 1 already cover this.

How appropriate is the methodology by which mid-year population estimates are reached?

My answers to question 1 already cover this. We consider the outputs from the latest (summer 2007) revised MYEs to more accurately reflect population growth for our area than the earlier methodology for the period 2002-05, and seem to be closer in line with housing growth in the area. However, we have no specific evidence to support this other than consideration of past rates of change.

What are the most appropriate ways to ensure a common level of accuracy in mid-year population estimates for different areas and what steps should be taken to improve the accuracy in areas where population is hard to measure?

Consistency of method across all areas is one of the valuable characteristics of nationwide datasets such as the MYE. Therefore any particular improvements in hard-to-measure areas would affect this.

What is the right process for making revisions to mid-year population estimates so as to meet the needs of users?

My answer to question 1 already covers this. I consider the current method to be useful and considered, and carefully thought through. The recent change allowed us to make local users aware of impending change, and the relative scale of change, before we had finally-agreed figures to use.

THE ROLE OF SURVEY AND ADMINISTRATIVE DATAWhat role can and should be played by survey data in the compilation of population statistics?

Various administrative datasets can form a useful check against which the MYEs can be assessed, but each will have its own issues and problems.

What role can and should be played by data from General Practitioners and other health service data in the compilation of population statistics?

In the past, GP data has tended to overestimate the total population when compared with the MYEs, although movements on and off local registers are more reliable, and indeed form another key dataset for us (migration moves to and from county/districts).

What other data sources can and should be used in the compilation of population statistics?

Council tax registration, electoral registration, GP registers, income tax records, child benefit, pensions data, National Insurance Number registrations, Workers Registration Scheme, school census data.

4. COOPERATION WITH STAKEHOLDERSHow effectively has the Office for National Statistics cooperated with stakeholders with an interest in and information relating to population statistics and how can cooperation be improved by the Statistics Board and by its Executive Office?

ONS seem well aware of other data methods and data sources, and they keep local authority contacts informed through the Statistical Liaison Officers network. However their website needs significant improvement, as a mechanism for communication with users.

November 2007

Memorandum from the Association of Census Distributors

EXECUTIVE SUMMARYE1. The Association of Census Distributors ( ACD ) was set up fourteen years ago to negotiate with the

Census Offices on behalf of organisations whose business is the value-added dissemination of Census data. ACD members now supply many thousands of organisations, primarily in the private sector, but also including public sector bodies.

E2. Initially, ACD members were seen as the “poor relation" to other Census user sectors, but over the last decade, the OPCS/ONS consultation mechanism has developed very well in this regard. However, there are still occasions when it seems that central government departments can exert undue influence on the consultation process—a recent example being the imposition of questions designed to address the issue of inward migration and the intentions of recent immigrants.

E3. We would question the suitability of these “migration" questions for inclusion in the Census; they also have the effect of displacing other questions which we believe are more appropriate to the Census method of data collection, and which are very important to most users, such as “labour force" questions (qualifications, industry of employment, etc).

E4. Therefore we would urge that space on the Census form is reserved for questions that have the greatest utility and are most likely to be answered well, thus providing good “raw material" for analysis.

E5. We are pleased to note ONS' expressed intention to revisit and develop its vision for an Integrated Population Statistics System (IPSS) which would link results from the 2011 Census to administrative records and to its imminent Continuous Population Survey. If achieved, this system would solve two drawbacks of the “traditional" Census, viz. the production of data once in a decade, and a two-year lag in obtaining results.

E6. One of the key elements of the IPSS model is for a “high quality address register, covering all properties in England and Wales" We regard the failure in 2005 of the National Spatial Address Infrastructure, and the recent statement by Communities and Local Government on 1 June 2007 that “the Department does not intend to carry out any further work on the NSAI at this time" as a very important opportunity missed. Surely, such an address database is crucial for many national purposes—in the context of the Census, its absence represents a huge risk to the success of the 2011 Census—and we would urge those responsible within government to readdress this issue with great urgency.

SUBMISSION1. The role of the Association of Census Distributors (“ACD")

1.1 The ACD is an association of organisations licensed by ONS to distribute Census data to third parties. The Census data are invariably distributed in value-added form by ACD members, as a part of products such as small area estimates, neighbourhood classifications, targeting products, etc. Census data are often combined with other data, and sometimes illustrated by maps. Analytical software is provided in most cases. Current members of the ACD are:

@UL@Acxiom UK@/UL@

@UL@Beacon Dodsworth@/UL@

@UL@CACI Information Solutions@/UL@

@UL@EuroDirect Database Marketing@/UL@

@UL@Experian Business Strategies@/UL@

@UL@MapInfo Corporation@/UL@

1.2 There is vast experience of Census analysis within these organisations, some individuals have experience going back to the 1970s. ACD members play a full role in the Census user “community"; they meet with

ONS Census Customer Services twice per annum, are represented on the Business and Professional Interests Advisory Group, and have close links to the MRS Census and Geodemographics Group.

1.3 ACD members supply their products to both public sector and private sector customers, as already observed. The public sector has traditionally been seen as the key recipient of Census data (unsurprising, given the role of the Census in informing public policy, and resource allocation in local government) and for many years the private sector seemed to be the poor relation. However, this situation is changing, albeit slowly. These days the models and other products developed from Census data perform a vital role in targeting and location issues in the private sector, and users (and third-party suppliers such as ACD members, who represent their customers' needs ) wish to play their full part in specifying what they need. ONS' consultation mechanism has developed well over the last decade in this respect; but there are still occasions when it seems that Central Government Departments “call the shots", for example with the imposition of new questions related to the issue of inward migration (see later).

2. The role of the Census

2.1 As the Association's name suggests, the UK Census of Population is central to the activities of ACD members. Although many other data sources are now available (and used) by ACD members, the Census provides that key “snapshot" of the total population at a point in time, and at small-area level. Census data could be said to provide the key to the geodemographic products produced by ACD members, which are then provided to thousands of organisations. The range of customers encompasses small firms up to huge multi-national companies, in all business sectors, and including many public sector organisations (police, emergency services, Local Authorities, Heath Service, Government Departments). In practice, the ACD members form an interface between ONS and the vast majority of private sector users where Census data are concerned.

2.2 Not only are sales forecasts for (for example) superstores built on the basis of geodemographic data, but so—using similar principles—are models for resource allocation. Clearly the accuracy of the underlying Census data is key to this. In our experience, Census data have been generally reliable, although the problems in some major cities in the 2001 Census showed evidence of defects. A reliable address database would have contributed to the solution of that problem—of which, more later.

2.3 We were pleased to note, in the recently-published ONS consultation document “ONS Statistical Work Programme 2008-2012, (Annex C11)" that ONS are planning to refresh their vision for an Integrated Population Statistics System (IPSS) that was originally formulated in 2002-03. The ACD was very enthusiastic when the idea for the IPSS was first published. In a nutshell, the vision is for a comprehensive system which links into the 2011 Census initially, but also takes feeds from administrative data and survey data. The survey data in question refers to the Continuous Population Survey where ONS surveys such as the General Household Survey, Labour Force Survey, Expenditure and Food Survey, etc. will be combined to collect common data, which could feed into the IPSS. On this vision, the 2011 Census would be the last decennial Census. The paper produced in 2003 made the point that the IPSS “would provide highly accurate small area statistics much more frequently and quickly than is possible at present". Thus it would meet the two current drawbacks of the “traditional" census, which only produces data every 10 years, and then with a two-year delay before data are available for use.

2.4 Significantly, one of the key elements of the IPSS proposal was that “a high-quality address register, covering all properties in England and Wales, is needed". It is little short of tragic that the attempt to create such a national address register with the National Spatial Address Infrastructure, foundered (news release from Ordnance Survey, 11 August 2005, and statement by CLG on 1 June 2007). It was evident to all users of these data that a definitive national address dataset was essential; witness (for example) the costly attempts by ONS to produce a “combined" address database for the recent Census test (13 May 2007). This task will obviously be far greater when planning for the Census in 2011. We are not sure whether this issue falls within the terms of reference of the Treasury Sub-Committee; but if it does so, then we would urge the Sub-Committee to use its influence on Parliament to have this issue revisited. It cannot surely be beyond the powers of the Department of Communities and Local Government to find some resolution between Ordnance Survey and the Local Government Association.

3. Questions to be asked in the 2011 Census

3.1 We have noted with concern that new questions, essentially intended to address the issue of inward migration to the U.K., and how long recent migrants intend to stay in the U.K., have been added to the list of questions previously intended.

This has had the effect of causing doubt as to whether some questions asked in previous Censuses, and vital to geodemographic analysis, may now be asked. We have in mind questions such as:

@UL@— qualifications@/UL@

@UL@— industry/name of employer@/UL@

@UL@— size of (workplace) establishment@/UL@

@UL@— hours worked@/UL@

These questions provide invaluable information for both neighbourhood

classifications (which themselves provide the basis for much of the work on area estimation and targeting) and socio-economic classifications. Surely it is not only the ACD members that would lament the absence of this information—both Central and Local Government would find certain tasks much more difficult without it.

3.2 Looking at some of the “migration" questions, experienced Census users question whether a decennial Census is really the best source for such information. One obvious point is that the information apparently required will not actually be available until Census results are published in 2013! That hardly meets the need, surely? Also, recent migrants are difficult to enumerate—elusive, not necessarily keen to respond to a Census form, and not necessarily able to state how long they will stay in the U.K.

In summary of this point, we would urge that space on the Census questionnaire is reserved for questions that have the greatest utility and are most likely to be answered well, thus providing good “raw material" for analysis.

4. Mid-year population Estimates

We have no specific comments about mid-year estimates, other than the observation that the current shortcomings is migration statistics are a contributor to inaccuracies; and that use of administrative records is probably the best solution to this problem.

5. Statistics Commission Report No 36

We welcome publication of Report No 36 “Counting on Success; The 2011 Census—Managing the Risks, November 2007, published by the Statistics Commission.

This is a very sound document and makes a number of excellent recommendations. We endorse the Statistics Commission's advice.

6. Harmonisation between Census Offices

We note that the three Census Offices—ONS, GROS and NISRA—have undertaken to harmonise output of the 2011 Census so far as is possible. This is crucial to members of the ACD, who deal with UK-wide Census data as a matter of course. The lack of harmonisation in 2001 caused many problems in handling the Census data. Details of these problems may be found in a joint submission by the ACD and the MRS Census & Geodemographics group, which can be accessed on www.mrs.org.uk/networking/cgg/downloads/users_view_2001census.pdf

November 2007

@HR25@

Memorandum from the London Councils

INTRODUCTION AND SUMMARY1. London Councils represents all 32 London boroughs, the City of London, the Metropolitan Police

Authority and London Fire and Emergence Planning Authority. We develop policy, lobby and run a range of services designed to improve life for Londoners.

2. London Councils welcomes the opportunity to respond to the questions raised by this Inquiry because concerns about the robustness of population estimates and projections have particular relevance for the capital. The use of inaccurate population data has been a long standing concern to London's 33 councils. For example, the Association of London Government raised concerns about the accuracy of the mid-year estimates in response to the 2004-05 provisional local government finance settlement.

3. The accuracy of the ONS' population estimates and projections is vital to local authorities because the statistics play a key role in determining the level of central government funding allocated to each local authority in England. Despite this. it is widely acknowledged by experts in both the public and private sectors that the current population statistics are not fit for this purpose. Work undertaken or commissioned by local authorities has shown large discrepancies between official and local authority estimates. This has significant implications as undercounted population figures reduce the level of Revenue Support Grant (RSG) local authorities receive and, therefore, result in the misallocation of resources for vital local services.

4. Since 1990, the change to the balance of funding has meant that councils are more reliant on central government funding. Therefore, the impact of inaccurate and unreliable data on council funding is critical because it can result in a mismatch between the level of resources available to a local authority and the actual needs of the local community. Councils now have very few avenues open to them to compensate for a reduction in the level of their grant due to inaccurate population data. This means that the communities in areas disproportionately affected by undercounted population figures are directly penalised. This is particularly true of deprived areas with high numbers of residents from minority ethnic groups where undercounting is especially prevalent. These local communities can least afford to lose government grant.

5. In addition, the definition of a resident for population purposes excludes a number of groups of people living in local authorities and who are using local authority services. In particular, short-term migrants and “part-time" Londoners are currently excluded from the ONS' population statistics. This means that local authorities do not receive any funding for these groups despite providing them with local authority services. There is an urgent need for statistics on these groups and for further information to be collected on them in the 2011 Census.

6. London Councils recognises that there are numerous other uses of population statistics by central government, local government and other users. However, the focus of this submission is on local authority level estimates and projections and the impact of these on the funding of local authorities by central government.

1. USES AND DEFINITIONSWhat are the uses of population estimates?

7. Local authorities use population estimates to plan and budget for local services, to develop local policies and deliver local solutions for their diverse communities.

8. Population estimates and projections play a vital role in determining the level of central government funding allocated to each local council. London Councils' response particularly focuses on this impact because the right level of funding is critical to an individual council's ability to make good local decisions and to deliver the appropriate level and quality of services that its community needs.

How far do current population estimates meet the needs associated with those uses?

9. It is widely acknowledged that the current population estimates fall a long way short of meeting the needs of their users because the estimates of internal and particularly international migration are not reliable. These concerns are shared by both the public and private sectors.

10. In July 2006 Professor David Rhind, Chairman of the Statistics Commission, told a seminar organised by Slough Council: “There is no dispute—current local migration data is inadequate for many key government purposes including grant allocation."

11. The Governor of the Bank of England, Mervyn King, told the House of Lords Select Committee on Economic Affairs: “We do not have the ability to measure accurately, at present, the size of the population."32

12. The Audit Commission pointed out “current local population projections and diversity data do not fully reflect the recent increases in migration."33

13. A further way in which the population estimates and projections do not meet needs is their failure to include short term migrants, “part-time" Londoners and illegal migrants (described in more detail below). Local authorities do not receive any central government funding for these groups despite providing them with services. The exclusion of these groups from the population statistics has a major impact on funding in London.

What are the effects of inaccuracies or inadequacies in such estimates?

14. The recent introduction of three-year Local Government Finance Settlements has seen population estimates replaced by population projections in some parts of the distribution formula. Inadequacies in the population estimates are perpetuated and compounded in the population projections. The problems of undercounting population numbers are frequently concentrated in deprived localities with transient populations and a higher representation of people from minority ethnic backgrounds. In addition, some international in-migrants, including refugees, asylum seekers and illegal migrants may arrive in this country with complex needs that require a range of local authority support. Missing out on grant because of inaccurate population estimates only compounds local authorities existing problems.

How appropriate is a definition of the population based on the usually resident population in the context of the needs of the users?

15. The definition of a resident from overseas for population purposes, currently defined as a migrant intending to reside in the UK for at least a year, has become increasingly important over the last few years due to increasing numbers of short-term migrants in the UK. This is a particular issue for London, as demonstrated by recent experimental statistics published by the ONS which suggest that 40% of all short-term migrants live in London. Therefore, the current definition of a resident fails to meet the need for population statistics for funding purposes because they exclude a significant proportion of people who are living in the capital. This means that local authorities that experience influxes of short-term migrants do not receive funding for people who are using their services. Thus, there is an urgent need for estimates of short-term migration at the local authority level, or for them to be included in existing population statistics.

16. A further inadequacy with the current definition of a resident for population estimates and projections is the failure to include people who live in London for part of the week but are resident elsewhere for population purposes also known as “part-time" Londoners. This includes those that either own or rent a home in the capital and who perhaps return to a family home for the weekend or part of the week and also those who stay with family or friends in London during the week. These “part-time" Londoners may spend the majority or a significant part of the week living in London but because they are registered elsewhere they are excluded from the population statistics. Therefore, despite the fact that they use local authority services local authorities do not receive any funding from central government for them.

17. These omissions also mean that local councils cannot accurately plan for the delivery of services as the level required will be higher than that presumed by the population statistics. Only when these, and inaccuracies in the population statistics, are addressed will local authorities receive the funding required to provide services to all residents living within the local authority.

2. THE ROLE OF THE CENSUSHow does the Census contribute to the creation of population statistics?

32 Evidence to the House of Commons Select Committee on Economic Affairs, 31 October 200633 Crossing borders: responding to the local challenges of migrant worker, Audit Commission, 2007

18. The Census is the starting point from which population estimates and, consequently, the projections are derived. It is also used as a benchmark to check the rolled forward mid-year population estimates. Therefore, the validity and reliability of base population figures from the Census is a prerequisite for, although not sufficient to ensure, accurate population estimates and projections. Inaccuracies in population figures from the Census were a particular problem following the 2001 Census when a number of authorities, including Westminster, successfully challenged their Census population figures and subsequently had them amended.

19. In addition, data from the Census is used to allocate international migrants from sub-regional geographies to local authorities, contributes to the estimate of international out-migration from local authorities and is used to create sub-regional groupings of authorities in London. Given the extent of migration since 2001, and in particular from 2004 onwards, it is extremely unlikely that current settlement patterns of international in-migrants match those in 2001.

20. The Statistics Commission, the Audit Commission and the Bank of England have all expressed serious concerns about the quality of population estimates based on Census information.

Is the current frequency of Censuses appropriate and, if not, how frequent should Censuses be?

21. Population statistics are likely to get further and further away from the true population in a local authority the greater the distance between the estimate and the base year ie the Census. Therefore, there are grounds for increasing the frequency of the Census so that the base for population estimates can be updated more regularly. Re-basing the population estimates every five years might avoid the large discrepancies between the rolled forward mid-year estimates and the Census that were experienced following the 2001 Census. However, London Councils does not currently have a formally agreed view on this issue.

What lessons have been learnt from the Test Census carried out on 13 May 2007?

22. The London borough of Camden took part in the Census Test in 2007. This was in part due to the large undercount experienced in the 2001 Census as a result of a number of factors but including the fact that Camden has a high proportion of “hard to count" groups such as students, a highly mobile population, a large number of gated developments as well as language and cultural barriers. Authorities involved in the Census Test and the ONS will be best placed to outline the lessons learnt from the test but some of the key lessons that have emerged were the value of checking the ONS address database to local sources of addresses, and the importance of early information and publicity leading up to the Census.

What steps should be taken to increase the rate of responses to the Census, particularly in hard-to-reach areas?

23. A co-ordinated approach needs to be taken to find a range of solutions to tackle undercounting in specific population groups. ONS needs to pilot and evaluate different approaches to tackle recognised hard to reach groups to develop a variety of approaches to suit the needs of different population groups.

24. Working closely with local authorities ONS needs to consider appropriate ways of tackling the problems of undercounting people living in households in multiple occupation and households where no English is spoken or is spoken as an additional language. Consideration also needs to be given to the head of household as this may be open to different cultural interpretations and may be an issue in households of multiple occupation.

25. ONS needs to focus on the changing demographic dynamics of local communities and particularly take account of the impact of migration from Accession countries.

How effective are plans to measure the extent of non-response to the 2011 Census?

26. London Councils is unable to comment on the plans for dealing with non-response in 2011 with any authority, however, every effort should be made to maximise the response rate to the 2011 Census. In addition, it is imperative that all problems that were experienced during the 2001 Census are resolved such as, the lack of Census enumerators in some areas to ensure that factors that led to low response rates in 2001 are not repeated in 2011.

To what extent is there a trade-off between the length of the Census form and its role in providing population information?

27. There is a trade-off between the length of the Census form and the potential for asking further questions on any subject not just population unless further resources can be sought to fund an extra page to be added to the Census form. Unless this is achieved careful consideration must be given to which other questions can be sacrificed to make way for additional questions on population which are clearly a priority.

28. The Census needs to capture robust and reliable population information. Thought needs to be given to avoiding questions which may be open to different interpretations as these could compromise the quality and accuracy of Census data, or those that reduce response rates to the Census.

What questions should be included in the 2011 Census?

29. The Data Management and Analysis Group at the Greater London Authority coordinate borough input on the content of the 2011 Census. However, from London Councils' perspective, with a focus on funding, there is a definite need for a question in the 2011 Census that would provide further information on second homes and residences. As noted earlier, these people may actually spend the majority of their week in one local authority but be recorded elsewhere for population purposes. In addition, a question that would allow ONS to capture data on short-term migrants would also be invaluable given the difficulty in producing these figures at the local authority level and the need for this measure for funding purposes.

30. More information is needed about the important issue of population mobility. The current questions fail to capture the extent of movement in areas with very volatile populations and large number of frequent movers (who make several moves in a 12 month period). These population groups are also more likely to be undercounted.

3. MID-YEAR POPULATION ESTIMATESHow accurate and useful are the mid-year population estimates for England and Wales, including information available for local authorities and Strategic Health Authorities?

31. As this response illustrates, there are widespread concerns about the accuracy of mid-year population estimates in both the public and private sectors. Work by, or commissioned by local authorities comparing ONS population statistics with other sources of data has shown large discrepancies between official and local authority estimates. For example, in Brent the results of a detailed data matching exercise suggests that the ONS has undercounted their resident population by over 17,000 residents. A similar study in Enfield confirmed that the ONS estimates were undercounting at least 3,300 residents.

32. Other boroughs have looked at a range of data sources to indicate that the population statistics are an under-prediction of their resident population. For example, in Newham comparisons with Flag 4 data and National Insurance Number Registrations suggest that the current population estimates are undercounting the number of international migrants in the borough. Although differences in definition exist between these data sources, the scale of the differences seem to suggest that there are issues with the level of international in-migration to that area as estimated by the ONS. Perhaps more damning is a comparison of resident pupil data with the population estimates. Resident pupils in the borough, which are themselves an undercount of the relevant resident population because they exclude children in private schools, are 750 pupils higher than the equivalent estimate of children of that age in the borough estimated by the ONS.

33. This underestimation of local authority populations is by no means unique to London. Even the Office of National Statistics (ONS) recognises that international migration is the most difficult element of the population statistics to accurately measure. Indeed, in a letter to the office of the Deputy Prime Minister 34, a senior ONS official asks the government department to consider “recognising the particular uncertainty for those parts of the country that are affected by relatively high levels of migration in general grant distribution."35

How appropriate is the methodology by which mid-year population estimates are reached? What are the most appropriate ways to ensure a common level of accuracy in mid-year population estimates for different areas?

34. The ONS has recently changed the way it allocates international migrants to regions which was previously based on the International Passenger Survey (IPS). The ONS are now using the Labour Force Survey (LFS) to determine how many international migrants are arriving to live in London each year. Although the IPS

34 Now the Department for Communities and Local Government (CLG)35 Letter from Glen Watson, Director of the Social Reporting and Analysis Group (ONS) to Lindsay Bell (CLG), Dec 2005

suffers from some well documented limitations, the LFS also has some serious inadequacies including its failure to survey people living in communal establishments, its extremely small sample size and the low response rate to the survey in London compared to elsewhere in the country. Given the acknowledged undercounting in London due to high levels of transience and the scale of migrants who do not appear in official statistics, it is a matter of serious concern that this one methodological change has led to a population estimate that is over 60,000 people lower than the estimate produced by the previous methodology. These new population figures are counterintuitive and do not match experience on the ground as highlighted above.

35. Worryingly, these inaccuracies in the population estimates are perpetuated and compounded in the population projections. This approach is clearly inadequate and is an issue which needs to be addressed as London prepares for a further inflow of migrant workers in the run-up to the 2012 Olympics. Many of them will need access to local housing and will use local services (including translation services) but the methodology for projecting population figures fails to take account of them.

What steps should be taken to improve the accuracy in areas where population is hard to measure?

36. The Inter-departmental Taskforce on Migration Statistics made a number of recommendations on ways in which the population statistics could be improved but these, for a number of reasons, have not yet been implemented. Where there are recognised problems with the accuracy of population data for some communities or resident groups, it would be sensible to draw upon other data sources to gather further information or to quality assure the population statistics. London Councils feels there is scope to consider a range of material including:

@UL@(i) National Insurance registrations by overseas nationals;@/UL@

@UL@(ii) The Workers Registration Scheme;@/UL@

@UL@(iii) The Pupil Level Annual School Census;@/UL@

@UL@(iv) Council Tax and Housing Benefit information;@/UL@

@UL@(v) The Electoral Register;@/UL@

@UL@(vi) Local Authority and housing association data;@/UL@

@UL@(vii) Other local authority data;@/UL@

@UL@(viii)Utilities data;@/UL@

@UL@(ix) Local surveys;@/UL@

@UL@(x) NHS data;@/UL@

@UL@(xi) GP registers (although the quality of this information is known to be variable) and other health service data.@/UL@

37. These data sources will each have limitations but can at the very least be used to highlight whether there are any potential issues of undercounting, and taken in combination may provide more insight into the extent of any inaccuracies.

What is the right process for making revisions to mid-year population estimates so as to meet the needs of users?

38. It would be appropriate to make a revision to population statistics where the ONS, demographers and users with local expertise are all in agreement that the revision represents an improvement in the accuracy of the population statistics because there is evidence of undercounting or another factor which has compromised the quality of the existing data. This needs to be balanced with an aspiration to minimise the number of sets of population estimates and projections for any one year. Given the problems with the current population projections and estimates it seems likely that further revisions will be needed although the ONS have ruled out making any further changes in the next two years.

4. THE ROLE OF SURVEY AND ADMINISTRATIVE DATAWhat role can and should be played by survey data in the compilation of population statistics?

39. London Councils has already highlighted a number of recognised problems with survey data that are currently used in the production of population statistics. For example, the LFS does not survey people living in communal establishments, it is based on an extremely small sample size and the response rate to the survey in London is particularly low compared to elsewhere in the country. The fact that data derived from other surveys, such as the IPS give a different picture of how many migrants have settled in each region questions the validity of these data sources when the two measures are trying to capture the same entity.

40. As highlighted previously, there are other data sources that London Councils feels could contribute to the compilation of population statistics:

@UL@(xii) National Insurance registrations by overseas nationals;@/UL@

@UL@(xiii) The Workers Registration Scheme;@/UL@

@UL@(xiv) The Pupil Level Annual School Census;@/UL@

@UL@(xv) Council Tax and Housing Benefit information;@/UL@

@UL@(xvi) The Electoral Register;@/UL@

@UL@(xvii) Local Authority and housing association data;@/UL@

@UL@(xviii) Other local authority data;@/UL@

@UL@(xix) Utilities data;@/UL@

@UL@(xx) Local surveys.@/UL@

41. Each of these data sources have individual limitations and there are also definitional issues such as length of stay and de-registration but these sources could at the very least be used for quality assurance purposes. That is, to highlight whether there are any potential issues of undercounting and in combination may provide insight into the potential extent of any inaccuracies.

What role can and should be played by data from General Practitioners and other health service data in the compilation of population statistics?

42. London Councils believes that, when used in conjunction with other information, GP registers and other health service data have the potential to increase the robustness of population figures. However, we would recommend that GP registers are not used independently and should always be used with other material to ensure that specific population groups who are less likely to register with GPs are not disregarded.

43. It should also be recognised that more mobile population groups and some migrant groups are less likely to appear on GPs registers and other health data.

What other data sources can and should be used in the compilation of population statistics?

44. As mentioned earlier, London Councils believes that a range of material could be used collectively to gather more detail about the population of areas where there is recognised undercounting. The full range of sources that could improve the accuracy of population figures is set out in the previous section in answer to the question about data accuracy.

5. CO OPERATION WITH STAKEHOLDERSHow effectively has the Office for National Statistics cooperated with stakeholders with an interest in and information relating to population statistics and how can cooperation be improved by the Statistics Board and by its Executive Office?

45. The ONS made genuine attempts to engage with local government as part of the process of changing the methodology for estimating international migration in the population statistics in 2007. However, wider engagement with stakeholders should have started much earlier in the process so that the ONS' thoughts on potential changes to the methodology could have been shared at an earlier stage with local authorities and other users of the statistics. In addition, this would have allowed for a period of consultation with stakeholders who have demographic expertise and knowledge of their local areas and who could have usefully contributed to the process and methodology.

46. Concerns have been raised about the Central and Local Information Partnership (CLIP) Population Sub-Group of which London Councils is a member. The papers discussed at these meetings can be freely circulated to non-members, and are placed on the CLIP website. However, it is our impression that the proceedings of the group are perceived as opaque. London Councils has tried to address this within London by circulating papers and notes of the meetings but perhaps this needs to be done on a more consistent basis to other authorities.

47. The CLIP Population sub-Group is also not particularly representative of local authorities. Attempts have been made to address this but there is a definite need to ensure that this group's membership is representative of all regions and tiers of authority whilst maintaining demographic and statistical expertise on the group.

48. The government should encourage partnerships between provider and users of statistics to improve data accuracy because it is in everyone's interest to have reliable and valid population figures.

November 2007

@HR25@

Memorandum from South Tyneside Council

SOME KEY CONSIDERATIONS : REBUILDING THE FOUNDATIONThis issue has reached a “crunch" point and it's been a long running problem for at least 16 years since the

1991 Census.

1. Following the release of the 1991 Census results there were numerous revisions to the population totals.

2. In the lead up to the 2001 Census this problem was going to be resolved by the “One Number Census", it was not. Following the release of the 2001 Census results disparities emerged in some local authorities like Westminster and Manchester, where population numbers were seriously under counted. These issues were resolved in the authorities favour after an extensive checking exercise.

3. At present the population of England & Wales is experiencing a large amount of volatility, while natural change is largely stable there are large international population flows, especially from the newer EC member states like Poland. For many users of population statistics, this has led to a growing “creditability-gap" between the ONS “world-view" and the reality that many local authorities face.

4. This reflects very badly on the national statistics body—Office of National Statistics. The perception, from a user perspective, is that the issue is not under control and the problem is getting worse. In areas like ours, in South Tyneside, where previous crisis have not arisen, there is a perception that official statistics and reality are diverging.

5. In 2001 we would be fairly sure that the Borough population was the 152,785 as counted by the 2001 Census but hand on heart we cannot say whether the current estimated population is indeed 151,000 in 2006.

6. There appears to be no short-term improvement in matters. Indeed recent revisions to the population estimates, due to the revised procedures for reallocating international migrants have given quite a different picture.

7. Up to publication of the revised mid year estimates earlier this year, South Tyneside's population trend was one of improvement and slowing of the historic population decline so common with many urban authorities, this was viewed with optimism, could a new era have begun as a result of the recent regeneration efforts? Indeed, elsewhere in the Tyne & Wear conurbation, population increases could be hailed as the success long hoped for.

COUNTING THE POPULATION: REBUILDING THE FOUNDATION (CONT.)8. Yet was this just an apparent illusion, the revised population estimates now show population

declines, some of which are very significant, for most of Tyne & Wear.

9. In short with the current population counts we do not know where we are, Tyne & Wear authorities have become so concerned with the recent revised estimates that they have decided to “health warn" users when the statistics are used, highlighting the concerns of the Tyne & Wear LA's and the joint information unit Tyne & Wear Research and Information. Increasingly time is spent trying to validate ONS population figures be they estimates or projections.

10. This is an inefficient use of resources—the suite of population products: Census, population estimates and population projections is fast becoming “not fit for purpose" and not the “gold-standard" ONS hopes.

11. Yet there is an indication of a solution here—activity at the Government / ONS and local levels, this will be examined in a little more detail later.

12. We must say that there is a great of good work being undertaken by ONS—new small area estimates for example, but without a good solid foundation all this work is in danger of being wasted.

13. In short we do not know accurately how many people there are and their characteristics—there is a core of easily identifiable sections of the population but there is an increasing sub-group, which is difficult to enumerate, this has been recognised internationally by all Census taking bodies.

14. Urgent action needs to be taken—the next scheduled “reality check" is the 2011 Census—this is just too long & the problem too severe.

15. An accurate headcount of the population should be undertaken at more frequent intervals than 10 years. This should be a low cost option because all that will be asked is:

@UL@(a) How many people live at this address?@/UL@

@UL@(b) Their Sex@/UL@

@UL@(c) Their Age—with Date of Birth@/UL@

@UL@(d) Whether usually resident or visitor@/UL@

@UL@(e) Possibly either or both National Insurance Number or national Health Number.@/UL@

All households must be included.

16. How is this to be implemented?

@UL@(a) A stand alone ONS survey or@/UL@

@UL@(b) A survey conducted in partnership with Local Authorities—perhaps as part of the regular work of the authority such as the compiling of the Electoral Roll.@/UL@

@UL@(c) A similar approach already occurs in Scotland where there is a “Voluntary Population Survey" included with the annual Electoral canvas of voters, the process enumeration is extended to all household members. Kadham Jallab of Tyne & Wear Research and Information suggested this idea. Apparently 17 Scottish local authorities already operate this scheme.@/UL@

17. There is a real urgency to establish a correct full population base now. This Intercensal head count must be implemented within the next year or two before the 2011 Census to re-establish the creditability of all population statistics.

18. It four years to 2011 Census then it will take two/three years to get results we cannot really wait to 2013/2014 to get an updated population base.

19. Defining the Population

South Tyneside's supports the concept of a usually resident population, with the ability to distinguish between who usually lives at an address and who are visitors.

20. Population Estimates

The only real issue of concern is the new methodology in redistributing international migrants, the Tyne & Wear authorities think that this methodology is flawed, see the City of Sunderland's and Tyne & Wear Research and Information's submissions.

21. Use of Administrative Sources

@UL@(a) It is South Tyneside's view that administrative sources are not capable of replacing Census/Headcounts of population—they are a useful supplement however.@/UL@

@UL@(b) We have found that children & pensioner numbers from HMRC/DWP are reasonably accurate but numbers for working age persons (derived from NINO's) is grossly over inflated.@/UL@

22. The Census

@UL@(a) This is essential. It is a comprehensive reality check where the real-world intrudes into the theoretical models of population.@/UL@

@UL@(b) Its main value—all data is set to a “bench mark" at a specific date—there is national consistency, the geographic levels are on the same basis—allowing effective comparison. @/UL@

@UL@(c) There are issues in comparing across time—definitions of variables shift from Census to Census@/UL@

@UL@(d) South Tyneside support the use of a consistent geographic basis, based around the current 2001 Output Areas.@/UL@

23. Ethnicity & Disability between Censuses

There is a real need for accurate estimates for ethnicity and disability between Censuses.

24. Summary

@UL@(a) Following an increasing level of concern, over a number of years, among the user community there is a real need to restore the creditability of population statistics produced by Office of National Statistics. @/UL@

@UL@(b) What is required is the rebuilding of users confidence in an accurate estimate of current population levels to at least district level. Current methods of estimating the population at district level cannot give that guarantee.@/UL@

@UL@(c) It is suggested that a simple head count be conducted at a frequency greater than 10 years, possibility even annually until confidence is restored.@/UL@

@UL@(d) A means of undertaking this, at limited cost, is suggested in co-operation with local authorities, building on examples of best practice elsewhere in the UK.@/UL@

@UL@(e) We thank the Committee for this opportunity to be able make comments on this very important issue.@/UL@

November 2007

Memorandum from Milton Keynes Council

1. EXECUTIVE SUMMARY1.1 Milton Keynes Council believes an increased frequency and improved accuracy of the census would

assist the local government and planning community in ensuring appropriate resource is secured for infrastructure and services in areas of high population growth.

1.2 In forming local estimates of populations, Councils such as Milton Keynes look to the ONS for expert guidance. Milton Keynes Council see opportunities for the ONS to expand their service offering in this regard through networks such as the Statistical Liaison Officers, or other user groups.

1.3 Milton Keynes Council believes council tax records and housing numbers/projections could provide an additional valuable tool in understanding population statistics.

1.4 Keynes Council supports the exploration of alternative methods for the distribution and completion of Census forms, including the use of the internet. There is potential in such methods to increase return rates and improve administrative efficiency. It could also assist in an increased frequency and accuracy of census.

2. THE ROLE OF THE CENSUS2.1 Population statistics are used by the Department for Communities and Local Government to

determine the financial settlements of local authorities. The census provides a “baseline" against which to base future projections, after adjustment for undercounts.

2.2 However, for a city such as Milton Keynes, one of the fastest growing in the UK, often census statistics are out of date within months of being collected. Indeed, by the end of its cycle, census data is over ten years out of date. This has resulted in local government funding settlements beneath that of its entitlement in terms of population. Milton Keynes Council believes an increased frequency and improved accuracy of the census would assist the local government and planning community in ensuring appropriate resource is secured for infrastructure and services in areas of high population growth.

2.3 The Council also call for Government to apply a consistent methodology across Government to use of population statistics for funding services. For example, Milton Keynes Council were recently extremely disappointed to discover that the Department for Children, Schools and Families made little or no ackowledgement of the cities growing population when cutting our three year Basic Needs Allocation from £30 million to £10 million. This has caused particular concern amongst the local community.

3. MID-YEAR POPULATION ESTIMATES3.1 Mid-year population estimates are important to Local Government, as they determine the level of

funding received. Errors in the estimates can lead directly to problems in funding local services.

3.2 Milton Keynes Council believes the current methodology used for mid-year estimates are appropriate but may not suit all local circumstances. Hard to record transient populations include the refugees, armed forces and students. Means of recoding these groups locally need to be employed.

3.3 In forming local estimates of populations, Councils such as Milton Keynes look to the ONS for expert guidance. Milton Keynes Council see opportunities for the ONS to expand their service offering in this regard through networks such as the Statistical Liaison Officers, or other user groups.

4. THE ROLE OF SURVEY AND ADMINISTRATIVE DATA4.1 Milton Keynes Council believes council tax records could provide an additional valuable tool in

understanding population statistics, for example about numbers of second homes. Moreover, the use of wider data sources, included housing data should be used. Given that the Government has set housing growth numbers for each area, it should be possible for ONS to project likely household growth for local authorities. If nothing else, housing projections and targets could be used as a “reality check" for ONS figures in high growth areas.

5. THE DISTRIBUTION AND COMPLETION OF CENSUS FORMS5.1 Milton Keynes Council supports the exploration of alternative methods for the distribution and

completion of Census forms, including the use of the internet. There is potential in such methods to increase return rates and improve administrative efficiency. It could also assist in increased frequency and accuracy of census.

November 2007

@HR25@

Memorandum from the London Borough of Brent

INTRODUCTION1. This submission by the London Borough of Brent seeks to answer two of the questions posed by the

Sub-Committee, as follows:

@UL@— How accurate and useful are the mid-year estimates for England and Wales, including information available for local authorities and Strategic Health Authorities?@/UL@

@UL@— What other data sources can and should be used in the compilation of population statistics?@/UL@

2. Our evidence is based on the revised ONS estimates published in August 2007, information available to the authority on population trends from data sources other than ONS, and work commissioned by the authority from Professor Les Mayhew of CASS Business School on use of other administrative sources to estimate the population of the London Borough of Brent.36 A copy of Professor Mayhew's report is attached to this submission.

ACCURACY AND USEFULNESS OF THE ONS MID-YEAR ESTIMATES3. Our experience of the ONS estimates is that they do not measure population movements in Brent

accurately. This is based on the following evidence:

@UL@(a) The mid-year estimates are not consistent with historical trends of population growth in Brent. Population grew by 28,800 between 1991 and 2001, from 240,800 to 269,600. This was made up of natural growth in population of 20,400 (births less deaths), which was supplemented by a net inflow of population of 8,500 (people moving in less people moving out). Natural population growth was a further 13,200 between 2001 and 2006. But ONS estimates show a reversal of migration trends with net outflow of population from Brent of 11,400 over the same period. This is not supported by other evidence of population movement in Brent. Appendix A shows ONS figures on population movements within Brent between 1991 and 2006 together with the make-up of these movements between natural growth and net migration.@/UL@

@UL@(b) Movement in mid-year estimates is not consistent with movement in data from other sources. This is shown by comparisons between movements in ONS population estimates and movements in numbers of children in receipt of child benefit and older people in receipt of pensions. Appendix B compares movements between 2003 and 2006 in ONS estimates for children aged up to 15 and HMRC information on children aged 0-15 in receipt of child benefit. Appendix C compares movements between 2002 and 2006 in ONS estimates of people of pensionable age and DWP information on people in receipt of a pension. Both show significant unexplained differences between London boroughs in the extent to which ONS estimates of population match movements taken from other sources.@/UL@

@UL@(c) There are also inconsistencies between actual numbers the ONS includes in particular age groups and data from other sources. Appendix D compares 2006 ONS estimates for children aged up to 15 and HMRC information on children aged 0-15 in receipt of child benefit in 2006. Appendix E compares 2006 ONS estimates of people of pensionable age and DWP information on people in receipt of a pension. Again both show significant unexplained differences between London boroughs. For Brent, the differences are significantly greater for children than pensioners and this fits in with the findings in Professor Mayhew's study which showed a greater discrepancy in number of children and people of working age between his and ONS estimates than in number of older people.@/UL@

@UL@(d) Information used by ONS on international migration is not consistent with other data sources. For the past five years, Brent has had amongst the highest number of number of non-UK residents registering for national insurance numbers (NINOs) in the country. There are various reasons why these might not be the same as the number of international migrants coming into an area (eg they exclude children and other people not registering for work and they include people who may have only come for a short stay) but again consistency across boroughs in the comparison with ONS figures on international inward migration would be expected. Appendix F compares NINOs with ONS inward migration estimates for 2002-06. There is no obvious relationship across boroughs between the number of NINOs and ONS estimates on international inward migration.@/UL@

36 Brent population estimation, household composition and change: A comparative analysis between 2005 and 2007. Les Mayhew and Gillian Harper (Mayhew Associates Ltd.), November 2007.

@UL@(e) The work commissioned from Professor Mayhew which matches administrative data sources in Brent established Brent's “confirmed" population at March 2007 as 289.1k, some 17.7k above the ONS mid-2006 estimate of 271.4k. Professor Mayhew's methodology and the information on which he based his results are included in the attached report.@/UL@

4. The examples in paragraph 3 above, the graphs in the appendices, and the work carried out by Professor Mayhew provide evidence that ONS estimates of population in Brent are wrong. The lack of trust in the estimates leads us to use alternative data sources for our community and service planning including:

@UL@(a) Greater London Authority population projections which are more consistent with other data sources;@/UL@

@UL@(b) Data matching information using Professor Mayhew's Neighbourhood Knowledge Management approach;@/UL@

@UL@(c) Other data available to individual services.@/UL@

We are also now developing the use of the MOSAIC data-base, which uses a number of private sector data-sources to map population characteristics, and our own client index which is in the process of being implemented and will allow real time data matching.

5. Our principal concern is not how the council uses the ONS mid-year estimates—we don't—but how the government uses them, particularly in funding decisions. Our best estimate is that the ONS estimates currently under-count Brent's population by in excess of 17,000, based on the data matching exercise carried out by Professor Mayhew. The local government Formula Grant system allocates approximately £500 for each additional person in the borough. A difference of 17,000 therefore amounts to an estimated £8.5m loss in the needs element of Formula Grant. The interaction of the needs element of Formula Grant and the “grant floor" mechanism means that the impact on actual grant is significantly less (in practice, Brent's Formula Grant is so far below the “grant floor" that even if the population were to be correctly measured, the likelihood is that the council would still be at the “grant floor"). It does mean however that in the longer term, unless ONS estimates are improved, the council will continue not to have its true population reflected in Formula Grant.

6. Formula Grant is not the only grant that central government allocates to local government. There is often less transparency in the way other specific grants are allocated to councils and therefore it is unclear how much the council might be losing as a result of government using underestimates of Brent's population to allocate grant. It is expected that this lack of transparency will continue for those specific grants replaced by Area Based Grant. Population figures are also used to fund other local services such as health. The distortions in funding therefore apply across the range of local services.

7. There are other ways in which central government uses ONS data which can lead to distortions. Performance measures, for example, often use population as the denominator. This means that there could be significant inaccuracies in data used by government to measure performance of different councils, for example, in disposing of household waste or data used by the Audit Commission to compare unit costs of services.

OTHER DATA SOURCES THAT CAN BE USED8. We believe that there is a mass of administrative data which could be used to provide regular and

up-dated population estimates. If used well, these data would produce more accurate and useful information than ONS mid-year estimates.

9. The report by Professor Mayhew sets out a number of data sources he has used with details of their relative usefulness.37 These are as follows:

@UL@(a) GP register;@/UL@

@UL@(b) Register of births and deaths;@/UL@

@UL@(c) Electoral register;@/UL@

37 Annex D in the report by Les Mayhew and Gillian Harper.

@UL@(d) Pupil data-base (PLASC);@/UL@

@UL@(e) Housing waiting list (HWL);@/UL@

@UL@(f) Housing benefit database;@/UL@

@UL@(g) Council tax database;@/UL@

@UL@(h) Local property gazetteer.@/UL@

10. Brent Council is currently developing its own client index which will give real time matching of its own data sources including the above data sources (except GP registers) and others held by the council eg people who use libraries and sports facilities, social care clients, users of building control and planning services, and so on.

11. Other government departments also hold data on individuals which, if combined in an overall data-matching exercise with local government and health data, would give a very complete picture of the population—for example, child benefit, pensions, tax, national insurance, and so on.

CONCLUSION12. Our evidence suggests that there are serious inaccuracies in the ONS estimates which have not been

corrected by the revisions made to the methodology in April 2007 and reflected in the revised estimates published in August 2007. Part of the reason may be that the Census itself is not 100% accurate. But the main reason appears to us to be that the ONS are failing to measure migration accurately.38 This has led to increased mistrust locally in ONS data and use of alternative data sources on which to base strategic decisions about community and service development locally. At a national level, use of ONS data is distorting government funding decisions and also provides a poor information base on which to make other national policy decisions. We believe that the government should consider radically different approaches to measuring population which might produce more accurate and useful data than current ONS estimates.

/home/PKB/DATA/386539/FOLIO146.EPS>

/home/PKB/DATA/386539/FOLIO147.EPS>

/home/PKB/DATA/386539/FOLIO148.EPS>

/home/PKB/DATA/386539/FOLIO149.EPS>

/home/PKB/DATA/386539/FOLIO150.EPS>

/home/PKB/DATA/386539/FOLIO151.EPS>November 2007

@HR25@

38 We have not set out reasons in this submission for the migration data used by ONS being inaccurate. However it is our view that a major reason for the inaccuracy is that the International Passenger Survey and Labour Force Survey are not suitable for identifying population movement down to local authority area for a variety of reasons, including sample size and data collection methodology. We believe that this problem is reinforced by the failure of ONS to record first onward moves where international migrants have not registered with a GP at their original destination.

Memorandum from Slough Borough Council

SUMMARYMigration has had a major impact at a local level in Slough. Councils have long been at the forefront of

managing these impacts in terms of local economies, community cohesion, cultural differences and continuity of service provision—not just in recent years.

What is different in the current waves of migration to those gone past is not so much in scale and speed of migration (although this has been unprecedented since EU expansion in 2004) but in the failure of statistics to keep pace and reaction of central government to the management of the impacts resulting from migration.

Slough Borough Council has been responsible for implementing innovative services, conducting research and community cohesion programmes to meet the challenges of the impacts of migration since the turn of the century. The town has successfully integrated migration since the 1920s.

History of migration:

@UL@— 1920 migration from Wales, the North, Scotland, Ireland and the South to take advantage of growing local economy.@/UL@

@UL@— Polish migration in the second world war (130,000 Poles arrived in Britain during the first few years after the war).@/UL@

@UL@— Windrush and Indian subcontinent immigration in 50s, 60s and 70s.@/UL@

@UL@— Current EU A8 and A2 migration.@/UL@

As a result the town has the most diverse ethnic mix outside London with 37% of the population being classed as Black or Ethnic Minority, at the last census.

In a report entitled Colour and Communities examining the effects and impacts of migration from Indian sub-continent and the West Indies in 1964 many of the impacts highlighted then mirror those being faced by the council today.

@UL@”The fact remains that Slough has a successful history of accepting large numbers and great varieties of people from other places. Past experience has demonstrated Slough's ability to assimilate newcomers without severe or lasting difficulties, and in virtually all cases the newcomers have added something of genuine value to the life of the town. Perhaps it is not too unrealistic to think this could happen once more..." William H. Israel, Council Officer reporting on migration in 1964.@/UL@

There are many direct impacts resulting from significant migration to a local authority area such as pressures to schools coping with the influx of new intake, labour market distortions, housing, policing, fire prevention and refuse collection. However, the most significant impact remains an indirect one resulting from poor management systems within the UK and an unresponsive centrally controlled mechanism for local government funding.

Poor measurement of local populations in “hyper-diverse" communities has serious consequences to the management of migration at a local level. The failure to track migration accurately to and within the UK results in inaccurate population estimates with corollary under funding for those areas where official statistics prove inadequate. This in turn results in pressures to all services, community cohesion risks and poor central and local planning of services. The council has published incontrovertible evidence of a population undercount in the town due to the inability of official statistics to measure current migration movement accurately.

It is tempting in the debate around impacts of migration to look solely at the visible effects on the ground. Although this is important they should be viewed with consideration of the indirect impacts of poor management of data and statistics.

The following evidence illustrates these points which results from a series of independent studies and seminars the council has conducted into the impacts of migration to the town since 2002 (see Appendix A).

1. USES AND DEFINITIONSWhat are the uses of population estimates, and how far do current population estimates meet the needs associated with those uses? What are the effects of inaccuracies or inadequacies in such estimates?

Population estimates are not meeting the needs of users. It is now widely recognised that the methodology by which mid year population estimates are compiled have not kept pace with significant migration to the UK. As a result central government users are relying on outdated and inadequate information on which to base formula grant funding for local authorities and to determine policy.

At the 2001 Census Slough's population grew by 14% to 120,600. However, census returns were low at 86% and there is a possibility this growth underestimated the true population of Slough

Between 1991 and 2001, Slough experienced the ninth fastest population increase of any local authority in the country. The official population of Slough is now reported to have decreased since the 2001 census during a period when the town has seen a significant increase in migration. This is due to inadequate measurement of migration used in mid year estimates since the 2001 census

Mid Year Estimates are determined by data for the rate of births, deaths and net migration in local authorities. According to the ONS Slough's population has declined considerably since 2001 by 4,000 people or 3.3%. The 2004 mid year estimates puts Slough's population at 117,600

The ONS estimates there were only 300 net international migrants to Slough in 2004. National Insurance numbers issued in Slough for the 18 month period between 1 April 2004 and 30 September 2005, broken down by nationality show that of the 9,048 new national insurance numbers issued in Slough over an 18 month period to October 2005 only 150 (1.7%) were for British citizens. By contrast, over 41% were received from Polish and Dutch (Somalian) nationals.

Slough's local data sets such as electoral register, housing stock, school rolls and service pressures all suggest a strong growth in population. The council estimates the true population to be between 125-135,000.

Slough Borough Council commissioned the Greater London Authority Data Management and Analysis Group (DMAG) to conduct a study of its population compared to ONS statistics. The analysis found that when the DMAG team's figures and the ONS projection for 2006 are compared the ONS figure could be underestimating the actual population level by as much as 7,000 people

The Office for National Statistics uses methodology to determine migration numbers that simply is not fit for purpose. The International Passenger Survey is conducted at a handful of ports of entry and samples only 0.2% of what is believed to be all passengers arriving in the UK. (250,000 people).

Of these 250,000 interviews only 3,100 were identified as migrants in 2003 (1.2%). The numbers are then apportioned to local authority level using formula based analysis. The room for error is enormous.

Warnings have been constant and consistent about the need to improve migration statistics:

@UL@”Shortcomings in measuring migration, for instance, led to difficulties in getting robust results from the 2001 Census, the main source for allocating over £100 billion to local authorities and the NHS each year. Work by the Statistics Commission three years ago indicated that there was very wide agreement across Whitehall that statistics relating to migration, both internationally and within the UK, need urgently to be improved. But thus far we have seen little evidence of real progress being made." Professor David Rhind, Chairman Statistics Commission in a letter to Tony McNulty MP Home Office Minister 8th May 2006.@/UL@

Local government funding

Slough estimates it will lose £15million in grant up until the next census in 2011 because of inaccurate population estimates. This is only the tip of the iceberg as the authority will also lose funding for the growth of our population.

Ministers may argue that all authorities are guaranteed a minimum percentage increase in general grant, and that this should protect authorities, including those that have population under estimates. This response, however, is inadequate. The guaranteed general grant increase for Slough, an area we believe evidence shows is growing rapidly,

would be no higher than other areas where population is not growing. The expected increase in grant for the next spending review period will not be sufficient to respond to growing pressures on the range of local authority services

Slough Borough Council has asked for a specific grant to be made available for those areas that are suffering an undercount due to inadequate population data. This could be constructed using proxy data such as local data sets (electoral register, school rolls etc.), the Workers Registration Scheme and National Insurance Numbers.

If this does not happen, before three-year grant settlements are decided later this year, the Council will be inevitably experience a further period of under funding which is likely to result in negative community cohesion issues.

Chairman of London Councils Councillor Merrick Cockell, recently said: “The government is still not taking this issue seriously enough. We must get things right before the three year funding settlement kicks in. These inaccurate population counting methods used by the ONS do not reflect the actual patterns of migration in London. The government needs to get them up to scratch quick or London's councils will be locked into an inadequate finance settlement for the next three years. This would put London's already stretched council services in danger of being severely overburdened."

2. THE ROLE OF THE CENSUS@UL@— How does the Census contribute to the creation of population statistics? Is the current frequency of Censuses appropriate and, if not, how frequent should Censuses be?@/UL@

@UL@— What lessons have been learnt from the Test Census carried out on 13 May 2007? What methods should be employed for the distribution and completion of Census forms, including the use of the Internet?@/UL@

@UL@— What steps should be taken to increase the rate of responses to the Census, particularly in hard-to-reach areas? How effective are plans to measure the extent of non-response to the 2011 Census?@/UL@

@UL@— To what extent is there a trade-off between the length of the Census form and its role in providing population information? What questions should be included in the 2011 Census?@/UL@

@UL@— To what extent should Census questions and statistical information be coordinated across the United Kingdom and how effective are plans to ensure such coordination?@/UL@

@UL@— What forms should the outputs of the 2011 Census take, how and when should they be made available and how should they be integrated and coordinated with other information on population?@/UL@

The council is concerned that the census will be able to cope with enumerating its complex communities. This concern is heightened by the fact that the ONS has not used Westminster City Council as a test authority citing: “our methods might be sufficiently good enough for more typical cities".

We strongly believe that the 2001 census did not accurately enumerate the Borough's population. As a consequence, any methodology to assess inter-census populations using the 2001 census as a base have failed the Borough. We believe the allocation of migrants using the town's 2001 census as a baseline has further exacerbated the population undercount.

The council will commit to making sure that the census has the best opportunity of success in the town through extensive local preparations and by helping the ONS wherever possible. However, there remain serious concerns over how enumerators will cope with the range of languages in the town, overcrowded accommodation, long census forms, imputation of records in hyper-diverse communities and the ONS' current funding and recruitment issues.

If mid year estimates are not significantly adjusted between now and the 2011 census we believe confidence in the census and official population estimates could be fatally damaged.

3. MID-YEAR POPULATION ESTIMATES@UL@— How accurate and useful are the mid-year population estimates for England and Wales, including information available for local authorities and Strategic Health Authorities?@/UL@

@UL@— How appropriate is the methodology by which mid-year population estimates are reached?@/UL@

@UL@— What are the most appropriate ways to ensure a common level of accuracy in mid-year population estimates for different areas and what steps should be taken to improve the accuracy in areas where population is hard to measure?@/UL@

@UL@— What is the right process for making revisions to mid-year population estimates so as to meet the needs of users?@/UL@

Housing

The fast pace of the changing patterns of our communities has created more pressure on space and resources and is significantly stretching the council (Slough was the 2nd most overcrowded town in the country at the 2001 census).

The council has evidence of a rapid increase in the numbers of households of multiple occupation (HMOs): Slough's own database now holds in excess of 1,050 records of small HMOs. Officers calculate that it will cost over £400k of new funding to license Slough's HMOs in line with the new Housing Act 2004 requirements.

“Sheds with beds" (as described in a recent Panorama investigation Migration: How we lost count July 2007) have been constructed in the back gardens of properties in Slough. These temporary structures house large numbers of migrants on their arrival to the town. These structures do not have planning permission for residential use. The council needs specific funding in order to deal with the enforcement of regulations and with the consequences of “closing down" these inappropriate and dangerous forms of accommodation.

However, because the government use flawed population data in its own projections of housing need the official estimates for Slough show some perverse results. The Department for Communities and Local Government (DCLG) has published39 projections of households for England and the Regions up to 2029.

The figures for Slough, when compared with the England, London and South East totals are surprising and disappointing. According to the latest DCLG figures:

@UL@— The number of households within Slough is projected to remain static from 2004 to 2016.@/UL@

@UL@— The number of households within Slough will only increase by 3,000 between 2004 and 2029.@/UL@

According to the DCLG projections, Slough will experience the 7th lowest increase in its household population between 2004-2029 of any local authority in the country. Slough is the only Unitary Authority in the Country projected to receive a 0% increase in its number of households until 2016. Only five40 other authorities are in a similar position.

These housing projections show a complete disregard of the reality on the ground of the impacts of migration to places such as Slough. As more “sheds with beds" appear in the town central planning using flawed data suggest the town needs zero extra households. As such the concerns of the council and bids for extra resource have fallen on deaf ears.

“But as DCLG has had no evidence from the ONS that the population estimates in Slough or any other similar authority are under-estimated, there is no objective case for considering a specific grant." Letter from Phil Woolas, Minister for Local Government, to Cheryl Coppell, Chief Executive SBC on 22nd June 2006.

Education

The council has responded innovatively to the sharp increase of children new to the UK accessing our schools. It was proving increasingly difficult to place new migrant children, particularly at secondary level, into our schools as a lack of recorded educational need of the children meant they were largely an “unknown quantity". It is essential to place children in an education stream where they can succeed and where existing pupils will not be disrupted. In response to this and in partnership with our head teachers we established the Slough Assessment Centre

39 http://www.communities.gov.uk/index.asp?id=1002882&PressNoticeID=237440 Teesdale, Christchurch, Spelthorne, Burnley, and Isles of Scilly

at Beechwood School at which new migrant children spend time having their educational needs assessed before mainstreaming.

During the summer holiday period 2006, 89 secondary aged pupils arrived in Slough from other countries. The Assessment Centre can only cater for a maximum of 8 pupils at a time, although 4-6 provides a more effective teacher: pupil ratio.

The nature and mix of migration in an 18 months period is demonstrated on school rolls. 888 pupils from non English speaking countries moved onto Slough school rolls over that time. 200 were from Poland, 185 from Pakistan, 104 from Somalia and 91 from India. The remainder were from other African, European or other Asian countries. Two primary schools in Slough have had to take in 60 Somalian and 50 Polish children respectively in just one term.

The centre also provides support for newly arrived parents. Many families new to the country have complex needs and require a level of information, advice or support that cannot usually be met by schools. By consulting new arrivals and their parents about their experiences and needs, the Assessment Centre has been able to provide comprehensive and accessible information. The cost of this, not least translation expenses, is considerable.

This single centre costs £92,000 a year to run but is proving to be insufficient. The council estimates it needs additional staffing for the Secondary Assessment Centre to enable more pupils to be admitted during peak times of the year. Two Primary Assessment Centres are also needed to improve the admissions process for younger new arrivals, help them settle in more quickly to the English educational system and provide comprehensive, accessible information for parents.

Unaccompanied children

From 1 January to April 2007 it is estimated 400 newly arrived Romanian Roma individuals (including children) arrived in Slough. Families shared often overcrowded and sub standard accommodation.

Eighty-eight unaccompanied Romanian Roma children between the ages of 10-17 years presented to Slough children's services requesting support in the same period. A temporary dedicated team had to be set up by Children Services. The Roma team provided support to fifty-two children. Thirty-six were refused support as they have been assessed as being over the age of 18 or the council located family members in the UK.

All children claimed they travelled to the UK by Lorry, the amount paid for this varied, usually around 75 Euros. Most said they left family in Romania or other EU countries and came to the UK for a better life. Each child was individually assessed and children's services sought to re unite children with their families as soon as possible.

Six young people had babies of their own and seven were pregnant. A high proportion of children had dental problems, including small children who are dependants. Eight children had been implicated in criminal activity such as theft, mugging or begging. A small number of children had been seen cleaning car windscreens at traffic lights.

The rapid reaction team cost the council and estimated £250,000. This swift action to deal with an unprecedented and dramatic event was essential in maintaining public trust and managing impacts of a very visible migration pattern.

The management of this impact should be considered in the financial context of the council. The town's population is still considered by official statistics to be shrinking. The effect of this erroneous population count on the council's funding is severe with an estimated minimum loss of £5-6million in funding to date. The council has managed through four years of under funding by driving efficiencies and using balances. It can no longer do this making rapid reaction of this kind unlikely in the future.

4. THE ROLE OF SURVEY AND ADMINISTRATIVE DATA@UL@— What role can and should be played by survey data in the compilation of population statistics?@/UL@

@UL@— What role can and should be played by data from General Practitioners and other health service data in the compilation of population statistics?@/UL@

@UL@— What other data sources can and should be used in the compilation of population statistics?@/UL@

Slough Borough Council is a High Ethnicity Authority Special Interest Group (HESIG) member and as such share its concerns regarding the use of survey data in the compilation of population estimates.

HEASIG member authorities has concern about the use of the International Passenger Survey and the Labour Force Survey (LFS) in particular, in relation to its coverage of the population.

The LFS in particular suffers from significant weaknesses as a data source for allocating migrants, partly due to the small sample size of migrants in the survey and partly as HEASIG authorities generally have a much lower response rate compared to other regions. The ONS itself has identified that certain Household types are under-represented in the Labour Force survey as profiled below.41

Households underrepresented in the LFS include:

@UL@— Those rented from Housing Association or with job.@/UL@

@UL@— Those in converted or shared accommodation.@/UL@

@UL@— Households with three or more cars or vans.@/UL@

@UL@— Those comprising only one adult, aged 16-19.@/UL@

@UL@— Those with head of household aged 16-24.@/UL@

@UL@— Those with only single head of household.@/UL@

@UL@— Those with the head of household born in Commonwealth countries.@/UL@

There is also evidence, for example, that individuals in hostels and housing of multiple occupation —disproportionately more likely to be lived in by international migrants—may be under-sampled and that any follow up surveys with highly mobile individuals which are intended to improve the tracking of interviewees—may not be practical.

Similarly it is also apparent that the LFS sample size in relation to international migrants is lower than that from the International Passenger Survey which does not suggest that using this data source necessarily offers any improvements over the current arrangements. Moreover, many residents living in communal establishments are not covered by the sample, and the coverage of short-term or recently arrived migrants is not comprehensive and thought to be poor. The quality of the final survey estimates are, of course, ultimately dependent on the accuracy of the population data used for grossing the survey sample, which themselves are highly problematic, particularly in HEASIG member authorities.

HEASIG also has concerns regarding the robustness of the International Passenger Survey. The International Passenger Survey is conducted at a handful of ports of entry and samples only 0.2% of what is believed to be all passengers arriving in the UK (250,000 people).

Of these 250,000 interviews only 2,800 were identified as migrants in 2004 (1.2%). The number of emigrants leaving the country is determined on an even smaller sample of just 750 in 2004. For example, the estimates for Pakistan, quite an important source country, were based upon 231 interviews of immigrants and 6 interviews of emigrants. The estimate of a net 3,000 inflow from the Caribbean was based on the difference between 28 interviews in and 6 interviews out. While the ONS has since taken steps to improve sampling the results of these will not feed into the population figures used to distribute local government funding until 2011-12.

HEASIG believes that the ONS can make use of other data to inform their population estimates. For example, local authorities have records of council tax properties (including certain estimates of vacancy and second residences), new housing developments, houses in multiple occupation and children in schools. These sources could be compared as time series with ONS estimates to determine potential problems of consistency of estimates of change since 2001. While the electoral register does not cover all adults of overseas origin it may also be used as a comparator to estimated changes in the adult population. With regards to the number of migrants, the ONS can use

41 Source:—Labour Force Survey User Guide—Volume 1 : Background and Methodology 2003

the number of national insurance numbers issued to identify potential level of discrepancies between migration data based on GP registrations.

Although there has been some movement recently towards a plan of action to address the deficiencies in migration estimates the Office for National Statistics does not estimate it will bear fruit for between 5-10 years. 42

The council would like to see substantial investment in a radical solution to measuring migration rather than the current intermediate efforts using the Labour Force Survey to supplement the International Passenger Survey. The council considers this to be “tinkering" round the edges of a much more fundamental problem. The revisions have failed to address the undercount in Slough.

The ONS has recently published “improved" migration estimates using the Labour Force Survey to supplement the IPS. However, this is another flawed data set as it misses many migrants. It does not survey communal properties such as hostels and is a small sample size of the local authority areas. The results do not reflect what is happening on the ground.

The Labour Force Survey has been used to distribute migration around the UK. Using the new methodology means that Slough has actually seen a decrease of 800 net international in-migrants since 2002. However, the new figures show that less people are leaving the town making a net growth in total population of 1,100 people since 2002.

The council is also concerned that at a time when there is an obvious need for increasing funding to the Office for National statistics in order for it to cope with this challenge the organisation is losing experienced staff, relocating to Newport and undergoing tough funding restriction.

“It is ironic, and a source of concern, that the ONS can't produce statistics on its own performance. Both the ONS and HMRC need to start listening to their users and working with them to alleviate the problems—real and perceived—arising from the efficiency programme." Treasury Select Committee July 2007

Public and Commercial Services union general secretary Mark Serwotka said: “One of the central aims of the efficiency programme was that services wouldn't suffer, yet day in day out we are seeing backlogs and services damaged." Treasury Select Committee July 2007.

Slough Borough Council also supports the Statistics Commissions views that the use of survey data is not the answer to improving population estimates and that a more radical review of methodology is urgently needed.

“There is currently no satisfactory source which can provide the raw information, at national and local levels, that is required for statistical purposes. Sample surveys, of the kind typically carried out by statistical offices, are not the answer."

“There are no quick or cost-free answers. Government needs to decide whether the issue is of sufficient importance to justify the commitment of substantial resources to address the technical issues and statutory obstacles. Other countries have solved the problem through the use of integrated population and household registers. There is no reason why the UK could not do the same given time and resources." Statistics Commission Lords Economic Affairs Committee Written submission.

5. COOPERATION WITH STAKEHOLDERS@UL@— How effectively has the Office for National Statistics cooperated with stakeholders with an interest in and information relating to population statistics and how can cooperation be improved by the Statistics Board and by its Executive Office?@/UL@

Slough Borough Council has had regular meeting with the ONS concerning its issues with the mid year population estimates. Although the National Statistician has raised the issue of limitation of its data with government there seems to be a lack of accountability for the use of these “limited" statistics.

42 Interdepartmental Task Force on migration http://www.statistics.gov.uk/about/data/methodology/specific/population/future/imps/updates/downloads/Inter-departmental_Task_Force_on_Migration_Statistics.pdf

The council has concerns that there is a lack of consultation around the revisions put forward by the ONS. It is recognised that the ONS attends the Central and Local Information Partnership (CLIP) twice a year to discuss improvements to population estimates but this is failing to communicate in an affective way with a wider audience. As a result revisions are often presented as a fait accompli.

The council would like to see an extended consultation group of local authority users that are able to block amendments or revisions through request to the Statistics Board.

@ANAME@APPENDIX A

Evidence base:

1. Slough Seminar: There's no accounting for some people

@UL@www.slough.gov.uk/mycouncil/articles/11421.aspx@/UL@

@UL@Evidence presented by:@/UL@

@UL@— Peter Goldblatt, Director Centre for Demography ONS@/UL@

@UL@— Professor Les Mayhew, Mayhew Associates@/UL@

@UL@— Andrew Cooper, Populus Ltd.@/UL@

@UL@— John Hollis, Data Management & Analysis GLA@/UL@

@UL@— Professor David Rhind, Chairman Statistics Commission@/UL@

2. Slough Seminar: Migration Works November 2006 www.slough.gov.uk/mycouncil/articles/11421.aspx

@UL@Evidence presented by:@/UL@

@UL@— Tony Travers, LSE@/UL@

@UL@— Professor John Salt UCL@/UL@

@UL@— Danny Sriskandarajah, IPPR@/UL@

@UL@— Ted Cantle, Institute of Community Cohesion@/UL@

@UL@— Russell Pask, The Research Unit@/UL@

3. Populus qualitative research of new migrant communities July 2006

4. Slough migrant Study November 2006: Survey of new migrants, employers and stakeholders

5. Slough Community Cohesion Audit September 2006

6. DVD of Panorama documentary on Migration in Slough

November 2007

@HR25@

Memorandum from Islington Council

INTRODUCTION1. This document represents Islington Council's submission to the House of Commons Treasury

Subcommittee “Counting the Population" enquiry. The following sections consider the specific questions in respect of which the submission of responses and supporting evidence have been invited. We also note that London Councils and the High Ethnicity Authorities Special Interest Group (HEASIG) have submitted responses to the enquiry and we share the concerns raised in those responses.

2. The particular issues of concern to Islington can be summarised as follows:

@UL@— Inaccurate estimates—eg in-migrants and their adverse impact on population projections and grant funding@/UL@

@UL@— Failure to count part time residents and short-term migrants, or reflect the impact of migration on local services, which leads to significant cost pressures for Councils@/UL@

@UL@— The poor response rate in the last census@/UL@

@UL@— The use of Labour Force Survey data in population estimates which we believe to be flawed@/UL@

@UL@— Ensuring key data that impacts on local government funding is included in the next census—eg general population, ethnicity, employment, housing, education, languages and income@/UL@

@UL@— Making use of local data sources would be beneficial—eg National Insurance number registrations and property register data@/UL@SECTION 1—USES AND DEFINITIONS

What are the uses of population estimates and how far do current population estimates meet the needs associated with those uses?

3. Population projections are used by the government as a key indicator in the distribution of Formula Grant to Local Government. Islington Council receives approximately £150 million each year.

4. Another important use, from the Council's point of view, is in planning for long-term sustainable population growth, and estimating the land needed for new housing as part of current revision of Local Development Framework.

5. However it is highly questionable whether the existing estimates are fit for purpose, given the unexplained population reduction in the 2001 census, and the subsequent amendments leading to £1m additional grant in 2003-04. The Audit Commission report on economic migration also notes the problem of data flaws making planning more difficult.

What are the effects of inaccuracies or inadequacies in such estimates?

6. Inaccuracies and inadequacies contribute to adverse grant settlement outcomes, and localised cost pressures resulting from higher than expected migration, which has a potential impact on community cohesion. There is also an impact on local services such as schools, health services, translation services, social care planning and people without access to funds.

7. This is also an issue for planning and development control as errors in population data affect calculations used for estimating land needed for future housing developments. New homes targets are likely to be at odds with projected population changes. This is a big issue for the Council, particularly as we are updating our Local Development Framework.

How appropriate is a definition of the population based on the usually resident population in the context of the needs of the users?

8. Usually resident population data is not appropriate for dealing with the costs related to high numbers of visitors in certain areas. A particular issue is the new concessionary fares scheme, where more non-resident visits, as well as the number of eligible residents will influence the costs of this scheme. Significant numbers of second homeowners / weekday residents in some areas also mean that usually resident population is not an appropriate measure.

9. Non-residents also have a significant cost impact on services such as waste collection and street cleaning, although this is hard to quantify. Specifically Islington has an estimated daily commuter inflow of 58,500 (among the highest in London), a significant night time economy in the Angel / Upper Street area which impacts on licensing costs, and has attendances of up to 60,000 for Arsenal football matches at the Emirates Stadium which has associated street cleaning and other costs.

10. The usually resident population definition also excludes short-term migrants. This is a particular concern as significant service pressures arise as a result of short-term migration, eg costs of supporting short-term migrants without access to funds.

How does the quality and range of population statistics in England and Wales compare with that available in Scotland, Northern Ireland and internationally?

11. Without details of what is available elsewhere it is difficult to compare England and Wales with other countries. The main problem is the impact of quality and range on local government funding and on cost pressures as a result of perceived or actual shortcomings. In particular the quality and range of data used for estimating the in-migrant population, results in undercounts, and is a major concern for all London boroughs, as London receives a large proportion of the countries in-migrants. There is also a need for ward or neighbourhood data for planning purposes.

SECTION 2—THE ROLE OF THE CENSUSHow does the Census contribute to the creation of population statistics?

12. The key issue is the use of the census as a baseline for midyear estimates. Although no longer used directly for local government funding these do influence funding projections, therefore the flaws in census data are of concern.

Is the current frequency of Censuses appropriate and if not how frequent should Censuses be?

13. The main issue with the census is not the frequency but the quality of the data. If there are rapid changes in data that can be made readily available, such that it renders existing information inadequate, there may be a case for additional censuses if the benefit outweighs the cost.

14. However if the outcome of the census is flawed as a result of poor response rates or weaknesses in the methodology used, as was the case in 1991 and 2001, the value of the census itself may be questioned. It may be possible to gather and maintain information more effectively by other means in the longer-term, including national and / or local registration schemes, if introduced.

What lessons have been learnt from the Test Census carried out on 13 May 2007?

15. The lessons learned from the census test, and in particular from the Camden test, will be useful to us when they are published, hopefully by the end of the year.

What methods should be employed for the distribution and completion of census forms including the use of the Internet?

16. The main issue for us surrounds the management of the census, ensuring that the process is well co-ordinated, forms are issued, and low response rates are picked up quickly and addressed effectively. This was a major problem in 2001, when Islington had the 9th lowest response rate nationally.. Both electronic and manual methods are needed, while the former may be more convenient for many, manual approaches, including face-to-face contact, are important to ensure that non-respondents are followed up.

What steps should be taken to increase the rate of responses to the Census, particularly in hard to reach areas?

17. There is a need to involve local councils in planning and make use of local knowledge. Councils can use council staff, publicity, community groups and commit resources to help improve response rates, although some help from the government would be desirable. This will allow areas and population groups where poor response is a problem, to be targeted effectively.

How effective are plans to measure the extent of non-response to the 2011 census?

18. It is not clear exactly how this can be measured. However the Council maintains a local land and property gazetteer, which is regularly kept up-to-date and can help in identifying non-respondents. It is unfortunate that a good quality land and property register like the one in Islington does not exist at national level.

To what extent is there a trade off between the length of the Census form and its role in providing population information?

19. We would expect there to be a trade off. There is a need to strike a balance between the two and focus on questions that provide most the useful information and add value to the exercise.

What questions should be included in the 2011 census?

20. The most important questions are those that provide information that is used for calculating government funding allocations, such as population, ethnicity, employment, housing, education and qualifications and travel to work data. More information on other factors relevant in measuring deprivation, which may have an impact on local cost pressures, would also be useful. Questions used in the 2007 test covering income, languages spoken and housing would be particularly welcome.

To what extent should census questions and statistical information be co-ordinated and how effective are plans to ensure such co ordination?

21. Ideally there should be as much co-ordination between the census and other statistical information sources as possible, with government departments, the Office of National Statistics (ONS) and local councils working together to ensure that this happens. In particular there should be co-ordination between the census and comparable surveys such as the household survey and labour force survey.

22. There is good cross working between the Greater London Authority (GLA) and the ONS though, it is less clear whether co-ordination with local government is currently effective.

What forms should the outputs of the 2011 census take, how and when should they be made available and how should they be integrated and co-ordinated with other information on population?

23. Outputs should include regional, authority, ward and locality level data. They should be made available on as a timely basis, consistent with the accuracy specified by users. In London, the provision of data to GLA statistics group and borough corporate planners is crucial, though circulation to individual Councils is also desirable. The outputs must be integrated with estimates and projections, and should be linked with local property databases and relevant health service and central government databases.

SECTION 3—MID-YEAR POPULATION STATISTICSHow accurate and useful are the mid-year population estimates for England and Wales, including information available for local authorities and strategic health authorities?

24. The Council has had serious concerns in the past about mid-year population estimates. The adverse impact of initial estimates on the Council's funding projections, after the 2001 census has only been partially addressed by subsequent revisions. The recently “improved" mid-year estimates remain flawed. Modification of estimates using the labour force survey has had an adverse impact in London and, for example, reduced the population in Islington by 2000 people despite local evidence to the contrary.

25. Studies in Westminster have highlighted the subdivision of private rented property and even the illegal subletting of council property and residential use of commercial premises, not picked-up through official population estimates. This is believed to be an issue in Islington too.

How appropriate is the methodology by which mid-year population estimates are reached?

26. The methodology for allocating in migration in particular, remains flawed, and the apparent reduction of in migration to London as a whole is implausible when compared to local evidence such as new NI numbers issued.

What are the most appropriate ways to ensure a common level of accuracy in mid-year population estimates for different areas and what steps should be taken to improve the accuracy in areas where population is hard to measure?

27. There is a need to ensure that the methodology used is soundly based and that no bias exists. There should be a comparison with other data sources and local evidence to check that the methodology is intuitively as well as statistically sound. It should be possible to make reasonable adjustments where mobility and other localised factors have an impact.

What is the right process for making revisions to mid-year population estimates so as to meet the needs of users?

28. All changes should be evidence based, and their timing should be linked to revision deadlines, such as the 3-year local government finance settlements. Changes to projections should be allowed in exceptional cases so that amending reports to Formula Grant can be issued if required. Revisions could also be a basis for exceptional special grant payments to meet cost pressures. The Local Government Association estimates that £250 million is needed to address local cost pressures, such as translation services, community cohesion, housing, and schools, along with the impact of short term and illegal migrants not reflected in existing estimates, as well as high levels of population mobility.

SECTION 4—THE ROLE OF SURVEY AND ADMINISTRATIVE DATAWhat role can and should be played by survey data in the compilation of population statistics?

29. Use could be made of data sources such as patient registers, National Insurance (NI) numbers issued, council tax and property registers, and student [enrolment?]data. However all data sources should be robust. The Council has concerns about data sampling and coverage, for example with the labour force and passenger surveys. In particular the passenger survey sample, although being improved, is still quite small. The labour force survey does not adequately cover communal establishments, for example, which is also of concern.

What role can and should be played by data from General Practitioners and other health service data in the compilation of population statistics?

30. General Practitioners data is a useful source, however while initial in-migration is picked up, subsequent moves are not. There is a problem of list inflation varying from place to place, and in areas such as inner cities not all in-migration is picked up as GP lists are often full.

What other data can and should be used in the compilation of population statistics?

31. NI number registrations and property register data are both useful sources of data. The latter picks up unauthorised property conversions and / or divisions which could lead to a higher population. The former is an indication of actual population increases, locally based on the level of new NI numbers issued.

SECTION 5—CO-OPERATION WITH STAKEHOLDERSHow effectively has the Office for National Statistics co-operated with stakeholders with an interest in and information relating to population statistics—and how can co operation be improved by the Statistics Board and its Executive Office?

32. Historically there has been good co-operation with certain stakeholders such as the GLA demography team, but less good co-operation with local councils. There has been consultation with councils regarding the 2011 census and improvements to mid-year estimates, although it is debatable whether concerns in London have been taken on board. There is a need for proper consultation with councils and council representation in the decision making process, in particular London Councils and LGA involvement is key.

OTHER ISSUES33. Population projections are one of the most important issues for the Council, given their key role in

government grant allocation.

November 2007

@HR25@

Memorandum from Wandsworth Council

USES AND DEFINITIONSWhat are the uses of population estimates and how far do current population estimates meet the needs

associated with those uses? What are the effects of inaccuracies or inadequacies in such estimates?

1.1 Population estimates are used extensively within local authorities to plan services and allocate resources at the local level. Estimates are essential for local authorities to understand the demographic composition of the population they serve and provide services tailored to the needs of local residents. Estimates and projections based on them are also important for identifying trends and changes in the population structure over time in order to plan future service provision. Population estimates feed into all areas of service delivery including health, education, social services, transport planning, infrastructure provision, waste and leisure services, tackling deprivation and crime and strategic needs assessments. Inaccuracies in population estimates and projections based on them impede a local authority's ability to provide and plan services to meet the needs of local communities.

1.2 Estimates which use the same base data and are comparable across local authorities are essential to the fair allocation of resources nationally. Office for National Statistics (ONS) estimates are used as denominators for local performance indicators (eg crime, unemployment) at both local authority and sub-local authority level and are a key component of the Indices of Deprivation, which are used to bid for funding. Sub-national population projections incorporate ONS estimates and any inaccuracies in the population estimates will therefore translate into the projections, which are used by Communities and Local Government (CLG) in the allocation of the local authority grant settlement. Inaccuracies in population estimates and projections can deny local authorities access to funding to which they may be entitled and prevent the targeting of resources to areas in need of investment.

1.3 The components of change (births, deaths, internal migration and international migration) that determine the year-on-year change in population estimates are also useful to local authorities. Births and deaths data feed into local health, education, registration services and cemetery plans. We have no concerns over the accuracy of estimates of natural change. Migration (both internal and international) is a key component of change in the population of a local authority. For areas of high mobility, the scale of annual turnover in population can place a large burden on local authorities. Large demands are placed on administrative functions (eg electoral registration, council tax, GP registration) where change is high and planning for future service provision, eg health, education, social services, is more difficult than for relatively static populations. In addition to causing problems for determining the overall population estimate, inaccuracies in the migration components of population estimates also potentially lead to under-estimation of the rate of change in the population within a local authority, which has significant implications for areas of high turnover. Furthermore, inaccuracies in the international migration component of population estimates have implications for funding additional services and adapting service provision to meet the needs of migrants. It is well documented that international migrants can place additional demands on local authority services such as education, housing and translation.

1.4 Current population estimates fail to meet the needs associated with the uses outlined above where there are doubts over accuracy of the estimates of both internal and particularly international migration. Confidence in the estimates and their ability to facilitate fair allocation of resources is low at present, particularly since EU Accession. The experiences of local authorities on the ground are not reflected in the statistics and many consider the ONS population estimate for their authority to be significantly lower than the true population.

How appropriate is a definition of the population based on the usually resident population in the context of the needs of the users?

1.5 The definition of usually resident population is inappropriate in the use of population estimates for local authority service planning. Many international migrants are not counted in the official estimates as they may indicate that they do not intend to stay in the country for more than 12 months if captured by the International Passenger Survey (IPS). In practice, there is often a continual churn of short-term migrants, using the same services (but with more changes) as someone who is “usually resident". Short-term migrants are an important element of the resident population and should be counted. In urban areas like London they are resident in greater numbers and need to be taken account of in local authority funding, particularly in areas of high turnover.

1.6 The recently published ONS experimental short-term migration estimates, although a welcome development towards estimating the true population, do not reflect the scale of migration in London that is evident from other administrative sources of data eg national insurance registrations. Further work is needed to improve these

estimates and the sample sizes of international migrants entering the country need to be improved considerably to place greater confidence in the data.

1.7 The definition of a “usual resident" excludes people from official counts if they are resident for only part of the week or part of the year. Similarly to short-term migrants, such part-time residents place demands on local authority services and impact on housing stress and transport systems. Second home owners can reflect a significant proportion of the total population in large employment districts and areas of tourism and second homes may be clustered together. In the 2001 Census, 15,800 household spaces in London and 600 in Wandsworth were recorded as second residence/holiday accommodation, equivalent to 0.5% of household spaces. The distinction between second residence/holiday accommodation and vacant accommodation for unoccupied household spaces was based on information provided by the enumerator. It is therefore possible that some second homes were miscoded as vacant. In a 2007 survey of new housing developments in Wandsworth, 69 (4%) of respondents stated that the dwelling surveyed was their second home and a further 28 (2%) of respondents stated that the dwelling surveyed was their main home and that they had a second home elsewhere.

How does the quality and range of population statistics in England and Wales compare with that available in Scotland, Northern Ireland and internationally?

1.8 The quality of population statistics is inferior to many European countries who have well established population registers which are linked to other administrative sources. However, it is accepted that population statistics produced by ONS are estimates and as long as the estimates are based on sound methodology and local authorities are happy that they reflect the position on the ground, estimates can be fit for purpose. The current methodology and data sources used by ONS in the production of population estimates have proved to be unfit for purpose. Urgent action is required to overcome the current deficiencies in the accuracy of the estimates.

1.9 The range of estimates produced by ONS are generally good and improving. The production of experimental estimates at geographies below local authority level (eg ward) is welcomed, together with the experimental ethnic group estimates. However, the macro-level estimates need to be accurate in order for estimates derived from them to be useful. The timeliness of these data releases could also be improved upon in order to facilitate the use of one mid-year estimate for all purposes at a particular point in time. For example, at present the latest available mid-year estimate is for 2006, whilst for the experimental ethnic group estimates, the latest data is for 2005.

THE ROLE OF THE CENSUSHow does the Census contribute to the creation of population statistics? Is the frequency of Censuses appropriate and, if not, how frequent should Censuses be?

2.1 The Census is the only comprehensive source of population data with detailed cross-tabulations of demographic information for small areas. In the absence of a population register, the Census provides a population base from which to model and estimate population in inter-Censal years and project change. Information on age, gender, migration and ethnicity can be used to provide further detailed estimates and projections.

2.2 A five-yearly Census would enable population estimates to retain closer contact with reality. However, this would be welcomed only if the depth, quality and coverage of the Census was maintained. The resource implications of a more frequent Census, together with declining response rates to all types of survey may be reasons to support the retention of a decennial Census. Resources could then be concentrated on achieving an accurate count of the whole population less frequently. Investment in improving the data sources used in producing the estimates of the migration, both internal and international, should facilitate greater accuracy of estimations over the inter-Censal years.

2.3 The Census is used as a base for population estimates produced by the ONS. Any problems with enumeration in the Census are therefore carried through to subsequent estimates, which can have significant impacts on the quality of population data over 10 years. The population estimates for Wandsworth have been revised several times since the 2001 Census to correct for Census under-enumeration in hard-to-count areas, an undercount of young men aged 25 and 34 and most recently as a result of work on estimating the number and geographical distribution of international migrants. The quality of the Census, particularly in terms of coverage and response, is therefore crucial to avoid the resulting uncertainty in population estimates and the need for revisions which can impact on both funding received and precision of local authority service delivery planning.

2.4 The quality and coverage of the 2001 Census differed significantly between different areas of England and Wales. The ONS needs to adapt a more flexible approach for 2011, concentrating resources in those areas which are hardest to count. A 100% Census nationally would be the only way to ensure a sound base for inter-Censal population estimates. A one size fits all approach is not considered appropriate to achieve this, as some areas have a high degree of stability whilst others have a very mobile population and large migration flows. These factors need to be considered in conjunction with identifying other traditional hard-to-count areas eg areas where deprivation or language act as barriers to participation.

What lessons have been learnt from the Test Census carried out on 13 May 2007? What methods should be employed for the distribution and completion of Census forms, including the use of the Internet?

2.5 The test of posting-out Census forms produced some positive results and this delivery method should be adopted for the 2011 Census, with field staff concentrating on hard-to-count areas. However, the address list issue still needs to be resolved or there will be under-enumeration as in 2001, particularly in urban areas. For local authorities with a good Local Land and Property Gazetteer, this should be used as a base as many local authorities invest significantly in their Gazetteers and place confidence in them. In Wandsworth, the OS address file has proven to be frequently inaccurate and it is considered the use of this product would greatly undermine the validity of the Census results. The quality of the address list will directly impact on the quality of the Census and local authorities should have full access to the address lists to be used in order to perform extensive address checking and cleansing where necessary. The ONS should utilise local knowledge and address systems. This also applies to identification of communal establishments and gated developments as electoral registrars have built up considerable experience of the local context.

2.6 Any internet option would need to provide value for money and undergo thorough testing. As the cost is likely to be considerable, the money may be better spent on a fourth page of questions. Each household would still need to receive a paper copy of the form.

What steps should be taken to increase the rate of responses to the Census, particularly in hard-to-reach areas? How effective are plans to measure the extent of non-response to the 2011 Census?

2.7 The Census form itself should stand out from other “official" mail and advertising is crucial for residents to understand why their return is important, how it will benefit them and their local area, and also how the information will be used. There is generally increasing apathy towards surveys, concern over confidentiality and fear of “linked systems" for those who do not want to be identified. A clear and simple message needs to be relayed that reassures people and encourages them to complete and return the form. Engaging with schools in this process would be highly beneficial, particularly for some hard-to-count areas. The help-line should also have sufficient capacity to deal with demand. The problems with the help-line in 2001 should not be repeated.

2.8 The proposed bar-coding of questionnaires and response envelopes should facilitate the tracking of responses and enable follow-up field staff to concentrate where most needed and this is welcomed. However, the problems of 2001 must not be repeated whereby enumerators did not have enough time to follow-up, particularly in hard-to-count areas.

2.9 The definition of hard-to-count areas needs to include areas with a high proportion of young, single men (aged 18—35). A review of Census outputs following the 2001 Census led to substantial revisions to the population estimates in Wandsworth and other local authorities with a significant proportion young males, due to a recognised under-count. The hours worked by field staff in these areas would need to mirror the days/hours most likely to find young men at home.

2.10 Hard-to-count areas should also include those with high numbers of apartments. There was concern in Wandsworth at the last census that the number of such properties coded as vacant, particularly in social housing blocks, exceeded the numbers we knew from our administrative records. This suggests that enumerators failing to get a response chose to code properties as vacant rather than continue to try to get a response. The Census enumerator payment system could perhaps be structured to reward persistent enumerators rather than those who code high numbers of properties as vacant.

2.11 With an accurate address list and sufficient follow-up resources targeted in the right areas, the Census should be aiming for 100% response.

To what extent is there a trade-off between the length of the Census form and its role in providing population information? What questions should be included in the 2011 Census?

2.12 In the Census test, a four-page questionnaire did not significantly impact on response rates compared with a three-page questionnaire. The opportunity to gather such detailed information about the population currently only presents itself every 10 years. This opportunity should be utilised to ask all questions for which there is a demonstrated need for information. The questions at risk include language, carers, qualifications and industry. The loss of these topics would be very regrettable as there is no other data source available and with the exception of language (a welcome new addition), they are questions that have to date provided detailed information for local authority service planning and provision.

2.13 A decision is still to be taken over the inclusion of a question on income in the 2011 Census. Initial results from the 2007 test show only a 2.9% difference in response rates between questionnaires that included a question on income, and those that did not. If the question were to be made voluntary as the Religion question was in 2001, the issue of impact on response rate might be removed. Wandsworth Council conducts regular surveys of new housing developments. The questionnaires include a question on household income and taking part in the survey is voluntary. In the 2004 survey, 90% of over 2,000 respondents answered the question on income. This rose to 93% of 1,900 respondents in 2007.

2.14 Income data would be invaluable for local authority service planning and would facilitate the identification of concentrations of poverty and affordable housing need. The provision of actual data on income rather than proxy data for cross-tabulation with other variables is important at the local level, particularly for identifying small pockets of deprivation, which may be hidden in densely populated areas such as Wandsworth.

To what extent should Census questions and statistical information be coordinated across the United Kingdom and how effective are plans to ensure such co-ordination?

2.15 Within England, it is important that the same questions are asked everywhere to maintain the provision of comparative data, particularly with regard to funding bids and the Indices of Deprivation. Whilst it would be preferable to co-ordinate the majority of questions asked across the UK, there is a demonstrated need for additional locally specific questions in other areas, for example Welsh speakers in Wales. The need for such data in one area should not compromise the collection of other, more relevant data in England.

What forms should the outputs of the 2011 Census take, how and when should they be made available and how should they be integrated and co-ordinated with other information on population?

2.16 Local authorities need detailed outputs at Output Area (OA), Super Output Area (SOA), ward, borough and regional level in order to plan services and undertake meaningful spatial analysis. The OA-SOA hierarchy has been confirmed by ONS as the stable geography to be used in 2011. Whilst this stability is welcomed, local authorities also have a strong need for ward level data and as some data sets were not published at SOA level in 2001, there would be no reliable comparator data for some outputs if ward level tables were not produced in 2011. Outputs at national level (England/England and Wales) are also needed for comparison and funding bids. Outputs need to be flexible and only rounded at the point of publication. The ONS needs to facilitate local authorities to amalgamate data to local areas without compounding errors from disclosure control. Local authorities, researchers, academics etc should have greater access to raw data products (via licence) than the public.

MID-YEAR POPULATION ESTIMATESHow accurate and useful are the mid-year population estimates for England and Wales, including information available for local authorities and Strategic Health Authorities?

3.1 National level population estimates are considered to have the highest degree of accuracy as the births, deaths and internal migration components of change are relatively easy to measure. There are, however, doubts over whether the IPS samples capture the true scale of migration into the country. It is also difficult to assess how many people leave.

3.2 At a local authority level there is far greater cause for concern, particularly for areas of high mobility (see paragraph 1.3) where GP registration data may not fully capture internal migrant movements and accurate information on internal migrants is currently non-existent.

3.3 Concern over the accuracy of GP registration data as a measure of internal migration particularly affects areas with a large young adult population like Wandsworth as this group (particularly young males) are most likely not to register with a GP. In a 2007 survey of new housing developments in Wandsworth, 270 (7%) of respondents were not registered with a GP and a further 630 (17%) were still registered with a GP at their previous address. 37% of males and 26% of females aged 20-29 were either not registered or registered at a previous address. For the 30-39 age group, this fell to 27% for males and 15% for females.

3.4 The degree of confidence in international migration statistics at a local authority level is particularly low at present as recent trends in international migrant inflows seen on the ground are not reflected in the statistics. The mobility of migrants has increased significantly since the Census, particularly post EU Accession and other administrative data sources such as National Insurance registrations reflect this. There is concern over the figures for total in-migrants nationally due to the sample size and coverage of the IPS. At a local level, there is further uncertainty over the figures as the sample size lacks robustness at a local authority scale and the data is modelled to smaller geographies. The IPS was not designed for the purpose of estimating migration and greater investment is needed to monitor cross-border flows more closely before e-borders becomes operational.

How appropriate is the methodology by which mid-year population estimates are reached?

3.5 The current methodology lacks the level of data needed to accurately measure migration (see paragraphs 3.3 and 3.4) and relies on data sources which are not fit for purpose.

What are the most appropriate ways to ensure a common level of accuracy in mid-year population estimates for different areas and what steps should be taken to improve the accuracy in areas where population is hard to measure?

3.6 An urgent review of the systems required to provide data should be undertaken to enable accurate estimates to be provided. Current plans to improve data on international migration are probably both inadequate and will be implemented too late to make a real difference for a considerable number of years.

3.7 In the current vacuum of reliable information, estimates at the local level need to be validated using other available administrative data sets eg pensions databases, school Census, national insurance registrations, GP flag 4 registrations, electoral register data. Methodology should be developed to alter the basic estimates for areas with large migrant populations and highly mobile populations in line with recent trends identified in administrative data sources. The DWP datasets which link NI numbers to employment, tax, residence, pensions, benefits etc should be investigated further as this data could be invaluable to population estimates.

3.8 There is a real need to concentrate resources in areas with high levels of change. To achieve a degree of standardisation of approach, ONS could supplement currently available data by conducting local surveys where relevant. The ONS should develop a flexible, consistent methodology to examine estimates in relation to survey and administrative data across areas of high turnover. Local authorities are currently conducting their own research into population estimates using available data sources eg national insurance registration data. A more comprehensive analysis of data across areas is necessary to inform the direction of population estimate methodology.

What is the right process for making revisions to mid-year population estimates so as to meet the needs of users?

3.9 Although numerous revisions are not desirable, it is better to try and improve the estimates at a local authority level than continue using inaccurate estimates. The ONS should review administrative data sources and validate the estimates, concentrating on area of high mobility and migration. Once their findings have been published and consulted on, a full set of revisions back to 2001 should be made.

THE ROLE OF SURVEY AND ADMINISTRATIVE DATAWhat role can and should be played by survey data in the compilation of population statistics?

4.1 One of the main problems with current population estimate methodology is the use of survey data to estimate and distribute the international migration components. There is concern over the figures for total in-migrants nationally due to the sample size and coverage of the IPS. At a local level, there is further uncertainty over the figures as the sample size lacks robustness at a local authority scale and the data is modelled to smaller geographies using Labour Force Survey (LFS) and New Migration Geographies (NMGs) which are based on Census data.

4.2 The LFS has coverage issues and relatively poor response rates in London and the sample size is not considered adequate to be used for this purpose. In addition, the recognised problems with under-enumeration in the 2001 Census in Wandsworth and other hard-to-count areas casts doubt on the validity of using Census data for geographical distribution of migrants.

4.3 The successful use of any survey data for population estimates is dependent on the accuracy of the population data used for grossing up the sample data. For local authorities who doubt their mid-year estimates reflect their true resident population, there is additional reservation regarding the use of such sample data to accurately measure population change.

4.4 Survey data should only be used directly within population estimate methodology if the survey is specifically designed for the purpose of measuring population change and the sample size is sufficiently robust to be confidently applied at a local authority level. This is not the case for either the IPS or the LFS. Survey data that does not meet these criteria should only be used for identifying trends and validating data from other administrative and survey sources.

What role can and should be played by data for General Practitioners and other health service data in the compilation of population statistics?

4.5 GP data is a valuable source of data for validating other data sets and identifying trends. However, there are problems inherent with it in terms of measuring actual stocks and internal flows due to delays in re-registration, particularly for young, single males (see paragraph 3.3). The quality of data is possibly better for older people and children who are more likely to be in regular contact with their GP.

4.6 However, in terms of “flag 4" registrations which record the initial registration of international migration, the data could be used more extensively in population statistics. This component of GP datasets has the potential to be used for better distribution of international migration in-flows to the local level. Further research is necessary to establish whether this data can be successfully incorporated within ONS methodology, but for areas like London where the LFS and NMG are not considered to be robust for distribution of IPS in-flows, large improvements could be made using this data.

What other data sources can and should be used in the compilation of population statistics?

4.7 Population estimates need to be validated against administrative data sources in order to ensure trends in population estimates during inter-Censal years are accurate at the local level. The ONS need to develop a methodology which can be applied to population statistics nationally to validate their existing data, but also with enough flexibility to treat areas with a high degree of change differently if necessary. Potential useful data sets include council tax (particularly to identify change in vacant properties, second homes, students and single people), new housing development, pensions databases, school Census, national insurance registrations, electoral register data (including European registrations) and DWP records.

COOPERATION WITH STAKEHOLDERSHow effectively has the ONS cooperated with stakeholders with an interest in information relating to population statistics and how can cooperation be improved by the Statistics Board and by its Executive Office?

5.1 Local authorities are bound to use official population estimates for grant settlements, funding bids and performance indicators. ONS mid-year estimates also underpin the Indices of Deprivation. It is vital that local authorities support the methodology used to derive the estimates and can therefore have confidence that they have access to all resources they are entitled to in order to best serve their residents. Presently, many local authorities do not have this confidence. The official population estimates have not kept up with the pace of change in population mobility since the 2001 Census, particularly with regard to international migration since EU Accession.

5.2 The ONS are relatively good at keeping users informed of their current thinking and developments/changes in population estimate methodology. However, they are less willing to accept criticism from data users and often cite that they are doing the best they can with the resources available. The importance of population estimates at the local level and their significance for funding and targeting services has only recently been recognised by the ONS, due to media coverage and political interest. However, local authority users of the data have raised concerns over the quality of population estimates repeatedly and for a long time now. If the ONS need extra resources in order to produce sound estimates of the population, they should secure Government investment to do this.

5.3 It is important that ONS continue to liaise with local authorities and other population data users and develop stronger links with statisticians and other experts who use the data. It is also important that they act upon the concerns of data users in order to improve methodology now, in order to back-correct mid-year estimates since the 2001 Census and avoid large discrepancies when the 2011 Census is conducted.

November 2007

@HR25@

Memorandum from Hertfordshire Census Users' Group

The Hertfordshire Census Users' Group is comprised of representatives from Hertfordshire County Council, the ten district councils, Hertfordshire Constabulary, Hertfordshire Health Authority, the Learning and Skills Council and Hertfordshire University

EXECUTIVE SUMMARYIn general, we support the observations of the Local Government Association, in their general points about

the role of the Census and Mid Year Estimates. We offer the following observations specific to Hertfordshire.

@UL@— An accurate census is essential for benchmarking population, household and dwelling characteristics, which enable monitoring of the annual population estimates against dwellings provision. @/UL@

@UL@— The 2001 Census provided a plausible estimate in terms of households and dwellings, when compared to our own counts of dwellings stock. From what we understand about the relation between dwellings and population, the population estimates from the 2001 Census are also plausible.@/UL@

@UL@— Greater flexibility in Census outputs would add value to the use of census data. This is being explored by ONS.@/UL@

@UL@— While it is difficult to measure the annual post-censal population estimates against other data, the trend in recent population estimates is plausible in comparison with dwellings provision.@/UL@

@UL@— Comparison of the recent trend in Mid Year Estimates with the coresponding trend in patients' register populations shows the patients' register rising more rapidly in Hertfordshire. This may be accounted for by differences in the definition of population. Patients' Registers have potential to be useful indicators of population counts and trends, with some caveats.@/UL@1. USES AND DEFINITIONSHow appropriate is a definition of the population based on the usually resident population in the context of the needs of users?

1.1 The definition is appropriate, provided that recent substantial changes in migration patterns can also be accounted for. ONS are themselves attempting to estimate short-term migration, but at present these estimates are only available at national level. Information obtained from administrative data about National Insurance applicants and the Workers Registration Scheme overlap with these definitions. Some local authorities have an additional burden placed on them by short-term visitors who are students and tourists, and these need to be considered separately, or in a range of population outputs based on different definitions.

2. THE ROLE OF THE CENSUSHow does the Census contribute to the creation of population statistics? Is the current frequency of Censuses appropriate, and if not, how frequent should Censuses be?

2.1 AT BROAD GEOGRAPHIC LEVELThe census provides a benchmark for population every ten years, but it also provides a view of the relation

between dwellings, households and population which helps to monitor the estimating procedures between Censuses. This is available chiefly through statistics such as the distribution of household types, the average household size, and the average occupancy rate of dwellings. As discussed below, the expected relation between dwellings and populations is one of the keys to assessing the plausibility of population estimates between the censuses.

2.2 At fine geographic level

The census is the sole nationally consistent source of statistics on population, households and dwellings and their characteristics at local level.

What steps should be taken to increase the rate of responses to the Census, particularly in hard-to-reach areas? How effective are plans to measure the extent of non-response to the 2011 Census?

2.3 Plans to measure the extent of non-response and to adjust for it were first made in the 2001 Census through the Census coverage survey. These were inadequate, since they assumed independence of the Census and the

Census Coverage Survey. Hence it is essential to maximise Census coverage, but also to measure and adjust for under-coverage. To do this, it is crucial that the census engages local communities, and field officers who have local knowledge, and commitment to the successful outcome of the census. This seems more difficult to achieve when out-sourcing the recruitment of field staff. The quality of address registers is crucial in the enumeration process, and improvement is needed to ensure that these are consistently completed to a high standard by the time of the next census.

What forms should the outputs of the 2011 Census take, how and when should they be made available and how should they be integrated and coordinated with other information on population?

2.4 Census outputs have historically been regarded as a series of pre-formatted tabulations, and in the 2001 Census they were made available widely by internet and other sources. This led to changes in the procedures for disclosure control late in the Census planning phase. As a result, some census outputs were rendered almost unuseable, particularly at local level. There is scope to look towards different ways of making census data available to different customers. In particular there is a need for specialist users to be able to design their own tables to be run on census data. This would not necessarily require individual census records to be made available outside the Office for National Statistics (ONS), although this could be managed under licence as other ONS data is. ONS have consulted on the delivery of output during 2007. These provisions would increase the value that could be derived from the census data.

MID YEAR POPULATION ESTIMATESHow accurate and useful are the mid-year population estimates for England and Wales, including information available for local authorities and Strategic Health Authorities?

3.1 This is a difficult question to answer from a local authority's perspective, because there are no established alternative statistics. Herrtfordshire County Council (HCC) have used local population surveys, the Health Authority's Patient Register, and dwellings-based estimates of population change to try to monitor the Mid Year Estimates. All have had limited success.

3.2 Local Population Survey

In Hertfordshire in 1997, HCC conducted a local survey based on the electoral register, the results of which were statistically uncertain due to low response rate of around 60%. The unadjusted results were in reasonable accord with the Mid Year Estimates for 1997. In the 1980s a sample survey achieved a response rate of around 90%, but response bias is recognised as a problem even at this higher level.

3.3 Health Authority Patients' Registers

HCC have also tried to use administrative data, particularly the Health Authority Patients Registers to estimate population.

In 1997, the Patients Register (after eliminating incomplete records) showed a similar population to the contemporary unrevised ONS Mid Year Estimate. In 2006, the similarly adjusted Patients Register has around 50,000 persons more in the county than the Mid Year Estimate suggests.

This is partly due to the improvement of the quality of postcodes on the register, but this accounts for only about 20,000 additional person records in Hertfordshire. Downward revisions of the pre-2001 Mid Year Estimates in the light of the 2001 Census results accounts for a further 20,000. Allowing for these, over the past decade the numbers on the Hertfordshire patients register have risen by about 10,000 more than the Mid Year Estimates.

The reasons for this disproportionate increase in the patients' register are not clear, but among them could be the arrival of increasing numbers of short term international migrants. If these people were to register and then leave the country, they could remain on the patients register indefinitely.

3.4 Accuracy of Census

We believe that the Census provided a reasonable count of the population in Hertfordshire in 2001. The counts of dwellings and households were broadly commensurate with our own records from Planning Information databases. The 2001 Mid Year Estimate for Watford was subsequently adjusted to include an additional 600 persons,

mainly young adult males. This adjustment was made because of information from the ONS Longitudinal Survey which was only considered after the census results were published.

3.5 Dwellings-based estimates

The increase in population indicated by the Mid Year Estimates between 2001 and 2006 is broadly in line with what we would expect on the basis of the number of additional dwellings provided. This is supported by local models and by forecasts based on the actual dwellings completions in the period using the Chelmer model (true for Hertfordshire, but not necessarily for all of Eastern Region, the area to whose extent the model was run).

4. THE ROLE OF SURVEY AND ADMINISTRATIVE DATAWhat role can and should be played by data from General Practitioners and other health service data in the compilation of population statistics?

4.1 As suggested in the previous answer, patients registers have potential to provide an indication of both the absolute population stock in a particular year, and also of the change in population year to year or for other periods. However, comparisons over recent years suggest a greater rate of population growth than implied by changes in the dwelling stock and our understanding of the current trends in household size. This is partly due to improvements in the quality of the address records, so that fewer records are now discarded because of inadequate location details. It may also be that other factors have affected the pattern of household composition to create a much greater average household size than is expected. Such factors could include pressure from high house prices, and the presence of short-term migrant workers living in multi-person shared households.

4.2 However, whatever the causes, the disparity between the patients' register and the other estimates contributes to uncertainty about all the sources. HCC use the Hertfordshire Patients' Register to apportion the Mid Year Estimate to small areas (below district level) to produce small area population estimates. This process is tantamount to constraining the Patients' Register population to the district Mid Year Estimate.

4.3 Comparing these small area estimates with the census populations reveals some anomalies. Areas with prisoners or school boarders are affected, as these are two groups that appear to be excluded from the Patients' Register.

What other data sources can and should be used in the compilation of population statistics?

4.4 The present estimating process is blighted by the fact that it comprises a number of elements, most of which (births, deaths and to a lesser extent internal migration) are reliably measured at local level, but one of which (international migration) is not. The international migration estimates could be informed by the NI number applications and the Workers Registration Scheme data from the Home Office. The overall LA population estimates and their annual change could be informed by changes in dwellings stock combined with an understanding of the level and trend of average household size. Use of HESA and PLASC data could help enumerate the student populations. It is recognised that the questions asked might need to be reviewed to enable best use to be made of these datasets.

November 2007

@HR25@Memorandum from the London Borough of Hammersmith and Fulham

EXECUTIVE SUMMARYPopulation data is extensively used by this local authority and our partners. It forms a fundamental part of

our service planning and feeds through to a range of key performance indicators. It is also has a significant impact on local authority funding. Population data is a key driver of the Formula Grant distribution process and a number of specific government grants.

The London Borough of Hammersmith and Fulham has significant concerns about the accuracy and reliability of the existing population data produced by the Office for National Statistics (ONS). This authority does not consider that population data has kept pace with current levels of migration, particularly the population movement associated with the enlargement of the European Union. Current ONS estimates for Hammersmith and

Fulham are at odds with a range of other data sources, such as that for new National Insurance Registrations, and our local experience.

The Council welcomes the beneficial contribution that migrants make to the local economy and to the social and cultural diversity of the area. But inaccurate population data impacts on community cohesion and places a strain on local services. Funding and services are not delivered where they are most needed.

Our detailed concerns with current population data are set out in the submission. A number of improvements are also suggested, such as taking account of short-term migration and giving recognition to the costs associated with high levels of population mobility. Serious consideration should be given to undertaking a mid-term census for city areas where population mobility is high and there is evidence of significant short-term changes. This Council has detailed a number of actions that can be taken to improve the response rate to the next census.

The important role that population data plays for local authorities does not appear to be reflected within the current procedures for producing population estimates. Sample sizes regarding migration levels are remarkably small for such critical information. Local authority funding can go up or down by several million pounds based on a handful of interviews. The Office for National Statistics needs to be adequately resourced if confidence is to be restored in population estimates and an effective dialogue held with local authorities and other public bodies.

1. INTRODUCTION1.1 Hammersmith and Fulham is a small and densely populated west London borough with a population

of 171,400 people (Office of National Statistics [ONS] Mid Year Estimate 2006). It is a popular place to live and work with over 70% of local people satisfied with the area (2006 Best Value national survey). The Borough offers a range of cultural attractions in the three town centres of Hammersmith, Shepherds Bush and Fulham and on the Thames-side. It has a net revenue budget of £180 million.

1.2 Between 2001 and 2006 the population is estimated to have grown by a moderate 1.2%, which is a lower increase than Inner London (4.0%) or Greater London (2.6%). The Greater London Authority's 2006 projections indicate that the Borough's population will grow at a steady rate in future years, but these projections will be revised in the light of recent ONS revisions to population estimates. The current data suggest a high projection of 184,800 in 2011 and 189,000 in 2016 and a low projection of 181,000 in 2011 and 183,000 in 2016. Over the next 10 years, the largest percentage population increases are projected in the 40 to 54 age group, followed by the 5 to 19 age group. This growth in population and the changing age distribution will place new demands on local public services such as education, health and housing.

1.3 The Borough has a relatively young and ethnically diverse population with a higher proportion of young adults aged 25-39 (37%) than London and the rest of the country. 27% of households consist of a single person under pensionable age and only 22% of households contain dependent children. Just over one in five residents are from non-white ethnic backgrounds, 5% were born in Ireland and there is a well-established Polish community. Some ninety different languages are spoken in local schools. London's place as a world city means that the Borough will continue to be home for many diverse groups of people, of different nationality, ethnic origin, religion, and culture.

1.4 Hammersmith and Fulham has a very visible presence of Eastern European nationals on its streets. By far the greatest majority are Polish. This reflects the fact that the Borough has a long and proud history of Polish migration to the area and a wide range of Polish businesses and services have developed in Hammersmith and Fulham as a result.

1.5 The Polish Cultural Centre, within sight of Hammersmith Town Hall, is the largest Polish cultural institution outside Poland. The close historical tie between Poland and Hammersmith and Fulham is demonstrated by the fact that there is a Polish eagle on the Council's mayoral regalia. The Council is proud of the established Eastern European community, which is an essential part of the fabric of the borough's life, and welcomes the beneficial contribution of the new accession state nationals to the local economy and to the social and cultural diversity of the area.

2. THE USES OF POPULATION ESTIMATESUses

Service Planning

2.1 Population projections, including age group data and trends, form a key part of the Council's, and our partners, service planning. Some examples are set out below.

2.2 Local Development Framework (LDF). The LDF is the key strategy for co-ordinating council, and other partners' strategies and private sector development, it is therefore essential to have realistic population and household projections. In order to plan for housing provision and the future need for infrastructure, such as transport, water and drainage and for services and facilities, the Council needs detailed estimates for specific age groups such as children, elderly, as well as household numbers and types.

2.3 Although the LDF is primarily concerned with the next 10-15 years, the London Plan now covers the period up to 2026 and therefore projections are needed for a rolling 20 year time period.

2.4 One of the issues at the Examination into the soundness of the LDF Core Strategy is likely to be whether the provision of housing and other services and facilities take adequate account of the population and household projections for the local area.

2.5 At a more detailed level, the government's guidance on Strategic Market Housing Assessments bases new household formation rates on ONS and Communities and Local Governments (CLG) population and household estimates.

2.6 Schools. Population data is used within school organisation plans to ensure that an adequate number of school places are provided. In particular this authority is using population data in developing our plans for Building Schools for the Future—a major programme of investment in the secondary sector. This work will be followed by planning for the primary sector.

2.7 Health. From a health perspective it cannot be stressed too much how important population projections are. For example, a projected ageing of the population means that the Primary Care Trust (PCT) is projecting an increase in long-term conditions and is putting in place plans to address this. Likewise, the gap between male and female life expectancy is higher than in England as a whole, an area the PCT wishes to explore further from a health inequalities angle.

2.8 Commissioning intentions are based on a combination of past, current and projected activity data, considering population estimates, funding flows, service developments and policy changes (whether from national agencies or developed locally). From the PCT perspective, it is important not only to know the future population estimates, but some estimate within these of the breakdown of age, gender, disability, ethnicity etc. as the prevalence of certain conditions may well be dependent upon this. For example, the current increase in childhood obesity between Reception (24%) and Year 6 (40%) is made more frightening in the context that the number of under the age of 14 is set to increase.

2.9 Adult Social Care. Population data is used to inform our adult social care planning and demand forecasts. A new duty has also been placed on Councils and PCTs to conduct health and social care needs assessments. These will inform the planning of care provision and will need to be undertaken every three years. Population data will form a key part of the assessment process.

Performance Framework

2.10 Population data is widely used within Performance Indicators. Over recent years the Government has required local authorities to prepare a range of performance indicators. These have ranged from Best Value Performance Indicators and Local Area Agreements to the new list of 198 national indicators (which are currently being consulted on) which will underpin the new performance framework for local authorities and local authority partnerships. Where the indicator is expressed as a rate, the denominator is usually derived from a population estimate produced by the ONS.

2.11 This authority also uses population data when reviewing our performance. For example it is a key element when benchmarking our services against other local authorities. It should enable meaningful cost and value for money comparisons to be made as well as judgements on performance delivery.

Resources

2.12 Population data has a significant impact on local authority funding. It is an integral part of the Formula (general) Grant Distribution process and impacts on a range of specific grants. In 2007-08 Hammersmith and Fulham will receive £111 million of formula grant—this represents 60% of the Council's budget requirement.

2.13 A new 3-Year Local Government Finance Settlement is expected to be announced in early December 2008. It will cover the period 2008-09 to 2010-11. The Formula Grant payable to individual authorities will be calculated using a four block model that takes account of a Relative Needs Formulae (RNF), relative resources, a central allocation based on a per head amount and a “floor" damping block. The largest block is that relating to the RNF. Population data is a key driver of the RNF and central allocation calculation.

2.14 The population data set that will be used within the forthcoming Local Government Finance Settlement is the revised 2004 sub-national population projections (SNPP). This ONS data set projects forward the population to 2029 using a revised estimate for 2004 as a base. For Hammersmith and Fulham the 2008 population projection (175,800), using the revised 2004 SNPP data, is lower than the figure (177,200) currently assumed within the 2007-08 Local Government Finance Settlement. This does not seem credible to this Council given the recent upsurge in migration associated with the new EU entrants and other conflicting data sources, such as the sharp increase in new National Insurance Registrations.

2.15 Hammersmith and Fulham has expressed repeated concern about the population data used within the Formula Grant process. This has covered not only inadequacies with the methodology used to estimate international migration at a local and national level (particularly regarding the increase in migration associated with the new EU entrants), but also our concerns about the failure of grant formulae to take account of short-term migrants, people who live and work in London for part of the week but are formally resident elsewhere, illegal migrants and the cost associated with high levels of population mobility.

EFFECTS OF INACCURACIESService Planning

2.16 Inaccurate population data may well result in poor service planning. The Council will not be putting in place appropriate plans to meet future needs. Examples of potential problems could include a shortage of school places or inappropriate service delivery models for health and adult social care. For example in mental health services there are a number of people without resource to public funds who are not in existing population projections but make service demands.

Performance Indicators

2.17 Inaccurate population data will mean that performance indicators will be inaccurate. This will damage the reliability of Government set indicators and the delivery of government set targets. External assessment of local authority service provision will be flawed. The use of benchmarking data will be undermined.

Government Grant

2.18 Inaccuracies can mean that Government grant support fails to keep pace with the demand for public services. For Hammersmith and Fulham the Council's historical ties, along with the various Polish businesses, services and community organisations, have acted as a magnet for new Polish migrants arriving in the Capital. The Council has welcomed the beneficial contribution made by the new accession state nationals to the local economy and to the social and cultural diversity of the area. But this increase in population has not been recognised through the Formula Grant system.

2.19 The increase in inwards migration is borne out by recent data regarding National Insurance (NI) registrations. For 2006-07 Hammersmith and Fulham had the fifth highest number of NI registrations by overseas nationals, as a proportion of its population, in the whole country. Last year, 9,310 foreign nationals registered for NI numbers in the borough—5.2% of the total borough population. This evidence is in stark contrast with the formula grant system which is implying that our 2008 population will be lower than in 2007. This is not credible particularly given that the 2007 Estimate took no account of the impact of the new EU entrants.

2.20 New migrants access the full range of local authority services as well as having particular impact on homelessness and advice services. The use of inaccurate population data has resulted in the formula grant system failing to adequately recognise spending pressures.

3. POPULATION DEFINITION3.1 This authority considers that the usually resident population should remain as the basis of the

population count but that this is, on its own, inadequate. There must be a count of “visitors" as well. The exclusion of visitors and short-term migrants (people staying for less than a year) leads to an understatement of service need. This issue is increasing in importance. There is clear evidence of a change in employment patterns associated with both the new EU entrants and a greater freedom of movement globally. It is increasingly common for workers to reside in the Borough for several months and then return to their country of origin for several months. The current population definition is not capturing this change.

3.2 Consideration should also be given to data capture on population mobility. This is an important issue for many authorities. Research undertaken by the London School of Economics (LSE) on behalf of London Councils indicates that there are several London boroughs where population mobility exceeds 35%. For Hammersmith and Fulham data from the last census suggested that more than one in five households move in any one year. This has a significant impact on service provision and cost. The LSE research concluded that there are many situations where population mobility imposes little or no additional costs but there are types of mobility that impose additional costs—these vary from the need for translation services to specialist support for the homeless. The research suggested that such costs could amount to £100m for London alone. Data capture on population mobility would assist service planning and enable consideration to be given on how the associated costs can be funded.

4. THE ROLE OF THE CENSUSFrequency

4.1 Population mobility and levels of international migration have increased significantly in recent years. Moreover it is often the case that such fluctuations emerge at short notice due to wider global events such as refugees from the Iran/Iraq war in the 80s, refugees from the former Yugoslavia, asylum seeker fluctuations, and A8 migrants in the period since 2004. The current arrangement whereby the census is undertaken every 10 years is failing to adequately capture such short term fluctuations. The net impact is that local authorities are faced with an increased demand for services that is not recognised through funding allocations.

4.2 A mid-term Census should be seriously considered for city areas where population mobility is high and there is evidence of significant short-term changes

Response Rate

4.3 This authority considers that a number of actions can be taken to improve the response rate to the census. These include:

@UL@— More local liaison with local authorities with hard-to-reach populations eg on detailed knowledge of addresses and local ethnic and other communities. Address lists must be validated from local knowledge within the local authority.@/UL@

@UL@— Resources should be put in to local authorities by ONS for locally targeted publicity especially in areas with hard-to-reach populations such as this.@/UL@

@UL@— Adequate remuneration for enumerators where these are to be used, and reasonable workloads@/UL@

@UL@— Clear instructions to residents on the mandatory nature of the Census @/UL@

@UL@— More resources and more time devoted to follow-up. The response rate in 2001 could have been much better if a thorough follow-up on unreturned forms had been allowed.@/UL@

@UL@— Make available an adequate range of languages to be used in translations of the Census form@/UL@

@UL@— More flexibility in the ONS operation to adapt to local issues that can arise@/UL@

@UL@— An adequate helpline service is necessary@/UL@

@UL@— The use of enumerators and liaison with the local authority to identify vacant properties and second homes@/UL@

The Census Form

4.4 The Census form needs to be 4 pages long to include the key questions that local authorities want. The cost of the increase of one page is small compared to the overall cost of the Census operation and the use of 4 pages would actually be less confusing to the respondent.

4.5 This authority has submitted responses to the ONS consultations on the Census content. We believe that a new question on household income is very important. It feeds in for example to housing needs assessments. Other new questions on language spoken and nationality are desirable.

UK Co-ordination

4.6 No comment.

Outputs for 2011

4.7 This authority considers that:

@UL@— Outputs should be in electronic form compatible with propriety software, especially SASPAC.@/UL@

@UL@— Outputs should be prompt. A year after Census date is a reasonable target.@/UL@

@UL@— Disclosure control should be minimised and should not compromise the fitness for purpose of the data. In 2001 OA level double geography data was unusable, and the validity of Ward level data was also in doubt.@/UL@

@UL@— Outputs on a Ward basis should be continued as well as SOAs@/UL@5. MID-YEAR ESTIMATESAccuracy

5.1 Mid Year Estimates are of vital importance because they are the only official annual estimates of the population of local authority areas. But there are major concerns over accuracy. The recent revisions to the 2002-5 figures, as a result of changing the way of estimating international migration, demonstrate this.

5.2 There is currently no way of properly allocating international migration to a local authority area. The sample used in the International Passenger Survey (IPS) is too small (the total national sample in 2005 was only 2,965 in-migrants and 781 out-migrants), and the same is the case for the Labour Force Survey (LFS), which is now being used to determine geographical distributions of in-migrants and where data is only available for groupings of local authorities.

5.3 Furthermore, official estimates measure only long-term migrants, ie those staying for a year or more. Short-term migrants, of which there are many within the A8 flows, are completely missing. Workers' Registration Scheme data for Hammersmith & Fulham for the period May 2004 to March 2007 indicates that 37% of A8 migrants registering were intending to stay for less than 12 months, so that over one third of A8 migrants would be missing from official counts. Short term migrants, of course, still use public services.

5.4 The Office for National Statistics (ONS) recognises these deficiencies and has been working on their improvement. The necessary improvements, however, have not been fully implemented and an interim solution has been devised by switching from IPS to a combination of IPS and LFS, which substitutes one set of inadequate data for another and there is still no measurement of short-term migration.

5.5 Improvements are proposed to the IPS but there is no evidence that this will improve the method of distributing data between local authority areas. Again, there are proposals to make national estimates of short-term migrants but no method of allocating these to local authority areas.

5.6 The net result of the recent change has been a disaster for Hammersmith and Fulham, in terms of Formula Grant, in that, as a result of adjustments to estimates of international migration, the population of the Borough in mid 2006 is estimated to be 8,500 fewer than the previous estimate for 2005. The adjustment to previous Mid Year Estimates 2002-05 has resulted in a net loss to the Borough over that period of 8,900 people, caused by reductions in estimated in-migration of 4,400 and increases in estimated out-migration of 5,400.

5.7 The absurdity of the figures is that the revised official data now suggests that overseas migration to the Borough was less in 2005-06 (a total of 5,680) than before A8 accession (between 6,600 and 7,000 a year in the 2001-04 period). In terms of net gains the new data suggests a figure of only 1,180 compared to a figure which is more than twice as high, 2,580, in 2001-02. This is plainly wrong in the context of the recent wave of A8 migration.

5.8 The main indicators of A8 migration are National Insurance (NI) registrations and data from the Workers Registration Scheme. Since 2004, 4,330 A8 workers have registered for NI in H&F and 4,080 have registered on the WRS. This can be contrasted with the fact that there were only 200 NI registrations from A8 countries in 2002-03, so we know that there has been a very substantial increase in A8 migration. Yet in the latest population estimates by ONS the net gain to the Borough due to international migration for 2005-06 is estimated to be at the same level as pre-accession, in the years 2001-03.

5.9 Among London Boroughs, Hammersmith & Fulham ranks 5th highest in numbers registering on WRS between May 2004 and June 2007 and 12th on NI registrations. In the ONS estimates of international in-migrants 2005-06, however, Hammersmith & Fulham ranks 14th and in terms of net international migrants as a percentage of total population, Hammersmith & Fulham ranks as low as 19th among London Boroughs.

5.10 The gap may indeed be even wider than this. NI and WRS data has known limitations. Not everybody registers—the self employed for example are not required to register for WRS, and of course dependants are not registered in either scheme. An even greater drawback may be that the place of initial registration may not reflect current place of work or residence. A person registering in Westminster, for example, may actually soon after be living or working in Hammersmith & Fulham. So it may well be that registration data may under-count numbers for this reason, as well as the fact that not everyone registers at all in the first place.

5.11 The indications are, therefore, that for the migrants that ONS purports to measure, ie those staying for 12 months or more, the estimates are likely to be erroneous. Furthermore, there are a large number of migrants staying for shorter periods who are not measured at all.

5.12 As well as reductions in estimated in-migration, ONS has increased the number of estimated international out-migrants so that the Borough has the 7th highest figure among the London Boroughs: in terms of out-migrants, as a % of total population, Hammersmith & Fulham ranks 4th. This is not so much based on hard data but on a regression model using the 2001 Census. There is no common sense reason for supposing that 5,400 more people have left the Borough over the 2002-05 period than was previously estimated. So here again ONS estimates are likely to be erroneous.

5.13 Given the obvious substantial increase in the numbers of Polish migrants in the Borough over recent years, this reduction in the estimate is clearly wrong and it will result in a significant shortfall in the Council's Formula Grant from next year. The Council is facing an increasing burden on its own services, and on those provided by local voluntary sector agencies which it funds, as a result of increased migration from Eastern Europe, yet the Government's inadequate and inaccurate data collection on the local migrant population will increase the strain on the Authority's annual budget.

Improving accuracy

5.14 Resources must be given to ONS to improve the surveys on which population estimates depend.

5.15 In addition work must be done on investigating checks on accuracy of estimates by triangulating against administrative records held by public bodies (see below).

The Role of Survey and Other Administrative Data

5.16 There is a range of local administrative data that should be investigated. By combining sources, a good picture of the size of local populations can be built up—there are examples in London that demonstrate this—

but work is needed on improving data and, in particular, on facilitating data sharing. Clearly data matching requires the sharing of names and addresses and there are data protection issues that arise.

5.17 For example, GP data is a valuable source but GP lists often need “cleaning up". The local authority could help in this process by matching data with records it holds but there are data sharing issues and the degree of data sharing varies greatly from area to area.

Co-Operation With Stakeholders

5.18 There is a greater degree of co-operation of ONS with local authorities now compared to previous periods of time but this needs continual advance.

November 2007

@HR25@

Memorandum from the Local Government Association

SUMMARY@UL@— The Local Government Association (LGA) welcomes this timely initiative. There is wide concern amongst councils that official statistics on population, and particularly international migration, do not reflect experience at the local level.@/UL@

@UL@— Councils use population estimates to inform service provision, performance monitoring, policy development and community leadership. Their share of grant from central government is substantially influenced by population estimates. Thus inaccuracies in population estimates inhibit the best use of public resources.@/UL@

@UL@— The current problems with population estimates at local level stem from two inter-related sources. The first is inaccuracy in the 2001 Census, attributed largely to problems in “hard to count" areas. The second concerns the counting (and allocation to local level) of international migrants. The latter problems partly relate to definition (only those intending to stay for at least 12 months are counted) and partly to methodology (which involves a top-down approach beginning with the national estimate).@/UL@

@UL@— The Census remains a key source for information about localities. An increasingly mobile society blurs key concepts of population, household and residence. This makes the Census more difficult to undertake, but increases its importance as an aid to understanding local economies and communities.@/UL@

@UL@— In a 2004 report on the problems associated with the 2001 Census, ONS identified a number of measures to reduce the likelihood of recurrence in 2011. These remain relevant.@/UL@

@UL@— There are a number of administrative sources which provide some information at a local level on migration and/or population. There is scope to enhance these and to use them, at the least, as a check on official population estimates and ideally as part of the estimation methodology.@/UL@

@UL@— A particular concern of councils is the extent of short-term migration, which feeds population churn and demand on services. At present official estimates—and funding regimes—do not deal well with this phenomenon.@/UL@

@UL@— On the whole ONS staff engaged in producing population statistics are helpful and accessible to stakeholders in local government. However, other government departments also have a role to play in delivering good quality population figures, and their officials are not always so accessible. The Statistics Board should take a prominent role in co-ordinating improvements in population statistics for the public good, ensuring that all appropriate government departments play their part.@/UL@1. INTRODUCTION@UL@— The Local Government Association (LGA) promotes better local government. It works with and for member authorities to realize a shared vision of local government that enables local people to shape a distinctive and better future for their locality and its communities. The LGA aims to put local councils at the heart of the drive to improve public services and to work with government to ensure that the policy, legislative and financial context in which they operate, supports that objective.@/UL@

@UL@— The LGA welcomes the committee's interests in this matter. There is now wide concern amongst local authorities that the official population estimates, and especially the element accounting for international migration, do not reflect experience at local level. We welcome steps initiated by the Office for National Statistics to find long term remedies to address these weaknesses. However, we believe that, alongside this, action is necessary to make better use of available data to understand recent and rapid local population changes resulting from migration; and to ensure best possible response to the 2011 Census. @/UL@

@UL@— This is because the evidence clearly shows that the changes are widespread and are placing short and, increasingly, longer term demands on local public services that are not reflected in funding formulae for these services because the formulae rely on out of date and inaccurate population figures.@/UL@

@UL@— Our response is structured around the questions asked in your call for submissions. @/UL@2. USES AND DEFINITIONS2.1 What are the uses of population estimates, and how far do current population estimates meet the needs associated with those uses? What are the effects of inaccuracies or inadequacies in such estimates?

@UL@— Local authorities use population estimates intensively and extensively in four broad ways:@/UL@

@UL@— To plan service delivery.@/UL@

@UL@— To formulate policy.@/UL@

@UL@— As denominators for indicators to measure and evaluate performance (both as required by central government and as a means to ensure effective service provision).@/UL@

@UL@— In their community leadership role, to ensure that citizens are empowered with good quality local information to use in their own decision-making.@/UL@

@UL@— In addition, population estimates are used by central government in the funding formulae to determine the allocation of central grant amongst authorities.@/UL@

@UL@— It follows that inaccuracies and inadequacies of population estimates can result in@/UL@

@UL@— Less than optimal use of resources in service delivery and policy formulation, to the detriment of service users specifically and the effectiveness of public expenditure generally (though some types of service provision require less precise population estimates than others) @/UL@

@UL@— Misleading performance indicators, with potentially serious consequences for the reputation of authorities and potentially inappropriate remedial action. For example, an underestimated population could produce a falsely high employment rate and crime rate, leading to the diversion of resources and effort from labour market initiatives towards crime initiatives.@/UL@

@UL@— Confusion amongst citizens about the characteristics of their localities@/UL@

@UL@— Misallocation of central grant to local authorities, with consequences for service quality.@/UL@

@UL@— It is difficult to quantify these effects. However, an illustration of the possible scale is provided by the fact that, in making its case for the 2011 Census, ONS estimated that the absence of census data could result in the misallocation of central grant to the tune of £2.8billion over 10 years. This only refers to the possible effect arising from distribution of central grant. There are other costs arising from the misallocation by councils themselves stemming from erroneous population data, and misallocation arising from inaccurate performance indicators.@/UL@

@UL@— A number of councils have substantial misgivings about current population estimates. In some cases these endure from, and reflect, concerns about an undercount at the 2001 Census. In other cases the concern is centred on the migration elements of the estimates, and in particular their failure to reflect increased levels of international in-migration since 2004. These misgivings are generally supported by quantitative evidence from other official sources. In Annex A we present three examples from local authorities. These are chosen to give a flavour of the range of concerns, and to illustrate the sources used to justify concerns. The reasons for the problems with the 2001 Census and with current estimates, and possible solutions, are explored at various points in this submission.@/UL@

@UL@— A further effect of inaccurate population estimates is that they can reduce public trust in official statistics as a whole. If citizens feel that official population estimates are out of step with their own experience they may be more inclined to give credence to other speculations about population and migration, to the detriment of community cohesion.@/UL@

2.2 How appropriate is a definition of the population based on the usually resident population in the context of the needs of the users?

@UL@— The usually resident population definition is one amongst many population definitions of potential use to local authorities. We recognise the value in having a single standard definition which can be consistently applied across the country. Ideally, though, the population estimates system would provide enough flexibility to allow authorities to calculate populations against a range of definitions to inform service provision and policy development.@/UL@

@UL@— Concepts such as “usually resident"—and indeed “population" and “household"—are becoming harder to define as society becomes more mobile. For example, more fluid household structures result in children splitting time between separated parents; and in couples sharing two (or more) residences in different locations. The expansion of higher education has resulted in many more towns experiencing a sharp distinction between term-time and vacation populations. As the number of second homes and holiday homes rises, the concept of “usual residence" becomes blurred. As the ability to work remotely from one's primary workplace grows, so does the balance of time spent between a home located near to the workplace and a second home. However, all these factors serve to underline the need for the census, in order to understand the nature of modern populations and behaviours and to arrange and target public services to meet people's needs; and in order to provide meaningful denominators for performance measures. @/UL@

@UL@— The range of alternative definitions of population would ideally include population present, day/night populations, term-time and holiday populations, weekday and weekend populations.@/UL@

@UL@— A particular difficulty with the usually resident definition concerns the treatment of migrants. The current system only counts those migrants who state the intention to remain in the UK for at least 12 months at the time of entry. There is considerable evidence that many migrants come for a shorter period (but nevertheless make demands on services); and that others stay for longer than they had originally intended.@/UL@

2.3 How does the quality and range of population statistics in England and Wales compare with that available in Scotland, Northern Ireland and internationally?

@UL@— Scandinavian countries and the Netherlands, amongst others, rely heavily on continuously-updated population registers as a basis for population estimates, rather than the mix of census and surveys as in the UK.@/UL@

@UL@— The advantages of these systems are that they are cheaper, provide up-to-date basic data, and can provide good quality local statistics.@/UL@

@UL@— The disadvantages are that they can be inflexible, limited in the range of data items collected, and are dependent on public acceptance.@/UL@

@UL@— The Scandinavian official statistics systems are rated by their international peers as being amongst the best in the world. They also command high levels of public trust (in contrast to the UK system).@/UL@

@UL@— It does not necessarily follow that the UK should adopt a population register approach because there are disadvantages as well as advantages, and cultural issues to be considered. Even so, it is sensible to monitor how these things are done elsewhere. ONS' intention to develop a business plan for an Integrated Population Statistics system, which would presumably examine approaches taken elsewhere, is therefore welcome.@/UL@

@UL@— In comparison with the arrangements in England and Wales, those in Scotland are perceived to offer the following advantages;@/UL@

@UL@— earlier production of population estimates and projections by about four months@/UL@

@UL@— population and household estimates carried out within the same branch, facilitating cross-checking@/UL@

@UL@— greater use of health service data in producing small area population estimates@/UL@

@UL@— opportunities for councils to quality assure small area population estimates@/UL@

@UL@— Practitioners recognise that some of the advantages in the Scottish system stem from it being a far less populous country with only 32 councils. On the other hand lower staffing levels in the statistical offices limit the range of analysis carried out.@/UL@3. THE ROLE OF THE CENSUS3.1 How does the Census contribute to the creation of population statistics? Is the current frequency of Censuses appropriate and, if not, how frequent should Censuses be?

@UL@— The census acts as the basis from which population estimates are rolled forward. It is therefore a crucial benchmark. The census is, of course, much more than simply a benchmark. It is the richest source of information about the economic, social and demographic composition of small areas. It provides an understanding of the linkage between population and dwellings at a small area level, which is increasingly important as the number of second homes and holiday homes rises, and as short-term migration and shifts in the nature and frequency of home-working result in the partial use of dwellings. Such understanding is all the more important since second homes, vacant properties and partially used dwellings are geographically concentrated.@/UL@

@UL@— There are two basic problems with the use of the Census as a benchmark. The first is that, by definition, the census becomes out of date as soon as it is published; and the second is that errors present in the census remain, and are sometimes compounded, through the estimation period. @/UL@

@UL@— The longevity of errors present in the census is illustrated by the fact that some of the difference between 2001 Census figures and population estimates were attributed by ONS to the over-estimation of undercount at the 1991 Census. An adjustment of 351,000 was made to the 1991 base to compensate.@/UL@

@UL@— That adjustment went some way to bridging a 1.1 million gap between the 2001 Census estimates for England and Wales and the rolled forward 2001 mid-year population estimates. Other adjustments were made, described at Annex B and involving both reduction in the original mid-year estimates and increase in the 2001 Census-based figures. Even so, a difference of more than 200,000 remained unexplained. A number of councils (see Annex A) remain convinced that the 2001 census, and the population estimate based on it, did not reflect their actual population; and that their use as a basis for subsequent estimates has perpetuated an undercount.@/UL@

@UL@— Other sources are needed to capture inter-censal population change, such as the substantial international migration that has occurred since 2001 and has changed the size, shape and distribution of the population. Moreover, the geographical distribution patterns for migrants evident in 2001 are not a good guide to distribution patterns for more recent migrants.@/UL@

@UL@— Some countries, such as New Zealand and Eire, hold censuses every five years and this approach is apparently popular there with both national and local bodies. But any census is resource-intensive, requiring a build-up to census day and intensive effort during the analysis and production period. We are not persuaded that a 2016 Census would represent the best use of limited resources as compared with, for example, exploration of greater use of administrative data systems.@/UL@

3.2 What methods should be employed for the distribution and completion of Census forms, including the use of the Internet

@UL@— Ideally fieldwork should exploit latest technological tools to the full, facilitating the gathering of real time data with instantaneous feedback to local field offices. This might allow for more effective cross-checking and better identification and hence coverage of hard to reach groups. A number of local authorities have adopted such technologies to deliver their services and the Census offices would benefit from engagement with them.@/UL@

3.3 What steps should be taken to increase the rate of responses to the Census, particularly in hard-to-reach areas? How effective are plans to measure the extent of non-response to the 2011 Census?

@UL@— In its review of the problems which had affected the 2001 Census, ONS identified implications for the 2011 Census. Briefly, the implications involved:@/UL@

@UL@— address issues: a more accurate and up-to-date address list is needed before the census questionnaires are distributed. Close working with local authorities had been shown to produce better address lists, but had been resource-intensive;@/UL@

@UL@— enumeration issues: centring on the need for a good management information system;@/UL@

@UL@— estimation issues; hard-to-count areas need to be identified more accurately, using administrative data. Quality assurance processes need to be improved, drawing on local knowledge of available sources;@/UL@

@UL@— contingency plans: ONS observed that contingency plans need to be more locally based; and@/UL@

@UL@— resource balance: the census takers need to be able to switch resources more swiftly to deal with difficulties.@/UL@

@UL@— This list provides a worthwhile agenda for an improved 2011 Census. It is largely reflected in the recommendations made by the Statistics Commission in its review of the 2001 Census in Westminster. In particular, we endorse the need to work closely with local authorities, and the value of administrative sources of information.@/UL@

@UL@— We are concerned that ONS should not simply concentrate its resources in areas which were hard to count in 2001. International migration since 2001, particularly from the A8 states, has been geographically dispersed. There are likely to be language difficulties in enumerating these populations which should not be underestimated. Moreover, there is a pressing need, in the absence of other reliable statistics, for the Census to enumerate this element of the population accurately. In short, we expect that the proportion of the population that will be hard to count will be greater in 2011 than in 2001, and that hard to count elements will be geographically dispersed.@/UL@

@UL@— In order to address this issue, and for a number of other reasons such as establishing accurate address lists, it is vital that ONS engages councils in support of the 2011 census. @/UL@

3.4 To what extent is there a trade-off between the length of the Census form and its role in providing population information? What questions should be included in the 2011 Census?

@UL@— ONS reports that response rates in the Test census were not significantly different between households issued with a three-page questionnaire and those that received the four page version.@/UL@

@UL@— We are therefore attracted by the possibility of having a four-page form for the 2011 census. This would mean the form could carry questions on such topics as qualifications, carers and industry which were covered in 2001 but cannot be fitted onto a three-page form as well as new questions designed to provide a better estimate of migration and therefore population.@/UL@

@UL@— We understand that the cost of a fourth page would add some £20million to the cost of a Census which will, in total, cost around £500million. The extra cost is not negligible, but the value of the extra information collected would be substantial. A fourth page would increase the number of data items collected at the census by around a third but would increase the overall cost by no more than 5%. Moreover, it would obviate the need for separate exercises to collect data on topics which are likely to be omitted from the three-page form.@/UL@4. MID-YEAR POPULATION ESTIMATES4.1 How accurate and useful are the mid-year population estimates for England and Wales, including information available for local authorities and Strategic Health Authorities?

@UL@— We have noted above that local authorities use population estimates for a number of purposes, and not all purposes demand the same level of accuracy. Generally, though, estimates are useful only if they are reasonably accurate. If the official estimates are insufficiently accurate, councils may have to incur the extra expense of estimating the size of elements of their population—older people, primary-age children or those in particular ethnic groups, for example—to help plan service provision or to monitor their performance. @/UL@

@UL@— Shortcomings in current population estimates stem, in some areas, from undercounts at the 2001 Census. As explained previously, and as illustrated at annex A, concerns remain despite amendments which were made to the Census-based 2001 population estimates for some local authorities in 2003 and 2004. @/UL@

@UL@— These concerns have been supplemented by doubts about estimates of international migration, and its allocation to localities. The fact that migration is (using currently available methods) relatively difficult to measure is compounded by the fact that, for most of the last decade, it has been the major component of population change. @/UL@

@UL@— The inadequacy of estimates of international migration has many facets, but can be summarised as:@/UL@

@UL@— Inappropriate definition—only those expressing the intention to stay for at least 12 months are counted.@/UL@

@UL@— Inaccuracies in estimates at the national level.@/UL@

@UL@— Inappropriate allocation of in-migrants to the local level.@/UL@

The last of these issues is examined in the next section.

@UL@— Some councils have indicated that problems with internal migration estimates are long-standing, and pre-date the more recent uplift in international migration. In some instances this is likely to be connected with student and Armed Forces populations. The area case studies carried out by ONS during 2007 should enhance understanding of student populations and we hope that internal migration estimates will improve in this respect. Some observations about the current methods of estimating internal migration are made in the next section. @/UL@

4.2 How appropriate is the methodology by which mid-year population estimates are reached?

@UL@— The migration element of the method poses the greatest problems. The international element of the migration estimate is problematic because it uses a top-down approach, beginning with a national estimate of international migration and apportioning that figure to localities. While the methodology was supplemented recently with Labour Force Survey data, the fact remains that a survey with small sample numbers at local authority district level is combined with assumptions about the likely behaviour of migrants based on patterns evident from the 2001 census to calculate an outcome. The resultant figures form a dataset which is but one of a number of possible outcomes. @/UL@

@UL@— As noted above, international migrants are only counted in population estimates if their stated intention on arrival in the UK is to remain for at least 12 months. Experience suggests that many migrants stay (or state that they intend to stay) for shorter periods. Shorter-term migrants also place demands on local services. A report commissioned by the Local Government Association (LGA) from the Institute of Community Cohesion argued that population “churn"—that is, the rapid turnover of people in an area—can place particular and significant demands on services, for example education. The LGA argues that a special contingency fund should be set up by government into which councils could bid to address service pressures arising from population churn.@/UL@

@UL@— The treatment of internal migration is based on NHS data, the accuracy of which depends on migrants registering with GPs. There is considerable evidence of delay in registration, particularly amongst groups such as young men, who have a high propensity to migrate but a lower perceived need to use medical services. In combination, the migration methodologies produce estimates which are not felt to be right because they do not square with local experience, which in many cases can be supported by quantitative evidence from other sources.@/UL@

@UL@— ONS has gone some way in recognising the shortcomings of existing population and migration estimates. Its draft work programme for 2008-12 sets out plans to increase spending in this area significantly. However, a better system depends on the willingness of other government departments to offer practical and financial support, so that best use can be made of administrative sources.@/UL@

4.3 What are the most appropriate ways to ensure a common level of accuracy in mid-year population estimates for different areas and what steps should be taken to improve the accuracy in areas where population is hard to measure?

@UL@— This is a difficult question but it goes to the heart of the problem with official population estimates. The present top-down approach is justified on the basis that it applies a consistent methodology in an even-handed way. But the evidence indicates that, in a significant number of areas, it produces results which are simply inaccurate and do not accord with other data.@/UL@

@UL@— A number of authorities make a compelling case that their population estimates are incorrect (see Annex A for examples). They draw on a number of sources that include:@/UL@

@UL@— specially-commissioned surveys or research;@/UL@

@UL@— data from General Practitioner services;@/UL@

@UL@— data from utilities;@/UL@

@UL@— statistics on housing completions;@/UL@

@UL@— schools census data;@/UL@

@UL@— registrations for National Insurance numbers;@/UL@

@UL@— the Workers Registration Scheme; and@/UL@

@UL@— contacts with organisations which serve hard-to-count elements of the population.@/UL@

@UL@— None of these sources, on its own, is currently likely to provide an entirely accurate and comprehensive population estimate. Some are not replicable across the country as a whole. Each has known limitations. Each source is likely to produce different figures for any one area. Few demographers would argue that any specific one of these sources should immediately take the place of the sources currently used in population estimates methodology. However, in combination they provide a reality check for the official estimates. Moreover, each might be enhanced to provide more pertinent data for population estimation. @/UL@

@UL@— In a report to the Greater London Assembly, Rees and Boden, population statistics experts from the University of Leeds, argue that “the differences that exist between datasets are the reason why the data should be used, as the development of a common view of statistics will not only provide the basis for the most informative picture [of international migration] but will also provide the basis for establishing precise relationships between them."@/UL@

@UL@— The LGA's recently commissioned report from the Institute of Community Cohesion recommends that a working group involving local government practitioners and others should be set up to consider how the various administrative sources which provide statistics on international and internal migration might be enhanced, and how they might then be incorporated in the population estimates methodology. We understand that the University of Leeds has been commissioned by the Economic and Social Research Council to work on the construction of a National Migrant Databank. We anticipate that the Inter-departmental Task Force Report on Migration Statistics will also prompt an examination of the range of administrative sources. We would expect these strands of work to be complementary. @/UL@5. THE ROLE OF SURVEY AND ADMINISTRATIVE DATA5.1 What role can and should be played by survey data in the compilation of population statistics?

@UL@— We have already commented at some length in the previous section about the way that survey data are used in compiling population statistics. Our observations here are therefore brief. @/UL@

@UL@— Surveys are the cornerstone of ONS' output generally, and make a substantial contribution towards population estimates. In comparison with censuses, surveys offer significant savings. They exploit the statistical reality that reasonably accurate information can potentially be produced by examining a small element of the population of interest.@/UL@

@UL@— Thus surveys can provide good quality data at national level.@/UL@

@UL@— They are not so good for enumerating “rare populations". So, for example, long-term migrants (those intending to stay for more than 12 months) form a small element of all passengers who pass into or out of the country through UK ports. Hence ONS boosted its sample of migrants within the International Passenger Survey in order to produce more reliable data.@/UL@

@UL@— Similarly, survey estimates which are relatively robust at the national level will have wider margins of error at finer geographies. The International Passenger Survey records around 3,000 long-term in-migrants at the national level. There are nearly 400 unitary or lower-tier local authorities in England and Wales. It is clear that the numbers destined for each authority will be small, and that estimates based upon such numbers will be subject to a substantial margin of error. @/UL@

@UL@— In producing more robust local estimates we are faced with a choice between a substantial boost in survey sample sizes; use of other survey and census sources relating to locality; or greater use of administrative data sources. The current approach is nearest to the second of these options. We think that the third approach merits further consideration. @/UL@

5.2 What role can and should be played by data from General Practitioners and other health service data in the compilation of population statistics?

@UL@— Statistics held by the NHS could potentially provide richer information than is currently used in compiling estimates. For example, the individual level data contains information on country of birth, first UK address of registration and information on returning migrants which at present does not carry through to the aggregate statistics. As Rees and Boden's report referred to above notes “as a comprehensive registration process that encapsulates a large majority of the UK population, including children, [NHS data] has to be made available as a key alternative source of new migrant statistics at a local level." @/UL@

@UL@— We understand that the NHS' Connecting for Health Initiative aims to produce a new personal demographic system holding summary information on everyone registered with the NHS. Its potential for tracking internal and international migration at local level should be exploited. @/UL@

@UL@— Nevertheless the usefulness of figures from the NHS does depend on registration. Migrants in particular are likely not to register for a variety of reasons—unfamiliarity, use of Accident and Emergency Services rather than General Practitioners, their general good health and their high levels of mobility. It would be unwise, therefore, to rely exclusively on this source of administrative data.@/UL@

5.3 What other data sources can and should be used in the compilation of population statistics?

@UL@— We have already mentioned most of the relevant sources. They are:@/UL@

@UL@— National Insurance Number registrations.@/UL@

@UL@— Pupil-level Annual Schools Census.@/UL@

@UL@— Workers Registration Scheme (and any points-based system which replaces or supplements this).@/UL@

@UL@— Higher Education Statistics Agency datasets.@/UL@

@UL@— Electoral register.@/UL@

@UL@— None of these is specifically designed to measure all migrants, or all population. They are by-products of systems designed for other purposes. This is why it is important to consider how these sources might be enhanced to produce statistics which would better serve the compilation of official statistics.@/UL@6. COOPERATION WITH STAKEHOLDERS6.1 How effectively has the Office for National Statistics cooperated with stakeholders with an interest in and information relating to population statistics and how can cooperation be improved by the Statistics Board and by its Executive Office?

@UL@— On the whole those engaged in producing population statistics are helpful and accessible to stakeholders within local government. They appear to recognise the importance—and the mutual benefit—of consultation. This has not always been the case. The change is welcome. But if population statistics are to improve then ONS cannot act alone; and some other government departments have further to go in recognising that they have stakeholders outside central government. @/UL@

@UL@— The Statistics Board needs to ensure that other government departments recognise the fact and legitimacy of stakeholders outside central government. @/UL@

@UL@— In particular, the Board's support may be required where a demand for changes to administrative systems to contribute towards better population statistics are perceived by departments as being unaligned with the primary purposes of those systems.@/UL@

@UL@— It must also be prepared to put weight behind requests for data linkage and administrative record access, and be bold in arguing the case for use of individual data to produce aggregated (and therefore anonymised) statistics in the face of opposition.@/UL@

@UL@— Finally, some users have expressed fears that the quality of service from ONS' population statistics function may be compromised by the efficiency savings required of ONS, combined with the impact of the relocation from London to Newport (albeit that the population statistics function is located in Titchfield and not affected

directly). Should the service be damaged, the savings from the efficiency programme could be outweighed by the inefficiencies stemming from having population statistics which are not as reliable as they could and should be. @/UL@REFERENCES

Institute of Community Cohesion (2007), Estimating the scale and impacts of migration at the local level, Local Government Association

Office for National Statistics (2004), 2001 Census local authority population studies: full report, ONS

Rees P and P Boden (2006), Estimating London's new migrant population: stage 1—review of methodology, Greater London Authority

Statistics Commission (2004), Report No 22: Census and population estimates; and the 2001 Census in Westminster: final report, Statistics Commission

@ANAME@Annex A

NEWHAMThe following extract is taken from the London borough of Newham's Financial Report 2006-07 (p20)

Mid Year Estimates

Mid-Year Estimates (MYE) are the “official" population figure for the borough, provided by the Office of National Statistics. These are produced annually, arriving in late August/early September, and should be used for any year-on-year comparative figures.

The Mid-Year Estimates are re-based following each Census, and then each year births, deaths and migration are calculated to produce another estimate. Following the 2001 Census, the previous decade's estimates were revised to adjust the 1991 Census population from overimputation, and the error in subsequent years following the original adjustment to the 1991 Census.

Newham's population, according to the 2005 Mid Year Estimate is 246,200—a fall for the third year in succession, with a loss of 1,500 people. This is a reduction from the loss seen in the 2004 MYE, when there was a fall of 2,900, but is still at odds with Newham's own administrative data which shows an increasing number of dwellings, an increasing number of electors and the PCT also reports an increasing number of people registered with GPs.

The Office of National Statistics primarily ascribes the loss to outward migration of residents to other parts of the country, and this has been partially offset by a high birth rate and inward international migration.

This apparent population decline of 1,500 residents is partially explained by the problematic methodology used by ONS. The MYEs are produced by a standard methodology in order to be “fair" to all local authorities. There is a national population estimate control figure so if for any reason the ONS can be persuaded to increase the estimate for one area, they have to make compensating reductions elsewhere.

LEICESTERA recent Cabinet paper (“The city's case for proper funding for its population", July 2007) observed that:

@UL@”there is considerable evidence that these [population] estimates understate the actual population, and the city is losing out on grant as a result. This, we believe, is largely due to undercounting of new migrants. Whilst it is impossible to be precise, we believe that present funding arrangements could undercount the City's population by some 20,000 people."@/UL@

The paper goes on to identify three key elements which render problematic the population projections used for 2008-11 grant calculations:

“the 2001 census faced problems counting particular groups; there is very limited data available to estimate migration between census dates; and the forward-looking projections are based on past trends, take no account of recent developments, and have been inaccurate over the most recent years for which data is available."

The characteristics associated with an undercount at the 2001 Census are identified as:

@UL@— multi-occupancy;@/UL@

@UL@— unemployment and deprivation;@/UL@

@UL@— young men, particularly in cities;@/UL@

@UL@— students in shared houses;@/UL@

@UL@— transient populations (who may not regard themselves as “resident");@/UL@

@UL@— language difficulties; and@/UL@

@UL@— large households.@/UL@

The contention that the 2001 Census produced an undercount was investigated through checks against the electoral roll and GP records.

The paper argues that “the settlement for 2007-8 underestimated the city's population by 5,500 solely as a result of using out-of-date projections, and before other data deficiencies are considered." This is attributed to the projections being influenced by “the apparent dip in population around the time of the census" but omitting migration-driven population growth in 2005 evident in ONS' own estimates.

A range of sources are examined to indicate the nature of the increase in migrant population, including:

@UL@— registrations with GPs;@/UL@

@UL@— National Insurance numbers issued to foreign nationals;@/UL@

@UL@— Workers Registration Scheme data;@/UL@

@UL@— the proportion of pupils in city schools described as “White European" (but not British or Irish);@/UL@

@UL@— attendance at the city's Polish Catholic church;@/UL@

@UL@— Child Benefit payments; and@/UL@

@UL@— Annual School Census (adjusted for residence).@/UL@SLOUGH

The Council's publication Migration works: harnessing the benefits of migration (2006) notes that “population estimates suggested that Slough's resident population was approximately 111,000 people at the year 2000", but that census data led to an amendment of the 2001 mid-year estimate to 120,600, implying a 14% rate of population growth during the 1990s, four times the national average. However, notwithstanding the substantial implied growth, the report observes that there is a risk that the census “continued to underestimate the town's actual population", noting that it has “many characteristics that make it one of the areas where the population is hard to count: it is ethnically diverse; has a relatively young population; and has high levels of mobility and overcrowding within its boundaries."

The focus of the report's concern is, however, on the period since 2001, when “there has been a sudden shift in the levels of migration attributed to the Slough area. Instead of continuing the previous upward trend, ONS population estimates and projections now suggest Slough's population is rapidly decreasing." The report contrasts this with an increase of 1,800 in the number of dwellings on the valuation list between 2002 and 2006.

The report identifies estimates of internal migration, and particularly international migration, as the greatest area of concern within the population estimates. It cites National Insurance registrations by non-UK nationals, which in 2004-05 stood at 2,570 compared with the ONS international migration estimate of 1,560.

Other evidence is drawn from:

@UL@— Annual School Census;@/UL@

@UL@— child benefit figures;@/UL@

@UL@— GP registrations (for children and the population as a whole);@/UL@

@UL@— birth rates;@/UL@

@UL@— dwelling completions; and@/UL@

@UL@— sewage flow.@/UL@

@ANAME@Annex B

THE 2001 CENSUS UNDERCOUNTThis annex has been compiled with reference to the ONS and Statistics Commission publications cited under

“References"

1. ONS was aware that the population was likely to be harder to count in 2001 than at previous censuses. It had prepared for this by trying to identify hard-to-count areas and concentrating resources there; and by instituting a “One Number Census" approach which supplemented the initial census with a Census Coverage Survey designed to compensate for any undercount.

2. Even so, the census figures announced for England and Wales in September 2002 were some 1.1 million (2 per cent) lower than the mid-year population estimates for 2000, which were based on previous censuses.

3. ONS explained part of the difference by, effectively, reducing the earlier population estimates. The 1991 census base was reduced by 351,000 (the error being attributed to an overestimation of the undercount at that census) and a further reduction of 305,000 was made to compensate for an overestimate of international migration in the decade since 1991.

4. A number of local authorities had challenged the census figures for their areas. A programme of local authority studies was initiated. One strand involved address matching, carried out in Westminster and Manchester, where the discrepancies between census and population estimates had been greatest. A second strand was based on an analysis of variables relating to all 376 local authority areas. An initial selection of nine authorities for further study was enhanced with 23 more authorities as a result of consideration of a wider range of variables.

5. The studies resulted in the addition of 26,200 to the Manchester population estimate, and 17,500 to the Westminster estimate. A further 64,000 was added to the population estimates for 13 of the other 30 authorities studied. The largest addition was 7,800 (Derby) and the smallest 1,500 (Hartlepool). Five of the 13 authorities treated in this way are in the North East. These increases were explained as compensating for localised enumeration difficulties and the shortcomings of pre-census identification of hard to count areas.

6. A net addition to estimates of 168,000, affecting 68 authorities, was made as a result of a separate analysis (known as the “longitudinal study adjustment") to account for an undercount of men, chiefly in the 25-34 age range.

7. This left an unexplained difference between the Census figures and the previous population estimates of 209,000.

8. The adjustments made as a result of the local authority and longitudinal studies were applied to mid-year population estimates, but not to the census figures themselves. This is justified on the basis that the studies do not provide evidence about the characteristics of the people missed in the census, though ONS accepts that “research on non-response generally, as well as common sense, suggests that the characteristics of the people missed by the census are likely to be different from those included."

9. ONS identified a number of actions to reduce “the risk that, when population estimates based on the 2011 Census become available, users of population statistics are once again faced with large differences between these and the mid-year population estimates based on the 2001 Census." Broadly, these involve:

@UL@(a) Identifying the implications for the 2011 Census.@/UL@

@UL@(b) Continuing to work with local authorities.@/UL@

@UL@(c) Assessing the implications of the migration reviews.@/UL@

@UL@(d) The 2007 Special Population Study.@/UL@

10. The elements of the first of these are discussed in the main body of this submission. The third has been supplemented by the work programme proposed in the Inter-departmental Task Force Report on Migration, again discussed in the main body of the submission. The 2007 Population Study has yet to be published, but will not provide the “improvement in the estimates for those areas where current population estimates are seen as less reliable", as envisaged in a recommendation from the Statistics Commission. Local authorities are liaising closely with ONS with the aim of securing a more reliable census in 2011.

11. In 2003 the independent Statistics Commission reviewed the problems connected with the 2001 Census in the borough of Westminster. In 2004 it updated this report alongside a wider consideration of the census and population estimates.

12. In relation to the 2001 Census, the Commission concluded that

@UL@(a) Inadequate address lists contributed to problems.@/UL@

@UL@(b) Estimation of international migration needs to be improved markedly.@/UL@

@UL@(c) Government departments do not know, or do not publish the rationale for, their own data requirements.@/UL@

@UL@(d) Co-operation across the public sector is needed to improve population estimates.@/UL@

@UL@(e) Disclosure control methods are a source of concern to Census users because they reduce the value of the data.@/UL@

13. The Commission recommended, looking forward to 2011:

@UL@(a) Government departments, local authorities and other public bodies should commit to working together in the planning and execution of the 2011 Census.@/UL@

@UL@(b) ONS should draw together public sector expertise in demography.@/UL@

@UL@(c) Targeted studies or surveys should be conducted in selected areas ahead of 2011 with a view to improving the population estimate for the most problematic areas.@/UL@

@UL@(d) Improvement of the quality of migration data should be addressed by ONS and the Home Office.@/UL@

@UL@(e) The creation of a robust and continuously updated address register should be a priority for government, led by ODPM [now CLG].@/UL@

@UL@(f) Government departments should assess and publish their own requirements in relation to the census.@/UL@

@UL@(g) ONS should develop one or more alternative measures of population in addition to “usually resident".@/UL@

@UL@(h) ONS should take account systematically of the concerns of local bodies about the accuracy of population estimates.@/UL@

@UL@(i) ONS should lead government-wide consideration of a common approach on disclosure control methods.@/UL@

14. In relation to population estimates, the Commission noted that “there is considerable potential to make greater use of the administrative information that is now available to improve the quality of population estimates and reduce the reliance on household surveys."

15. In relation to the Westminster case, it observed that “though several London boroughs offered to help carry out the Census based on their local knowledge, ONS took the view that this would not be appropriate because authorities had a financial interest in the population count. While statistical work must not be open to influence by anyone likely to benefit—or lose out— depending on the results, the loss of local knowledge exacerbated other problems."

November 2007

@HR25@

Memorandum from the LGA High Ethnicity Authorities Special Interest Group

INTRODUCTIONThe LGA's High Ethnicity Authorities Special Interest Group (HEASIG) is a special interest group of the

Local Government Association and represents many of the most diverse communities in England. The following local authorities are members of the group:

Barking & Dagenham, Birmingham, Brent, Croydon, Ealing, Enfield, Hackney, Hammersmith and Fulham, Haringey, Harrow, Hillingdon, Hounslow, Islington, Kensington and Chelsea, Lambeth, Lewisham, Newham, Redbridge, Slough, Southwark, Tower Hamlets, Waltham Forest, Westminster and Wolverhampton.

The Treasury Sub Committee has invited written evidence as part of its inquiry into Counting the Population and this document sets out the HEASIG group response. Several of our individual member authorities will also be submitting their own responses.

The following sections consider the specific issues that will be examined in the inquiry and this response focuses on the issues relating to local government finance.

1. The uses of population estimates, the definition of the population and the impact of any inaccuracies or inadequacies in population estimates;

@UL@— What are the uses of population estimates, and how far do current population estimates meet the needs associated with those uses? What are the effects of inaccuracies or inadequacies in such estimates?@/UL@

@UL@— How appropriate is a definition of the population based on the usually resident population in the context of the needs of the users?@/UL@

@UL@— How does the quality and range of population statistics in England and Wales compare with that available in Scotland, Northern Ireland and internationally?@/UL@

The population estimates and projections produced by the ONS play a fundamental role in relation to the distribution of public resources geographically across England—particularly in relation to local authorities and primary care trusts.

For example, population projections are used as the primary client indicator for the distribution of formula grant (£26 billion in 2007-08) to English local authorities and population data is also the key driver for specific grant funding streams for children's and adults social care, regeneration funding via the Neighbourhood Renewal and Safer Stronger Communities funds and Supporting People funding for vulnerable adults to allow them to live in their own homes. Population estimates are also used in the deprivation component for the allocation of the £28 billion Dedicated Schools Grant (the main source of schools funding) by the Department for Children Schools and Families and to target resources for youth services and anti social behaviour.

As around three quarters of the income of English local authorities is provided via government grant and subsidies population data plays a key role in determining the resources available to provide local authority services in each area.43 A failure to utilise accurate population data can therefore leads to a mismatch between the costs which each council incurs in delivering services and the funding it receives which therefore leads to pressures on council tax levels and reserves and eventually to restrictions on access to services (eg via thresholds applied to eligibility for social care) or more general reductions in front line service provision.

As our member authorities serve some of the most deprived urban areas of the country and thus receive additional top up funding for factors such as deprivation and density the potential impact of an inaccurate population count for our member authorities is even greater. Indeed some of our member authorities in inner London, for example, receive average levels of formula grant of up to around £1,000 per resident so a 1,000 undercount in population terms is equivalent to a shortfall in funding before damping of £1million.

43 74% of the revenue expenditure in 2007/08 is funded by central government resources (inc. business rates). Source: Local Government Key Facts: England, November 2007 http://www.local.communities.gov.uk/finance/stats/keystats/key2007.pdf

Many HEASIG authorities remain concerned about the robustness of population estimates in general, and the new population estimates and projections in particular which were released by the Office for National Statistics in August and September 2007. The scale of the changes, as well as the number of revisions that have been made since the publication of the 2001 Census, illustrate just how difficult it is to create reliable population estimates. However frequent revisions and retrospective amendments to previously published data also undermines confidence in the robustness of population data and thus the perceived fairness of local authority grant allocations.

Our members have particular concerns in relation to ONS methodology on migration estimates and our concerns are listed out in part 3 of this response. However there are also wider definitional issues which affect the accuracy and comprehensiveness of the current population data.

Of particular to concern to our member authorities is that the current definition of resident population only includes data on the number of resident in an area for more than 12 months and takes no account of the mobility of the wider population who may reside in different local authorities during the working week compared to their usual place of residence for census purpose.

As a result short-term migrants (ie people who plan to stay in the UK for less than 12 months) which will include international students, economic migrants and individuals on secondments are currently excluded from the population data used to allocate local government resources. This is of particular concern to ethnically diverse authorities whose attractiveness to migrants for work purposes is significant and whose populations generally have higher levels of mobility than experienced in other parts of the country.

Not only do these individuals utilise local authority services during their stay—including the majority of environmental, protective and cultural services as well in many cases social care and housing services—they also generate administrative costs for revenues, benefits and social care services linked to population turnover. This is of particular concern to our member authorities as many of us are located in the capital and major urban centres such as Birmingham, where high levels of population mobility exist. Recent research by the London School of Economics44 has estimated that the costs to local authorities of mobility in London alone are around £100 million per annum.

At present, there is no data on short-term migration at local authority level, though the ONS has produced experimental data at regional level. However, we believe these estimates appear to be significant underestimates compared to what our member authorities are seeing on the ground.

In HEASIG's view it is imperative that the ONS are given sufficient resources to produce robust estimates of short-term migration for publication alongside their mid-year estimates, which include only long-term migrants. This will enable appropriate planning and funding of the additional services likely to be required, some of which are extremely expensive to provide. Most short-term visitors to the UK are on family visits or tourists and are putting resources into the economy by paying for their stays of a few weeks.

On the other hand there are many short-term economic migrants who are staying for several months and who are presenting for housing, medical, legal, translation and other services. There are also short-term study visits to the UK. These students require accommodation, mainly in the private sector, and add pressure to local housing capacity. We strongly believe that the ONS needs to conduct more research in this area in order to produce robust local data on short-term migrants.

2. THE ROLE OF THE CENSUS@UL@— How does the Census contribute to the creation of population statistics? Is the current frequency of Censuses appropriate and, if not, how frequent should Censuses be?@/UL@

@UL@— What lessons have been learnt from the Test Census carried out on 13 May 2007? What methods should be employed for the distribution and completion of Census forms, including the use of the Internet?@/UL@

@UL@— What steps should be taken to increase the rate of responses to the Census, particularly in hard-to-reach areas? How effective are plans to measure the extent of non-response to the 2011 Census?@/UL@

@UL@— To what extent is there a trade-off between the length of the Census form and its role in providing population information? What questions should be included in the 2011 Census?@/UL@

44 Population Mobility and Service Provision, LSE Report for London Councils (February 2007)

@UL@— To what extent should Census questions and statistical information be coordinated across the United Kingdom and how effective are plans to ensure such coordination?@/UL@

@UL@— What forms should the outputs of the 2011 Census take, how and when should they be made available and how should they be integrated and coordinated with other information on population?@/UL@

The foundation of all UK population statistics is the decennial census and it is the only source for estimates/statistics for small areas. It has been used as a benchmark every decade to check on the estimates produced the ONS between censuses. If the frequency of censuses was reduced to five years, in theory, it would improve the accuracy of population statistics as they would be brought into line more often and it would avoid financial shocks to local authorities. However, it still would not produce accurate estimates for authorities with high level of population mobility.

We strongly believe that the ONS should target extra resources for the 2011 census in hard to count areas which have had a low Census response rate in the past. These tend to be areas where diverse communities exist and they are also likely to contain a large proportion of population whose first language is not English. These resources should be invested in appropriate training, seeking to ensure that enumerators are representative of the communities they are enumerating, ensuring that sufficient copies of the census forms are available in a range of languages and that translation and interpreting services are provided to ensure recent migrant communities are contacted, engaged and counted.

Table 1 below shows the ten areas with the lowest response rates to Census 2001 of which eight at HEASIG members and the other two are bordered predominantly by HEASIG members.

Table 1

AREAS WITH LOWEST RESPONSE RATES IN 2001 CENSUS

Kensington and Chelsea * 64%Hackney * 72%City Of London 74%Westminster * 74%Tower Hamlets * 76%Hammersmith and Fulham * 76%Camden 77%Southwark * 77%Islington * 78%Lambeth * 79%* HEASIG members

This illustrates the extent to which resources for the 2011 Census need to be tailored to the needs of particular hard to reach communities and the ONS will need to ensure that sufficient follow up with fieldwork in hard to count areas is undertaken following the initial enumeration to ensure that the information from the survey matches with the reality at ground level.

HEASIG is pleased that the ONS is carrying research and pilot studies to improve the methodology upon which the 2011 census will be carried out. Considering the impact of inaccurate population counts on the funding of local authorities and PCTs and the delivery of efficient and appropriate services, we believe that the ONS (or its successor organisation) should put in place a mechanism for reviewing anomalous counts in partnership with key stakeholders, including the local authority. These reviews should take place early in the ten-year census cycle, perhaps as early as within two years of the census taking place. Local authorities should be allowed to call for a review. It is in the review process that we believe it is most appropriate for the wide variety of administrative data sets may be drawn upon in order to cross-reference with the census data.

3. MID-YEAR POPULATION ESTIMATES@UL@— How accurate and useful are the mid-year population estimates for England and Wales, including information available for local authorities and Strategic Health Authorities?@/UL@

@UL@— How appropriate is the methodology by which mid-year population estimates are reached?@/UL@

@UL@— What are the most appropriate ways to ensure a common level of accuracy in mid-year population estimates for different areas and what steps should be taken to improve the accuracy in areas where population is hard to measure?@/UL@

@UL@— What is the right process for making revisions to mid-year population estimates so as to meet the needs of users?@/UL@

HEASIG recognises that the Office for National Statistics' current population estimates appear to be at odds with other datasets which imply much larger levels of population growth in many HEASIG member authorities:

Nationally there is mounting evidence from alternative estimates, the Statistics Commission, the Audit Commission, the Governor of the Bank of England, and indeed the ONS that suggests that the population estimates significantly underestimate the scale of international in-migration nationally. These national estimates are based on the IPS which is a survey of over a quarter of million passengers. Of those sampled, just under 3,000 were in-migrants.

The likely underestimation nationally resulting from this approach means that insufficient numbers of migrants will be allocated to each region, sub-region and consequently local authority in England. The effect of this underestimate is more significant for areas where international migrants form a larger part of the resident population. In addition, the international migration figures are an underestimate of the population because the figures exclude unauthorised or illegal migrants. The Home Office estimated that there were up to 570 thousand unauthorised immigrants living in the UK in 200145. A substantial proportion of these illegal migrants are likely to be based in HEASIG member areas for the reasons outlined above and perhaps also because large urban areas and existing communities provide a place for illegal migrants to remain unnoticed.

Indeed a number of our member authorities have commissioned work which suggests that the ONS's previous and revised estimates under estimate the level of population in their borough or have locally generated datasets which diverge from the ONS's figures.

@UL@— Brent and Enfield have undertaken detailed and extensive data matching exercises of individuals to addresses within the authority to produce estimates of their population. In Brent, the recent results of this work demonstrated that the ONS had undercounted the borough's resident population in mid 2006 by 17,700 residents or 6%. Similarly, the study in Enfield confirmed that the previous ONS estimates were undercounting the borough's population at least 3,300 residents. @/UL@

@UL@— Newham has looked at a range of data sources to indicate that its population estimates are an under prediction of their resident population. Comparisons with Flag 4 GP registration data by recent arrivals to the UK and new National Insurance Number Registrations suggest that the current estimates are undercounting the borough's population. Of even greater concern is a comparison which they have undertaken of resident pupil data from the schools census with the borough's ONS population estimates for the 5-16 age group. The number of pupils resident in Newham attending secondary schools—which are themselves an undercount of the relevant resident population because they exclude children in private schools and other pupils not educated in maintained schools settings—was 805 higher in 2005 and 1,436 higher in 2006 than the equivalent estimate of children of that age in the borough estimated by the ONS.@/UL@

@UL@— Hounslow has also undertaken research which demonstrates that its in migration according to the ONS since 2001 is lower both than its Flag 4 GP registration data and new National Insurance Number issues to foreign nationals.@/UL@

@UL@— Slough has pointed to a range of evidence including pupil numbers, the number of children in families receiving child benefit, GP and national insurance registrations, the increasing number of dwellings with no corresponding decrease in household size which suggests that the population is much higher than currently estimated by the ONS. Slough also commissioned the Data Management and Analysis Group (DMAG) of the Greater London Authority to produce independent population estimates. DMAG concluded that Slough's population was underestimated by up to 6,000 residents. Once again the new ONS population figures only go a small way to addressing this undercount in Slough.@/UL@

45 Woodbridge, J. (2005). Sizing the unauthorised (illegal) migrant population in the United Kingdom in 2001. Home Office Online Report 29/05.

@UL@— Despite the fact that Barking and Dagenham is at the forefront of delivering the government's sustainable communities agenda and is experiencing unprecedented levels of new house building the ONS is reporting that its population has fallen by 1,200 since 2001. According to its pupil roll data, however, the borough has seen the largest percentage increase in the number of its school children for whom English is not their first language in the country in the last five years which is indicative of significant levels of new migration. Similarly its GP registers have increased by 6,000 in the last three years alone and the number of new National Insurance numbers issued to foreign nationals resident in the borough is exceeding the ONS's in migration figures to the borough by up to 1,300 people per annum. This means that the borough is losing out on valuable funding which could be used to support community cohesion initiatives because its population estimates fail to reflect the reality on the ground.@/UL@

@UL@— Independent research conducted for the Council has found that the City of Westminster has over 13,000 illegal migrants within its boundary at any one time and that around 11,000 short-term migrants annually are “hidden" as they are not registered in official statistics. The research also identified that because migration is complex and has diverse characteristics in different areas of the country the current one size fits all measurement of migration and assumptions of migrant behaviour driving government policy is failing to address the needs of local communities in ethnically diverse areas.@/UL@

HEASIG has significant concerns about the ONS' methodology on International migration estimates. International migration estimates, which include only people who move to a country other than that of their usual place of residence for a period of at least a year, are currently based on estimates of Total International Migration allocated on the basis of the International Passenger Survey (IPS). International in-migrants are then allocated to local authorities on the basis of the distribution of in-migrants according to the 2001 Census and thus any errors in the census will be compounded in the subsequent population estimates.

The ONS has since conducted a review into international migration and this has resulted in some significant revisions to HEASIG member authorities' population estimates for the years 2002 to 2005. The key element of the changes introduced by the ONS estimates relate to the use of the Labour Force Survey to determine the allocation of migrants at regional level and changes to the clusters of authorities used to allocate migrations at sub regional level and in London as part of the process for distributing international inward migration between the 33 boroughs.

HEASIG has reservations about the proposed use of Labour Force Survey data by the ONS to allocate migrants between regions because of its low sample sizes and apparent failure to effectively target populations in hostels and housing of multiple occupation. There is no evidence to suggest that this offers any material improvement to the current methodology and in any case the LFS has a smaller sample of international migrants than International Passenger Survey which it is designed to supplement and replace.

Before the results of the ONS research were published, HEASIG argued that to improve population estimates the calculation should take into account a range of available data sources including National Insurance and local authority council tax and electoral registrations. This has not been taken forward in a consistent and robust fashion by ONS and as a result the group remains concerned about the level of robustness of the revised population data for many of its member authorities.

HEASIG is particularly disappointed that the changes made to the methodology for allocating migrants only deal with the redistribution of a fixed quantum of international migrants at national level, based on estimates from the International Passenger Survey (IPS). There are a range of concerns about the robustness of the IPS for measuring the quantum of migration—particularly its small sample size and possible inadequate sampling of modes and times of entry which would appear to be particularly important for recent international migrants to high ethnicity authorities (eg late night arrivals on low cost airlines from central and eastern Europe and from the Indian sub continent).

Many key decision makers have raised similar concerns that the current migration estimates are not fit for purpose. David Rhind, Chairman of the Statistics Commission, stated that “there is no dispute—current migration data is inadequate for many key government purposes" including grant allocation. The Audit Commission have also reported that the “current local population projections and diversity data do not fully reflect the recent increases in migration". The Governor of the Bank of England, Mervyn King, has also stated “that we do not have the ability to measure accurately at present, the size of the population". There is also significant anecdotal evidence to suggest that the current quantum is understated which suggests that the government and ONS need to revisit this issue as a matter of urgency.

4. THE ROLE OF SURVEY AND ADMINISTRATIVE DATA@UL@— What role can and should be played by survey data in the compilation of population statistics?@/UL@

@UL@— What role can and should be played by data from General Practitioners and other health service data in the compilation of population statistics?@/UL@

@UL@— What other data sources can and should be used in the compilation of population statistics?@/UL@

HEASIG has concerns about the use of the International Passenger Survey and the Labour Force Survey (LFS) as the bases for determining the quantum and distribution of migrants because of their limited coverage.

The LFS in particular suffers from significant weaknesses as a data source for allocating migrants, partly due to the small sample size of migrants in the survey and partly as HEASIG authorities generally have a much lower response rate compared to other regions. The ONS itself has identified that certain Household types are under-represented in the Labour Force survey as profiled below.46

Households underrepresented in the LFS include:

@UL@— Those renting from Housing Associations or whose housing comes with their jobs.@/UL@

@UL@— Those in converted or shared accommodation.@/UL@

@UL@— Households with three or more cars or vans.@/UL@

@UL@— Those comprising only 1 adult, aged 16-19.@/UL@

@UL@— Those with head of household aged 16-24.@/UL@

@UL@— Those with only single head of household.@/UL@

@UL@— Those with the head of household born in Commonwealth countries.@/UL@

There is also evidence, for example, that individuals in hostels and housing of multiple occupation —disproportionately more likely to be lived in by international migrants—may be under-sampled and that any follow up surveys with highly mobile individuals which are intended to improve the tracking of interviewees—may not be practical.

Similarly it is also apparent that the LFS sample size in relation to international migrants is lower than that from the International Passenger Survey which does not suggest that using this data source necessarily offers any improvements over the current arrangements.

Moreover, many residents living in communal establishments are not covered by the sample, and the coverage of short-term or recently arrived migrants is not comprehensive and thought to be poor. The quality of the final survey estimates are, of course, ultimately dependent on the accuracy of the population data used for grossing the survey sample, which themselves are highly problematic, particularly in HEASIG member authorities.

HEASIG also has concerns regarding the robustness of the International Passenger Survey. The International Passenger Survey is conducted at a handful of ports of entry and samples only 0.2% of what is believed to be all passengers arriving in the UK (250,000 people).

Of these 250,000 interviews only 2,800 were identified as migrants in 2004 (1.2%). The number of emigrants leaving the country is determined on an even smaller sample of just 750 in 2004. For example, the estimates for Pakistan, quite an important source country, were based upon 231 interviews of immigrants and 6 interviews of emigrants. The estimate of a net 3,000 inflow from the Caribbean was based on the difference between 28 interviews in and 6 interviews out. While the ONS has since taken steps to improve sampling the results of these will not feed into the population figures used to distribute local government funding until 2011/12.

46 Source:—Labour Force Survey User Guide—Volume 1 : Background and Methodology 2003

HEASIG believes that the ONS can make use of other data to inform their population estimates. For example, local authorities have records of council tax properties (including certain estimates of vacancy and second residences), new housing developments, houses in multiple occupation and children in schools. These sources could be compared as time series with ONS estimates to determine potential problems of consistency of estimates of change since 2001. While the electoral register does not cover all adults of overseas origin it may also be used as a comparator to estimated changes in the adult population. With regards to the number of migrants, the ONS can use the number of national insurance numbers issued to identify potential level of discrepancies between migration data based on GP registrations.

5. COOPERATION WITH STAKEHOLDERS@UL@— How effectively has the Office for National Statistics cooperated with stakeholders with an interest in and information relating to population statistics and how can cooperation be improved by the Statistics Board and by its Executive Office?@/UL@

HEASIG recognised that the ONS has participated in Central and Local Information Partnership (CLIP) meetings to enable central and local authorities to work together to develop effective statistical information and has arranged meetings with individual local authorities which had particular concerns relating to their population estimates. We hope that ONS will seek to engage effectively with representatives of local government (including local authority special interest groups) to ensure the success of the 2011 census in hard to count areas.

In conclusion HEASIG considers that If the ONS is to be able to perform its duties effectively in the future then additional investment needs to be provided by the Treasury to facilitate the more widespread use of local and national government administrative datasets in the development of national and sub national population estimates and projections to supplement the existing use of the International Passenger and Labour Force Surveys. This would enable more robust data which reflects the complexity of both short and long term international migration patterns to and from the UK and the increasing levels of mobility of the country's population both in terms of residency and employment. Until these changes are put in place resources for key public services will not be targeted to those areas with the greatest level of need.

November 2007

@HR25@

Memorandum from the London Borough of Merton

1. USES AND DEFINITIONSWhat are the uses of population estimates, and how far do current population estimates meet the needs associated with those uses? What are the effects of inaccuracies or inadequacies in such estimates?

Population estimates are used throughout the council to anticipate the potential need for services that are targeted at specific demographic groups. For example nursery and schools places, housing, social services and planning rely on accurate population estimates. In addition population figures are used to help ensure surveys represent the population.

However, the estimates are (i) inaccurate, (ii) out of date and (iii) not detailed enough in terms of age, gender, ethnicity, faith, language, disability etc to enable accurate and real-time mapping of needs.

The population estimates provide much needed updates to the census data, although the inaccuracies mentioned above may lead to an overall lack of funding for services, plus council departments being unable to adequately plan for changes in population.

How appropriate is a definition of the population based on the usually resident population in the context of the needs of the users?

Usually resident population does not reflect the change in population throughout the seasons. For example, in Merton Wimbledon fortnight sees a huge increase in visitors—the Metropolitan Police estimate that 400,000 people visit Wimbledon village during Wimbledon fortnight, over double the number of residents in the whole borough of Merton.

Similarly, a significant number of Merton school and college students are from out of borough and this is not at all captured by or reflected in the “usually resident" population data.

Usually resident population also does not pick up on residents who routinely spend several months abroad but return for part of the year. This includes older residents who live abroad and return to their UK base for treatment.

How does the quality and range of population statistics in England and Wales compare with that available in Scotland, Northern Ireland and internationally?

Not able to respond to this question.

2. THE ROLE OF THE CENSUSHow does the Census contribute to the creation of population statistics? Is the current frequency of Censuses appropriate and, if not, how frequent should Censuses be?

Whether the frequency of the census is adequate depends on the quality of the interim-year data sets, which are currently inadequate. Were they improved a ten-year census interval may remain adequate.

We would nonetheless recommend a more robust year 5 update to pick up on any major trends in between major censuses.

What lessons have been learnt from the Test Census carried out on 13 May 2007? What methods should be employed for the distribution and completion of Census forms, including the use of the Internet?

Not able to respond to this question.

What steps should be taken to increase the rate of responses to the Census, particularly in hard-to-reach areas? How effective are plans to measure the extent of non-response to the 2011 Census?

Not able to respond to this question.

To what extent is there a trade-off between the length of the Census form and its role in providing population information? What questions should be included in the 2011 Census?

Language spoken at home is included in the Welsh census, but this would also be useful in the English census, particularly in London where there are a large number of non-English speaking families.

To what extent should Census questions and statistical information be coordinated across the United Kingdom and how effective are plans to ensure such coordination?

Not able to respond to this question.

What forms should the outputs of the 2011 Census take, how and when should they be made available and how should they be integrated and coordinated with other information on population?

The full census results needs to be published online. Currently some of the more detailed tables are only available on request, or through specialist software such as Saspac.

3. MID-YEAR POPULATION ESTIMATESHow accurate and useful are the mid-year population estimates for England and Wales, including information available for local authorities and Strategic Health Authorities?

Based on calculations using local information sources, such as register of electors, languages spoken in schools, and national insurance registrations local estimates are that the population is at least 10,000 higher than the ONS 2006 mid-year population estimates.

Additionally, smaller area statistics are needed from the mid-year estimates, ideally super output area level or electoral ward. This would allow the council to address deprivation and demographic changes within wards.

How appropriate is the methodology by which mid-year population estimates are reached?

The mid-year estimates need a better calculation of emigration as well as in migration. The labour force survey is inappropriate for estimating migration and calculating population.

What are the most appropriate ways to ensure a common level of accuracy in mid-year population estimates for different areas and what steps should be taken to improve the accuracy in areas where population is hard to measure?

Mid-year data should draw on a wider number of sources (see A1 above) to arrive at a better calculation of population. This may produce a range rather than a point figure. Some detailed local work on which data set, or combination of data sets, are likely to provide the most accurate figures could be undertaken.

What is the right process for making revisions to mid-year population estimates so as to meet the needs of users?

See above.

4. THE ROLE OF SURVEY AND ADMINISTRATIVE DATAWhat role can and should be played by survey data in the compilation of population statistics?

Survey data could be useful in understanding the biases in reporting/ measurement likely in given cohorts of the population, but the methodology needs to be standardised throughout the country to ensure consistency.

What role can and should be played by data from General Practitioners and other health service data in the compilation of population statistics?

GP data is invaluable, although a correction needs to be applied for areas where high proportion of the population use private health care providers and in areas of high population churn where registering and deregistering may be delayed.

More detailed data could also be collected from GPs, such as ethnicity.

What other data sources can and should be used in the compilation of population statistics?

The mid-year estimate should use available statistics on the full range of encounters people have with official bodies. These would include child benefit, national insurance, sickness benefits, job seekers allowance, attendance allowance etc.

The accuracy of these estimates could be verified by techniques such as capture-recapture methodology on a small sample to improve accuracy of estimates and reduce double counting.

5. COOPERATION WITH STAKEHOLDERSHow effectively has the Office for National Statistics cooperated with stakeholders with an interest in and information relating to population statistics and how can co-operation be improved by the Statistics Board and by its Executive Office?

The ONS have involved stakeholders in a series of events to gather views on methods. Merton would welcome continued dialogue on new methods for calculating the population.

November 2007

@HR25@

Memorandum from the London Borough of Newham

1. USES AND DEFINITIONS@UL@— Current population estimates and projections are inadequate, as they do not appear to accurately reflect international migration, despite the recent review. @/UL@

@UL@— The population base needs to be segmented, to include short-term residents who are users of local services and infrastructure, as well as residents.@/UL@

@UL@— The uncertainty in the population estimates must be made explicit. ONS should lead in ensuring appropriate use of their statistics by others.@/UL@

@UL@— Population turnover should be quantified, as it impacts on the costs of providing services as well as community cohesion.@/UL@2. ROLE OF THE CENSUS@UL@— A fourth page for the Census is strongly supported, to meet information needs for the next decade for policy and resource allocation. @/UL@

@UL@— A five-year Census should be considered for London, to ensure that resources are accurately allocated in the intra-Censal period.@/UL@

@UL@— Clear publicity is required for the Census to succeed, publicising its anonymity and its role in public spending allocation.@/UL@

@UL@— Disclosure control needs to be sensitively applied to ensure that the data is not rendered unsuitable for use, as it was in the 2001 Census@/UL@

@UL@— Census data should be released for SOAs and wards as the sub-district levels (both are needed), and ownership of the geography needs to be resolved so that the Census can be geographically analysed without additional costs.@/UL@

@UL@— Addresses from OS and NLPG need to be integrated to ensure the greatest Census coverage, with clear plans for follow-up for non-response.@/UL@3. MID-YEAR POPULATION ESTIMATES@UL@— Poor migration data has undermined confidence in the estimates, and current plans to improve estimates are going to take some time to implement.@/UL@

@UL@— Government departments with data of use to ONS for improvement should be encouraged to assist with its early implementation to address inadequacies.@/UL@4. THE ROLE OF SURVEY AND ADMINISTRATIVE DATA@UL@— Survey data samples for London need to be increased to make it robust at sub-borough level.@/UL@

@UL@— Research should be undertaken in cooperation with relevant government departments to identify how administrative data can be used to strengthen population estimates and provide “reality checks".@/UL@

@UL@— GP registration data needs to be seen as a core dataset for population estimates for both internal and international movements, and used in conjunction with National Insurance Number data (NINO) for assessing current migration flow patterns for international migrants rather than relying on Census data alone.@/UL@5. COOPERATION WITH STAKEHOLDERS@UL@— Concerns about the population estimates need to be addressed more speedily, with the impacts of their use recognised. It is hoped that the Statistics Board will maintain an oversight of these concerns to ensure that they are addressed.@/UL@

@UL@— If SOA data are to be used for performance assessment and resource allocation, their accuracy and fitness-for-use needs to be made clear by the Statistics Board.@/UL@NEWHAM'S WRITTEN EVIDENCE SUBMISSION TO THE TREASURY SUB COMMITTEE'S INQUIRY INTO “COUNTING THE POPULATION"Submission from London Borough of Newham

1. USES AND DEFINITIONSWhat are the uses of population estimates, and how far do current population estimates meet the needs associated with those uses? What are the effects of inaccuracies or inadequacies in such estimates?

1.1 Population estimates are used for allocating resources, performance monitoring, developing policy and monitoring the effectiveness of policy and change.

1.2 Current population estimates are inadequate for these purposes because of their timeliness, accuracy and geography.

1.3 Census data is used for analysing population characteristics and is delivered only once a decade and this is far too long a period as changes takes place so rapidly. The Census takes too long to be delivered, so that by the time information is received it is woefully out of date.

1.4 Population estimates are used primarily to identify age and gender of constituent populations, but are only re-based every 10 years and the intervening methodology is inadequate to accurately estimate annual migration change. Because there is so little information available about migration (internal and international), it is impossible to obtain accurate data about age, gender or ethnicity in the intercensal period.

1.5 Another inadequacy of the estimates is that they do not count all people who are eligible for and access publicly funded services, because they count only those who are seen as “residents" and anyone who lives in or intends to live in an area for less than a year is not included in the population count.

1.6 The effect of the statistical inaccuracies in areas with high levels of population movement is that they will receive less funding per capita than those with more settled populations.

1.7 Performance assessments will be distorted as the population denominators used rely on faulty estimates, and as a consequence policy-making becomes less rooted in evidence.

1.7 The lack of robust population estimates for areas within a local authority diminish the ability to monitor change, shape policy or evaluate policy effectively.

1.8 While it is difficult to quantify the level of inaccuracy, there is a range of administrative data sources that consistently support the view that London local authority population estimates are clearly not reflecting local population changes. The growing discrepancy between official population statistics and the evidence of administrative data as well as local knowledge has resulted in a loss of confidence in ONS data.

1.9 As reliance on smaller area statistics has grown, so too has the unease at the use of the official population statistics, particularly when used for performance assessment and, more crucially, financial settlements.

How appropriate is a definition of the population based on the usually resident population in the context of the needs of the users?

1.10 Population estimates need to be produced for a variety of population counts, rather than only “resident" population. The Census includes anyone who returns a Census form, and can include people who do not meet the usual definition of “resident" (at least one year) if they chose to submit a form.

1.11 Areas with large turnover of population who are resident for much shorter periods still have costs attached to that short-term population, and it is essential that an estimate of this population is produced to be able to quantify the costs associated with them (for example, the administrative costs of changing account names, electoral registration, or introducing children to schools outside of normal start dates).

1.12 Short-term residents have clear impact on the business of local government, with use of housing, schools, translation services, healthcare, etc, and they are an essential part of planning for service delivery. Population estimates need to be produced for both long-term and short-term residents in order to be able to create effective policy and services.

1.12 In order to effectively address community cohesion, a clear understanding of the population dynamics is needed—which includes the level of “churn" in an area as well as the impact of the turnover on one's perception of an area. Current “resident" population figures will exclude some people simply on the basis of their length of stay, but for some purposes anyone in the area is of interest. Emergency planning will need an idea of the total number of people in an area, including short-term visitors as well as those who are currently “resident" in an area.

How does the quality and range of population statistics in England and Wales compare with that available in Scotland, Northern Ireland and internationally?

1.13 Not known.

2. THE ROLE OF THE CENSUSHow does the Census contribute to the creation of population statistics? Is the current frequency of Censuses appropriate and, if not, how frequent should Censuses be?

2.1 The Census adds a huge amount of richness to population statistics, providing a range of demographic characteristics such as ethnicity, employment patterns and family structures that are not so comprehensively available from any other source. The frequency of the Census is inadequate in areas of rapid change but probably suitable for areas with little population turnover. London needs a Census at least every 5 years, if one is to be held.

What lessons have been learnt from the Test Census carried out on 13 May 2007? What methods should be employed for the distribution and completion of Census forms, including the use of the Internet?

2.2 One lesson that does not appear to have been learned from the Test Census is that we need to have an integrated use of maps and addresses to effectively carry out the Census and to analyse it geographically. The current situation with OS “owning" geographic information needs to be resolved. Such data should be seen as a national resource and utilised freely for policy and analysis, rather than this current situation of conflict around who “owns" address and postcode data as well as the geographic boundaries used for analysis. The Census coverage needs to be as full and accurate as possible as well as easily analysed, and to do this, we need to have easy access to the underlying geographic information.

2.3 The method for distribution of Census forms needs to be determined by the needs of the area. Areas that are difficult to enumerate would be best served by personal delivery of forms to ensure that hidden households are enumerated. More settled areas can effectively be enumerated with post-out/post-back of census forms. There needs to be clear publicity about the Census to ensure maximum response, and clear follow-up plans for areas with low enumeration along with adequate resources to undertake follow-up.

2.4 Experience from the US has found that Internet Census forms are not cost effective, and it is preferred that funds are put into more adequate coverage (particularly in community languages), providing more follow-up coverage, or in providing an additional page of questions, rather than providing an expensive web-based form for data collection.

What steps should be taken to increase the rate of responses to the Census, particularly in hard-to-reach areas? How effective are plans to measure the extent of non-response to the 2011 Census?

2.5 There needs to be a clear collaboration between local authorities and ONS about the identification of properties to be enumerated, and the numbers of dwellings inside them. At present, there are two address lists—those derived from Ordnance Survey and those from the Local Land and Property Gazetteers (LLPG). Both probably have value, but could be strengthened with collaboration, and this would lead to more confidence in the address list for the Census post-out. Without a complete address list, it is feared that the post-out will miss hidden households or make it easier for some to simply opt-out of the Census. If forms are to be posted-out using only one address source, then more intensive follow-up will be needed to ensure that all households have been covered.

2.6 Enumeration of hard-to-reach areas will need to have local knowledge of the population, its language needs and how best to engage with them. Targeted means of raising awareness of the Census and how the data collected is used is likely to increase engagement. A well-resourced publicity campaign for the Census is essential.

2.7 We do not have full information on plans to measure non-response for the 2011 Census, and so cannot comment. It cannot be stressed enough that there is a clear need for adequate follow-up for non-response to the Census or for partial responses to Census questionnaires.

To what extent is there a trade-off between the length of the Census form and its role in providing population information? What questions should be included in the 2011 Census?

2.8 At present, the plans for a three-page Census do not adequately reflect the need for information, and a fourth page should be added to make the Census as valuable as possible. The additional page should include health/disability, language, carers, industry, qualifications, full-time/part-time working and income questions that are not covered by current surveys.

2.9 The Census has the ability to provide data for topics that are not comprehensively covered by any other source, and which are essential for policy development in an area of dynamic change. With adequate publicity stressing the use of the data, respondents will understand the use made of their responses, and therefore aid compliance.

To what extent should Census questions and statistical information be coordinated across the United Kingdom and how effective are plans to ensure such coordination?

2.10 There is less need for national comparability than to be able to obtain local comparators. It is reasonable to be able to adequately compare within a country, or to be able to compare different urban environments, but comparing Newham with Newark, or with Newry in Northern Ireland, is not essential for the borough's needs. Trying to compile a single Census form makes the information less useful, as some questions will not be suitable for all areas. For example, ethnicity needs to be much more detailed in a dynamic urban environment than it needs to be in a settled rural environment, but it is not cost-effective to ask for detailed ethnic groups in areas that are not very diverse. Disclosure control will diminish the detail anyway, and broader categories will be more useful in less diverse areas. It would be better to shape the Census to be able to reflect the population more accurately, while ensuring that aggregations can be used for comparisons.

What forms should the outputs of the 2011 Census take, how and when should they be made available and how should they be integrated and coordinated with other information on population?

2.11 The outputs should be geared towards the audience, to enable the most suitable analysis. Rather than imposing disclosure control at the outset, and therefore making small area data useless (see workplace statistics at ward level and below and Census Area Statistics tables for OA level from the 2001 Census), there should be a means of controlling the way in which material is reported in order to prevent disclosure. Data made available at OA level could be aggregated, and then subjected to disclosure control by licensed users.

2.12 Disclosure control rendered some of the Census unusable, and limited the use of other aspects of it, and must be implemented in 2011 with a clear understanding of user needs and intended outputs. OA-level data was not very useful in 2001, as the small geography was potentially disclosive, but there was no mechanism for obtaining the data at a more useful small geography that would no longer hold the threat of disclosure. Disclosure control was also the reason given for Census Area Statistics tables not being provided at SOA level—when the OA CAS tables were rendered useless by disclosure control but users were unable to access the data at a small area that would have been more useful and not so heavily affected by disclosure control.

2.13 Census material needs to be made available at a variety of geographies. At the very minimum, SOA, ward and district level data should be made available. If there are issues about disclosure through differencing, then basic information should be made available for wards, while cross-tabbed material would be available at SOA lower-level and middle-level (depending on sensitivity) that could be used to describe wards or other political boundaries.

2.14 The geographies used for Census should be tied to the boundaries currently used for reporting a range of other administrative data, so that change can be monitored over time. SOAs are the current geography for reporting a range of benefits-related statistics by Neighbourhood Statistics, but these require robust and believable population denominators which are initially based on the Census and annually updated.

2.15 That is currently not the case for SOA populations, as they are currently constrained to mid-year estimates (MYE) which, in some instances, lack credibility. Updated SOA populations need to reflect physical

development, and require a data feed regarding physical change and housing development at SOA level. This should be possible with the National Land and Property Gazetteer or local authority information on Council Tax properties coupled with communal establishment data from the Census, so that the population changes within the local authority are rooted in development changes.

3. MID-YEAR POPULATION ESTIMATESHow accurate and useful are the mid-year population estimates for England and Wales, including information available for local authorities and Strategic Health Authorities?

3.1 The current MYEs for London local authorities are unable to accurately reflect the population, particularly the migrant population and non-“resident" population in areas with high levels of population mobility. All people living in an area place demand on public services of some sort, whether they are deemed to be “resident" or not.

3.2 For Newham, as an area that has traditionally been attractive to international migration, the ONS population estimates have become more and more detached from the reality of the borough's population. As of 2006, the MYE estimate is now over 12,000 people lower than the GLA population estimate (which models population flows to new housing development).

3.3 Trends that are arising from the MYE populations are quite questionable and undermine confidence in ONS estimates. For example, while household sizes are supposed to be declining nationally, the central London boroughs that have been subjected to large revisions have had household sizes that were growing larger, given the population increase and the housing development in these boroughs. In contrast, the increase in new dwellings in Newham has not apparently been occupied by a commensurate increase in population with an increase of 919 dwellings resulting in a loss of 1,500 people in 2006. This is simply not credible.

3.4 Data from other administrative sources (DFES resident pupils, GP registrations from people coming from abroad, National Insurance Number registrations, Electoral Register), when it is all showing a contrary picture to that provided by ONS, also indicates that there is a problem with the estimates.

How appropriate is the methodology by which mid-year population estimates are reached?

3.5 The current methodology suited a time when populations were much less mobile, and no longer can adequately cope with the range of population movements, settlement patterns or new household structures. It may well suit areas where there is a lower level of population movement, but is inadequate for areas with a high mobility.

3.6 The methodology for determining migration (both internal and particularly for international) completely undermines the accuracy of the population estimates. The resulting migrant count bears little relationship to administrative data, and has undermined confidence in the estimates.

What are the most appropriate ways to ensure a common level of accuracy in mid-year population estimates for different areas and what steps should be taken to improve the accuracy in areas where population is hard to measure?

3.7 After identifying those areas where it is known that population is hard to measure, reference should be made to administrative data to ensure a “reality check" for the population estimates that have been produced, if the current modelling is to be used. ONS should then be able to make estimates of the range of potential error, so that any users of the estimates will be able to assess their fitness-for-purpose.

3.8 An alternative methodology should be considered that will rely upon administrative or survey data to estimate populations in a way that is more current, but perhaps less “comprehensive". It would be useful if a pilot could be done to be able to assess the accuracy of an administrative method vs. the traditional methodology following the next Census.

3.9 It may be that more government resources are required to improve the quality of local administrative data for the purpose of population estimation. Local data is more regularly collected and updated and provides a more reliable “current" picture of population in an area, but is not seen as sufficiently robust for ONS use. An investigation into which local data sets would be appropriate for comparability across areas would be helpful. National quality standards could be set for local authorities and other government agencies for maintaining their administrative data for use in this way.

What is the right process for making revisions to mid-year population estimates so as to meet the needs of users?

3.10 The best way is to have no revisions at all—to get the method right in the first place.

4. THE ROLE OF SURVEY AND ADMINISTRATIVE DATAWhat role can and should be played by survey data in the compilation of population statistics?

4.1 In order to have more timely population statistics, it is essential to develop a survey that gives confidence down to middle-level SOA, in order to provide local authorities with data that can be effectively used for monitoring and performance assessment as well as for service planning.

4.2 Any survey needs to be sufficiently robust down to, at the very least, local authority level for some basics—eg migrants, but should be to a lower level for population characteristics such as ethnicity, household structure, tenure, employment, etc. Survey data could be modelled using administrative data to provide small area population estimates.

4.3 The survey needs to cover the entire population and include communal establishments in the sampling to produce more accurate population data.

What role can and should be played by data from General Practitioners and other health service data in the compilation of population statistics?

4.4 The GP registration data is the single source of data that provides the greatest coverage of the population, and should be used not only for health services delivery but also to provide a strong element of population statistics. The stock data needs to be regularly cleaned, as it is known that the registers constantly over-represent the local populations due to movers not changing their addresses in a timely manner.

4.5 If robustly recorded, this data could be used to provide age/gender/ethnicity statistics, which are the key data needs as well as additional information on disability levels in an area (it is impossible to find comparative statistics on disability).

4.6 The flows data from the GP register is the core dataset used for internal migration, and the data collected on new international migrants should also be utilised. More resources should be made available to maintain “clean" records, as it is the key resource for estimating migration and this would then improve the stocks data as well.

4.7 It is essential that GP registration data retains information on new registrations from abroad, as this is the only administrative source that will cover a large proportion of migrants, and any new Department of Health information system should be designed to include essential demographics for not only their business, but for the wider government information needs.

4.8 It is believed that due to the sampling of the International Passenger Survey (IPS), many international migrants are not accurately allocated to local authorities, but once they register with a GP they are then counted as out-migrants from a district. Research should be undertaken by ONS to identify what the relationship is between new international registrations to the GP list (“Flag 4s") and their “resident" status.

4.9 Mayhew Associates are currently undertaking work for Newham, using the GP register to obtain an alternative population estimate, and this will be made available when it is finished. We also hope to have a particular study of Flag 4 registrants, to identify the proportion of these which have become resident in the ensuing year.

What other data sources can and should be used in the compilation of population statistics?

4.10 The Pupil Level Annual School Census (PLASC) can provide some information on movement that may not be picked up by the GP registration data. This also provides a source of comparative data to quality check the MYE. At present, Newham currently has 1,500 fewer secondary school age children in the MYE than exist in the DFES data of Newham resident secondary school pupils attending state schools.

4.11 National Insurance Number applications (NINO) provide an indication of the settlement patterns of new international migrants and could be used to help apportion IPS data inflows. Rather than using settlement

patterns from the increasingly out-dated Census data, NINO registrations offer a current insight into where people are choosing to settle.

4.12 Additional information needs to be made available to be able to filter out records that are no longer in use—eg GP registers need to be regularly cleaned and tax records might be able to identify NINO accounts that are no longer in use, presumed to have returned to their home country. This information could then be used to help model outflow data.

4.13 The Home Office should be identified as a partner agency in getting better information about both new migrants to the country and those who are leaving the UK to settle elsewhere. The Home Office should be encouraged to hasten the introduction of key changes that will enable ONS to have better data for their population estimates.

4.14 It is essential that a swift move to utilisation of the National Land and Property Gazetteer (NLPG) across government is promoted. This would enable much more effective use of local authority's own administrative data and make it more robust.

4.15 Use should be made of local authority data on housing development and other changes to dwellingstock. This can include Council Tax data (for both stock, flows and vacant/second properties), as well as planning data on development plans, environmental health and planning data on Houses in Multiple Occupation (HMOs).

4.16 Administrative data should be used as a “reality check" to the population estimates and projections produced by ONS, to ensure that the statistics produced are credible.

5. COOPERATION WITH STAKEHOLDERSHow effectively has the Office for National Statistics cooperated with stakeholders with an interest in and information relating to population statistics and how can cooperation be improved by the Statistics Board and by its Executive Office?

5.1 Newham's experience with raising concerns about the population estimates since 2001 has been marked by a lack of response. Our concerns, and the evidence to support those concerns, have been acknowledged, but the outcome has not changed. While it is understood that ONS is focussed on creating a nationally comparable data set rooted in a standard methodology, this appears to be completely at odds with the needs of individual local authorities for reliable, robust and ultimately believable population statistics for resource allocation and policy-making.

5.2 The ONS standard methodology has been hidebound and unresponsive to changes in modern migration patterns and definitions, resulting in counts of people in an area that bears little relationship to demands on local services and which ultimately do not robustly support the creation and monitoring of high quality services.

5.3 It has been very useful to have the Statistics Commission providing scrutiny of the use of statistics, and it is hoped that the new Statistics Board will continue in this role. As statistics, and the interpretation of statistics, become more widely used in government, it is important to ensure not only that the statistics are robustly created but also that they are not misinterpreted and misused. Newham continues to be concerned about the use of population estimates—particularly at a low-level geography—to assess performance. Current population estimates will amplify the error in statistics reported at SOA level, and it is imperative that the use of such statistics for performance and resource allocation be given careful consideration.

November 2007

@HR25@

Memorandum from Luton Borough Council

EXECUTIVE SUMMARY@UL@(i) Luton Borough Council welcomes this opportunity to submit written evidence to the Treasury Sub Committee into “Counting the Population".@/UL@

@UL@(ii) The Council have identified several uses of ONS population estimates including:@/UL@

@UL@(a) funding allocation/service planning and delivery;@/UL@

@UL@(b) informing policy development and decision making;@/UL@

@UL@(c) as denominators for performance monitoring indicators;@/UL@

@UL@(d) the basis for population forecasts and sub estimates, and;@/UL@

@UL@(e) share of grant from central government is heavily influenced by our population statistics.@/UL@

@UL@(iii) The census provides the most comprehensive and sought after socio-economic data locally. Its broad question base and thorough geographical coverage allows for meaningful and in-depth analysis which is not available through other means.@/UL@

@UL@(iv) As part of the 2001 census preparations, the ONS should be actively engaging with stakeholders to ensure the highest possible response rate. We believe local authorities are key stakeholders in the census process and that we are best placed to lead on local issues which may affect the census outcome. We welcome greater inclusion in the census process.@/UL@

@UL@(v) The major issue with population estimates is their inability to adequately count or allocate at local authority level international migration numbers. Population estimates are increasingly diverging from local observations. In part, this can be reconciled by definitional issues around the term “resident", however, there is sufficient evidence to suggest that the methodology (mainly due to the data which feeds into it) is inadequate in measuring international in-migration.@/UL@

@UL@(vi) The top-down approach to population estimates is in-flexible. There needs to be a process in place for allowing local authorities to challenge mid-year population estimates where there is a wealth of local knowledge administrative data which supports this.@/UL@

@UL@(vii) The ONS should look to incorporating some administrative data in their processes when producing population estimates. Ideally this data can be incorporated to enhance methodology, but can be used as a quality check at the very minimum.@/UL@

@UL@(viii) Short term migration, which does not appear in the mid year estimates, provides a particular concern to us. Short-term migrants have access to and place demand on our public services, yet these figures are not reflected in our population base and therefore go largely ignored and unfunded by central government.@/UL@

@UL@(ix) We recommend that there is an overhaul in the ONS approach and that there is more meaningful engagement with local authorities on demographic matters especially the census and mid year population estimates. The ONS should be using local authorities as critical partners given that we are major stakeholders of the data.@/UL@

Given our issues with population estimates, Luton Borough Council has decided to contribute to the evidence base for this sub-committee. We recognise the critical importance of accurate population statistics in which everyone has confidence such that we can avoid speculation and misrepresentation and the adverse impact on community cohesion that this may have. All efforts should be made by the ONS, the Government and other stakeholders to ensure the publication of accurate population figures.

Luton Borough Council welcomes the opportunity to feed into the debate on population statistics in the UK:

1. USES AND DEFINITIONS1.1 Luton Borough Council has identified several uses of ONS population estimates:

1.1.1 Population estimates form the basis of local authority funding. It is vital that estimates are accurate to ensure fair distribution of funding geographically. At present, if the mid year estimates are underestimated, as we have suspected for some time in Luton, then there is an adverse effect on our level of funding and hence the level of services we can afford to provide and maintain. Similarly, in areas where there is an overestimate of the population, there is a pro rata increase in funding and a greater likelihood of those local authorities delivering and maintaining effective services. It is imperative that the population estimates are accurate and reflect what is happening at grass roots level or there will be (and are) major inconsistencies in funding between authorities.

1.1.2 Population estimates are used within the Council to distribute resources, inform policy and decision making and provide a contextual framework within which all services operate. A lack of accuracy in the official statistics can and does have severe impact in each of these areas. We have noticed that provisions made for services given the official population figures have been inadequate, such as reception year school places where the take-up far exceeded the projection.

1.1.3 The official estimates are used as denominators when calculating some performance indicators for the purpose of performance measurement—whether statutory or otherwise. Inaccurate population estimates can result in an authority either deemed to be over-performing or under-performing based on its population overestimate or underestimate respectively. This can have a further knock on impact on funding and/or inspection regimes.

1.1.4 ONS population estimates form the basis of locally maintained population forecasts and sub estimates. These forecasts are only as good as the data that feeds into them, therefore it is important that the bases are correct (with an acceptable margin of error). We believe our population to be underestimated and as a consequence we have had to look to other data sources (riddled with their own inaccuracies and inadequacies) to supplement our knowledge of the local area. This does not aide us in forecasting and planning for the future.

1.1.5 Through the research and information role, local authorities facilitate the dissemination of population estimates to the general public. People and businesses have access to the information which empowers them and allows them to make decisions they deem appropriate in their personal or business environment. An implication of this is that investment decisions are made on inadequate evidence.

1.2 At present, estimates are produced for the resident population, as per ONS definition:

@UL@”The estimated resident population of an area includes all people who usually live there, whatever their nationality. People arriving into an area from outside the UK are only included in the population estimates if their total stay in the UK is 12 months or more. Visitors and short-term migrants (those who enter the UK for 3 to 12 months for certain purposes) are not included. Similarly, people who leave the UK are only excluded from the population estimates if they remain outside the UK for 12 months or more. This is consistent with the United Nations recommended definition of an international long-term migrant. Members of UK and non-UK armed forces stationed in the UK are included in the population and UK forces stationed outside the UK are excluded. Students are taken to be resident at their term time address." @/UL@

Given the climate of increased international migration in recent years, it would seem sensible to readdress this definition of “resident population". From a local authority perspective, our services are not limited exclusively to the “resident population", but the population which is physically here at any given time. In addition, we are unconvinced that students are allocated effectively to local authorities, especially given the large international student population in Luton (although this is not an issue unique to Luton). There is a case for developing Mid Year Estimates which consist of two elements; the “resident" population, and; the “visitor" population (short term migration). This is the approach being adopted by the ONS thought the IMPS project, and whilst we welcome it, we do believe this should have started sooner.

1.3 There are lessons to be learned about the quality and range of population statistics available in Scotland, Northern Ireland and internationally. This is an area which needs careful consideration and we would be in favour of gaining a better understanding of all the alternatives leading to consensus on the solution.

2. THE ROLE OF THE CENSUS2.1 Luton Borough Council recognises the value of the Census. It is the most comprehensive socio-

economic dataset available and as a result remains the most sought after information locally. We support, as a

minimum, a long term commitment to having a decennial census. Six years on, we believe the data from the 2001 census to be so out of date in terms of a detailed reference about our population. We will continue to work unguided until we get more accurate and detailed information about the make-up of our residents, which will be after the 2011 census. As a consequence, it would aide local demographers if the frequency of the census was increased, and hence our ability to properly measure change at a more meaningful interval and geographical level. With increased neighbourhood initiatives, it would prove invaluable if we could measure regularly at a neighbourhood level, rather than every 10 years as at present. An alternative approach to increasing the frequency of the census would be for the ONS to make greater use of administrative data, based on the assumption that it is available and in a format which allows for statistical manipulation.

2.2 The Test Census will provide many lessons which will prove useful. From our perspective, given undercount in the last two censuses, the key issue is one of co-working to engage with the public. We believe that the ONS needs to work closely with local authorities to gain a better understanding of issues pertinent to individual areas to ensure a high response rate overall. Localities where response has been low in the past need to be targeted and we believe that local authorities are in a better place to facilitate this than the ONS. We would welcome greater local authority inclusion in the census process. In addition, efforts need to be made to ensure the census adequately captures new international migrants (in particular from the A8 countries), many of whom are not sufficiently covered in other official statistics.

2.3 Response in the 2001 census in Luton was low, at 86% (19th lowest).

@UL@— Luton has many hard-to-reach components which make its population difficult to measure, such as a relatively young and ethnically diverse population and a large student presence. In addition, there are significant communities settling in Luton that have not been identified by the official statistics (such as those from A8 countries), who will need to be consulted and engaged to ensure they understand and contribute to the census process. @/UL@

@UL@— For the 2001 census, the ONS promoted an increased working with local authorities in engaging the community. The borough council is disappointed that this did not happen and believe this was a missed opportunity to help increase response rates.@/UL@

@UL@— Issues affecting engagement, such as language, need to be pre-empted and the local authority can assist the ONS on this. @/UL@

@UL@— In preference to any other source, we strongly recommend that the National Land and Property Gazetteer (NLPG) is used as the comprehensive address database and that the addresses are drawn as close to the census as practicable. At local level, ONS should be liaising with local land and property gazetteer (LLPG) custodians.@/UL@

@UL@— There needs to be adequate provision for responses to be made to the census in a variety of formats (ie whether hard copy or electronically), and in varying languages/scripts and font size.@/UL@

@UL@— It is anticipated that co-working between the ONS and local authorities during the 2011 census can and will help alleviate some of the issues and prove to increase the response rate. @/UL@

2.4 We recognise there is a trade-off between length of the Census form and its role in providing population information. Whilst we welcome the inclusion of questions relating to month and year of entry, nationality, citizenship, national identity and second residence we believe that restricting the census to a three page format is short-sighted. Questions relating to qualifications and carers should not be sacrificed to accommodate these new additions. We are in favour of extending the questionnaire to a fourth page, even at a cost of the quoted £20 million. Indeed, we believe there are still further questions which need to be included in the census, such as questions relating to disability as these will assist public bodies in meeting their duties under the DDA. We also feel that an income question would be useful analysis and that effective engagement on the census may remove resistance to this question.

2.5 The layout of the Census form needs to be reviewed in light of best practice on survey design. Is the design and layout of the census form encouraging a good response rate?

2.6 As best practice, it would be most useful if the Census questions and the statistical information were coordinated across the United Kingdom. At local authority level, however, our primary concern is data connected to our authority, our audit family and our government office region.

2.7 Output from the 2011 would be in great demand, therefore we suggest the data be made available as soon as possible. We support the one-off delivery of all census data and not the phased approach as was the case in 1991 and earlier.

2.8 ONS should also improve their processes on the migration and “travel to work" (TTW) matrices such that when this data is released the suppression methods do not yield the data meaningless.

3. MID-YEAR POPULATION ESTIMATES3.1 The accuracy of the mid-year population estimates has been questioned in Luton over recent years

given our large student population and the increased level of international migration experienced on our streets. For a town like Luton, the ineffective measures for migration currently in place are having an adverse effect in our mid-year estimates. According to the ONS, our population took a dive in Mid-2004 but has now recovered causing a net increase of only 900 people since 2001. Yet, our service managers are dealing with unprecedented demands. Our own estimates show that our population is far larger than the mid year estimate. This is partly due to definitional issues around the term “resident", but also because of an underestimate in Luton's international in-migration. We believe the latter is a derivative of the inaccurate estimates of international in-migration at national level, compounded by continual misallocation at the local level.

3.2 Whilst we have no issue with the general methodology if the data is correct, we are sceptical of the data which currently contribute to the components of change relating to migration. We believe that the International Passenger Survey is not fit for purpose despite measures to enhance the survey, and that GP re-registration as a tool for measuring internal migration is full of anomalies and prone to error. We suggest more robust measures be put in place which will effectively measure both aspects of migration. The top-down approach for distributing international in-migration numbers is unsound and needs a drastic overhaul.

Once accuracy issues are ironed out, the mid-year population estimates would be more useful to local authorities by including an ethnicity and ward level (or upper/middle layer super output area) analysis to compliment the age and gender data already available. This is the type of data we are increasingly being asked to supply in an attempt to measure our population profile and changes to it. Given the level of neighbourhood level working and the extent to which this expected of local authorities, analysis at a smaller geographical area is being requested more and more, yet the official estimates are inflexible and do not deliver on this front. Where we are asked for neighbourhood demographic data at present, we are having to rely on output from the 2001 census which we know is out-of-date as a context for the borough.

3.3 We believe there needs to be a thorough review of the data sources for the components of change relating to migration. Once the inaccuracies within these data streams are removed, the mid year estimates will begin to reflect true population. There is a wealth of administrative data that can be used to compliment the mid year estimates which include PLASC, WRS, NINo, housing completions, waste management data, local survey data, utility usage data, etc, and the ONS needs to recognise their value and seek to incorporate some of these in their methodology. If nothing else, these data sets have assisted Luton Borough Council in determining our reaction to the mid year estimates.

3.4 Revisions to mid-year population estimates are necessary from time-to-time in the event of more accurate data becoming available. From a local authority perspective, it would be useful if all the data were made available at the same time rather than a staggered dissemination, as was the case for the 2002-2005 revisions. It is more efficient to analyse the changes caused by revisions if all the data is released in one go.

3.5 We have reservations about the methodology used to calculate the mid-year population estimates in that once the national/regional totals are fixed, there is no scope for further alteration as a result of identified inadequacies in the data at the local level. In effect, the methodology does not allow changes in one local area without having a corresponding opposite affect in other areas. We question whether this is appropriate and whether a bottom up methodology could be developed.

3.6 At present, mid-year population estimates are published and hence available for use. Given the well documented criticisms of the mid year population estimates each year, we suggest that the ONS make the figures available to local authority demographers on a confidential consultation basis such that these figures can be reviewed

in light of local knowledge (essentially use local authorities as a critical friend). For this to work, it would require a more flexible approach by ONS but, if successfully implemented, would give all stakeholders greater confidence in the mid-year population estimates.

4. THE ROLE OF SURVEY AND ADMINISTRATIVE DATA4.1 At present, the International Passenger Survey is the major survey affecting the mid-year estimates.

We are unhappy with its scope and believe the margins of error to be too great. This survey places too much onus on intention of individuals rather than actual recorded behaviour, and as a result, the output suffers detrimentally. We firmly believe this survey needs to be overhauled and a more robust and appropriate mechanism to measure international migration be put in place of it. Any survey superseding the IPS needs to adequately cover all ports of entry and ensure individuals are checked in and out of the country, ensuring full stay locations are recorded. This will assist in regional and sun-regional allocation of individuals to local authority areas in a more meaningful fashion.

4.2 In our experience, GP register data can provide a wealth of demographic data providing it is made available to local authorities. GP registration data should record date of birth (and hence age), gender, residence postcode (to assist in measuring populations at ward or output area), and ideally record ethnicity. This would allow continual monitoring between census of basis population size and characteristics. However, discussion with the Luton PCT suggests that some of the information may not be as reliable as anticipated and GP lists are prone to inflation. The effect of international migration on GP list is also an area which needs to be explored further. Indication suggests, for example, that A8 workers are less likely to engage with the NHS via GP's and more likely via Walk in Centres. If all these issues can be tackled, then GP data may prove invaluable as a mechanism for population statistics.

4.3 There is a plethora of administrative data available to us, unfortunately it is not always accessible or in the most useful format. However, in the absence of anything better, Luton Borough Council has used administrative sources to gain a better understanding of our changing population:

@UL@4.3.1 Pupil Level Annual School Census (PLASC)@/UL@

@UL@4.3.2 Child benefit data@/UL@

@UL@4.3.3 New national insurance numbers (NINo) data@/UL@

@UL@4.3.4 Workers registration Scheme (WRS)@/UL@

@UL@4.3.3 Council Tax data@/UL@

@UL@4.3.4 Electoral Registration data@/UL@

@UL@4.3.5 University admissions data (HESA)@/UL@

@UL@4.3.6 Police authority statistics@/UL@

Luton Borough Council is willing to work with the ONS in developing methodology to use such data.

5. COOPERATION WITH STAKEHOLDERS5.1 We believe the ONS has to be slow to react to criticism of their population statistics from local

authorities. There appears to be no effective method to challenge estimates within a regular forum other than via media scrutiny and hype. Over recent months, there has been a more open approach adopted by the ONS in engaging with local authorities through workshops such as IMPS, although this has been in reaction to the ongoing discontent felt by many authorities. The ONS is too removed (despite having a regional presence) from the issues and experiences of demographers within local authorities.

5.2 There is a perception within local authorities that government departments work in silo's. If population statistics are to improve, then the ONS will need the cooperation of other government departments such as DWP, DCLG, DCSF and others. There has to be real emphasis placed on joining-up datasets if administrative data is to play a greater role in the production of population statistics.

5.3 We recommend that there is an overhaul in the ONS approach and that there is more meaningful engagement with local authorities on demographic matters especially the census and mid year population estimates.

As mentioned previously, we believe that consultation drafts of mid-year population estimates should be published and local authorities views sought. For this to be successful, the ONS will need greater acceptance of robust local evidence that challenges their figures and methodologies that allow changes to the figures.

5.4 ONS need to increase the level of local authority involvement in the census processes especially regarding community engagement. This is on the basis that local authorities will have their own strategies for community engagement based upon local experience and will be able to engage local strategic partnerships and the voluntary sector. A “one size fits all" model from ONS will probably be unsuccessful.

November 2007

@HR25@

Memorandum from the Metropolitan Police Service

The Metropolitan Police Service (MPS) will not be making a detailed submission in relation to each of the areas being considered by the Sub-Committee, not least because we're not in a position to provide a detailed response to questions such as “The role of the census".

However, the MPS is a key stakeholder in the use of population data, and we therefore want to take this opportunity to outline our position in relation to the first area under consideration—the uses of population estimates, and the impact of inaccuracies or inadequacies in those estimates.

Key decisions around police funding and performance management are based on population estimates. The presentation and management of police performance statistics, used by Home Office, HMIC and others to assess Force and Basic Command Units (BCUs), are based on offending rates per thousand population. Clearly any significant inaccuracies in the population estimates have the potential to adversely affect Police Forces at both BCU and force level, in terms of resource allocation, financial planning, and performance management.

Engagement with communities is also integral to the continued support for policing in the UK. Reliable estimates of the make up of the population are therefore of paramount importance in order to facilitate continued engagement, particularly with new communities.

We have had sight of the submissions being made to the Sub-Committee by both the Greater London Authority (GLA), and Newham Borough. We support the thrust of their submissions but more specifically, we feel that the following points are particularly noteworthy:

@UL@— alternative information sources (GP registrations, school records etc) tend to suggest that there are significant inaccuracies with the current estimates;@/UL@

@UL@— as London accounts for about 40% of net international migration, it is far more likely to be affected by inaccuracies in relation to this area—which is considered to be one of the least reliable elements of the population estimates;@/UL@

@UL@— it is vital that a means of taking short-term migration into account is included within future estimates;@/UL@

@UL@— a targeted approach to improving the accuracy of the census and population estimates, with reference to alternative data sources, would appear to be a sensible way forward.@/UL@

In summary the accuracy and reliability of official population estimates are crucial if confidence in them amongst end users is to be retained, and scarce resources are to be appropriately targeted. This is especially important for the MPS in terms of the population and development expansions, which we are expecting to occur in East London and the Thames Gateway development. There is also the question of the 2012 Olympic and Paralympic Games, and this is important more generally for the police service at a time when Government is requiring us to achieve significant cashable improvements in efficiency and productivity as well as further reducing crime and improving the quality of the service we provide to the public.

November 2007

@HR25@

Memorandum from the Equality and Human Rights Commission

1. The Equality and Human Rights Commission is pleased to respond to the Treasury Sub-Committee's invitation to submit written evidence to its inquiry into Counting the Population.

2. The Commission is a non-departmental public body (NDPB), established under the Equality Act 2006, which opened on 1 October 2007. It brings together the work of the three previous equality commissions, the Commission for Racial Equality, the Disability Rights Commission and the Equal Opportunities Commission, in relation to discrimination on the grounds of race, disability, sex and transgender status and takes on responsibility for the other aspects of equality: age, sexual orientation and religion or belief, as well as human rights.

3. The new Commission will be working to eliminate discrimination, reduce inequality, protect human rights and build good relations, ensuring that everyone has a fair chance to participate in society. Two of our priorities will be to undertake a programme of authoritative research and statistics projects, and by 2010 to prepare our first “state of the nation" report. These reports will be produced at least every three years and will monitor society's progress towards equality, including statistical indicators where appropriate.

THE ROLE OF THE CENSUS4. Once results are published, hopefully in time for the second state of the nation report in around

2013, the 2011 Census should provide a rich source of equality statistics. The Commission will be looking to the 2011 Census to provide information that is not available from surveys, such as:

@UL@(a) population estimates for each aspect of equality (sex, age, ethnic group, religion or belief, disability and sexual orientation) and their distribution across Britain;@/UL@

@UL@(b) population estimates for combinations of equality characteristics, eg older people by ethnic group and disability;@/UL@

@UL@(c) measures comparing groups in terms of different aspects of inequality or disadvantage, including: qualifications, employment, full-time/part-time work, access to jobs across occupations and industries, socio-economic status, income, and related characteristics such as: caring responsibilities, marital / civil partnership status, and language;@/UL@

@UL@(d) interactions between equality characteristics and different aspects of inequality and disadvantage.@/UL@

5. The above requirements call for some questions to be asked which are currently only proposed for a fourth page of individual questions (qualifications, industry, carers and language) and some not currently expected to be recommended for inclusion (sexual orientation and income).

6. The Commission therefore urges the provision of sufficient resources to fund a fourth page of individual questions in the 2011 Census and the inclusion of questions on sexual orientation and income in addition to those currently proposed for a fourth page. Space could be found for these by deleting other questions, eg either the question on term-time address or on second address, since these clearly overlap, and the question on intention to stay in the UK, which seems inappropriate for a Census since it requires a prediction of future behaviour.

7. In view of its GB-wide remit, the Commission considers the coordination of the England and Wales and Scotland Censuses extremely important so as to ensure the comparability of Census results across Britain.

8. We would also wish to see the 2011 Census results produced as soon as possible after Census day, consistent with the need to avoid major amendments and revisions after publication. Outputs from the Census should be made available in suitable formats for the wide range of potential users and made fully accessible, eg websites should comply with the latest accessibility standards.

POPULATION STATISTICS IN GENERAL9. The Commission would like to see population statistics covering a much wider range of population

groups. In order to fulfill its remit, it will require statistics on the population size and composition of all equality groups, ie by ethnic group, religion, disability, sexual orientation and transgender status, as well as by sex and age. Anticipated uses include:

@UL@(a) Monitoring the size and composition of these “equality groups".@/UL@

@UL@(b) Comparative figures required by public organisations to comply with their responsibilities under the disability, gender and race equality duties.@/UL@

@UL@(c) For comparison with other data sources.@/UL@

@UL@(d) To provide grossing factors for use with survey data.@/UL@

10. We therefore welcome the work carried out by ONS to develop population estimates by ethnic group, and would like to see this developed further to produce regular estimates by ethnic group and by other equality characteristics.

11. The Commission would welcome the development and production of regular series of population estimates on all equality groups, beginning with the regular production of estimates by sex, age and ethnic group.

12. As regards the definition of the population, the Commission has two concerns in relation to the “usually resident population" definition. One is that there is no regular breakdown of the population into those living in households, those living in communal establishments and those living elsewhere. Of particular concern is the lack of information on people living in communal establishments, which particularly affects both young adults and older age groups in the population. The other is a concern over the ability of the estimates to incorporate the effects of migration and to keep pace with an increasingly mobile population.

13. We therefore welcome the inclusion in ONS' Work Programme for 2008-12 of the project “Improving Migration and Population Statistics (IMPS)", but note that cross-Government funding is required in order to take forward all the recommendations of the Migration Statistics Task Force.

14. The dependence on the Census for many population statistics results from the limitations of alternative sources, eg in terms of the range of estimates (see 4. and 9. above). This is despite two serious drawbacks: the delay in the publication of results and the ten year gap between Censuses. Since it is recognised that there would be major cost implications in increasing the Census frequency, eg moving to five-yearly Censuses, we would encourage ONS to investigate alternative methods of producing more timely and accurate population statistics, making use of administrative or survey data where appropriate. The proposal for an Integrated Population Statistics System (IPSS) project would be a welcome first step.

COOPERATION WITH STAKEHOLDERS AND USERS15. The ONS has been making significant efforts to understand users' needs for better population

information, and we hope to see cooperation with stakeholders and consultation with users made a top priority by the new Statistics Board.

November 2007

@HR25@

Memorandum from the Economic and Social Research Council

OVERVIEWThe Economic and Social Research Council (ESRC) welcomes the opportunity to provide evidence to the

Inquiry.

The ESRC is the UK's largest funding agency for research and postgraduate training relating to social and economic issues. It provides independent, high quality, relevant research to business, the public sector and Government. The ESRC total expenditure in 2006-07 was £159 million and at any one time the Council supports over 4,000 researchers and postgraduate students in academic and research policy institutes. It is the largest Research Council in terms of the community it serves, accounting for over 25% of those staff returned in the last Research Assessment Exercise, across a wide range of disciplines.

For the UK science base to contribute fully to economic development, quality of life and public services, social science research is crucial. This is because human progress depends not only on natural science, engineering and technology but on the social and economic conditions that shape them and their subsequent implementation. Work supported by the ESRC contributes in many ways to the social, political, economic and intellectual life of the UK and beyond it. It directly informs public policy and professional practice across all sectors of the economy. It is therefore essential in meeting the Government's economic objective of “building a strong economy and a fair society, with opportunity and security for all".

Increasingly, the information which social scientists need to draw upon to provide evidence for policy and practice is from large scale national resources. These include national census and survey data, or data which are generated via administrative records, such as hospital visits, social benefits received or schools attended. Census, survey and administrative data, when suitably organised, stored, documented and made accessible for research purposes, form part of the “national infrastructure for social research".

ESRC RESPONSE

1. USES AND DEFINITIONSWhat are the uses of population estimates, and how far do current population estimates meet the needs associated with those uses? What are the effects of inaccuracies or inadequacies in such estimates?

How appropriate is a definition of the population based on the usually resident population in the context of the needs of the users?

How does the quality and range of population statistics in England and Wales compare with that available in Scotland, Northern Ireland and internationally?

Academic researchers make extensive used of UK population estimates and particularly census data. The ESRC Census programme delivers census data to students and staff in UK higher and further education institutions and currently has over 35,000 registered users. Use is made of all of the major census outputs including area statistics, digital boundaries, samples of anonymised records, interaction data and longitudinal studies. Research uses include both investigation of the substantive themes available in the census datasets and also the use of census and other population estimates as the basis for denominators in studies of substantive themes not covered by the census such as rates of health, crime, deprivation, economic activity, migration and demographic change. In many instances, major academic research projects in these domains directly and indirectly inform public policy making by central and local government. Error and uncertainty in the base demographic counts jeopardizes every layer of derived research and the impacts are greatest when interest focuses on small population subdivisions—both geographical and statistical. The range of information is comparable between the countries of the UK but there are important differences in the statistical disclosure rules adopted in Scotland in 2001, with the result that far richer results are available in Scotland than elsewhere. There are also methodological differences between the countries which reflect independent decision-making processes and serve to frustrate many attempts at the production of UK-wide comparable analysis.

2. THE ROLE OF THE CENSUSHow does the Census contribute to the creation of population statistics? Is the current frequency of Censuses appropriate and, if not, how frequent should Censuses be?

The census is currently central to the creation of population statistics. At present there are no alternative sources which enable the construction of acceptable estimates of population which must include both demographic and geographical detail. Similarly, no other sources offer the degree of population coverage which was achieved by the 2001 census nor the range of population topics which can be included in the census form. However, internationally some very relevant comparator countries (eg US, France, Netherlands, Scandinavia) have moved away from conventional census enumeration while those with more frequent censuses (eg Republic of Ireland, Australia, New Zealand) have not yet experienced the level of enumeration difficulty seen in the UK in 2001.

What lessons have been learnt from the Test Census carried out on 13 May 2007? What methods should be employed for the distribution and completion of Census forms, including the use of the Internet?

The 2007 Census Test demonstrated the enormous importance of having a high quality address base in order to achieve a successful enumeration. This is particularly the case with regard to the strategy of mail-out of census forms. It is of concern that the use of the internet did not form part of the 2007 test and international experiences are not especially helpful as there has been such change during the years since most major comparator censuses. New Zealand's 2006 Census perhaps offers one of the most relevant comparators and was seen as “a longer-term investment in improving participation in and data quality from the census."1 Government is already investing significant resource in communicating with citizens via the internet, and it offers an essential new channel in the collection of census data. There are likely to be significant new benefits and challenges regarding both the implementation and public perception of Internet collection. It is imperative that this channel be thoroughly tested in 2009 due to its potential cross-cutting impacts on other aspects of enumeration planning.

Proportionally, the small impairment observed in the 2007 test as a result of including an income question does not imperil the overall response rate to any greater degree than that which will result from several separate aspects of inadequate address listing.

1. http://www.stats.govt.nz/census/about-2006-census/methodology-papers/internet-and-field-collections.htm

What steps should be taken to increase the rate of responses to the Census, particularly in hard-to-reach areas? How effective are plans to measure the extent of non-response to the 2011 Census?

One of the single most important aspects of improving response rate to the census is to have a single definitive national address list. This issue is well understood by census agencies internationally who are actively engaged in the enhancement (eg USA) or creation (eg South Africa) of such lists to facilitate the creation of high quality population statistics. ONS are currently hampered by the continued failure of both collaborative initiatives and central government to resolve the current situation in which the National Land and Property Gazetteer and the Ordnance Survey Mastermap Address Layer offer competing address solutions with different strengths and weaknesses. ONS proposals to undertake a complete national address check as part of the census fieldwork are rational given their requirement for the best quality list for enumeration and the apparent weaknesses of the existing lists but would be unnecessary in an environment in which government acted seriously to facilitate the creation of a single, definitive national address list. If ONS are effectively to undertake the creation of a temporary third national address list for census enumeration purposes, it is imperative that the new address information be incorporated into a single national product rather than continuing the present unacceptable situation in which competing products exist within the public sector with different strengths and weaknesses, a complex IPR environment and high costs and disbenefits to users.

To what extent is there a trade-off between the length of the Census form and its role in providing population information? What questions should be included in the 2011 Census?

The complexities of household structure and multiple residences make it necessary to include a good range of questions to ensure adequate capture of the resident population base. There has been strong and consistent demand from academic users for an income question to be included in the census: an enormous amount of effort by academia and central government has gone into the construction of proxy measures for deprivation in the absence of income and no administrative alternatives are available which offer a comparable coverage of income distribution. The single best solution is not to build different proxy measures but to ask an income question. A census is not however, the best vehicle for the capture of the dynamic phenomenon of migration however and the current proposals to include more migration questions to the detriment of other content such as employment information are less welcome.

To what extent should Census questions and statistical information be coordinated across the United Kingdom and how effective are plans to ensure such coordination?

It is strongly desirable that Census questions and statistical information be tightly coordinated across the UK and the agreement between the Registrars General to work together on matters of Statistical Disclosure Control is to be welcomed. It is readily accepted that there need to be some differences in questions and classifications in the different countries but it is of paramount importance that the definitions, methodology and outputs be as consistent as possible. While this consideration is of relatively low importance to geographically-based entities such as local authorities and health authorities, it is a major requirement for UK-wide commercial organizations and academic researchers who have requirements to produce UK-wide analyses. Central government has a strong desire to understand the processes and impacts of devolution and also the variations in (for example) health across the UK: many of the most basic analyses are undermined by unnecessary differences in method or presentation between the countries.

What forms should the outputs of the 2011 Census take, how and when should they be made available and how should they be integrated and coordinated with other information on population?

It is essential that the outputs of the 2011 Census be published primarily online and that a modern data architecture is adopted which incorporates all the key metadata with the data themselves. It is recognized that production of full census estimates, incorporating post-enumeration validation and correction, is a complex task and takes many months but delivery to users will be greatly aided by the use of internet-based dissemination systems and if at all possible, flexible tabulation systems which allow users to specify forms of output rather than having to wait for specific tabulations to be produced within a sequential outputs production cycle. Further, the underlying data needs to adopt contemporary web-based data standards, most likely some variant of XML, which will ensure the most flexible and widest possible use of the data: the most important point is to recognize that these output standards cannot helpfully be determined in detail until at least the time of the census itself, although much of the user consultation and preparatory database design should already be complete. Such an approach need not be a barrier to more traditional uses, but it is in some respects impossible to construct integrated metadata if it is not present as an integral aspect of the data on publication.

3. MID-YEAR POPULATION ESTIMATESHow accurate and useful are the mid-year population estimates for England and Wales, including information available for local authorities and Strategic Health Authorities?

How appropriate is the methodology by which mid-year population estimates are reached?

What are the most appropriate ways to ensure a common level of accuracy in mid-year population estimates for different areas and what steps should be taken to improve the accuracy in areas where population is hard to measure?

What is the right process for making revisions to mid-year population estimates so as to meet the needs of users?

The mid-year population estimates are extremely important in providing annual updates to the demographic base structure. However, the MYEs are not independent of the census counts and do not contain sufficient subdivision (demographic or geographic) for many research and policy purposes, hence census datasets continue to be used extensively even though they are known to be out of date. Given the demonstrable potential value of administrative data sources in tracking population stocks and flows, increased and integrated effort should be invested in proposals to combine and integrate these sources rather than to treat them as separate strands.

4. THE ROLE OF SURVEY AND ADMINISTRATIVE DATAWhat role can and should be played by survey data in the compilation of population statistics?

What role can and should be played by data from General Practitioners and other health service data in the compilation of population statistics?

What other data sources can and should be used in the compilation of population statistics?

It is anticipated that 2011 may offer the last opportunity to take a fully conventional census and that a range of enumeration challenges will continue to emerge which make operation of a full-scale conventional census unrealistic in 2021 and beyond. 2003 ONS proposals for an integrated population statistics system (IPSS) were

welcomed and are still considered to indicate an inevitable and desirable trajectory towards a multi-source future for demographic statistics. There are good grounds for expecting that data from multiple administrative sources could be combined in order to provide more up to date basic population counts and subdivisions but international experience, eg from Finland and the Netherlands, demonstrates the very long periods over which such systems need to be brought to maturity. At present none of the UK administrative systems such as data from GP lists, is conceived or created with statistical purposes in mind and statistical use of operational data requires large-scale acceptance of statistical requirements by the original data creators. ONS has been severely limited in its ability to actively explore these routes by the current limitations on interdepartmental data sharing and the new Statistics Act is to be welcomed for providing a framework within which such developments may be possible. In the light of these remarks it is therefore essential that 2011 census operations incorporate to the fullest extent provisions to cross-check census and administrative sources under conditions of the highest data security, to provide a “benchmark" position from which post-2011 strategy can be developed.

5. COOPERATION WITH STAKEHOLDERSHow effectively has the Office for National Statistics cooperated with stakeholders with an interest in and information relating to population statistics and how can cooperation be improved by the Statistics Board and by its Executive Office?

ONS has a range of consultation mechanisms for interaction with the academic community and there is currently good representation on working groups and advisory panels, in addition to the open consultation exercises conducted on different aspects of census planning. It is however somewhat frustrating to users of all sectors that there is no single, readily accessible consultation plan (eg on the ONS website) which identifies the scheduling and contacts for all census-related consultations allowing users to clearly identify the timing and scope of areas on which responses will be required in order to input appropriately to the census planning process. It will be important that the Statistics Board and its Executive Office do not simply add to the volume of separate census consultations rather than working with ONS to streamline and make as open as possible the mechanisms which exist.

FURTHER INFORMATIONIn drafting this response comments and advice was sought from members of the Council's Research

Resources Board and the Directors of the various investments that it supports. The ESRC Research Resources Board is responsible for ESRC investment in the resources required to support the Council's mission. The Board advises Council on the provision needed to ensure the long-term vigour and utility of the social sciences, and helps to guide the work of other funders. The Board invests in the resources needed to carry out high quality research, such as data, archives and advanced research methods.

November 2007

@HR25@

Memorandum from Westminster City Council

SUMMARYWestminster City Council is pleased to respond to the request from the Treasury Sub-Committee to set out

our concerns in relation to impacts of migration and in specific the impact caused by the use of inadequate population estimates and projections and the implications for the distribution of funding to English local authorities.

International migration levels to Westminster and surrounding boroughs have been traditionally significant, and recently published ONS figures, covering the years 2001-06 show Westminster as having the highest volume of international migration per 1,000 population in England & Wales. Westminster is uniquely placed therefore, to comment on the problems caused by non-robust national migration statistics on the resources of Local Authorities and the consequent impacts on the benefits of international in-migration.

Westminster has been at the forefront of research in a growing consensus within the public sector, that flaws in national migration figures and the current definitions and formulae used to drive local authority, police and health service resources, are so significant that they threaten mainstream services and community cohesion.

It should be noted that although there are direct impacts of migration to local authority areas such as the pressures to the private rented sector, overcrowded accommodation, rough sleeping, increased cleansing and refuse collection, the indirect impacts are greater still. Indirect impacts are those caused by the lack of accurate government management data. In the case of population estimates this is extremely urgent and will cause severe detrimental impacts in the forthcoming three-year grant settlements if not recognised.

CHALLENGES FACED BY WESTMINSTER AS A FIRST POINT OF ARRIVALWestminster faces particular challenges as it is the first point of arrival for a large proportion of new arrivals

from overseas. An estimated 2,000 migrants arrive at Victoria Coach station each week on coaches which originate on the continent. This figure excludes migrants arriving on train, tube, bus and coach services from the major London airports.

Since the 2001 census Westminster has seen rapid population growth linked to increasing levels of international migration. The following research results outline the issues raised by Westminster regarding current population estimates.

@UL@— Westminster is the number one destination for working migrants accepted through the Workers Registration scheme—over 16,000 have registered in Westminster between May 2004 to March 2007@/UL@

@UL@— Over 34,000 residents of Westminster received a New National Insurance Number between 2002-06—equivalent to 17% of our 2001 census population.@/UL@

@UL@— Around half the rough sleepers in central London are now A8 migrants (Council survey December 2006).@/UL@

@UL@— There has been a large rise in the numbers of migrants being supported through voluntary groups. Destitution amongst A2 migrants from Bulgaria and Romania who are not entitled to work permits is becoming an increasing issue in the City according to voluntary bodies in the Victoria and Pimlico areas who support refugees and migrants. @/UL@

@UL@— Our 2006 Housing Needs survey has identified increasing overcrowding and household sizes linked in significant part to a growth in housing of multiple occupation. This is adding further to the acute pressures on affordable housing in the City where 44% of children already live in overcrowded accommodation.@/UL@

@UL@— Thousands of migrants are not being counted within Westminster's boundaries. Independent research has found that Westminster has over 13,000 illegal migrants within its boundary at any one time and that around 11,000 short-term migrants annually are “hidden" as they are not registered in official statistics. 47 @/UL@

@UL@— Innovative research using ethnographic techniques conducted by ESRO suggests that some migrant communities in Westminster may be twice as likely to register for bank accounts and mobile phones than with the

47 Westminster Population Research. SQW September 2007.

state institutions such as with a GP or for National Insurance numbers that are used either directly or as proxies to estimate population figures@/UL@

@UL@— The ESRO research also found that the Office of National Statistics' definition of a “household" does not accurately reflect the complex living arrangements of migrants in the twenty first century. Westminster is increasingly finding multiple fluid households in properties across the housing sector.@/UL@

@UL@— Research conducted by SQW and Local Government Futures shows that official estimates are not “fit for purpose" in areas as diverse as Westminster. This is backed up by the fact that the Office for National Statistics, responsible for compiling population estimates, refused to include Westminster in a test of the forthcoming 2011 census that was undertaken in May 2007. The ONS said: “our methods might be sufficiently good enough for more typical cities".@/UL@

@UL@— Local Government Futures research revealed: “We have found no evidence to suggest that the combination of the City of Westminster's extreme characteristics have been considered, or addressed, by the current or proposed ONS methodologies for calculating mid year population estimates between 2002 and 2005."@/UL@

@UL@— At a population summit held by Westminster City Council and attended by approximately thirty other councils identified the non-measurement of short term migration as a major problem: “Short term migration has a significant impact on the provision of public services in many local authority areas but migrants who are identified by the International Passenger Survey as planning to stay in the country for less than 12 months are excluded from the existing population estimates used to distribute grant funding. This approach will by definition exclude many international students planning to stay in the UK for a single academic year or migrants from Old Commonwealth nations such as Australia and South Africa as well as individuals from A8 migrant states whose intentions are not firm when they arrive in the UK. @/UL@

@UL@— In the absence of an ONS short term migration dataset at local authority level (which is not expected until 2008 at the earliest) the government should introduce a specific grant for the CSR07 period to target resources at those authorities most affected by these communities. This grant could be distributed using proxy indicators such as WRS or NiNO data which are available at local authority level."48 @/UL@

@UL@— The summit also found a lack of clarity on the minimum standards of accuracy for population estimates and raised concerns about the relocation of the ONS to Wales linked to inadequate funding of migration estimates methodology. @/UL@1. USES AND DEFINITIONSWhat are the uses of population estimates, and how far do current population estimates meet the needs associated with those uses? What are the effects of inaccuracies or inadequacies in such estimates?

How appropriate is a definition of the population based on the usually resident population in the context of the needs of the users?

How does the quality and range of population statistics in England and Wales compare with that available in Scotland, Northern Ireland and internationally?

Annual population estimates and projections for local authority and NHS areas use the 2001 census as a baseline and adjust for estimated annual changes in the number of births and deaths as well as the impact of internal migration within the UK and international migration to and from overseas. It is the distribution of migration patterns and their impact on service pressures which is of significant current concern to many local authorities including the City of Westminster.

On 24 April the Office for National Statistics published the initial conclusions for its Improving Migration and Population Statistics (IMPS) study and released indicative changes to the previously issued population estimates for 2001 through to 2005 for each local authority area. Revisions to the 2004 based sub national population projections were released on 27 September in the light of this new methodology and these are expected to be used as the primary population sources for the three year CSR 07 local government finance settlement which will be published in December.

Use of Population Estimates and Projections to Distribute Government Funding

48 Westminster City Council Population summit 2nd July 2007.

In addition to the £26 billion allocated through the revenue support grant system, population data is also used to distribute the majority of local authority specific grants for social care services (eg children's services, carers grant, supporting people). It also influences the targeting of regeneration funding via the Neighbourhood Renewal and Safer Stronger Communities funds (now being delivered via local area agreements) and is the denominator for the primary deprivation indicator used to distribute the £28 billion Dedicated Schools Grant (the main source of schools funding) by the DfES. Funding allocations to Primary Care Trusts through the capitation formula are also primarily driven by population.

It is therefore essential that the underlying population estimates are robust and credible. Unfortunately Westminster City Council believes that the current estimates are failing to adequately quantify population levels amongst “hard to count" areas which have large levels of international migration inflows according to the current definition of a resident. This methodological failure is compounded by the failure to enumerate short-term migrants, or include the sizeable and growing number of people who have second homes/or rent in Central London during the week.

2. THE ROLE OF THE CENSUSHow does the Census contribute to the creation of population statistics? Is the current frequency of Censuses appropriate and, if not, how frequent should Censuses be?

What lessons have been learnt from the Test Census carried out on 13 May 2007? What methods should be employed for the distribution and completion of Census forms, including the use of the Internet?

What steps should be taken to increase the rate of responses to the Census, particularly in hard-to-reach areas? How effective are plans to measure the extent of non-response to the 2011 Census?

To what extent is there a trade-off between the length of the Census form and its role in providing population information? What questions should be included in the 2011 Census?

To what extent should Census questions and statistical information be coordinated across the United Kingdom and how effective are plans to ensure such coordination?

What forms should the outputs of the 2011 Census take, how and when should they be made available and how should they be integrated and coordinated with other information on population?

Problems with the 2001 Census

The foundation of all UK population statistics is the decennial census, but this foundation has been shaken over the last two decades. The 1991 census was considered a failure in national terms, largely as a result of the introduction of the Poll Tax (Community Charge) which provided an unwelcome incentive for some people to avoid completing the form, or for missing members of their household off the return in an effort to avoid paying this new form of taxation.

Whilst the Office for National Statistics (ONS) regarded the national performance of the 2001 census as successful, within London, and Westminster in particular, serious shortcomings were identified. In 2004, following a high profile campaign by Westminster City Council, the ONS accepted that their previously issued population figures were wrong for the borough. ONS subsequently produced a revised population total for Westminster of 203,000 for 2001, this revision was regarded as the best estimate available and added an additional 17,500 people to the original census figure.

However, Westminster City Council considers that due to problems with the enumeration process, the high population turnover, the diversity of its population and more generally the large numbers of hard to count groups resident in the City which the ONS revised estimate could only hope to partially address, this revised 2001 Census figure remains an underestimate. The ONS for example, made assumptions about household sizes of missed properties being commensurate with those enumerated, ESRO research suggests that actually those dwellings are likely to be home to a significantly higher number of residents.

The City Council's concerns are further heightened by the ONS's own admission that the methodology proposed for the 2011 census may not be adequate for hyper-diverse areas like Westminster.

“We were surprised to find that the City of Westminster had been expressly excluded from the 2007 Census Test because of its extreme characteristics. This exclusion, and the ONS's statement that “our methods might be sufficiently good enough for more typical cities" suggests that the ONS is seeking to develop a `one-size fits all' methodology with regard to local authority population counts." LG Futures “An Assessment of Population Data for the City of Westminster July 2007".

3. MID-YEAR POPULATION ESTIMATESHow accurate and useful are the mid-year population estimates for England and Wales, including information available for local authorities and Strategic Health Authorities?

How appropriate is the methodology by which mid-year population estimates are reached?

What are the most appropriate ways to ensure a common level of accuracy in mid-year population estimates for different areas and what steps should be taken to improve the accuracy in areas where population is hard to measure?

What is the right process for making revisions to mid-year population estimates so as to meet the needs of users?

Concerns About ONS Methodology for Estimating International Migration at Local Authority Level

International migration estimates, which include only people who move to a country other than that of their usual place of residence for a period of at least a year, are currently based on estimates of Total International Migration allocated on the basis of the International Passenger Survey (IPS). International in-migrants are then allocated to local authorities on the basis of the distribution of in-migrants according to the 2001 Census and thus any errors in the census will be compounded in the subsequent population estimates.

In the revisions published on 24 April 2007 linked to the publication of the results of its Improving Migration and Population Statistics (IMPS) study the ONS announced that it was intending to supplement the migration estimates identified through the IPS with data derived from the Labour Force Survey (LFS). In the ONS's view this was designed to improve the distribution of migrants between regions and also to individual local authority areas. The LFS is a quarterly sample survey of households living at private addresses in Great Britain and its purpose is to provide information on the UK labour market that can then be used to develop, manage, evaluate and report on labour market policies. The Labour Force Survey, however, has significant drawbacks as a tool for allocating migration as it has an even smaller sample size of migrants than the IPS and does not have any significant coverage of communal establishments or housing of multiple occupation. It is unclear how the LFS in practice deals with non-responsive households, and the methods used to sample the entirely non-English speaking households prevalent in Westminster are rejected by other academic institutions

Independent research commissioned by Westminster found:"We do not believe the recent proposals by ONS, to incorporate the use of the Labour Force Survey within their methodology for allocating inward international migration flows amongst local authorities in England, will more accurately reflect the extreme characteristics of the City of Westminster. Consequently, the new ONS proposals will not more accurately reflect the City's resident population. It is clear that, until the City of Westminster's extreme characteristics are reflected in the ONS methodologies, the mid- year population data produced for the City will always be inaccurate." Local Government Futures September 2007

The Office for National Statistics' rationale for using the Labour Force Survey in the formulation of TIM estimates is not robust, as the relevant Labour Force Survey migration sample sizes are actually smaller than those of the International Passenger Surveys. SQW September 2007.

The ONS also announced its intention to adopt a propensity to migrate model to produce estimates of out-migration at the local authority level and changed the basis for making assumptions about individuals who switch their intentions between being a short term visitor or a long term migrant. However this latter data is only informed by people leaving the UK and not by those switchers who decide to stay permanently.

The council contests the assumptions made by statisticians and politicians regarding migration. On too many occasions have seen public statements have been made as fact regarding the assumed behaviour of migrants which have not been backed up by research.

For example:

@UL@— That these figures are lower than previous estimates [revised population estimates] simply reflects the reality that many migrants to the UK who arrive in London quickly move on (including many of those you indicate as arriving through Victoria Coach station)." Letter from John Healey, Treasury Minister to Sir Simon Milton@/UL@

@UL@— The improvements that the ONS is making to its methodology makes the information more accurate and improves quality." Angela Eagle, Treasury Minister.@/UL@

@UL@— The Home Office used a report that suggested just 5-13,000 migrants would arrive in the UK on the accession of EU A8 countries. The Home Office paper, coordinated by the University College London economist Christian Dustmann, said: “Estimates for the UK range between 5,000 and 13,000 net immigrants a year." Guardian 2 Sep 2006@/UL@

@UL@— The Home Office says its best estimate of illegal immigration is 430,000 but the number could be between 310,000 and 570,000. Immigration Minister Tony McNulty said the figures were a “best guess". “By its very nature, it is impossible to quantify accurately, and that remains the case," he said. BBC Online 30th June 2006@/UL@

@UL@— Migration figures released by the ONS earlier this month [April] suggested that approximately 56,000 Poles entered the UK in 2005, although the Department for Work and Pensions has issued figures suggesting that over 170,000 Polish citizens applied for National Insurance numbers in the same year. Observer 29 April 2007@/UL@

Westminster's research shows that migration is complex and assumptions made will not fit migration patters uniformly throughout the country.

Poor management data impacts on local communities

The cumulative effect of the population revisions changes announced on 24 April in London was a loss in population of over 60,000 residents between mid-2001 and mid-2005, most of which is attributable to the introduction of the LFS to supplement the IPS at the regional level. Westminster saw a reduction to its previous migrant allocation of around 15,500 which the City Council believes is counterintuitive. This resulting relative loss of population does not reflect our experience on the ground and evidence from a number of boroughs and other authorities to the contrary.

Westminster City Council believes that they will lose up to £18 million before damping in funding per year because the government is not properly counting population. The potential loss of £18 million per year would be the equivalent of an £150 increase on top of the average band D council tax bill in Westminster.

However, last year the Council gave a commitment given to residents that they would not increase council tax by more than two per cent per annum until 2009-10, subject to the need to maintain financial prudence. The alternative would be facing difficult decisions in relation to social care services where there is a major threat to our future funding. The council says the mismanagement of migration will affect services as three-year funding for councils at the end of this year will lock them into settlements that disregard these “hidden" communities.

There is mounting evidence from alternative estimates, the Statistics Commission, the Audit Commission, the Governor of the Bank of England, and the ONS that suggests that the population estimates significantly underestimate the scale of international in-migration nationally. This reflects the weakness of using small sample survey data to generate information on the migration patters of hundreds of thousands of individuals.

Of those sampled by the International Passenger Survey in 2005, for example just under 3,000 were in-migrants as opposed to permanent UK residents. If the samples from the three largest airports (Gatwick, Heathrow and Manchester) are excluded only 79 in migrants were in fact interviewed by the ONS through the IPS in 2004 for all the other UK airports combined (including Liverpool, Stansted and Luton which have the largest proportion of low cost flights from central and eastern Europe). Similarly in 2005 only 17 migrants coming through the Channel Tunnel were interviewed under the IPS. Westminster questions whether these sample sizes—which will drive the data for the distribution of NHS and local authority funding over the CSR 2007 period—are sufficiently robust to measure the true quantum of migration into the UK.

The current assumptions about migration which are derived primarily from survey based methods need to be supplemented by a more rigorous review of local datasets held by local and central government as well as seeking evidence from the business and voluntary sector. This requires a more proactive approach by government departments to data sharing.

There has been widespread criticism of the methodology used by the Office of National Statistics to measure migration dating back to an investigation conducted by the Statistics Commission in October 2003—The 2001 Census in Westminster: Interim Report.

@UL@” . . . it is a fact that whatever the true population in Westminster on Census night, the population now—2 years later—could be significantly different. The churn in population in inner urban areas, and especially in Westminster, is high, with up to a quarter of the electorate on the electoral register changing annually. We know that methods currently used for measuring migration into and out of the UK, and between local authority areas, are unreliable. Particularly unreliable are the estimates of international emigration and immigration into and out of Central London. Without improved methods, up-dating population census figures is liable to error."@/UL@

The Statistics Commission has written to government departments alongside the Office of National Statistics to further reinforce the consequences of using “limited" population data.

“There is now a broad recognition that available estimates of migrant numbers are inadequate for managing the economy, policies and services." Letter from Karen Dunnell, National Statistician in May 2006 to four government departments.

“Until our research has concluded you may wish to consider how the estimates and projections are used and whether there is any scope for recognising the particular uncertainty for those parts of the country that are affected by relatively high levels of migration." Letter from Glen Watson Director Social Reporting & Analysis Group of the Office for National Statistics to Lindsay Bell Director, Local Government Finance CLG December 2005.

Short Term Migration

The existing population estimates and projections issued by the ONS only include migrants who intend to stay in the country for more than 12 months. Individuals planning to stay for less than 12 months are classified as short term migrants.

This approach will by definition exclude from the existing population estimates many international students planning to stay in the UK for a single academic year or migrants from Old Commonwealth nations such as Australia and South Africa as well as individuals from A8 migrant states whose intentions are not firm when they arrive in the UK. Until and unless these individuals leave the UK any switching from being a short term migrant to a “permanent" resident will not be identified through a change in the national quantum of population through the International Passenger Survey. If there has been an increasing trend in recent years for short term migrants to decide to remain in the UK permanently this may not be adequately captured therefore through the current methodology leading to the understatement of the nation's population.

Whilst the majority of short-term migrants may arrive in the UK without families so they make fewer demands on some of the more expensive forms of public services such as education and children's social care, they still use a range of local authority services which means that the authority incurs costs whether it be in relation to waste management, leisure provision or housing. In addition short term migrants who fail to find jobs, or lose their jobs and who are not entitled to or have only limited entitlement to benefits may become homeless and destitute or engage with the black economy.

The ONS has recognised the need to produce a robust short term migration dataset at regional and local authority level but this is not expected to be available until late 2008 at the very earliest. Experimental national estimates are however expected to be published in late September 2007. In view of the acute pressures being placed on local services by short term migration it is imperative that the production of a short term migration dataset is prioritised by the ONS—this should be supported by new resources from HM Treasury if required.

In the absence of a consistent ONS short term migration dataset at local authority level Westminster considers that the government should introduce a specific grant for the CSR07 period to target resources at those authorities most affected by this issue. This grant could be distributed using proxy indicators such as WRS or NiNO

data which are available at local authority level. It would be inappropriate for this issue to be ignored until the next spending review.

4. THE ROLE OF SURVEY AND ADMINISTRATIVE DATAWhat role can and should be played by survey data in the compilation of population statistics?

What role can and should be played by data from General Practitioners and other health service data in the compilation of population statistics?

What other data sources can and should be used in the compilation of population statistics?

What's wrong with the Labour Force Survey?

The Labour Force Survey (LFS) is a quarterly sample survey of households living at private addresses in Great Britain. Its purpose is to provide information on the UK labour market that can then be used to develop, manage, evaluate and report on labour market policies. It is not designed to be a measure of migration.

The ONS itself has identified that certain Household types are under-represented in the Labour Force survey as profiled below. Given that all these types (except for Car ownership) are over-represented in Westminster, how can ONS seriously guarantee LFS data accuracy for Westminster/ Central London?

(source—Labour Force Survey User Guide—Volume 1 : Background and Methodology 2003)

Households underrepresented in the LFS are:

@UL@Households in London@/UL@

@UL@Those renting from Housing Association or in job related housing@/UL@

@UL@Those in converted or shared accommodation@/UL@

@UL@Households with three or more cars or vans@/UL@

@UL@Those comprising only 1 adult, aged 16-19@/UL@

@UL@Those with head of household aged 16-24@/UL@

@UL@Those with only single head of household@/UL@

@UL@Those with the head born in New Commonwealth country.@/UL@

Although the ONS claim this is a large survey and gives better data than the International Passenger Survey it still only conducts on average 700 interviews with households in Westminster. This is far too small to ensure that migration is adequately measured.

We understand that the LFS sample of 55,000 responding households represents only 0.2% of the GB population, and that the ONS is having to boost sample sizes by aggregating London Boroughs. Given the unique nature of Westminster's dynamic population the council is very concerned about how much this cluster will still be representative of Westminster and our lack of opportunity to formally analyse it.

What's wrong with the International Passenger Survey?

The Office for National Statistics uses methodology to determine migration numbers that simply is not fit for purpose. The International Passenger Survey is conducted at a handful of ports of entry and samples only 0.2% of what is believed to be all passengers arriving in the UK. (250,000 people).

Of these 250,000 interviews only 2,965 were identified as migrants in 2005 (1.2%). The number of emigrants leaving the country is determined on an even smaller sample of just 781 in 2005. For example, the estimates for A8 countries, a critical source area, were based upon 78 interviews of immigrants and 19 interviews of

emigrants. Of the 2,965 migrants identified, 2,740 were identified at Heathrow leaving just 225 migrants identified at all other ports of entry.

The numbers are then apportioned to local authority level using formula based analysis. This is not an appropriate way to measure migration (see Governor of Bank of England below).

The ONS has recently published “improved" migration estimates using the Labour Force Survey to supplement the IPS. However, this is another flawed data set as it misses many migrants (see above).

In its response to the Lords Economic Affairs Select Committee September 2007 The Statistics Commission has highlighted that the current methodology being pursued by the Office for National Statistics may not be the approach required to deliver real improvements to migration estimates. This chimes with Westminster's criticism of the Labour Force Survey and International Passenger Survey.

“There is currently no satisfactory source which can provide the raw information, at national and local levels, that is required for statistical purposes. Sample surveys, of the kind typically carried out by statistical offices, are not the answer."

“There are no quick or cost-free answers. Government needs to decide whether the issue is of sufficient importance to justify the commitment of substantial resources to address the technical issues and statutory obstacles. Other countries have solved the problem through the use of integrated population and household registers. There is no reason why the UK could not do the same given time and resources."

Government cut backs when investment is needed

There is widespread concern that the Office for National Statistics is not being funded adequately enough in order to make the improvements necessary to provide confidence in migration estimates.

The ONS is working under revised budgets and undergoing efficiency programmes that include moving the whole of its statistical base to north Wales. The ONS estimate that only 10-15% of its London based staff will relocate.

The Bank of England in its evidence to the Treasury Sub Committee gave a stark warning about the loss of experienced ONS staff; “The relocation programme poses a serious risk to the maintenance of the quality of macro-economic data. If substantial numbers of ONS staff are unwilling to relocate, the loss of skilled individuals could have a severe impact on a range of statistics."

“It is ironic, and a source of concern, that the ONS can't produce statistics on its own performance. Both the ONS and HMRC need to start listening to their users and working with them to alleviate the problems—real and perceived—arising from the efficiency programme." Treasury Select Committee July 2007

Public and Commercial Services union general secretary Mark Serwotka said: “One of the central aims of the efficiency programme was that services wouldn't suffer, yet day in day out we are seeing backlogs and services damaged."

5. COOPERATION WITH STAKEHOLDERS@UL@How effectively has the Office for National Statistics cooperated with stakeholders with an interest in and information relating to population statistics and how can cooperation be improved by the Statistics Board and by its Executive Office?

Westminster believes that the consultation and engagement between ONS and the users of its statistics needs to be improved urgently. It is recognised that the ONS participates in the Central and Local Information Partnership (CLIP) meetings that facilitate debate on improving government statistics. However, the frequency of these meeting at twice per year is inadequate and the representation and communications of findings is deficient.

The ONS has issued a new revisions policy that seems to be built around the needs and capacity of the organisation rather than the users of statistics.

It is felt that the ONS has been “hung out to dry" by the government on the issue of improving statistics rather than fostering a collegiate working environment across departments to find a solution to this pressing problem. For example, the Interdepartmental Task Force has been set up by the ONS and makes comprehensive recommendations for improving population estimates including the use of e-boarders and ID cards. However, as far as the council knows this plan and the estimated costs of implementing it have no Ministerial accountability. The users of statistics are left in the dark about which aspects of this plan will be implemented and by when or how much government resource is being dedicated to it.

November 2007@/UL@@HR25@

Memorandum from the Local Intelligence Network Cornwall (LINC) Business and Economy Module

The Business and Economy Module brings together researchers and policy makers in who have an interest in economic research in Cornwall. The group's membership draws upon colleagues from the Southwest of England Development Agency, Government Office for the South West, Cornwall County Council and other local organisations. The group have played a key role in over-seeing the research that has directed plans for the Convergence Programme—a European Union development fund worth £450 million to Cornwall and the Isles of Scilly.

The group have been following developments with the 2011 census and welcome the opportunity to comment on the proposals. The particular question the group wish to respond to is:

@UL@— To what extent is there a trade-off between the length of the Census form and its role in providing population information? What questions should be included in the 2011 Census?@/UL@

The group would strongly support a forth page of questions if it is not proven to compromise the response rate. The particular value of this additional page would be the question on self-employment by industry. This question is critical to our work as it is the only source for this data. Without this question, Cornwall would know very little about the fishing, agriculture and construction sectors as there are very high levels of self employment within these sectors (as much as five times the number of employees). Furthermore, Cornwall has a very high level of self employment generally, and knowledge of this section of the workforce is extremely important for understanding the dynamics of our economy and informing decisions on funding.

November 2007

@HR25@

Memorandum from Barnet Council

1. INTRODUCTIONBarnet Council welcomes this Inquiry.

The ONS tells us that long-term international migration has been the largest component of population change in the UK since 1999. This change poses many challenges to local authorities including population turnover, greater diversity and greater social need, which are not currently factored into central government's funding of local authorities.

The ONS has recently revised up its projections for international migration and expects the rate of net international migration to increase over the next 20 years. It is certain that London in particular will continue to be the predominant destination.

This submission finds serious problems with both the ONS population model and the unfunded costs on local services from the growing international migration trends.

It recommends that improvements to the population model already recommended by the 2006 Migration Task Force report are accelerated, and that locally planned housing and employment growth should be incorporated into the model.

It further recommends that the Treasury gives urgent attention to the LGA's call for a new mobility fund, to recognise the growing costs felt by councils with high international migration and population turnover like Barnet.

2. WHY THE ACCURACY OF POPULATION FIGURES IS IMPORTANT TO BARNET2.1 Constantly changing estimates, and their financial impact

ONS figures play a key role in the distribution of central government resources to local authorities and health authorities. Barnet was one of the many councils that suffered a financial penalty when the 2001 Census results identified a much smaller population than the ONS had estimated for 2000, so that the estimates used in the local government grant formula were then revised down. The conclusion by ONS and a Middlesex University study commissioned by Barnet was that the ONS population modelling was at fault rather than the 2001 Census. 49 However, Barnet agrees with the Statistics Commission that problems with enumeration in hard-to-count areas and technical difficulties led to less reliable Census results than required (Statistics Commission, 2007, p7).

The ONS set up the Improving Migration and Population Statistics (IMPS) project to investigate problems with estimates, rather than the Census, recognising that migration is both the dominant component of population change and difficult to estimate due to the lack of robust data sources.

Barnet was selected as a case study for this project because of its high population density, high levels of migration and high proportions of young males, students and people from non-White ethnic groups—all factors that the ONS recognises affect the quality of its estimates for the borough. Barnet is also ranked 25th in the country for the proportion of international in-migration to total population in 2006—2.1%—according to ONS estimates.

Yet the first major output of this research programme was a set of revised estimates that actually took 6,600 (2%) away from Barnet's 2006 population estimate. Barnet does not believe that the Labour Force Survey, the cause of this decrease, is sufficiently robust because of its low sample size, low response rates in London, and exclusion of communal establishments such as hostels and halls of residence.

It is clearly important that local authorities can rely on the population figures they are given for the purpose of local planning, and the developments discussed above have all affected Barnet's confidence in the ONS figures.

2.2 The time lag between people arriving and the ONS counting them

The three year grant funding of local authorities over 2008-2010 will be based on 2004-based population projections. These figures cannot therefore take into account the rapid increase in immigration that has occurred since

49 In other areas however there was a clear Census undercount identified and in 2004 the ONS produced revised population figures for 15 local authorities, adding 104,000 people in total.

2004, particularly from the A8 countries. The table below shows that even under ONS estimates, Barnet's population has been growing at a faster rate since 2004.

Table 1

Year ONSestimate

Annualincrease

%increase

2001 319,481 2002 320,100 619 0.19%2003 320,800 700 0.22%2004 323,000 2,200 0.69%2005 326,100 3,100 0.96%2006 328,600 2,500 0.77%

The other major problem with the use of ONS projections for the three-year settlements is that they do not take into account planned growth, which in Barnet is hugely significant. In the next 20 years Barnet will produce over 32,000 new residential properties and at least 22,000 jobs which will bring a huge number of new people into the borough both as residents and employees, and make large demands on local services.

House building in particular is set to start accelerating dramatically in 2008—see chart 1. Thus ONS projections based on historical data are clearly not adequate for ensuring Barnet receives proportionate funding in time to meet this increased demand, and the consequences will be felt in both the quality and equity of service provision.

Chart 1

/home/PKB/DATA/386539/FOLIO263.EPS>2.3 The effect of short-term migration

The ONS's exclusion of migrants who enter local authority areas and use public services, but do not settle for a year or longer, is clearly penalising areas like Barnet which have high levels of migration. The ONS also excludes people who live in Barnet for part of the week but are resident elsewhere for population purposes. Central government funding does not cover these people because the ONS fails to count them.

The ONS recently released experimental estimates of short-term migrants at the regional level, which showed that around 40% of all short-term migrants in the year to mid-2005 spent the majority of their visit in London, which is almost 3 times higher than London's share of resident population (14%). This shows that London experiences a disproportionate number of short-term migrants so it is essential that the Government takes account of the additional pressures on local authorities in London when determining the allocation of grants.

If the London figure for short-term international migrants (here for 3 to 12 months) is allocated to the boroughs according to the same proportions as long-term migrants, Barnet has 3,977 short-term in-migrants, bringing its total international in-migration count to 10,848, or 3.3% of its total 2006 population.

This figure is obviously a very rough estimate, and does not take account of the number leaving the borough in the same year, but short-term migration is undoubtedly an issue that must be addressed.

2.4 Illegal immigrants

Migrationwatch, a voluntary, non political body set up in 2001, states that the UK population is likely to comprise 1.1% illegal immigrants (which includes the children of illegal migrants). Applying this to Barnet's 2006 estimate produces 3,614 illegal immigrants, who are not counted in our population figures but whose presence creates a demand on local public services. Yet since immigration in London is higher than nationally, the real figure is likely to be higher. Again, this is not recognised in government funding.

2.5 Evidence from other areas

There is a growing body of evidence to challenge the accuracy of the ONS figures and the reliability of the methodology:

@UL@— Two of our neighbouring councils have commissioned research based on local administrative data to demonstrate that the official estimates were short. Brent Council found a discrepancy of 2,000, whilst Enfield found a 3,300 shortfall.@/UL@

@UL@— In 2006, Newham had 1,436 more resident children in their schools than the ONS estimate of the number of school-age children in the borough. @/UL@

@UL@— Recent research commissioned by Westminister produced an estimate of 13,000 illegal immigrants living and uncounted in Westminster, based on international research into the numbers of illegal migrants coming forward when amnesties are declared. It also estimated a further 11,000 short-term migrants. In total this adds 10% to its population. It estimates that the council will lose up to £18million in formula grant funding per year because of inaccurate population figures. @/UL@

2.6 Widespread concern about the ONS methodology and statistics

@UL@— A letter from Karen Dunnell, National Statistician, in May 2006 to four government departments included the statement “There is now a broad recognition that available estimates of migrant numbers are inadequate for managing the economy, policies and services."@/UL@

@UL@— In a letter to the Times Newspaper on 15 May 2007, David Rhind, Chairman of the Statistics Commission said “It is certainly true that there is a real problem with statistics of migration into and around the country . . . The Statistics Commission has been arguing for four years for action on this matter."@/UL@

@UL@— A recent study by Slough Council showed 9 out of 10 chief executives of local authorities felt they are under funded because the ONS had failed to keep track of immigration.@/UL@

@UL@— In September this year, speaking at a Labour conference fringe meeting, Labour minister Sadiq Khan said there is “no real confidence" in official immigration figures, and that Whitehall has “failed to recognise the real problems" faced by councils when setting grants (reported by the Daily Mail, 24/09/07 and by the Telegraph, 25/09/07)@/UL@

@UL@— The ONS estimate the number of Poles entering the UK in 2005 was 57,000, but the National Insurance registration figure for new Polish workers between April 2005 and March 2006 was 171,380. Even though this includes three months of 2006, the discrepancy should not be anywhere near this large, and indicates that the ONS figures are flawed.@/UL@3. FINANCIAL IMPLICATIONS NOT COVERED BY LOCAL GOVERNMENT FORMULA GRANT3.1 The effect of churn

Britain's population is becoming increasingly changeable, mobile, and diverse, particularly in London. This poses considerable challenges to local service providers.

In Barnet, for every 6.5 residents in mid-2006, 1 resident had either left or joined the borough, which is 50,600 people or 15.4% of the 2006 population estimate. Barnet ranks 37 in the country for population mobility or “churn".

An LSE study for London Councils, published in February 2007 found that there are many types of mobility which impose additional costs on local authorities, and affect London disproportionately to the rest of the country. The costs include electoral registration and council tax administration, temporary accommodation and greater need for language provision. It can also throw up social cohesion management issues that need to be addressed.

Pupils joining a school at other than the start of a school year, often because they have moved into the borough from elsewhere in the country or abroad, are a particular concern. In 2006 9% of the borough's primary pupils and 5% of secondary pupils joined after the normal admission time. Local evidence demonstrates that mobile secondary school pupils in particular perform less well in public examinations than those who join their school at the start of Year 7. Yet mobility is not taken into account in the governments revenue support to local authorities. Barnet Council recognises the impact of mobility on schools and distributes additional support from its funds to schools with high levels of mobility—in 2007-8 this was £385,900.

Clearly new populations concentrated in a small number of areas creating new spending needs are difficult for the current, centralised system of funding to respond to quickly and it is essential that this problem is recognised

in re-designing local government financing. Mobility creates extra costs that are not recognised in the funding formula, and for which there are no additional government grants.

A report published this month by the LGA suggests that the impact of churn may be even more significant than the overall number of people.

3.2 A net financial benefit from migration

National research has demonstrated that international migrants bring a net benefit to the British economy, filling jobs that would not otherwise be filled, and creating new forms of employment, thus boosting productivity.

However it is unlikely that there has been a net benefit to local authorities, since they bear the brunt of service demands yet only achieve financial returns through central government formula and council tax accounts, and neither of these accurately match population growth.

Council tax accounts will not reflect the true growth in population because of the growth in overcrowding, a trend linked to international migration. Barnet's 2005 housing needs survey showed that overcrowding had increased since 2000 from 4.1% to 5.4% of households. Inevitable difficulties collecting council tax from short-term migrants (here less than a year) and failed asylum-seekers also leave a funding gap.

4. EVIDENCE THAT THE POPULATION MODEL IN BARNET AS ELSEWHERE IS NOT WORKING4.1 Local evidence of a growing migrant population

There are at least four data sets which provide a strong indication of international in-migration patterns. New National Insurance registrations by people previously overseas (NINo) and the NHS Flag 4 datasets are the closest to measuring international migration, but births to mothers born outside the UK, and patterns of pupils resident in Barnet speaking a language other than English are also useful.

Whilst we would not expect any of these data to identically match the ONS estimates of international in-migration in any year, the rates should be broadly consistent. Yet for all four indicators the rates of increase over the four year period suggest in-migration is growing much more strongly than the ONS estimates. The trend in the ONS in-migrant data shows a rate that has fluctuated, and is actually lower in 2005-06 than it was in 2002-03.

Table 2

COMPARING ONS INTERNATIONAL MIGRATION INFLOW FIGURES WITH 4 INDEPENDENT INDICATORS OF FLOWS INTO BARNET

Year ONS estimates Overseas nationals registering for NI number

Flag 4 (new GP patients from overseas)

Births to mothers born outside UK

Pupils living in Barnet with English as an additional language

2002-03 7,121 5,940 6,413 4,180 14,5882003-04 7,068 6,190 7,190 4,3342004-05 7,539 6,720 7,296 4,4822005-06 6,871 8,840 8,332 4,728 16,760% change -3.5% + 49% + 30% + 13% +15%

Data from the housing needs surveys50 conducted in 2000 and 2005 also show a 62% increase in the numbers of households moving to the borough from overseas over that period, from 1,387 to 2,248.

4.2 The origins of the growing migrant population

Data from the Barnet pupil level annual schools census (PLASC) shows that between January 2003 and January 2007, the number Eastern European pupils rose by 916 (162%) and Somali pupils by 536 (71%).

50 The housing needs survey is conducted according to government guidance and run every five years to a consistent methodology across London.

Over 2002-07 the NINo data shows a 582% increase from five Eastern European countries—predominantly Poland—and a 41% increase from Nigerians.

The trend of housing applications also shows that the largest increase in applications between 2004-07 came from Eastern Europeans, whilst “Black other" and “Somali" also showed large increases.

Over 2005-07, one of the Primary Care Trust's two local walk-in centres have seen a 28% increase in residents from White–Other backgrounds, and a 18% increase in residents from Black or Black British–African, compared with a decrease in residents from White British backgrounds.

Analysis of the ethnicity of patients attending the local general hospital's A&E department over the last year is inconclusive, though does suggest a small rise in the Other category, among which Polish and other Eastern European patients are most likely to be found.

4.3 Local costs from growing migration

4.3.1 Costs felt by the local authority

The costs of migrants visiting temporarily for work, study or leisure or moving here permanently are felt in terms of demands on public services, and the administrative costs of registering and deregistering people. This population is clearly growing, particularly as a result of EU enlargement.

Pupils entering Barnet schools from abroad increased 5% over 2004-07, from 1,617 to 1,695, and of these a steady proportion of around two thirds have had a non White-British ethnicity. The 2007 intake of pupils that previously lived abroad was 5.7% of the total school population. This compares with the ONS estimate that 2.1% of the total 2006 population arrived into Barnet from overseas.

Of all English/Welsh areas, Barnet has seen the ninth largest increase in the number of pupils speaking a language other than English between 2003-06, and the seventh largest percentage increase (PLASC Dept for Children, Schools & Families, and LGA, 2007, p40). In just the nursery schools the percentage increased from 23% to 31% over 2004-07.

Additionally, there is growing evidence of migrant workers seeking to operate in Barnet outside of local authority regulation. The numbers known to date are small but warrant further investigation. Typical is the investigation of the sale of counterfeit goods (trading standards) leading to closure of a house in multiple occupation with 10-15 beds (environmental health). This will involve language assistance throughout, and occasionally assistance from housing and social services. Evidence from other areas shows that these same people will be using the local hospital's walk-in centre or A&E rather than a GP.

Another cost is on the council as an employer; over the last four years the council's anti-fraud team have investigated 38 internal fraud cases with an immigration link, and 27 illegal workers have been dismissed as a result.

The number of asylum-seekers in Barnet, and people without recourse to public funds is small and stable, but they are increasingly costly for the council to support:

@UL@— In 2006-07 Barnet spent £37,000 looking after 54 unaccompanied asylum seeking children, in addition to the £648,000 reclaimed from central government. Despite the fall in asylum seekers nationally, the predicted net costs for this year are £216,000 because of an increase in 16-17 year olds for whom the costs paid by the council are not matched by government grant.@/UL@

The problem of failed asylum seekers that the Home Office hasn't deported but who cannot claim state benefits and must therefore be supported entirely by the local authority is predicted to have imposed the following costs on Barnet by the end of the financial year:

@UL@— £134,000 supporting families with children @/UL@

@UL@— £100,000 supporting adults with mental health problems @/UL@

@UL@— £376,101 supporting adults with physical and sensory impairments @/UL@

@UL@— £254,000 supporting young people leaving care without leave to remain, but towards whom we have a statutory responsibility under the Children (Leaving Care) Act. This is an increase of 21% from the previous year.@/UL@

4.3.2 Costs felt by the police force

The Police has reported a number of new and growing problems in Barnet linked with Eastern Europeans immigration including a growth in pick pocket activity, street drinking and disorderly activity, drinking and rough sleeping in two parks.

5. HOW POPULATION MODELLING SHOULD CHANGEImplement the 15 recommendations of the Inter-departmental Task Force (published December 2006) with

urgency. In particular:

@UL@— the ONS must produce estimates of short-term migrants at local authority level. (A1) @/UL@

@UL@— The International Passenger Survey (IPS) sample is too small, as is the Labour Force Survey (LFS). The Integrated Household Survey will replace the LFS in 2008, and it is vital that this and the IPS are improved in scale, in order to be reliable at the local level. The LFS needs also to include a communal establishment component so that hostels and halls of residence are included. (A1)@/UL@

@UL@— Use data from the Work and Pensions Longitudinal Study to quality assure and supplement ONS migration statistics. (C1)@/UL@

@UL@— Improve the quality of national administrative data sources—particularly PLASC, NINo, Flag 4, and higher education data on student numbers and term-time addresses. (C2, 3, 4)@/UL@

@UL@— Break down barriers to sharing and linking individual-level data across central government and integrate them more closely with the existing population model. (C5 & 6)@/UL@

Barnet would additionally urge that the ONS looks into:

@UL@— incorporating the data on planned housing and employment growth recorded by local authorities, and in London used by the GLA to produce population projections for the London Plan, a statutory planning framework; and@/UL@

@UL@— plugging the current data gap in pupil data, where there is no data about children in independent schools and those who are home schooled, in conjunction with the DCSF.@/UL@6. HOW GOVERNMENT FUNDING SHOULD CHANGE

Recent revisions to populations have been widely disputed, and in this context, they must not be used in the grant allocation formula until the problems have been remedied.

We support the LGA's call for there to be a new “mobility" fund against which authorities could bid. This fund should reflect not just the growth in the figures, but also the extent of population turnover, in recognition of the extra costs this places on local services. It must also incorporate short-term migration, and for this to happen, the ONS must receive the necessary resources to produce robust estimates of short-term migration at the local level.

In the absence of robust estimates for short-term migration, the government could use a combination of NINo and Flag 4 data to allocate this funding.

7. ATTRIBUTIONBarnet Council is happy for its response to be published and attributed in whole.

8. REFERENCES

Local Government Association (November 2007) Estimating the scale and impacts of migration at the local level

London Councils (February 2007) Population mobility and service provision (by Tony Travers et al at the London School of Economics)

Office for National Statistics (December 2006) The Inter-departmental Migration Task Force Report

Rhind, David (15 May 2007) Letter to editor of the Times Newspaper

Statistics Commission (November 2007) Report No. 36 Counting on Success The 2011 Census—Managing the Risks

November 2007

@HR25@

Memorandum from the London Borough of Camden

EXECUTIVE SUMMARYCamden is a diverse central London borough with a history of large population changes. The area has and

will continue to be one that experiences high levels of in and out migration, being an origin as well as destination for people moving within the UK and internationally. Camden is a vibrant place with a large student population, the 3rd largest concentration of jobs in London and a place with good communications: the terminus of 3 mainline railways at Euston, King's Cross and St Pancras, serving commuter and national lines to the Midlands, North West, North East and Scotland, as well as St Pancras International and Piccadilly Line links to Heathrow.

Concerns have been growing in London about the accuracy of the ONS mid year estimates and the population projections based on them. The main area of concern is the relative weakness of the migration data that has been fed into the estimates, particularly the distribution of international migrants to and from local areas.

A secondary issue surrounds the lack of clear information about visiting workers, often confused with migrants. Although ONS has gone some way to address some of the issues, there is need for more work on improving the statistics and to regain confidence.

Consideration of a mid-term census in order to pick up on rapidly changing population trends may reveal it to be appropriate, particularly for those areas experiencing such rapid change.

The mid-year estimates impact upon formula grant funding, being the population base for and providing trends for the subnational population projections that are used to determine that funding. Local authorities need to be confident that they are receiving the funding they deserve.

An important part of getting estimates of the population right is the success of the decennial census. Camden has taken part in the 2007 Census Test and consequently realises that much work needs to be done in the run up to the next census in 2011, by both ONS and local authorities. Camden calls upon ONS, government departments and local authorities to join together to plan and work towards a successful 2011 Census.

Camden is a diverse central London borough with a history of large population changes. The area has and will continue to be one that experiences high levels of in and out migration, being an origin as well as destination for people moving within the UK and internationally. Camden is a vibrant place with a large student population, the 3rd largest concentration of jobs in London and a place with good communications: the terminus of 3 mainline railways at Euston, King's Cross and St Pancras, serving commuter and national lines to the Midlands, North West, North East and Scotland, as well as St Pancras International and Piccadilly Line links to Heathrow.

1. POPULATION ESTIMATES—USES AND DEFINITIONS1.1.1 Uses of population estimates. The use of population estimates are many and varied, but mainly to

build a picture of the people residing in an area. Analysis of population estimates and projections are built into corporate and individual service planning to make sure that scarce resources are targeted in the right places.

1.1.2 With the increasing use of performance indicators to judge councils and services, estimates are vitally important in comparing place with place since they are used as denominators, often with “official" government sanction. This is well and good when the population estimates are accurate, but tell a different story if the population is under- or over-estimating either the total or parts of a population.

1.1.3 A concern amongst London authorities (especially central London boroughs) is that population as measured by the ONS mid-year estimates may not be a true reflection of reality. Over many years of scrutiny, it is felt that the methods employed by ONS in constructing the mid-year estimate is on the whole sound, that the portion of the data using the vital registration service on births and deaths is good, but that there is a weakness on the measurement of migration, both for moves within the UK (based on GP re-registration) and international flows (based on the International Passenger Survey and Home Office data), and in particular the distribution of international migrants to and from local areas.

1.1.4 For local authority areas like Camden and other central London boroughs with high population flows in and out, within UK and internationally, there only needs to be a small error to start a migration trend that may grow and be compounded with each year. Greater London Authority analysis shows there may be imbalances within London created by the inherent weaknesses of the international migration data, such that their projections for

London show some very sharp differences when calculating population allied to housing growth. Other indicators such as the General Fertility Rate can also reveal when the underlying population data may be suspect.

1.2.1 Population definition. The definition of “resident population" is an important one to hold on to. However further work is required to cover other areas of population. It must be stressed that the “migration" alluded to above does not include the often cited “visiting workers" (often described in the press and by others as “migrants") but who do not qualify for that status under the 1 year qualification demanded by UK migration statistics. There is obviously a need for more work to be conducted by ONS to better understand the effects of the transitory foreign workers on the local population and its affects upon access to housing and other services—and its associated costs. It would seem to be appropriate that these should be treated separately from the normal resident population. Recently published work conducted by the City of Westminster Council has started to pick this apart in their local authority area, but it is the kind of work that needs to be done nationally to understand this new phenomenon.

2. THE CENSUS2.1.1 Contribution of the Census. The Census is crucial to the creation of sound population estimates,

providing a decennial stock-take of population and household characteristics. The Census provides enough detail in statistics across a range of topics and geographies that allow the creation of a range of detailed population estimates, projections and forecasts down to local authority level; and further to ward and to Super Output Areas (SOAs). Local authority level mid-year population estimates created for the census year and adjusted to take account of census undercount give the most accurate estimate of population each decade.

2.1.2 There simply isn't an alternative to the coverage provided by the Census. No other administrative source or sample survey can give such detail, or to such a low-level geography. Indeed, most surveys require the Census (or estimates based on) it to gross-up survey results.

2.1.3 Census Frequency. In areas like Camden and other London/metropolitan areas that are experiencing rapid population change, it seems to be more appropriate to hold a mid-term census in order to pick up on rapidly changing population trends. For this reason the London Mayor had been pushing for a mid-term census for 2006.

2.2.1 2007 Census Test. Camden was one of 5 local authority areas to take part in the 2007 Census Test. Camden was chosen partly because of difficult to reach/count populations and because 2001 Census response rate in Camden was the 7th worst of English local authorities. The Council's active involvement in the Census Test has allowed a unique insight into the problems that need to be overcome before the census in 2011.

2.3.1 The key objective is to achieve high a census response rate, not just in terms of the overall return, but also of fully completed forms. Part of the ONS plan is to use address registers to post out census questionnaires and thereby saving money that can be better spent on targeted follow-up of missing and incomplete questionnaires. Camden is in the worst category of hard-to-count areas and fully supports this approach as one that stands to benefit our area and other metropolitan boroughs. However, this alone is not enough. There are other things that have to be done in combination.

2.3.2 Camden has learned from the 2001 Census and 2007 Census Test experience that a major problem is a lack of census awareness. It was evident that many people just do not know what the census is; what it is for; how the information collected is kept; and the uses made of the derived statistics. However, a glimmer of hope emerged from feedback from enumerators in the Test who found that by simply explaining to people who had not responded that census and census-derived statistics are used, eg for local planning of schools, hospitals and council services—was enough to encourage people to want to fill in the questionnaire.

2.3.3 A key lesson, therefore, is to encourage ONS and local authorities to make census publicity a priority. This may include national and local advertisements, engagement with community groups, through engagement in schools, including newcomers to Britain to explain that they also need to be included. This also needs to be balanced with the message (even more important now) that the Census is safe, personal details are protected for 100 years and that personal anonymity is protected by law.

2.3.4 Part of the approach to encourage people to take part is to give them more than one route to respond. The growing use of the internet and broadband has given people much wider access and through e-government initiatives, many council and government services are now available on-line. As long as confidentiality is assured and the web process is efficient and well managed (eg people can part-complete and come back, it's not clunky, etc.), then it may encourage take-up amongst crucial groups who normally tend not to respond, such as men in their 20s and 30s.

2.3.5 Local authorities have a vested interest in helping to ensure a good as possible count in the next census. Census-based population projections are currently used as a means of distributing revenue support grant and formula funding. ONS should take the lead and local authorities should work with them toward a common goal.

2.4.1 The length of the census questionnaire should be increased to 4 pages to ensure the retention of important questions on qualifications, full and part-time working, carers, industry of employment, an extended question on disability and limiting long-term illness (amongst others). Recent research undertaken by ONS suggests that inclusion of an extra page will not, necessarily, adversely affect response rates and combined with other changes to the questionnaire booklet might help to mitigate against estimated costs of the extra page.

2.5.1 Camden has no real view to the coordination of census across the national borders. National averages can be drawn with England and Wales for the majority of census data, though comparison against the England average is a common requirement in other datasets, eg with the Indices of Deprivation.

2.6.1 ONS has indicated that standard 2011 Census Outputs are due to be aimed at a local authority-level and SOA-level geography, rather than the LA and ward-level output providing standard outputs in 2001. One concern is that wards are much larger areas (in Camden averaging 5 to 8 SOAs per ward) which mean that some extremely useful, but complex tables, such as ethnic group when cross-classified with other census variables, might not be available at the lower level SOA geography due to issues of disclosure. There is also concern that the best-detailed data is not available for the elected councillors.

2.6.2 Compatibility. Camden is happy to use the SOA-level geography, but some concerns need to be addressed. Part of the use of the census is to be able to look back over previous decade to be able to analyse change. SOA-level tables from the 2001 Census have only been made available for the simplest hierarchy of data, the 2001 Census Key Statistics—a set of one-dimensional tables giving a summary across the 2001 Census topics (ie age or sex or ethnicity etc.). The important cross-tabulations (Census Area Statistics) provide the really useful information that provides the much needed detail have yet to be produced for 2001 by ONS (ie age and sex, ethnicity and age, etc.). Census Area Statistics by SOA for 2001 Census data need to be produced to provide compatibility with the proposed 2011 geography.

November 2007

@HR25@

Memorandum from the Association of Regional Observatories

The Association of Regional Observatories (ARO) represents the English Regional Observatories; observatories have extensive experience of using data and information to inform public service delivery, regional policy, strategy development and performance monitoring. ARO works closely with the English Regional Development Agencies and Regional Assemblies to maximise the quality and accessibility of the evidence base available to support regional policy.

ARO welcomes the opportunity to contribute to the Treasury Select Committee's inquiry into

counting the population. The information provided by the Office for National Statistics (ONS) around population estimates underpins much of the analysis undertaken by regional observatories, and hence has an impact on how that analysis is used to influence policy decisions.

 (I) USE AND DEFINITIONS OF POPULATION ESTIMATESPopulation estimates produced by ONS provide the foundation of much of the intelligence produced at sub-

national level. Estimates of population change are of particular interest both to regional observatories and to their partners in the regions, with considerable demand for further development and statistical releases of small area information.

ARO is not able to comment on the possible differences between population statistics in England and Wales against other devolved administrations or international comparators.

 (II) THE ROLE OF THE CENSUSCensus information is of central importance to regional observatories, and ARO supports the continued

priority placed on delivering outputs from the Census rapidly. ARO and its members have been involved in previous consultations from the ONS into the content and operation of the 2011 Census and look forward to working with ONS on future consultations.

 (III) MID-YEAR POPULATION ESTIMATESMid-year population estimates are also frequently used by regional observatories as part of the analysis

process. ARO is unable to comment on the methodology or make suggestions for further improvement.

 (IV) THE ROLE OF SURVEY AND ADMINISTRATIVE DATAARO welcomes the recognition by ONS that further use of administrative data (with appropriate safeguards)

can have a positive impact in validating population statistics and adding further detail, and regional observatories welcome the support offered for this position by the recent Statistics Act.

 (V) COOPERATION WITH STAKEHOLDERSARO continues to embrace opportunities provided by ONS to cooperate regarding population statistics (and

other statistical products) and welcomes the acknowledgement that regional observatories are able to contribute to the development process, recognising that occasions do arise where this contribution is not presently invited. In order to ensure that productive cooperation continues on the commencement of the Statistics Board, ARO would welcome the opportunity to continue to inform or brief the Board and its Executive Office on the concerns of the regional data community. It would also be a positive move for the Board to continue to be proactive in engaging regional data users in their early considerations of their priorities.

November 2007

@HR25@

Memorandum from the Medical Research Council

1. INTRODUCTIONThe Medical Research Council (MRC) welcomes the opportunity to provide evidence to the Inquiry.

The MRC is dedicated to improving human health through excellent science. Its work ranges from molecular level science to public health research, and is carried out in universities, hospitals and the network of its own units and institutes. The MRC liaises with the Health Departments, the National Health Service, other Research Councils, Government Departments and industry to take account of the public's needs. The results have led to some of the most significant discoveries in medical science and benefited the health and wealth of millions of people in the UK and around the world.

In the context of this Inquiry, the MRC supports a large amount of research on identifiable populations, as well as on individuals. Such populations are subsets of the UK population, selected either to be a representative subsample of the population as a whole or, more usually, as a defined subset—for example with a particular disease, from a particular area, of a particular age etc. It is thus important to ensure that, where possible, there is ethical interface between large datasets, and that there are accurate data about the total population when one wishes to sample from it. By achieving such complementarity, best value for money in undertaking large population-based surveys should be achieved.

This response therefore focuses on the census as a source of information for research. The MRC recognises of course that the census has many purposes and that these wider considerations need to be taken into account.

2. RESPONSES TO QUESTIONS 51

2.1 USES AND DEFINITIONSWhat are the uses of population estimates, and how far do current population estimates meet the needs associated with those uses? What are the effects of inaccuracies or inadequacies in such estimates?

Population estimates are used in medical research to estimate the prevalence of particular diseases or conditions, by multiplying up from data obtained from subsets of the population. These figures may then be used for resource predictions (eg in the NHS), but they may also form a rationale for studying particular health problems. Information about the sizes of particular subsets of the population (eg grouped by age, sex, ethnicity, religion etc) is also valuable in this context. The opposite is true also: researchers will judge the representativeness of a population in a particular study by comparing the features of that population with the census data. Clearly, inaccuracies or inadequacies in such estimates give misleading answers, but it is unclear currently what the “gold standard" is against which one may quantify inaccuracies (but see response in 2.4 below). [NB there are statistical techniques, such as Bayesian synthesis across data-sources, which attempt to address biases which are either internal (re study or data quality) or external (re representativeness)].

The ability to link information on the characteristics of the population as a whole (or as individuals), as obtained via the census, with morbidity and mortality (obtained from other sources, including ONS) is extremely valuable. Such linkages may be used to identify particular health problems, but also to generate hypotheses about cause and effect. In order to do this, it would be valuable to be able to link, say, a citizen's answer to Q10 on 2001 census to: survival outcome for that citizen on 30 April 2006 (say): alive, emigrated on date D1 before 30 April 2006, died on date D2 from ICD10 cause xxx on date D2 before 30 April 2006. Assuming that this sort of linkage is permitted, then a research team could receive a no-names file with year of birth, gender, answer to Q10, and the above survival data; or could specify analyses to be run by ONS on such a file. [Since death records are public documents, it is likely that linkage would be permissible]. Also, linkage of five-year survival outcome might be permitted with answers (simultaneously) to a series of census questions (Q10 to Q15), say.

How appropriate is a definition of the population based on the usually resident population in the context of the needs of the users?

This is generally appropriate but there are important exceptions. It is not valuable when undertaking research involving hard-to-reach populations such as travellers (in the sense of eg Romanies), illegal immigrants, drug-users, health tourists; and also people who enter and leave the country frequently

51 www.parliament.uk/parliamentary_committees/treasury_committee/tc210907pn77.cfm

How does the quality and range of population statistics in England and Wales compare with that available in Scotland, Northern Ireland and internationally?

Others are better placed to respond to this.

2.2 THE ROLE OF THE CENSUSHow does the Census contribute to the creation of population statistics? Is the current frequency of Censuses appropriate and, if not, how frequent should Censuses be?

From a medical research perspective, more frequent censuses would add value by providing more up to date information. This may be particularly useful when undertaking research on rapidly changing conditions and monitoring trends; examples include the spread of new infectious diseases, obesity, antibiotic resistance etc. Clearly there are resource implications, but more frequent censuses may not need to cover the whole population, but perhaps a random subset.

What lessons have been learnt from the Test Census carried out on 13 May 2007? What methods should be employed for the distribution and completion of Census forms, including the use of the Internet?

Even before the recent problems about lost data at HMRC, informed individuals were becoming increasingly alert to the potential for data to be “trafficked" electronically, and there may be resistance to e-census forms. The MRC would not wish attitudes to data privacy elsewhere to affect medical research's ethical capabilities.

What steps should be taken to increase the rate of responses to the Census, particularly in hard-to-reach areas? How effective are plans to measure the extent of non-response to the 2011 Census?

As full a response rate as possible is essential to ensure that the final data for the census are representative of the whole population. Hard-to-reach populations are those least likely to receive or complete a census form, and such populations typically have exceptional health problems.

To what extent is there a trade-off between the length of the Census form and its role in providing population information? What questions should be included in the 2011 Census?

The first question can only be answered empirically. On census questions, the census question team may wish to meet with a forum of the Research Councils to help frame well-chosen and well-phrased questions.

To what extent should Census questions and statistical information be coordinated across the United Kingdom and how effective are plans to ensure such coordination?

The MRC funds research throughout the UK. To have information that is consistent across the UK is therefore essential. That does not necessarily mean that all questions have to be asked of all citizens, but in our view there would need to be a good reason, with evidence (eg lack of compliance), not to.

What forms should the outputs of the 2011 Census take, how and when should they be made available and how should they be integrated and coordinated with other information on population?

For medical research, the sooner accurate census information can be made available the better. Data should be made available electronically in a form that is easy for others to manipulate.

2.3 MID-YEAR POPULATION ESTIMATESHow accurate and useful are the mid-year population estimates for England and Wales, including information available for local authorities and Strategic Health Authorities?

Accuracy should be assessed, and publicly qualified, by ONS

How appropriate is the methodology by which mid-year population estimates are reached?

What are the most appropriate ways to ensure a common level of accuracy in mid-year population estimates for different areas and what steps should be taken to improve the accuracy in areas where population is hard to measure?

What is the right process for making revisions to mid-year population estimates so as to meet the needs of users?

The mid-year population estimates are important in providing annual updates to the 10-yearly census. However, they do not contain sufficient subdivision (demographic or geographic) for many research purposes, so census datasets continue to be used extensively even though they are known to be out of date. To ensure a common level of accuracy, triangulation of methods is better than reliance on a single approach.

2.4 THE ROLE OF SURVEY AND ADMINISTRATIVE DATAWhat role can and should be played by survey data in the compilation of population statistics?

Our responses in 2.1 above highlight that answers to questions asked in the census are potentially predictive of mortality or morbidity risks in the following five years. So far as practicable, therefore, there is benefit in consistency between how questions are phrased in the census and in Health Surveys, Diet and Nutrition Surveys, Addictions Surveys etc in which a representative sample of citizens/householders/school-children is asked to take part. ONS might wish to discuss with the authors of such surveys how to maximise compatibility. [The MRC will seek to capitalise on census-tested questions by adopting their wording, when it is appropriate to do so, in Health Surveys or disease cohorts because census answers provide a national backdrop].

Survey data may also be used to test the completeness of census returns, or to highlight areas where there are inconsistencies.

What role can and should be played by data from General Practitioners and other health service data in the compilation of population statistics?

Great care must be taken about data from General Practitioners and other health service personnel. Many of such data are not administrative, but medical-in-confidence. Confidentiality of patient records is paramount. While completion of a census form is a legal obligation, patients divulge details about their lives to doctors under a bond of confidentiality. (Moreover, patients allow biological samples to be taken by doctors which, when tested, reveal information about the patient that s/he did not otherwise know). Current guidance is that the use of such data for research requires ethics committee approval and, unless anonymised, normally the consent of the individual.

If there are proposals to share data within Government in the public interest, in order to retain the confidence of the public, consideration should be given to establishing a register in which such data sharing is documented and which is open to public consultation.

What other data sources can and should be used in the compilation of population statistics?

The MRC has no particular suggestion to make.

2.5 COOPERATION WITH STAKEHOLDERSHow effectively has the Office for National Statistics cooperated with stakeholders with an interest in and information relating to population statistics and how can cooperation be improved by the Statistics Board and by its Executive Office?

The ONS works in partnership with the relevant Research Councils on population statistics. The MRC wishes to see these constructive dialogues continue, with a particular focus on ensuring that the type of information collected is relevant to present (and changing) needs, and thereby to facilitate medical and social research for the public good.

November 2007

@HR25@

Memorandum from Mr Philip Redfern

EXECUTIVE SUMMARY1. My Memorandum on Preparations for the 2011 Census was published in HC326-i and is

summarised here and in Section I below. It made five Recommendations. Recommendations nos.6-9 appear in this Memorandum. This text refers to England and Wales (E&W).

2. Since 1831 our main source of population statistics has been the decennial census of population, supplemented since WW2 by voluntary sample surveys and, to a limited extent, by administrative data.

3. The census. Since 1981 major changes have been made in census design, intended perhaps to emulate US practice. But these changes, coupled with changes in society and in attitudes to filling government forms, have increased levels of non-response and raised doubts about the census results. Population numbers, which are the primary objective of a census, have been challenged by local authorities and by independent researchers.

4. The post-1981 design changes which are having deleterious effects are as follows:

@UL@(i) There has been a big increase in the number and complexity of census questions.@/UL@

@UL@(ii) The doorstep visits of enumerators to deliver and collect census forms are being replaced by the use of the Royal Mail—”mail-back" in most areas in 2001 to be followed by “mail-out" in 2011.@/UL@

@UL@(iii) The decision was taken in 1991 not to process the information already collected about visitors (those present on census night but not “usually living at this address"). This was followed in 2001 by the decision not to collect any information about visitors. These decisions led to large numbers of residents of E&W being placed beyond census reach. In 2011 visitors will be asked 4 questions rather than the usual 30 or so; this will lead to distortions in population geography.@/UL@

@UL@(iv) ONS are unwilling to explore demographic checks on the census results that go beyond the inevitably shaky comparison with the “roll-forward" from the previous census benchmark. (Even that check was ignored in 2001.)@/UL@

5. For 2011 I propose a shortened census form, reinstatement of enumerators on the doorstep, and a census form that ask the same questions of visitors as of anyone else. On item (iv) I have proposed two demographic checks that “anchor" the national population estimates to the rock-hard data on births and deaths and state limits for those estimates.

6. There are other factors that put the future of the census beyond 2011 into doubt:

@UL@(i) Increasing costs (some £500 million for the 2011 census).@/UL@

@UL@(ii) The infrequency of censuses, coupled with the unreliability of the annual roll-forward as a method of updating the national and local population estimates.@/UL@

@UL@(iii) The burden on the public and increasing levels of non-response.@/UL@

@UL@(iv) The risks that a “Big Bang" census focussed on one day may be disrupted by public protest or by political or physical events.@/UL@

I support the Statistics Commission's view that the 2011 census should be the last on today's colossal scale. Other measuring instruments need to be developed.

7. The Commission suggests that one way forward would be a census on the lines of the US 2010 census; that will ask each person just 7 questions and will be supported by a large scale continuous sample survey giving in-depth analyses at county level. This Memorandum explores another way forward: a population register (Section II below).

8. A population register. A population register is a database containing only the core details of each resident of the UK and providing the base for all public agencies' personal records. Core details (also called contact

data) are name, address, sex, date and place of birth and a personal reference number. The primary purposes of a population register are administrative: to improve the accuracy of agencies' personal records so as to protect privacy; to cut costs of administration; to make identity theft more difficult; and to simplify the citizen's contacts with agencies. Population registers have been highly developed in the Nordic countries from the 1960s onwards and there are developments in the Benelux countries too.

9. Statistics are a spin-off from the register. In Nordic countries a set of registers of persons, buildings, businesses, etc is a central feature of their statistical systems, and in some of the countries the national statistical agency manages the population register. The population registers generate reliable annual population statistics, both national and local. Increasingly these countries are replacing a conventional census of population by an Administrative Record Census (ARC). Its key feature is that a person's data held in separate administrative registers are assembled using the personal reference number for linking, thus mirroring the personal data that previously were entered on a census form. Denmark took the first ARC in 1981. An ARC is much cheaper than a conventional census, can be taken more frequently, places no burden on the public and is not vulnerable to protest. But it has to be supported by surveys to handle topics not recorded in registers, eg household composition.

10. In April 2006 the government announced plans for a population register for adults resident in the UK, with the aim of improving contact between citizens and public agencies. The register will be based on the issue of ID cards. But this cannot come to full fruition and be useful for statistics until virtually all residents hold ID cards in 2020 or later. So, to yield earlier benefits, an ONS-led study recommended that agencies consider sharing contact data in the interim: this implies creating an interim population register to be linked later to the ID card-based register. But no person or agency is responsible for pushing the interim register forward and ONS have divested themselves of any involvement.

11. In my view the development of a population register here is inevitable: to improve the accuracy of personal records and so protect privacy, and for reasons of technology, economy and the convenience of the citizen. It will happen whether or not we have ID cards, whether or not ONS are involved, and despite today's worries about the security of large databases. It is unfortunate that statisticians, who in 1916 and 1969 were pushing for a population register, should lose interest and involvement at the moment when it is within their grasp.

12. ONS priorities. ONS resources are stretched and I suggest that too many are being diverted to work on migration statistics (Section III). There is an urgent need to engage intellectual resources of the right calibre to develop demographic checks for the 2011 census. Another area demanding high level resources is the planning of post-2011 population statistics. This includes examining the role of a population register, the part ONS should be playing in its development and the scope for accelerating current developments.

13. Cooperation with stakeholders is touched on in Section IV.

I. THE CENSUS OF POPULATION14. The last 20 years have been turbulent for population estimates for two reasons. First, society has

become more mobile, less accessible and less willing to respond to surveys. And increases in international migration have created major problems in the enumeration of cities and in estimating the migration component of the annual roll-forward. Second, ONS made serious errors in census design and execution in 1991 and 2001, and on present plans will repeat the errors in 2011.

15. There was a marked increase in non-response in 2001 in all types of area. A record 4 million people (71/2% of the population) were not entered on census forms. Even in the “best" local authorities (the unitary and county LAs with the lowest levels of non-response and together having 10% of the national population) the non-response rate was about 3%, compared with an average rate among all LAs in 1981 of under 1%. The next two paragraphs identify contributary factors.

16. Question content [HC326-i, Section I]. The 2001 census form asked more than twice the number of questions asked in 1981. In 2011 the census must concentrate on its prime objective: to count the population. ONS should slim the questionnaire and should not contemplate adding difficult questions, eg on income.

17. Form delivery and collection [HC326-i, Section II]. Postal return of the completed forms in 2001 (mail-back) meant that there was no doorstep help and checking of the kind previously provided by enumerators at collection, with adverse effects on response. In 2011 ONS propose to go a step further: forms will be delivered by post (mail-out) instead of by enumerators. [HC326-i, §§4 and 15, wrongly stated that mail-out was introduced in 2001.] But the Scottish 2006 test found that mail-out yielded lower response rates and so Scotland will retain delivery

by enumerators in 2011. Putting enumerators back on E&W doorsteps in 2011 hinges on securing the finance needed to recruit enough able people. ONS should reexamine this.

18. Who in the household has to be entered on the form—the opt-out. [HC326-i, Section III]. Successive censuses have varied the instructions on who was to be returned on the forms. In 1981 the form asked for a return of anyone “present at this address" and of anyone else “usually living at this address". The number of residents of E&W picked up as “present at this address" exceeded by 0.7 million the number picked up as “usually living at this address", and the former number was used to compile the population estimates.

19. In 1991 the censustakers did not process the information they had collected about visitors (those present on census night but not “usually living at this address"). Later analysis of the completed forms discovered that 0.5 million residents of E&W had been returned only as visitors—leading to the belated realisation that the census-based population estimates had been compiled in ignorance of the existence of such a large number of people. This processing blunder should have been foreseen from the 1981 experience.

20. In 2001 ONS went a step further when the census form asked for a return only of those “usually living at this address". The words in italics are fuzzy, so that the claim “I don't usually live here" could be stretched: for many this was a total opt-out from the census. It meant that, purely because of form design, up to a million residents of E&W did not appear on any census form and, again, the census-based population estimates were compiled in ignorance of these people's existence. They were beyond census reach. ONS still do not admit that this happened.

21. The planned census form for 2011 will ask those “usually living in this household" the full 30 or so questions and will ask visitors to the household only for name, sex, date of birth and usual address. Many of the kinds of people missed by the 2001 form will now be entered on a form. But a partial opt-out will still be given to (eg) John who shares a flat in Westminster: he can answer just 4 questions by giving his parents' address in Oxford as his usual address. ONS will undercount Westminster's population and overcount Oxford's. ONS will also have to invent John's personal details along with the personal details of up to a million others who fill in only 4 questions.

22. Demographic checks on the census results [HC326-i, Section VI]. Given the unprecedented levels of non-response, the large numbers of people beyond census reach and the census shortfall compared with the rolled-forward population estimates, there was an overwhelming case for examining whether the national population estimates for 1991 and 2001 could be reconciled with past figures of births and deaths whilst allowing for the substantial uncertainties about the stocks of migrants (ie people born here and living abroad (emigrants) and people born abroad and living here (immigrants)). The two demographic checks I devised—termed Demographic Analysis—provide population estimates (within stated limits) that are anchored to past figures of births and deaths. Both checks are cheap and demonstrably robust [HC326-i, §§10,31,32]. Both contradict the downward revision to the 1991 population estimates which ONS made in Feb 2003 in a bid to lend credibility to the 2001 census results. Only one of the checks has been applied to the 2001 population estimates and it shows a sex imbalance: 1/2 million too few males in relation to females. ONS decline to apply the other check.

23. ONS have brushed aside these checks. Their commitment to the 2001 census results, which had begun with spin ("the most accurate census ever"), left them no room to countenance a challenge to its accuracy. Neither watchdogs nor professional institutions have investigated the situation. There has been no audit to establish what went wrong in 1991 and 2001 and where responsibility lay: a collective failure of the nation's statistical apparatus.

24. Given what has happened, a full range of demographic checks is essential for the credibility of the 2011 census results. But in ONS's Strategic Agenda for Population Statistics (2007) the only demographic check at national level is the comparison with the population estimates rolled forward from 2001. The reliability of this comparison is doubtful because the figures for the 2001 base of the roll-forward are unreliable and disputed; and because the estimates of migration, both at national and local levels, are very uncertain and with present measuring instruments will remain so. The outcome of the comparison will be impossible to interpret. Even if there were no divergence between the census results and the roll-forward, it would be insubstantial evidence of credibility. The credibility of the 2011 results must be tested by other methods and the two demographic checks I devised (§22 above) should be developed to fill this role.

II. BEYOND 2011—A POPULATION REGISTER?

25. The Executive Summary has described what a population register is, its role, the statistical spin-offs including an Administrative Record Census (ARC), and the government's plan to introduce an adult register for the UK (§§8-11).

26. The government's plan stemmed from an ONS-led study of the options, The Citizen Information Project. The study recommended that the register be located in the Identity and Passport Service and be based on ID cards. But the plan has serious weaknesses. (1) It depends on the success of the ID Card Scheme. (2) It assumes that “citizens comply with providing changes in circumstances"—a change in our culture that is problematic. (Continental experience is that, to keep a population register and client agencies' registers up to date, there needs to be daily two-way exchanges of contact data between the population register and the client agencies.) (3) The limited patchwork of bilateral data-sharing that is intended to help us over the next 13 years or more will yield only a fraction of the benefits that a full population register could. (4) No one person or agency coordinates action: responsibility is dispersed. The present plan is a recipe for delay.

27. I propose that the sharing of contact data be speeded up and extended to all agencies holding large databases of personal records. Sharing might be done by linking to the contact data held by the Department for Work and Pensions using the National Insurance Number as the personal reference number. The resulting population register would cover all ages. As the contact data in the ID register built up, they would be linked in. All this would need funding, and a focus and driving force (an “owner") located in a non-policy department. The proposal would bring forward the administrative and statistical benefits of a population register from 2020 to 2012/13. It could usher in a new era of annual population statistics following the 2011 census, leading up to an ARC in 2021. But such an ambitious programme would require a drive and commitment that at present is lacking.

Recommendation 6: The Statistics Board and ONS should place a population register at the heart of their thinking and should play a leading part in its development. They should seek to accelerate implementation of the present plans for a register.

III. ONS PRIORITIES28. Currently, ONS resources are being diverted into an uncertain attempt to provide the better

migration statistics demanded by local authorities and the Bank of England. But small scale voluntary surveys (like the International Passenger Survey) and disparate administrative data are incapable of reliably measuring population numbers and flows. There can be no assurance that migration statistics can be radically improved ahead of 2011. Concentrating so many resources on this topic misjudges priorities.

29. Resources need to be applied to rejigging the census machine, including the aspects discussed in §§16-21. There is a particular need to explore and develop demographic checks for the 2011 census of the kind discussed in §§22-24. The only check in ONS's Strategic Agenda is the roll-forward—a feeble and widely discredited tool which anyway was ignored in the 2001 round. New resources need to be brought in.

30. I agree with the Statistics Commission that work on new methods for the post-2011 decade ought to proceed in parallel with work on the 2011 census. A population register is being created here based on the ID card register (§10) and at some time in the 2020s it will yield a dividend for population statistics. ONS need to be involved: they need to be principal actors not bystanders. In the meantime responsibility for developing an interim register is so dispersed among agencies that few benefits, whether administrative or statistical, will accrue for 10 years or more. That timetable is unsatisfactory and requires action from the Board (see Recommendation 6).

Recommendation 7: ONS should engage intellectual resources of the right calibre to develop demographic checks on the 2011 census.

Recommendation 8: ONS should begin planning the post-2011 future of population statistics.

IV. COOPERATION WITH STAKEHOLDERS31. My experience over a decade is that ONS have found it very difficult to engage with or to recognise

an outside researcher who puts forward critical but constructive comments. ONS are too inward looking.

Recommendation 9: The Statistics Board should encourage a more cooperative culture in ONS

November 2007

@HR25@

Memorandum from Tyne and Wear Local Authority Districts'

1. USES AND DEFINITIONSWhat are the uses of population estimates, and how far do current population estimates meet the needs associated with those uses? What are the effects of inaccuracies or inadequacies in such estimates?

Population estimates are used as an important measure of population change and as a basis for population projections.

Population estimates, as the basis for population projections, are used in the allocation of Central Government resources. The effect of under-estimating the population would be insufficient funding for the provision of services to meet the needs of local people. As a result, it is vitally important that population estimates are as accurate as possible.

Population estimates are used as denominators for a range of socio-economic and performance data eg unemployment, activity and take up rates. Accuracy is again important to ensure that the rates reported are valid and meaningful. Access to accurate small area population estimates is also important to enable effective data analysis at neighbourhood level. Currently, Census based estimates are still used for this purpose.

Population estimates are important in carrying out Needs Assessments of the resident as well as migrant population. Large inaccuracy could result in under or over provision of services resulting in the inefficient use of public resources.

Population projections are used to determine general demand for services, housing requirements and Revenue Support allocation from the Department for Communities and Local Government (CLG).

Inaccuracies and inadequacies in population estimates affect population projections which, in turn, lead to wrong grant allocation and consequential unfair adjustments to Council Tax, wrong population and household projections, leading to uncertainty about future service provision and housing requirements and difficulties relating to planning issues generally. Tyne & Wear local authority districts have little confidence in the latest revised 2004-based population projections.

How appropriate is a definition of the population based on the usually resident population in the context of the needs of the users?

It is also important to have figures in respect of short-term migrant population who would not normally be picked up in annual population estimates. This is important to ensure local public services can build up the capacity to meet their needs as well the resident population.

How does the quality and range of population statistics in England and Wales compare with that available in Scotland, Northern Ireland and internationally?

Unable to comment due to lack of a detailed knowledge of the quality and range of population statistics in Scotland, Northern Ireland and internationally.

2. THE ROLE OF THE CENSUSHow does the Census contribute to the creation of population statistics? Is the current frequency of Censuses appropriate and, if not, how frequent should Censuses be?

The Census is the most comprehensive data set of the resident population and forms the basis for future population estimates/projections.

A more frequent comprehensive Census would be desirable but there has to be a balance between level of detail and frequency so as not to become overly burdensome and thus affect public participation. We would suggest that within the existing 10-year Census period a simplified head count of the population should be considered. This is becoming increasingly important due to rapid changes taking place in the population profile in the country while serious inadequacies in existing administrative data sources limit their ability to measure this change. The head count needs only to include information on number of people in the household, date of birth and gender. However, the additional inclusion of an NHS registration number for each individual would provide a good link between the head

count and fuller information within health service data sets. The resources which would normally be used to produce the mid-year estimates in these years could be used for the head count. A regular head count of the population would provide a good basis for the subsequent mid-year estimate. Potentially, the method to collect this information could be via a short form (called, for example, a “civic registration return") attached to the electoral registration form.

What lessons have been learnt from the Test Census carried out on 13 May 2007? What methods should be employed for the distribution and completion of Census forms, including the use of the Internet?

The Test Census carried out in May 2007 highlighted problems relating to the mis-match of addresses. An intermediate head count of the population prior to the 2011 Census would generate more experience and confidence in the distribution and return of Census forms.

We would support the development of alternative methods for the distribution and completion of Census forms, including use of the Internet. Electronic forms could be an effective method of presenting some of the more detailed questions within the Census, particularly those that include a range of valid responses eg ethnicity.

What steps should be taken to increase the rate of responses to the Census, particularly in hard-to-reach areas? How effective are plans to measure the extent of non-response to the 2011 Census?

Targeted awareness raising campaigns should be used to improve understanding and perception of the benefits of completing the Census.

To what extent is there a trade-off between the length of the Census form and its role in providing population information? What questions should be included in the 2011 Census?

We recognise that there is a trade-off between the length of the Census form and its role in providing population information. Our suggestion of more frequent simplified population head counts during the 10-year Census interval would allow the Census itself to focus more strongly on the characteristics of the population than its number. The use of new technologies, including Internet could alleviate the issue of length of form versus ease of completion.

Tyne & Wear Research and Information, and the Tyne & Wear Councils individually, have already responded in some detail to previous consultation exercises relating to Census 2011 question content.

To what extent should Census questions and statistical information be coordinated across the United Kingdom and how effective are plans to ensure such coordination?

It would be a sensible approach to coordinate Census questions and statistical information across the United Kingdom. This is particularly important for areas adjacent to UK internal borders.

What forms should the outputs of the 2011 Census take, how and when should they be made available and how should they be integrated and coordinated with other information on population?

Output methods used in 2001 Census were acceptable. We are happy to receive key statistics as a priority, followed by more detailed tables.

The ability to generate customised tables should be available via the web.

3. MID-YEAR POPULATION ESTIMATESHow accurate and useful are the mid-year population estimates for England and Wales, incl. information available for local authorities and Strategic Health Authorities?

Mid-year estimates of population for 2006 and current revisions to mid-year estimates from 2002 to 2005 inclusive incorporate a new methodology by the Office for National Statistics (ONS) to assessing the local impact of international inward and outward migration.

The considerable increase in international migration—especially inward migration—experienced by the UK in recent years means that this component of population change has assumed much greater significance within local population estimates and projections. We have concerns regarding the accuracy of the latest published mid-year population estimates for the North East and Tyne & Wear (See Appendix 1).

How appropriate is the methodology by which mid-year population estimates are reached?

The Tyne and Wear local authorities and Tyne & Wear Research and Information have discussed at some length the appropriateness of the new methodology and have urged ONS to reconsider the accuracy of the latest published population estimates.

One of our main concerns relates to the use of the Labour Force Survey (LFS) for allocating national estimates of inward international migration to a regional distribution, which we feel has led to the North East being disadvantaged in comparison to more populous regions. This is because the limited sampling size of the LFS has not allowed it to identify large enough sub-samples of recent international migrants in the North East to provide a consistently reliable indication of their numbers in the wider population of the region, even when data is aggregated over several years.

While local authority district-level estimates of international inward migration outside Greater London are initially determined by a process that takes no account of the Labour Force Survey, those estimates are subsequently constrained to sum to the total estimated for the parent region. District population estimates are therefore ultimately affected by any deficiencies in the LFS-derived regional inward migration estimate.

We recognise that the introduction of a propensity to migrate model in the estimation of out-migration is an improvement over the previous method, which depended simply on the overall population size of an area. However, characteristics which are closely associated with propensity to migrate at the national level may not be such an accurate predictor at a local level, meaning that the resulting estimates for a local authority area could significantly over- or under-estimate its actual levels of out-migration.

What are the most appropriate ways to ensure a common level of accuracy in mid-year population estimates for different areas and what steps should be taken to improve the accuracy in areas where population is hard to measure?

As stated above, we have concerns regarding aspects of the methodology for calculating mid-year population estimates, including use of the International Passenger Survey (IPS) as a basis for the international migration component. We feel that there is potential for generating a more accurate estimate of this information from a combination of existing administrative systems eg the Worker Registration Scheme (WRS) for A8 migrants, and National Insurance Numbers (NINos) for non-UK citizens. However, given the inadequacies in these data we would suggest, more fundamentally, that within the existing 10-year Census period a simplified head count of the population should be administered on a frequent basis eg biennially. This is becoming increasingly important due to rapid changes in the population profile in the country.

What is the right process for making revisions to mid-year population estimates so as to meet the needs of users?

The impact of revisions to mid-year population estimates (which is often considerable in terms of revisiting and re-evaluating work already done) would be significantly reduced if the process of undertaking a simplified head count on a more frequent basis eg biennially were adopted.

4. THE ROLE OF SURVEY AND ADMINISTRATIVE DATAWhat role can and should be played by survey data in the compilation of population statistics?

Survey data has a role to play in the compilation of population statistics but these should be based on a sufficiently robust sample. As indicated in an earlier response, we have serious concerns regarding the use of surveys such as the IPS and LFS.

We also have some concerns over the consistency of surveys over time and the variability of return rates.

What role can and should be played by data from General Practitioners and other health service data in the compilation of population statistics?

We recognise the potential role of data from General Practitioners and other health service data in the compilation of population statistics due to their almost unique degree of entire population coverage. However, there are some issues relating to potential over-counts experienced using patient registers due to the deregistration process. Equally, we are aware of an under-representation in certain groups eg young adult males.

What other data sources can and should be used in the compilation of population statistics?

There is potential for using a combination of existing administrative systems eg the Worker Registration Scheme (WRS) for A8 migrants and National Insurance Numbers (NINos) for non-UK citizens (although the limitations of these data sources are also recognised).

The School and Early Years Census' could be used as a comprehensive source of information to measure the population of children and young people.

5. COOPERATION WITH STAKEHOLDERSHow effectively has the Office for National Statistics cooperated with stakeholders with an interest in and information relating to population statistics and how can cooperation be improved by the Statistics Board and by its Executive Office?

The 5 Tyne & Wear local authorities have collectively submitted a number of points to ONS for consideration throughout the consultation period following the issue of indicative revised mid-year population estimates. Initially, the ONS seemed reluctant to provide comprehensive responses to points raised, and in some cases further requests for information were necessary to prompt an adequate response.

November 2007

@HR25@@ANAME@APPENDIX 1

ESTIMATING INTERNATIONAL IN-MIGRATION—HOW THE NEW ONS METHODOLOGY DISADVANTAGES THE NORTH EAST

IN SUMMARYMid-year estimates of population for 2006 and current revisions to mid-year estimates from 2002 to 2005

inclusive incorporate a new approach by the Office for National Statistics (ONS) to assessing the local impact of international inward and outward migration.

The considerable increase in international migration—especially inward migration—experienced by the UK in recent years means that this component of population change has assumed much greater significance within local population estimates and projections.

Tyne and Wear local authorities and Tyne and Wear Research and Information believe that a key element of the new ONS methodology, the use of the Labour Force Survey (LFS) for allocating national estimates of inward international migration to a regional distribution, has led to the North East being disadvantaged in comparison to more populous regions. This is because the limited sampling size of the LFS has not allowed it to identify large enough sub-samples of recent international migrants in the North East to provide a consistently reliable indication of their numbers in the wider population of the region, even when data is “smoothed" by being aggregated over several years.

IN DETAIL

1. International in-migrant contacts recorded regionally per annum by the LFS—2000 to 2006

2000 2001 2002 2003 2004 2005 2006East 88 48 72 72 46 60 91East Midlands

19 24 45 41 42 49 65

London 226 147 219 165 172 167 162North East

18 18 8 6 24 19 24

North West

38 43 33 61 37 61 48

South 121 106 108 112 93 89 108

EastSouth West

45 43 53 67 45 76 55

Wales 15 19 17 11 13 20 13West Midlands

41 41 53 37 46 32 43

Yorkshire and the Humber

50 54 60 49 57 96 54

Source: Office for National StatisticsTable 1 above shows the annual number of relevant migrant contacts recorded by the LFS over the period

used by ONS to formulate their revised regional estimates for overseas in-migration. The obvious sparseness and very variable consistency of these returns at the regional level made some form of aggregation a necessity and ONS decided that rolling three-year combinations of data would provide adequate smoothing for their purposes.

However, as the tables below illustrate, this process has still been insufficient to constrain the degree of volatility in the North East sequence of data to an acceptable level, or indeed to anything like the levels observed for the other regions. Table 2 looks at the variation within the smoothed regional LFS returns for the succession of three-year combinations that were deployed to allocate estimates of overseas in-migration to the regions when ONS revised the 2002-05 mid-year estimates of population and prepared the current estimates for mid-2006.

2. Smoothed (three-year aggregations) LFS contacts by region 2000-02 to 2004-06

2000-02

2001-03

2002-04

2003-05

2004-06

Min Max % variation of Max over min

East 208 192 190 178 197 178 208 16.9%East Midlands 88 110 128 132 156 88 156 77.3%London 592 531 556 504 501 501 592 18.2%North East 44 32 38 49 67 32 67 109.4%North West 114 137 131 159 146 114 159 39.5%South East 335 326 313 294 290 290 335 15.5%South West 141 163 165 188 176 141 188 33.3%Wales 51 47 41 44 46 41 51 24.4%West Midlands 135 131 136 115 121 115 136 18.3%Yorkshire and the Humber

164 163 166 202 207 163 207 27.0%

The level of variation in these smoothed LFS returns for the North East, almost 110% within a five-year sequence, is more than seven times greater than that of the South East and several multiples of the level found within all but one of the remaining regions.

Even that exception, the East Midlands, falls into line with the others when attention is confined to the returns for 2002-04 (which informed the base-year estimate of the series of population projections proposed to be used by DCLG in the Local Finance Settlement for 2008 to 2011), 2003-05 and 2004-06, shown in table 3 below.

3. Smoothed (three-year aggregations) LFS contacts by region 2002-04 to 2004-06

2002-04 2003-05 2004-06 Min Max % variation of Max over min

East 190 178 197 178 197 10.7%East 128 132 156 128 156 21.9%

MidlandsLondon 556 504 501 501 556 11.0%North East 38 49 67 38 67 76.3%North West 131 159 146 131 159 21.4%South East 313 294 290 290 313 7.9%South West 165 188 176 165 188 13.9%Wales 41 44 46 41 46 12.2%West Midlands

136 115 121 115 136 18.3%

Yorkshire and the Humber

166 202 207 166 207 24.7%

The range of variation for the North East in this three-year sequence has been reduced to a little over 76% but this contrasts even more strongly with the other regions where the range is consistently less than one-third—and as little as one-tenth—as much.

Clearly, the actual levels of overseas in-migration to the North East region will have experienced year to year variation—no one would expect them to remain perfectly static. It is not credible, though, that actual variation would have occurred on the scale suggested by the changeability of the LFS returns, even after smoothing. Nevertheless, on the strength of these statistics previous assessments of international in-migrant moves to the region have been reduced by as much as 39.5% (in 2004) and 35.9% (in 2005). Over the entire period of the revised estimates—mid-2002 to mid-2005—their impact alone has reduced the previous estimate of the regional population by 8,500.

Under the new ONS methodology, local authority district-level estimates of international inward migration outside Greater London are initially determined by a variant form of its previous approach that takes no account of the Labour Force Survey. However, those estimates are subsequently constrained to sum to the total estimated for the parent region. District population estimates are thus ultimately affected by any shortcomings in the LFS-derived regional inward migration estimate. In the North East, the five districts of Tyne and Wear have seen their collective estimates of overseas in-migrant movements for 2004 and 2005 eventually reduced by over 50% as a result of this process, and it has also been responsible for removing 11,000 from their joint population estimates over the 2002 to 2005 revision period.

IS THERE AN ALTERNATIVE?Upon close examination, the revised ONS approach to regional allocation of international in-migration

estimates within England and Wales has shown enormous potential to misrepresent the extent of overseas movement into a comparatively very small region such as the North East, perhaps to a greater extent than even its preceding methodology. Clearly, though, neither approach embodies anything like the precise and reliable mechanisms that are needed to measure the external influxes of population now being experienced in 21st century Britain. This can only come from an integrated system of appropriate administrative data sources that effectively records the overwhelming majority, at least, of actual in-migrants. We should be maximising all future efforts towards achieving this.

In the meantime, given the determination of ONS to press forward with its revised methodology, it seems obvious that the LFS sample in the North East needs to be substantially boosted if the overseas in-migrant components of its returns are to reach levels of consistency and reliability that are comparable with those in other regions.

This cannot be achieved retrospectively, of course, but for now the pronounced relative inadequacy of the pre-existing sequence of LFS-derived data for the North East could be greatly reduced by merging it with that of an adjoining region, most obviously Yorkshire and the Humber. A joint Y&H and NE region would be broadly comparable in population terms with the current South East, London and North West regions, and sub-regional allocation processes for the duration of the merger would be unaffected aside from being ultimately constrained to a new combined regional estimate rather than two individual ones.

@HR25@

Memorandum from Cornwall County Council

Cornwall County Council is very pleased that the Treasury Sub-Committee is conducting this very important and topical investigation into population statistics. It wishes to make the following points, structured to fit the topics defined in Treasury Committee press notice no. 77.

The Council is aware that lengthy and technical submissions have been made, inter alia, by the Local Government Association, the Greater London Assembly and Nottinghamshire County Council covering the breadth of local government interests. This submission therefore focuses on a few key points which are either particularly important for Cornwall or reflect our perception of national priorities.

USES AND DEFINITIONSOn the definitions currently used, the County Council has been broadly content with the estimates and

projections produced by the Office for National Statistics (ONS) for Cornwall and its constituent districts, including the details on age structure and components of change. They are very useful for service and strategic planning and resource allocation. The increasing emphasis on small area experimental statistics is particularly welcomed.

The main concern we have is that the definition, which requires a year's usual residence to justify inclusion, is no longer appropriate to give a proper picture of the complex make up of the population. We welcome the effort ONS is putting towards producing statistics of short term migrants at local authority level but we think it essential that this work is accelerated.

The objective should be to derive sufficiently broad information for population estimates on a range of definitions to be produced at all spatial levels. As well as short term international migrants, we would like more attention to be given to refining statistics on other flexible elements of the population such as second home owners and students so that population estimates to suit all purposes are possible. This approach is necessary to cover the complexity of the shifting composition of the population of any area and is essential if population statistics are to be used fairly by Government as a factor in resource allocation. It is highly desirable to try to measure the stable population and all elements of the transient and incoming population—not just immigrants but also people like visitors, commuters and second home owners.

Nationally our impression is that Scotland is often several steps ahead of England in the amount and timeliness of output.

THE ROLE OF THE CENSUSThe Census is absolutely critical for its contribution towards population estimates but we are concerned that

financial constraints mean that enumerators of sufficient calibre can not be attracted in many metropolitan areas and therefore response rates fall unacceptably low. This means even identifying the total national population is very fraught and there are knock on effects for the areas where the Census is more successful.

Conducting a Census every 10 years is reasonable.

Our impression is that the Test Census has been broadly successful subject to some points below. We support the approach whereby the Census questions designed to identify the population on different definitions more accurately will be increased and we especially support the proposed introduction of questions to establish information about second homes, critical in areas like ours. One of our main criticisms of the 2001 Census was that it missed and undercounted second homes badly and we conclude that this became a significant area of difficulty for the Census not just in areas like Cornwall where the concentration of second homes is greatest but also in London boroughs like Westminster where the undercounting of the population has received much more publicity.

We also believe that the Test has provided evidence that it is almost as acceptable to the public to answer 4 pages of individual questions as to answer 3. We are very concerned that, if there are only 3 pages, information on critical topics, notably employment by industry, carers and qualifications, will be lost. Even if it is accepted that the changes in society during the 2000's mean that it is justified for the 2011 Census to give even more priority to demographic questions, we think it wrong for that to be at the expense of other important topics.

The industry question in particular is, we believe, the only source of statistics on small area employment in sectors with a significant element of self employment. The qualifications question is presumably very important with the Government's current attention on skills. With the growth in the elderly population, statistics on carers also seem to be of prime importance.

Bearing in mind that it a Census is only conducted once a decade, we believe that funding must be found to finance a Census with 4 pages of questions per respondent. Only then will enough information be obtained to improve the base of population estimates while retaining other key questions on other topics.

It is right that there is concern about increasing the rate of response. We consider that the main current issue relating to that is the absence of a comprehensive list of addresses which lists all properties and classifies them into the appropriate category. Probably the most disturbing aspect of the Test Census was the large number of additional addresses found, mainly but not exclusively arising from the fact that the sub-division of large buildings into flats and the like had not been identified before. This is a national problem, obviously greatest in the big cities, but not to be underrated elsewhere including rural areas where, as already discussed, it is questionable how accurately in previous Censuses second homes and holiday cottages have been distinguished from primary residences and vacant dwellings.

The apparent decision to use the post for both sending and returning Census forms in most areas is probably justified on financial grounds but one big disadvantage is that it will reduce the chance of enumerators finding missed properties and that is another big risk with the Census strategy.

In principle co-ordination and harmonisation of Census issues across the UK is desirable but bluntly we have been more impressed before with the approach to Census planning in Scotland than in England and Wales. In 2001 Scotland pioneered an important question on travel to school, for example, and its approach to disclosure control seemed much more appropriate and reasonable than that in England. Furthermore with the advance of devolution, we would sympathise with those in Wales who would argue for greater discretion to shape the Census to reflect what they perceive as their priorities.

As regards outputs of the 2011 Census inevitably technical advances will mean the internet can be used even more effectively than in 2001. The opening up of Census results in this way is wholly beneficial. However, ONS will probably admit themselves that this was partly at the expense of catering for the more serious user. We would hope that after 2011 public sector practitioners as well as academics will be given the opportunity under appropriately controlled conditions to access and manipulate finer Census data than the public as a whole are allowed to.

We also wish to take this opportunity to make the committee aware of the importance of Cornish issues in the Census, especially ethnicity, identity and language as well as second homes. Cornwall County Council assisted ONS in holding a number of meetings earlier this year when Ian Cope, the 2011 Census Director and colleagues met interested parties in Cornwall to discuss ways in which the Census could make it easier for the large proportion of the population of Cornwall who wish to do so to record themselves as being of Cornish ethnicity, regarding their identity as Cornish or the fact that they use the Cornish language.

Mid year population estimates

As already explained, we think the mid year estimates are very useful and, at local council level, acceptably accurate. We have concerns about the way that international migration is distributed and it is always necessary to test methodology for groups like armed forces and students. The supporting information local authorities receive is very valuable.

One new issue that deserves attention is the creation of an intermediate geography, perhaps parliamentary constituencies or community network areas, within the larger local authorities, including the proposed new rural unitary councils for which comprehensive demographic estimates can be produced. Without this initiative, there is a danger that the amount of information that goes into the public domain will gradually fall.

The main reason why the methodology may need fundamental review is to make it fit for purpose to produce the wider range of estimates on different definitions discussed above. ONS have a large number of projects in hand which should help towards this.

The role of survey and administrative data

Generally we think survey and administrative data are useful to test and double check population estimates; we do not think they can or should replace current and developing methodologies for producing estimates. Earlier in this decade the Government was keen on producing a population register and there was even discussion of this replacing conventional Censuses. The County Council was concerned about this and is relieved that thinking seems to have moved away from this.

Health data are very important, especially for producing inter-censal small area estimates and tracing migration volumes and patterns. It has been recognised since their inception that there are biases with the differential level and speed of reregistrations by people in different age bands and genders. There may be new biases associated with growing ethnic and national diversity and ongoing research is needed on the registration rates of international migrants.

Other data sources such as school pupil admissions and benefits data relating to children and elderly people will presumably continue to be useful in compiling and checking population statistics.

5. CO-OPERATION WITH STAKEHOLDERSONS is probably exceptionally good compared with other government agencies in consulting and discussing.

Their main problems seem to be that restrictions on staff numbers and turnover limit how responsive they can sometimes be.

November 2007

@HR25@

Memorandum from Worcestershire County Council

1. USES AND DEFINITIONSWhat are the uses of population estimates, and how far do current population estimates meet the needs associated with those uses? What are the effects of inaccuracies or inadequacies in such estimates?

1.1 It is notable that small uncertainties which are marginal in National terms for total population are significant when applied in the context of different definitions (eg age, nationality), geographies and categories. The importance and the difficulties in capturing these “marginal uncertainties" should not be underestimated.

1.2 Most particularly in recent years there is the impact of the most uncertain and volatile aspect of population estimation, migration. This is even more significant owing to public perception due to the visibility of “non-british" residents, and the impact of migrants demand on services being identifiable (eg owing to language).

1.3 This has led (rightly) to an increased concern for some authorities of perceived inaccuracies in population estimation. Considerable work has been done by authorities individually, collectively, and by ONS, to address the concerns. It should be noted that, in our view, this has deflected effort from other developments in ONS's programme of modernisation and systems development.

1.4 Of significance to us is the means of establishing a consistent source for population estimates data. Much of the information put forward in support of better population estimates cannot, for various reasons, be used consistently across the country (England, nor UK). We are satisfied that, within the resource limitations they have, ONS are pursuing the need for better data, it's proper analysis and use, with the right priorities. They have consulted their users on their programmes in this regard.

How appropriate is a definition of the population based on the usually resident population in the context of the needs of the users?

1.5 The usually resident population definition is one amongst many population definitions of potential use to local authorities and the more complex living arrangements of modern life mean that ONS must adapt to the changing situation. At present students are allocated to their term time address for 100% of the time even though they spend part of their time at their parents' address, while increasingly people have several residences sometimes abroad. If you spend three days a week at one residence and four with your family at another having one simple view of usual residence is out dated. To this of course must be added the much discussed topic of migrants and in particular short term international migrants. Rather than just consider the single issue of short term international migrants it is important to consider all these issues together to reach a definition of the usually resident population.

1.6 However we recognise the value in having a single standard definition which can be consistently applied across the country. Ideally, though, the population estimates system would provide enough flexibility to allow authorities to calculate populations against a range of definitions to better inform service provision, provide a relevant base for performance indicators and for policy development.

1.7 ONS preparatory work for the 2011 Census has identified many ways of defining a person, in excess of 40, 23 alone for UK nationals, most of which are valid for some purpose. In practice LAs use different definitions; a population count could support these by providing different options.

1.8 The range of alternatives would ideally include population present, day/night populations, term-time and holiday populations, weekday and weekend populations. This of course raises its own issue of the appropriate or incorrect use of any particular definition for any particular use. However, as mentioned below, the use of one definition does not prevent the misuse of the data.

1.9 A major concern over population estimates is their use (along with population projections) in resource planning and allocation for Government departments, in particular, but not solely, CLG. Bearing in mind also the time delay in preparing population estimates, and still more, population projections, our view is that a more responsive solution for the concerns that LAs have in obtaining suitable resources could be to use other sources of data, particularly service-based data, for short-term resource planning. This would leave the estimates and projections to provide an “underlying, long-term" basis for resource decisions; these are typically reviewed for periods of 2-5+ years (1-2 years following), with projections having terms of 25 years and more.

1.10 This is a matter of the fitness for use of (various) population counts.

1.11 Notwithstanding the need for accurate population information, CLIP Population has identified areas where the use of population data is inappropriate in the light of its genesis, assumptions and robustness. Examples that we have identified include the following.

1.12 Transport planning: A model used by the Department for Transport (DfT), called TEMPRO (for Trip End Model Presentation Program) has, in the past used as an input the trend-based ONS projection of population to derive households (an input to the model). This is increasingly being rectified by the use of “policy-based" projections, obtained through Regional Assemblies and (theoretically) aligned with Regional Spatial Strategies. There are still issues over currency, spatial disaggregation and the use of the modelling that are been addressed. An area of particular concern is the lack of data on the changing frequency of travel to work which has such a major impact on the results of the model. Estimating the proportion of the population who will travel four days or fewer to work each week is critical when calculating distance travelled for sustainability purposes or demand for travel.

1.13 NHS resource and CLG Financial Settlement decisions use trend-based population projections that do not reflect a “likely" scenario. This

1.14 All types of indicators increasingly use rates to illustrate year-on-year (or shorter period) change. In some cases these relate data from current administrative sources to out-dated, and “not fit for purpose" data. This is not to say the data are inaccurate, or the definitions unclear, but they are unsuitable for the purpose they are put to. (eg BV 17 uses as denominator a population that includes people not in work, when the numerator relates only to people in work.)

How does the quality and range of population statistics in England and Wales compare with that available in Scotland, Northern Ireland and internationally?

1.15 We would note that population-related statistics in England and Wales compare very favourably with those in the other home countries, with, as far as we are aware, two exceptions. These are, in Scotland census and other data are not subject to disclosure provisions, and that there are housing-based data sources, derived from administrative data, that prove very valuable to verify population data on a consistent basis.

1.16 While internationally there are official population statistics that are considered more robust and accurate, and command high levels of public trust, it is noteworthy that these are almost exclusively based upon a population register

1.17 Thus, as the LGA submission states, it is sensible to monitor how these things are done elsewhere. ONS' intention to develop a business plan for an Integrated Population Statistics system, which would presumably examine approaches taken elsewhere, is therefore welcome.

1.18 We note that there is ambivalence amongst local authorities on a register, but support for a more integrated, administrative data based system as proposed with the ONS Proposals for an Integrated Population Statistics System. This should deliver more robust statistics, but could not guarantee widespread acceptance. We also note that there is increasing public suspicion about collection of data on the individual, so a register system would at least have the advantage of being, in principal clear, transparent and in the open. It is important that a separation is clear between the administrative purposes of the register, and ONS's survey role.

2. THE ROLE OF THE CENSUSHow does the Census contribute to the creation of population statistics? Is the current frequency of Censuses appropriate and, if not, how frequent should Censuses be?

What lessons have been learnt from the Test Census carried out on 13 May 2007? What methods should be employed for the distribution and completion of Census forms, including the use of the Internet?

What steps should be taken to increase the rate of responses to the Census, particularly in hard-to-reach areas? How effective are plans to measure the extent of non-response to the 2011 Census?

To what extent is there a trade-off between the length of the Census form and its role in providing population information? What questions should be included in the 2011 Census?

What forms should the outputs of the 2011 Census take, how and when should they be made available and how should they be integrated and coordinated with other information on population?

2.1 The importance of the Census in the creation of population statistics cannot be over-estimated; it is crucial. There are two aspects of the Census which need to be recognised, first that it is a key method of collecting local migration and travel to work data both of which have population definition implications but most importantly it is the only way that a complete understanding of the linkage between houses and population can be achieved. With a significant proportion of the population having several residences and vacant properties concentrated in particular areas it is important that double counting of the population does not take place. If issues such as gated communities, absent households, short term migrants and persons abroad are taken into account it is clear that we are dealing with a complex situation which requires either a census or a register which can identify if double counting is taking place. It is easy to criticise the Census but there are even greater problems with the alternative data sources.

2.2 The only feasible alternative to obtaining even superficial information on the population (ie counts, age/sex, personal, not relationship information) at district level or below is a comprehensive, administrative register, and this would only provide part of the census statistics. Having said this, in terms of populations statistics, ie “simple counts", the currency and accuracy of information between censuses becomes increasingly important, such that currently, within 3 years of the release of census information, the dynamics of population change are problematic.

2.3 We note that “hard-to-reach" groups often correlate to the “hard to deliver services" areas and rightly demand more attention, which is an approach of ONS census planning. However, evidence that a non-geographic, systematic deficiency in the census enumeration, applying to all areas (for example the lack of response for certain aged males), would require approaches that address being “hard-to-reach" across the whole Census geography. In other words, “hard-to-reach" areas are areas where certain “hard-to-reach" groups are concentrated.

2.4 We would support the view that there is significant value of having 4 Census pages per person rather than 3. ONS have judged that response rates will not be affected, and the only deterrent is financial cost. We understand that this is the main difference between the 3 and 4 page versions. Less information would be gathered, in particular on educational qualifications and on industry/business of employer. These are both very important in getting an understanding of disadvantage.

2.5 Output from the Census should be for the usually resident population but more flexibility is required to enable population counts on different bases to also be provided.

3. MID-YEAR POPULATION ESTIMATESHow accurate and useful are the mid-year population estimates for England and Wales, including information available for local authorities and Strategic Health Authorities?

How appropriate is the methodology by which mid-year population estimates are reached?

What are the most appropriate ways to ensure a common level of accuracy in mid-year population estimates for different areas and what steps should be taken to improve the accuracy in areas where population is hard to measure?

3.1 The LGA submission generally covers this point and earlier paragraphs of this submission.

3.2 With regard to the last question, the revision of statistics creates difficulties for local authorities because of the discontinuity and transition between significantly different totals. Most noticeably in recent times, this occurred following the 2001 Census, although this was not the first instance, and more recent revisions of estimates and projections have repeated the problem. A solution, but not one we would favour, is that the population estimates are shifted by only limited amounts towards the “new" totals.

3.3 This is an intractable difficulty when thought of simply in terms of population estimates or statistics. It is the effects of the revisions that should be the focus of attention. We would particularly draw your attention to paragraph 1.8, with regard resource planning using consistent data across local authorities.

4. THE ROLE OF SURVEY AND ADMINISTRATIVE DATAWhat role can and should be played by survey data in the compilation of population statistics?

What role can and should be played by data from General Practitioners and other health service data in the compilation of population statistics?

What other data sources can and should be used in the compilation of population statistics?

4.1 It should always be borne in mind that the problems encountered by the Census, certainly when it comes to coverage, are experienced to an even greater extent by survey data. The particular strength of survey data relates to more complex questions where the presence of an interviewer is critical.

4.2 Data taken from GPs registers provides particular challenges. There are often issues relating to the removal of patients from a list and this can lead to local biases. For example some surnames particularly amongst the ethnic populations are very common leading to difficulties in deleting records.

4.3 We note that the work of ONS to develop an Integrated Population Statistics System would support many of the points made in the LGA submission about the benefits of administrative data. We also note, with reference to our earlier comments about the “marginal" factors, and the LGA's comments on “rare populations", that the collection of data on the scale and movement of migrants, especially international migrants, is the most critical element to be improved.

5. COOPERATION WITH STAKEHOLDERSHow effectively has the Office for National Statistics cooperated with stakeholders with an interest in and information relating to population statistics and how can cooperation be improved by the Statistics Board and by its Executive Office?

5.1 We would especially support the points made in the LGA submission that ONS resources need to be taken seriously if the concerns are to be addressed, we have seen the effects of limitations in our work liaising with ONS, and have on several occasions expressed concern about work streams having to stop or be suspended.

5.2 We would also reinforce the point that ONS needs cross departmental support in its efforts to improve population statistics. It is an area where co-operation in and thinking, planning and working will provide huge benefits. We gave our support to this approach presented to us by ONSCD at the last CLIP Population sub-group meeting.

December 2007

@HR25@

Memorandum from Manchester City Council

1. USES AND DEFINITIONSWhat are the uses of population estimates, and how far do current population estimates meet the needs associated with those uses? What are the effects of inaccuracies or inadequacies in such estimates?

National Census, Mid-Year Estimate (MYE) and Sub-National Population Projections (SNPP) have two uses, where type 2 is dependent on type 1:

@UL@1. Setting the level of Revenue Support Grant (RSG), where inaccuracies and errors can have a profound affect on the Local Authorities (LA's) budget over the next decade. In Manchester's case the 2001 Census undercount and the subsequent Census Study by ONS restored an estimated 30,000 persons to the 2001 MYE worth over £100 million over the lifetime of the Census. The Census itself remained unchanged and information in relation to households and population in Manchester remain undercounted thereby influencing the interpretation of data for policy making, resource allocation and planning;@/UL@

@UL@2. LA service planning, target setting, performance indicators, funding bids, Best Value evaluations, as a denominator for other variables such as life expectancy, Index of Deprivation and producing small area demographics for wards or other catchment areas such as Housing Market Renewal. Also used by many other external organisations, partnerships and voluntary groups for the same purposes.@/UL@

Inaccuracies and errors may have a profound effect at small area level where this can skew service planning or the take up of services, for example wrongly associating a small area as having a large elderly population.

How appropriate is a definition of the population based on the usually resident population in the context of the needs of the users?

The usually resident population counts all persons who live in residential accommodation. It also counts students at their term-time address, so where they spend most of the year, persons in communal establishments, long-term international migrants (LTIMs—persons staying 12 months or more).

It does not count visitors (person staying for less than 3 months, which includes tourists), or short-term international migrants (STIMs—persons staying for more than 3 months and less than 12 months).

A count of visitors would be useful for service planning reasons and for considering the funding/resources implications.

The only available data that indicates international migration is the National Insurance Number (NINo) registration to overseas nationals. However, NINo registration does not tell us how many are short-term and how many long-term, about those who do not register, or if they are still in Manchester, or even in the country.

International migrants contribute to the local economy and have a positive impact. The extent of that contribution considered alongside what actual social pressure they make on services is hard to quantify and requires more research.

Since STIMs are mainly from the EU, North America and Antipodes, are young and single, and tend to find vacant posts in the catering and leisure industry through workplace notices and word of mouth, it might be supposed that very little pressure is put on LA services. Some pressure may be put on private rented accommodation and refuse collection associated with this type of property, public transport, but it may help some bus routes that are not viable, hospital A&E (there is some anecdotal evidence that STIMs see no point in registering with a GP) and possibly the police, through increased emphasis on community cohesion. More research needs to be conducted into these and other areas of potential pressure on services.

ONS is presently working on producing a STIMs count in consultation with LAs. However, since STIMs data will come from the same source, the IPS (International Passenger Survey), it is unlikely that the methodology will be acceptable to those LAs who reject the new LTIMs methodology. A count for planning purposes would be very useful, but the affect which STIMs have on public services needs further research.

Clearly illegal immigrants are very hard to count, but those who arrived legally and overstay should have been covered from arrival by the IPS (from grossing up the IPS), or from Home Office asylum applications and are, therefore, already recognised by ONS' MYE methodology as included in the national count, although it would be useful to have a separate, and more detailed, explanation of how they are estimated and distributed to LAs.

Illegal immigrants who enter the country illegally are not included within counts

Rough sleeper numbers are very tiny in the overall context of the total population (current count is 7), often with a rapid turnover and subject to seasonal fluctuations, so would be difficult to include in a resident count. Residents who become homeless are already counted in the MYE and those from outside Manchester will be picked up like any other internal migrant when they register with a GP

A count of residents of a LA is probably the most robust count that can be made, but an attempt to count the total population present, if it could be broken down into its constituent parts, would be useful for service planning. The 2011 Census will attempt to count residents and persons present, which MCC supports.

How does the quality and range of population statistics in England and Wales compare with that available in Scotland, Northern Ireland and internationally?

There was a different level of geography in Scotland in 2001, with Data Zones being much smaller than Output Areas (OA) in England, so data was available at a more local level. MCC would have preferred the OAs in England to be redrawn to the equivalent size, or the current ones to be split in two, but ONS consultation with LAs this year on Small Area Geography accepted that current OAs should be kept for continuity and ONS have not accepted the idea of splitting OAs on the grounds of Statistical Disclosure Control (SDC).

Different SDC rules applied in Scotland in 2001, so that data was more reliable at local level. Clearly, individuals need to be guaranteed that they cannot be identified by Census output tables, however, it is important that the data used by LAs makes sense at OA level. The method used by ONS in 2001 did not always make sense. For example, there were OAs containing a number of residents in communal establishments, but no communal establishments and vice versa. ONS original intentions to release OA data was reduced to one table only and even that data was so modified as to be of little use. The General Register Office of Scotland (GROS) used a mainly pre-tabulation method of record swapping that Scottish LA users considered fit for purpose.

The three Statistical Agencies in the UK produced a preliminary paper on UK SDC policy principles in November 2006. The paper is a very good starting point for producing a new SDC methodology and is to be subject to consultation at some point. MCC supports this new approach and will participate in the consultation.

2. THE ROLE OF THE CENSUSHow does the Census contribute to the creation of population statistics? Is the current frequency of Censuses appropriate and, if not, how frequent should Censuses be?

The MYE is Census based and the SNPP is MYE based, so the Census is fundamental to both Type 1 and Type 2 uses outlined above. The Census is the only survey of the whole country at the same point in time and its importance to creating and understanding LA population profiles cannot be exaggerated. But because it happens only once every 10 years, some of the data collected starts to lose value as the years go by, especially for a LA like Manchester, with its large student population, young age structure and high levels of mobility.

Unless, or until, there is a move towards improving and enhancing the collection of robust national administrative data sets (such as the ONS proposal for an “integrated population statistics system" (IPSS) to replace the Census), a Census, or partial Census (as in Canada) every five years would be supported by MCC. However, it remains to be seen if an IPSS will be good enough to stand-alone or if it may still need support from a partial Census of some kind.

What lessons have been learnt from the Test Census carried out on 13 May 2007? What methods should be employed for the distribution and completion of Census forms, including the use of the Internet?

ONS have not reported yet on the 2007 Census Test.

However, they have reported on the Coverage of Address Registers used for the 2007 Test Areas. The Address Test based on 2006 data compared OS MasterMap ® Address Layer 2 and the National Land and Property

Gazetteer (NLPG). The results showed in the two Test Areas most like Manchester, an undercount in both Registers for Camden, and an undercount for Liverpool in the former and an overcount in the latter.

The 2001 Census undercount for Manchester was mainly the result of missed addresses, where the Census Study added 24,600 persons to the MYE, and an estimated 5,500 mainly young men were also missed in wholly absent, or partly absent households, so the lack of a comprehensive address database concerns MCC greatly.

The proposed methodology for delivery and return of Census forms, by post out and post back, is probably not suitable for areas of Manchester which are difficult to enumerate.

The Scottish 2006 Census Test showed, not surprisingly, that the response was much better with hand delivery of the forms and GROS have decided to go with hand enumeration. MCC would prefer hand delivery and collection, to be made by trained enumerators with local knowledge. We are working to improve our Local LLPG generally, and for the 2011 Census in particular.

MCC is committed to the local government initiative to enhance the quality of the NLPG to make it the definitive addressing source.

The extent to which ONS engage with local government to support the delivery of the Census will be enhanced by a high quality NLPG and ONS need to consider how best to support the development in conjunction with local government. Clearly, if the intention is to rely heavily on postal delivery this may tend to indicate a requirement to use postal systems derived addressing but this will not be universally the best solution.

Additionally, the extent to which geographical based systems can be used by enumerators as an aid when visiting addresses needs to be harnessed and, again, this would be enhanced through collaboration.

To see the selection of an address register as a choice between two rival products is likely to restrict access to valuable address data.

In our view, one critical point, learned from the address matching exercise following the 2001 Census, was the need to be clear how many dwelling spaces there were in each area because this forms the basis of fundamental tests of accuracy by establishing the total amount of forms which ought to be returned. Also, through effective address referencing, it enables those forms to be tracked so that none go missing as they did in 2001 and it facilitates the flexible deployment of enumerators to those areas exhibiting low levels of form return.

ONS will need to select an addressing strategy which does the job required and current arguments about commercial rights to address registers are inhibiting this choice.

Interactive use of the Internet will appeal to a growing number of people and should be tested to ensure accuracy can be achieved and to understand the impact on enumeration resources.

What steps should be taken to increase the rate of responses to the Census, particularly in hard-to-reach areas? How effective are plans to measure the extent of non-response to the 2011 Census?

@UL@1. Comprehensive address registers are the first essential;@/UL@

@UL@2. By ONS working closely with LAs, as they did in the Test Areas, to identify persons and areas that are hard-to-reach. This is best done by utilising LAs' local knowledge and expertise;@/UL@

@UL@3. By hand delivery and collection, at least in the hardest areas;@/UL@

@UL@4. That ONS utilise LA staff with their local knowledge as enumerators, give them better training, more incentives, better enumeration tool kits (particularly electronic support), better support from managers and supervisors, provided by LAs to ONS, and more time to collect and repeat collection visits than the 2001 Census methodology allowed;@/UL@

@UL@5. By having a rigorously thorough community liaison and publicity plan, developed by LAs, who know their own communities better than ONS, but utilising ONS support and resources;@/UL@

@UL@6. By an effective system of form tracking to ensure completeness of processing and enable the flexible use of resources.@/UL@

ONS plans for measuring non-response are not yet fully described, but the Coverage and Adjustment Strategy has been/is subject to consultation. The plans in theory are very good, but they rely on some elements that did not perform well in 2001. The proposal for a Census Coverage Survey (CVS) remains the same, but with some necessary improvements. The use of a dual system estimator remains the same, but it recognises that the Census and the CVS are not independent of each other. The One Number Census methodology used in 2001 was not all bad, but calling it “One Number" was. ONS clearly do not intend to make the same mistakes. The methodology showed ONS that the count for Manchester was well below the Confidence Intervals used in the Quality Assurance (QA) process, but ONS decided that the QA method was wrong, going against their own methodology that was agreed to by MCC. That must not happen again.

It is essential for LAs to have confidence in 2011, that there are procedures to appeal population estimates, on the basis of evidence of departure from procedures, or large departures from expectations. This latter is properly part of the quality assurance process, which is not yet described by ONS but will become an integral part of the adjustment of census counts for coverage. This process of QA and appeals procedure must be transparent to, and involve, LAs.

To what extent is there a trade-off between the length of the Census form and its role in providing population information? What questions should be included in the 2011 Census?

In all surveys there is a trade off between the length of the questionnaire and the response rate. The extent of the trade depends upon the relevance of the survey to the individual and the level of incentive, or punishment for non-completion.

MCC's first priority is to keep the Census questionnaire as short and as easy to understand, and complete accurately, by the respondent as possible. In order to achieve the best possible response rate and most reliable population count, MCC's view is that new questions should not be added at the expense of the response rate.

We favour the inclusion or enhancement of questions that clarify and expand population definitions and bases (for example, population present). This is essential in order to improve the reliability and validity of the Census as the base for the MYE and SNPP. We also support the inclusion of an Income Question, but only if the 2007 Test shows that it does not undermine the response rate. The 2006 Scottish Census Test showed that it did not affect the response rate.

We do not suggest adding more questions, unless some old questions that are no longer required to be answered are removed.

One well-worded question on qualifications is all that is needed, whether respondents have professional qualifications that they do not use for employment purposes is not required and could be seen as intrusive. This question in the 2001 Census in Manchester produced the highest level of non-response (22.6%). Also number of employees at the workplace caused considerable confusion. For example, is the workplace the whole organisation or the respondent's office? In Manchester, 18.6% of respondents did not respond.

The wording of questions, and the cognitive testing of the wording, is essential to interpretation and meaning and a limitation on questions requiring subjective answers is required, especially where the questions are used for the allocation of resources to LAs. For example the question on Long-Term Limiting Illness (LTLI) as ONS recognised from previous Census Tests by adding the wording “Include problems which are due to old age" recognises a difference in interpretation by age group. The large increase in LTLI between 1991 and 2001 in some areas, indicates both changing attitudes to disability which affects the way respondents interpret this question and also a subjective interpretation as some people see their illness, or disability, as limiting, long-term and severe, compared to others with the same illness, or disability, who do not.

MCC overall supports the Census topics staying as close as possible to the 2001 form, as continuity and comparability should be given a high priority for the Key Statistics. So many government-funded projects require performance measurement and this can only be done if some of the questions remain unchanged.

To what extent should Census questions and statistical information be coordinated across the United Kingdom and how effective are plans to ensure such coordination?

The three Statistical Agencies in the UK are preparing to coordinate the Census, which is no doubt a good idea, but except for getting LAs in England and Wales a better SDC, is not an issue of any great importance to MCC.

What forms should the outputs of the 2011 Census take, how and when should they be made available and how should they be integrated and coordinated with other information on population?

LAs have a need for timely, easily accessed, clearly formatted, statistically reliable, UK consistent data, for all geographies including wards and with a high ability to compare Key Statistics with the previous Census. However, Manchester had a ward boundary change in 2004 and OAs no longer fit to our new wards.

The National Statistics Small Area Geography (SAG) consultation this year showed that 36% of respondents (of which 54% were from LAs) wanted ward level geography, with 44% of LAs mentioning wards. No question was asked about wards in this consultation. If it had been, no doubt the support from LAs would have been higher. Without ward level data, statutory requirements such as producing Five Year Forecasts for the Boundary Commission of England would be even more difficult to do with any confidence than it is now. ONS have decided to maintain the OA and Super OA geography, but have not as yet guaranteed any ward level data. MCC require such a guarantee and we would like to see the 2001 Census re-issued to the same ward boundaries. We understand that this will have to be done by the best fit of OAs to wards, which, while not perfect, is what we had to do ourselves with 2001 Census Key Statistics data, or have no ward data at all.

From the consultation report on SAG, ONS consider that there will be no need to merge, or split more than 5% of OAs because of a change in population between the 2001 and 2011 Census. MCC took part in this consultation and for Manchester our view is that where there has been extensive land use change from industrial and commercial to residential and rapid development of dwellings, particularly in the city centre, and its concomitant population growth, some of our OAs will need to be redrawn or split. ONS have said that: “Findings from this consultation—and particularly the areas which users have identified as troublesome"; will be used for further research and so we expect that to include Manchester.

3. MID-YEAR POPULATION ESTIMATESHow accurate and useful are the mid-year population estimates for England and Wales, including information available for local authorities and Strategic Health Authorities?

The MYE is very important to LAs, as it has both type 1 and type 2 uses (see question 1 above). However, as a LA, we cannot be sure how accurate the MYE is. We have no reason to doubt the accuracy at national level. At LA and Primary Care Trust level, there is more room for doubt. The age cohort based methodology is good standard practice and the births and deaths data is robust. The areas of doubt arise with the way students are treated and migration, but all methodologies will have problems.

Consideration also needs to be given to the time lag between the publication of these statistics and their use, in particular for revenue allocation purposes. The RSG settlement for 2008-11 will utilise population projections based on 2004 data and before, even though mid year estimates have been published for 2006. In MCC's case these projections are undercounting the population of Manchester and will therefore reduce the level of revenue support available to the City for MCC, its partners and other agencies that work for the benefit of the whole community.

How appropriate is the methodology by which, mid-year population estimates are reached?

Students are such a large proportion of Manchester's population (12% in total, just under 70% of 19 year olds and just over 50% of 20 to 24 year olds in the 2001 Census) and they do not age, so that age cohorts do not apply to them. It is not clear from ONS, what part of the methodology for producing the MYE (if any) applies to keeping students constant and the SNPP does not deal with students at all. The MYE does not have access to student counts as Higher Education Statistics Agency (HESA) counts students by university, though they plan to count them at their term-time address in future and this data could be used to improve the way students are treated by the MYE. HESA data also needs to include data on nationality and address one year previously to identify overseas students. The methodology for counting students separately from the age cohort model needs to be reviewed and researched.

Migration includes internal migration and international migration.

Internal migration data is based on patient register and patient re-registration recorded in the National Health Service Central Register (NHSCR) and shows moves between Local Authorities, Government Regions in England

and Wales only. It does not include the rest of the UK (Scotland and Northern Ireland). There are problems with the NHSCR data, as it is more robust for children and elderly people, as they are more likely to register with a GP. It is poor on young adults and men generally who are less likely to register with a GP, and it suffers a level of list “inflation" as a result. Also, as the data is rounded to the nearest 100 for age groups, small variations are not shown and sub-totals do not add up to totals.

The Improving Migration and Population Statistics (IMPS) project developed a new methodology for international migration and applied it to the 2006 MYE and revised the 2002 to 2005 MYE and the 2004-Based SNPP. There is insufficient data to make a truly robust reliable count of LTIMs, everyone recognises this, including ONS. The total international migration count uses the International Passenger Survey (IPS), the Irish Republic Household Survey and the Home Office count of asylum seekers. Irish migrants are not covered in the IPS. MCC supports methods to improve the migrant count as outlined in the following section.

What are the most appropriate ways to ensure a common level of accuracy in mid-year population estimates for different areas and what steps should be taken to improve the accuracy in areas where population is hard to measure?

The existing MYE methodology is still probably the best to provide a nationally consistent estimate, though some improvements could be made to the base data sets.

HESA is mentioned in the section above.

The NHSCR data is the only reliable source nationally for internal movement and anything that could be done to improve the quality of the data would be very useful. (See Section 4 below).

The IMPS new methodology for counting LTIM, introduced this year by ONS, appears to be better than the previous methodology, and fairer. The revised MYE components of change resulted in Manchester both losing and gaining population, but gaining overall by the 2006 MYE. The new methodology means that more inward-LTIM are allocated to neighbouring areas than to Manchester, but more outward-LTIM are retained in Manchester. The means of counting Migrant and Visitor Switchers is improved, as is the method for counting outward-LTIMs by using a propensity to migrate model.

LTIM will be improved further next year when the Integrated Household Survey (IHS) replaces the Labour Force Survey (LFS). The LFS does not include residents in communal establishments, and because of its small sample size, it is only used below sub-regional level for London Boroughs. The IHS will have a much larger sample size, but it is also a household survey and some measure needs to be taken to include communal establishments residents.

However, the best way of improving the LTIM and the STIM count would be to make more resources available to increase the sample size of the IPS. However, consideration should be given to replacing it by the equivalent of 100% compulsory passenger embarkation cards, in and out, at all points of entry and exit. This form of compulsory monitoring existed prior to the UK joining the EU and could be reinstated through the EU e.borders project. Parts of the project should be funded and brought forward from 2010 and implemented in 2008.

What is the right process for making revisions to mid-year population estimates so as to meet the needs of users?

Through ONS continuous consultation, complete openness and clarity. ONS listening to LAs and explaining where users suggestions are capable, or incapable of being implemented, or require researching and the reasons why. Through meetings and workshops around the country, by email and website. Publicised as widely as possible using ONS existing email groups, Central and Local Government Information Partnership sub-groups, Local Government Association, Local Authorities Research and Intelligence Association, etc.

4. THE ROLE OF SURVEY AND ADMINISTRATIVE DATAWhat role can and should be played by survey data in the compilation of population statistics?

The use of available data needs careful consideration. Increasingly, administrative data is being compiled which can assist in validating survey result and the outcome of the Census. Following the 2001 Census, analysis of administrative data showed that the result was questionable, but at the time, this quality assurance result was rejected. Without vigorous representation by the Council this would not have been acknowledged in subsequent studies conducted by ONS.

The Census, IPS, LFS, and IHS are all surveys and without them it would not be possible to have any accurate population data. The Census is the most accurate survey because it attempts to cover 100% of the population, even though it is improbable that it has ever succeeded in the past or is ever likely to in the future, but smaller sample surveys are required for inter-Census years. The sample size depends on the resources available and all surveys that aid the national MYE and SNPP should have priority of resources.

However no existing survey other than the Census includes persons living in communal establishments and for Manchester this means missing large numbers of students in halls of residence.

What role can and should be played by data from General Practitioners and other health service data in the compilation of population statistics?

NHSCR developments phased in from 2005 to 2007 have, apart from changing the registers name from Central Registration System to Care Records Service, is a new electronic system of keeping patients records in one place to enable health workers to access a patients record on-line. However, according to NHS Connecting for Health, it will be “several years" before it is rolled out across England. The Personal Demographics Service (PDS) is an essential element of the NHS CRS. Every registered person in England will have an electronic record consisting of both demographic administrative data and medical data by 2010.

However, it is the perennial problem with all administrative databases, it is not designed to collect data for any other purpose and the demographic data held is very basic: name, address, date of birth, date of death (where applicable), sex, and preferred language. Unfortunately it does not hold any data that is considered to be “sensitive" such as nationality, ethnic group, religion, or date of arrival in the UK, so it is unlikely to be of any help in counting international migrants and its uses are limited. Nevertheless, if resources are committed to cleaning and maintenance it could help make an improvement to the population count at LA and small area level, and to internal migration. This may be in time to inform and to quality assure the 2011 Census population.

What other data sources can and should be used in the compilation of population statistics?

Other data sources have to be collected nationally and to the same standards. Most other data sources do not cover large enough sections of the population and because they are collected for other purposes, they have many problems associated with them that would need to be addressed before they could be used for counting the population. However, they could be used in a supporting, or quality assuring capacity when distributing population to LAs.

The datasets with most potential and/ or require further research as to their utility are:

@UL@1. Council Tax dwellings from the Valuation Office, which could be used to give an estimated population using an occupancy ratio;@/UL@

@UL@2. National Insurance Number allocations to overseas nationals, which could be used at the IMP New Geography sub-regional cluster level;@/UL@

@UL@3. Schools Census of languages spoken in pupils' homes could be used as a proxy for nationality.@/UL@

@UL@4. NHSCR Flag 4 added to the records of an international migrant when they first register with a GP needs further research as to how useful it is, as there appears to be some doubt about the reliability of data collection methods. However, it is not clear whether the changes to the system, currently being introduced, will include Flag 4.@/UL@

@UL@5. Workers Registration Scheme is probably less useful than NINo, as it only applies to EU countries.@/UL@

Other LA administrative data is dependent on the amount of resources allocated to their collection by each LA, for example the Electoral Register (ER), and quality is, therefore, variable and patchy nationally. Methods used to keep the ER clean and up-to-date are very good in Manchester and enabled MCC to use the ER as evidence of an undercount in the 2001 Census. Some other LAs, who considered their populations were undercounted were not able

to replicate MCC's evidence because their ERs were suffering from inflation as a result of the change to rolling registration.

The introduction of the Coordinated On-line Record of Electors (CORE) system may resolve the problem of patchiness. The aims of the Department for Constitutional Affairs, in introducing CORE in 2005 was to achieve: “improvements in the integrity of electoral registers; more efficient access to registration data for those authorised; support modernisation of the voting process; improvements in the efficiency of the electoral administration process; and provide a mechanism by which reports can be provided and/or research conducted efficiently on electoral registration data, based on a national dataset". When CORE is finally implemented across the country it will provide another source of data to help with quality assuring Census data.

Additionally, new legislative requirements will require more rigorous canvassing. This is an example of the need to keep the quality of administrative data under review and to anticipate improvements in that data in the run up to 2011.

However, as with all administrative data sets, the ER is not designed to provide population data. Adults entitled to vote sometime choose not to register, or many young adults are highly mobile and miss annual registration processes. ERs exclude children aged under 18 and non-EU foreign nationals. Nevertheless, the addition of the nationality question this year, could add a little to our knowledge of the population.

5. COOPERATION WITH STAKEHOLDERSHow effectively has the Office for National Statistics cooperated with stakeholders with an interest in and information relating to population statistics and how can cooperation be improved by the Statistics Board and by its Executive Office?

ONS has improved enormously since the 2001 Census undercount, but there is a concern that the speed with which they are able to establish the more practical cooperation in delivering an effective Census in 2011 is not in line with their stated intent of much greater involvement with LAs.

Recently, however, we are aware of the development by ONS of a strategy for engagement with Local Government together with the specific initiative to develop a regional approach to engagement through regional lead Chief Executives along the lines of the approach adopted for Regional Returning Officers. This initiative has our wholehearted support because the methodology is proven and it is vital that Local Government takes an active role in supporting the conduct of the 2011 Census.

More specific comments are:

@UL@1. ONS are conducting a wide range of consultation on the 2011 Census and the MYE, but how far they are prepared to take on the views of LAs is a moot point, which would be improved when ONS publish consultation comments, if they indicated their response and why.@/UL@

@UL@2. Another issue is the slowness of ONS to respond to any request, or enquiry outside standard data requests and on occasion months can go by before the results of consultation are known, often after ONS have made their final decisions and often deadlines are not always met. @/UL@

@UL@3. Another is ONS failure to disseminate widely enough to LAs, information on meetings and workshops that are important to LAs and therefore, can be easily missed.@/UL@

In global terms there needs to be a recognition throughout those at ONS charged with delivering an effective Census that LAs are not simply “user" stakeholders, but represent an essential resource which must be effectively harnessed to achieve the best outcome. This can be achieved without jeopardising ONS's delivery of independent results.

The Statistics Board could act as “ombudsman", or “OffStat", a last resort independent complaints commission for data users. Also it could act as a standing quality assurance committee.

Additional — Please list any further matters which you feel should be raised in the inquiry

The use of population data for the Comprehensive Spending Review and the Formula Spending Share, by which LAs are allocated Revenue Support Grants, should mean that ONS' first priority should be given to methods of

collecting population data, by committing the resources required to get the best possible results. Both national and local government spending crucially depends on this data, no other organisation, statutory or voluntary, have the same dependency and, therefore, ONS should give utmost priority to our needs. Government must make the necessary resource available to ONS to ensure that the research required, the enlargement of sample sizes, and the improvement of administrative data collection methods can be implemented as quickly as possible. In particular, MCC would like to see resource committed to the IPS, e.borders and the production of the 2006-Based SNPP to be substituted for the revised 2004-Based SNPP in determining current RSG.

MCC participated in ONS consultation on ONS' Statistical Work Programme for 2008-2012 earlier this year and commented on the need for reliable data for local geographies for policy work. MCC noted that in the coming years, all ONS products will be affected due to rapid changes in the global economy and population mobility. Correct measurements of economic and social changes need to be monitored and that is only possible with reliable statistics at a very local level, which are timely and regularly updated and can be linked geographically, such as by Lower Super Output Area. For example, recent in-migration has changed the socio-economic make-up of the city, particularly in some local areas, but local level information is very limited but necessary for local policy work. It is imperative to focus on improvements in the methods of distributing internal and external migration between local authorities in population estimates. Reliable information on population estimates is absolutely essential for policy development and to monitor progress, particularly at a local level.

December 2007

@HR25@

Memorandum from David Coleman, Professor of Demography at Oxford University

PREFACE@UL@”No gentleman who has sat any time in this House can be insensible of the utility of the law proposed; for seldom a session passes over but something happens in which it would be of singular advantage to the public and a great satisfaction to ourselves to have upon our table such authentic accounts as are proposed by this Bill to be laid yearly before parliament" @/UL@

Thos. Potter MP, proposing the bill to create a population register (An Act for Taking and Registering an Annual Account of the People . . . Geo III 1753).

SUMMARYAll aspects of population statistics in the United Kingdom are in an unsatisfactory state. Even the base

population remains uncertain. Despite every effort, the last two censuses have turned out to be unsatisfactory. Even the 2001 census, designed to be infallible, has had to be revised twice and its incompatibilities with other sources patched up with statistical Polyfilla. With present systems the degree of error is unknowable but possibly large. Inappropriate questions are asked, and necessary ones ignored. Immigration flow statistics are estimated on small voluntary samples of intended immigration and emigration, of incomplete coverage and high sampling error. Immigrants' destinations around the country are based initially on their stated intentions on arrival, naturally subject to revision. With these systems we cannot know who is in the country, legally or illegally, when they arrived, where they are or if and when they left. The number of illegal immigrants is anyone's guess although the government has given an estimate of about half a million. Internal migration and local population estimates are based on obsolete and often wrong census counts, sample surveys inadequate for local authority use and indirect and partial estimates from changes in doctors' registrations. Current huge migration flows quickly render estimates out of date.

While energetic attempts are made by ONS to rectify these acknowledged deficiencies, it is now clear that all these systems, never very effective, are now so unreliable that they have reached the end of the road of any effective radical improvement. The current system is beyond effective reform; investment in them has been inadequate. Only a comprehensive population register, possibly on the Scandinavian pattern, can restore accuracy and timeliness to population statistics at local and national level, and also tell us who has entered the country, where they are and when, or if, they leave it. Such a comprehensive register is not on offer: the “national identity register" of adults proposed for the identity cards is not the same thing

We have given ourselves a pervasive and expensive welfare state without the means to evaluate the condition, the needs or even the numbers of the population whose welfare is to be safeguarded and whose entitlements and security are to be protected. That is not a good way to enter the 21st century.

TIME FOR A FRESH STARTThis note considers the current state of UK population statistics: their accuracy and their fitness or otherwise

to serve public policy aims, what the aims should be in population-related matters. It makes some suggestions for short term improvements to existing statistical sources. But essentially it concludes that those systems (census, vital registration and surveys) are close to the limit of improvement and that a radical replacement is needed in the form of a population register.

THE CURRENT STATE OF POPULATION STATISTICSFew aspects of UK population statistics are in a completely satisfactory state, whether relating to the

numbers and condition of the people, at national or local level, the volume and purpose of movement of foreigners into, around and out of the country and their current numbers and distribution, lawful and otherwise, the pattern of fertility or the projection of population or its components. Some are so poor as to be barely usable.

In the interests of brevity, this submission will proceed directly to the case for a population register and some of the objections to it. It will be assumed that the controversies over the 1991 and 2001 censuses, the severe problems afflicting internal and international migration data, the consequent great uncertainty and disputes over local authority population size, are generally accepted as difficulties needing solution. The these problems and recent official responses to them are considered in some detail in an appendix to this note, which also comments on recent official responses and adds some other proposals for short-term improvement. For these that need no persuasion that “something must be done", all that can be ignored.

A CONTINUOUS POPULATION REGISTER

Short-term improvements aside, this submission proposes that existing systems should be brought together and connected with, or replaced by, a compulsory continuous population register for all UK citizens and non-UK citizen residents incorporating a unique person-number (PN). Such a system could be on lines long familiar in some Scandinavian countries, in Belgium, Israel and others (NCHS 1980) and in a less developed form in Austria, Finland, France, Iceland, Germany, the Netherlands, Switzerland and elsewhere (see Salt et al. 1994). A register would be a fuller development of some of the ideas put forward in the Integrated Population Statistics System (ONS 2003), most of the ambitious elements of which seem to have been dropped. However the first serious proposal for a modern register in the UK in recent times was made by Redfern (1990). The person-number would be assigned at birth to each child, or to each person naturalised or accepted for long-term residence, and would be unchangeable. It would be used as a unique identifier on all other records; NHS, DVLC, welfare payments and pensions and national insurance. A number of ways could be imagined of developing such a system (Birch 2007).

In most countries registers are kept by each local authority (counties, in Sweden) and centralised electronically in the national statistical office, the central records being updated regularly, often weekly. Usually there is a separate register for aliens (non-citizens). In the UK case, given present problems, that would be one of the most valuable innovations. However in Belgium there is one National Register including all Belgians and resident foreigners, together with Belgians living abroad. An extract from the register relating to foreigners only (for whom additional information is kept) constitutes a National Register of Aliens and Residence Permits. In the Netherlands the 489 municipalities are responsible for maintaining local registers of Dutch citizens and (separately) non-Dutch citizens. There is no central register as such, but a “virtual" register whereby all the municipal registers are linked by a secure electronic postbox maintained by the Ministry of the Interior. That linkage enables national totals to be derived regularly (Salt et al 1994; Poulain et al 2006).

Originally the registers were entirely local, in the case of Sweden dating from the 18th century. Centralisation only became possible in the 1960s with the development of electronic computers. Once that was possible, a huge advance was made in the maintenance of timely and accurate national demographic information for every kind of public policy need, also permitting considerable advances in epidemiological and demographic research.

A person number (PN) is typically a ten-digit or longer number indicating the date of birth and sex of the individual, a unique individual number and a check digit or digits that ensure that the PN has not developed an error. Thus in Sweden the PN 450410-1488 indicates a woman born on 10 April 1945 with the individual number 148 and check digit 8. All administrative records relating to that individual would have to bear the PN.

Even in the UK some elements of this system are already in place. There is already a universal person identifier. All births in the UK (and legal immigrants) have had a birth number assigned to them through the NHS system which tracks all their medical records and follows them as they move house around the country through the computerised NHS Central Register at Southport (Hornsey 1993). In the absence of much other data, it is one of the chief sources of information on internal migration, albeit an unsatisfactory one for obvious reasons. It is not, however, used outside the NHS. The Central Register can only give information on moves between the 100 (former) Health Authorities, and only when patients choose to (re)register. That can involve considerable delay.

However, each such authority holds a register of patients with local GPs. That Patient Register includes the NHS number, sex, date of birth and postcode for each patient. With these, the Office for National Statistics can create a register for the whole of England and Wales. Linking records using the unique NHS number, persons who change their postcode from one year to another can be identified, enabling much more detailed, if still incomplete and untimely internal migration estimates to be made for small geographical areas (Chappell, Vickers and Evans 2000; ONS 2003). The National Insurance number is a near-universal number increasingly used as a general identifier for persons over age 16, for tax and other purposes, well beyond the scope of its original intended function.

An opportunity thrown away

There are two previous examples of a compulsory registration in the UK, each accompanied by the issue of identity cards. Both were provoked by wartime emerge cy. The first (1915-1919), relating to the adult population only, was introduced amidst controversy about conscription in order to establish the number of men available to join the armed forces. That having been established (1,413,900) official and public interest in both card and register waned —there were no incentives for keeping or using it—and the system was put to sleep in 1919. A later proposal for a population register by the then Registrar-General, Sir Bernard Mallet (1929) was based upon much broader arguments, some demographic others related to the protection of individual entitlement. But it fell on deaf government ears. The next, much more successful emergency scheme arose from the establishment of the National

Register in September 1939, which had been in preparation since 1938 in the anticipation of war. That involved both a register and a simple identity card containing a few basic details including the unique identity code (eg ASEO3 followed by the surname). That work, using 65,000 enumerators to distribute and collect 12 million household schedules, was followed by the issue of identity cards to all individuals (47 million), except sailors and servicemen, from the Central National Registration Office near Southport. All that was completed between 5 September and 29 1939 (Registrars-Generals, 1939). Information collected was modest: name and address, sex, date of birth, marital condition and occupation. The unique birth number associated with the Register became, with modification, the NHS number mentioned above. This system worked quite well, not the least because the 1930s Registrar-General Sylvanus Vivian built into it an incentive of what he called “parasitic vitality". Possession of the ID card was linked with entitlement to the then rationed food supplies: no ID card, no rations (Agar (no date)). National Registration continued after the war without much controversy except for the use of the cards by the police, and some concern about “function creep" and only ended in 1952 following a change of government (food rationing did not end until 1954).

An opportunity missed

However, had far-sighted proposals made in parliament in 1753 come to fruition, Britain would have had a population registration system for the last two and a half centuries, which, if developed in the way that other similar schemes were in other countries, would have given us a far superior system for managing population, health and welfare data than we currently enjoy. The 18th century was a time of great population controversy. In the absence of a census or an understanding of the way in which the parish registers could usefully be analysed, there was great uncertainly between learned as well as ordinary people as to the size of the population, and whether it was increasing (as it was) or declining (Glass 1973). A number of remedies were proposed, of which Thomas Potter's was the most sensible. A annual local count of population, and a record of all marriages, births and deaths was to be taken on a secular, not a religious basis. The results would be centralized, providing annual local and national totals of population and vital events— a very suitable prototype for what is proposed here and similar to schemes initiated around the same time in Sweden and elsewhere. Despite some opposition on religious and libertarian grounds, the Bill passed in the Commons by 57-17. But in its Second Reading in the Lords, it was referred to a Committee of the Lords for a date outside the session. Accordingly it lapsed and was not revived in the next session. As a result England and Wales had to wait for a second-best system: until 1837 for the civil registration of vital events; a satisfactory census on modern lines (based on households) was not taken until 1841.

The new proposal

Apart from the PN linking all administrative records, the major innovation would be to create a register of all residents including the PIN and listing basic individual data: name, address of principal residence; citizenship, date and place of birth, marital status, age and sex. Following overseas practice, the obvious location and responsibility for maintenance would be with the local authorities. But it could be run centrally like the proposed National Identity Register for the Identity Card. It would be linked through the PN to administrative datasets such as the NHS Central Register, the National Insurance register, local authority rating lists and, where persons were eligible, the register of electors. Or it could replace some of those, for example the last-named. The person-number would be the passport for access to education, to NHS services, to the welfare system and to pensions. The register would not need to be associated with an ID card but it would be more useful if it were.. But with an ID card it would also be more difficult to introduce, given the reservations about that proposal on civil liberties, cost and other grounds (Enterprise Privacy Group 2005). A separate register sub-register would be needed for non-naturalised foreigners, linked to e-borders and the Home Office, containing additional information concerning immigration status. Some countries with PN systems such as Sweden have two or three stages of PN related to immigration status, showing the entitlement of the bearer to limited or full welfare and health care rights. In order to provide timely and complete national data, the information on the local registers would need either to be submitted regularly to a central register or connected with all the others in a decentralised network similar to that operating in the Netherlands. For information, the kinds of data kept in the Belgian municipal registers are listed in Appendix 2.

It should be noted that this register would be different from the National Identity Register proposed in conjunction with the Identity Card. The latter register would only exist to support the Identity Card, containing, it would seem, minimal information but capable of being linked to other administrative registers. Its multiple aims appear to be primarily to minimise crime, welfare and identity fraud, illegal immigration and the risk of terrorism. Accordingly only those aged over 16 would be included. It would have no demographic function and could not replace any existing demographic systems. The register proposed here is the basis a complete population system for public policy ends, which should be valuable in improving security, especially in respect of immigration, but is not created to that end. And it would not require an ID card. Some countries with population registers do not have ID

cards. The Swedish ID card was only introduced in 2005 and is not compulsory. The ID cards in Austria, Finland, France, Italy, Luxembourg, Netherlands, Portugal and Sweden are likewise voluntary. Denmark, although maintaining a population register with a person number, has no ID card at all.

The next section considers the advantages and notes some of the objections to national registration. A very useful and more complete review or these issues can be found elsewhere (Redfern 1990).

ASPECTS AND ADVANTAGES OF A POPULATION REGISTERAlready, in order to survive at an ordinary standard of living in a welfare state such as the UK, entitlement to

services depends on being listed in one register or document or another, at the moment each operating under a different system with a different form of identification.

The PN would not of itself confer any additional privileges, but would facilitate access to them and ensure that only the person entitled to them could make use of them. It would not itself contain confidential information. But through linkage it would ensure identification of each individual without confusion, on all existing administrative databases (NHS, National Insurance, electoral register etc). In broad terms, it would ensure that the person was certified as being known to the British State and was entitled to some or all of its services.

However parallel changes would be needed to tighten up the requirement to show entitlement to use local government and NHS services, which currently is very uneven or not required at all.

The PN would require, to generate population statistics, the development of local authority address databases with more information than the PN (for example name, address, age, sex, marital status and citizenship), to which the PN would be linked.

That would generate highly accurate national and local population data capable of being updated by linkage as frequently as necessary, providing true “one number" information.

Its use would require residents to register on moving permanently to a new LA and de-register from the old one. That would only make more systematic the current system for recording responsibility for local taxation.

Such regular national updates would replace the annual local and national mid-year population estimates obtained by rolling forward the data from the last census with subsequent births, deaths and migration.

It would give much better information on local migration and local population change than the present system, which relies on a rather ramshackle combination of sources, principally changes in NHS registration.

Because of their responsibilities for delivering education, LA housing, planning decisions and social services, local authorities are the most important consumers of population information and claimants on Treasury funding. At present many are badly served through the recent uncertainties of the census and the unprecedented pace of immigration, with which conventional local population statistics data sources cannot keep up.

Naturalised immigrants, or those granted indefinite leave to remain, would acquire a PN on the same or similar basis as others. A separate register or sub-section would be required for shorter term immigrants (those given leave to enter for more than three months). The PN would be recorded on the departure and arrival of international migrants.

By providing a secure basis for enumeration., the PN would ensure a much higher accuracy of population censuses, and would serve as a check on their accuracy, were any censuses to continue. Eventually, however, “censuses" would be entirely register-based, as in Sweden, and the sooner the better.

By comparison with data held in each local authority, the PN would eliminate duplication and error from the electoral register, which is known to omit many individuals and to include many who are not entitled to be on it. At present, the electoral register relies too much on unsupported statements on the annual canvass forms by the household reference person and those who are in charge of institutions, and for various reasons is far from complete.

A PN would be essential for the introduction of a compulsory, universal Identity Card (the only kind worth the effort of setting up) if it were decided to introduce it, although the card (which could be duplicated or faked, unlike the entries on the register), would require biometric data. Although an identity card requires a register, at least of a simple kind, a PN system does not require an identity card although the two tend to go together.

The PN would minimise or eliminate fraudulent use or duplication of National Health numbers or National Insurance numbers (The PN might replace the simple NHS number). It would, for example, detect and reduce “health tourism" as long as its use was enforced by NHS management.

The Population Register would record and establish each person's legal marital status. When a person's marital status changes because of marriage/remarriage or divorce, the new marital status cannot easily be linked into any existing database because the person's date of birth is not recorded either on the “marriage entry" made at the time of marriage or the divorce form (date of birth in conjunction with full name is the most common identifier used to link demographic and administrative records). Their absence prevents marriage and divorce events being linked into the ONS Longitudinal Study.

Because a number of benefits (and pensions) are dependant on legal marital status, a database providing up-to-date information would be invaluable.

“Living arrangements" are now diverse ie married and living together, married and living apart, co-residential cohabitation, living alone (without a partner)—and living apart together. These include some groups of individuals, some of them highly mobile or living in more than one household, who are especially prone to change households. That creates particular difficulties in making timely and accurate estimates of local population estimates. Accurate knowledge would also help to minimise benefit fraud by persons cohabiting or living apart together and sharing bills but claiming single parent status, and facilitate the collection of “rates".

Furthermore, a Population Register containing the legal marital status of each member of the population would confirm the declared marital status when individuals marry. The Superintendent Registrar may require evidence of divorce but it seems in many instances takes the person's word. It would also help to minimise repeated “sham marriages" entered into to facilitate immigration.

Many individuals have legitimate residency status in more than one place. A Population Register that recorded secondary residences in addition to the principal residence would facilitate alternative definitions of “population"—increasingly needed to reflect reality.

In the medium and long term the financial benefits would be considerable. A census might no longer be needed and possibly eventually some other large databases. The Netherlands held its last census in 1971 and relies on the population registers and surveys. In Sweden, where the last census was held in 1990, all “censuses" starting with the 2005 census will be based entirely on population registers (Bruhn 2001).

PROBLEMS OF A POPULATION REGISTERClearly whatever the eventual savings, the initial costs of a Population Register and the linkage of person-

numbers with all administrative data would be substantial.

Recent public-sector IT problems—the NHS computerisation, the Registrars' computerisation, difficulties with on-line tax forms, have weakened confidence in the ability of UK Government agencies to complete major IT projects.

Even more, public confidence in security is currently low following the way that HM Revenue and Customs, DVLA and other bodies have treated confidential data.

But it should be noted that these problems did not arise from a lapse in security in the actual computer systems, but from undisciplined negligence in administrative procedures for transferring confidential data from one official body to another; involving physical, not electronic data. Progress will depend on the rebuilding of confidence in official competence to repair error and avoid it in future, through a more successful track record in IT related projects.

However secure systems on a huge scale can be made to work. We take for granted that our credit card and PIN number would deliver money from ATMs or purchase goods in almost any country in the world in a few seconds, with several billion such transactions every year worldwide. In commercial systems there is and incentive for individuals to monitor the status of their account and generally cooperate with the requirements of the system. Public systems need an analogous incentive.

Furthermore we already rely on several huge computerised national datasets (eg for National Insurance), which work well, do not give rise to privacy problems and some of which already give each of us a unique identifier. Taken together they are more than three-quarters of the way to a national register.

Privacy issues are very important and provoke much opposition (Rose 1999, Watner and McElroy 2003). Even without a population register or PN, proposed systems raise problems. For example, the data to be collected through the current e-borders initiative would, at least in theory, enable the government and the security services to build up and possibly retain indefinitely all information on international movement by UK travellers; when, where to and for how long. Helpful to the security services, possibly alarming for the rest of us.

We should remind ourselves, however, that systems similar to that proposed have been introduced successfully, and have worked well for decades, in a number of European countries that consider their arrrangements to be the highest manifestation of human rights and freedom, and which certainly show no signs of obvious oppression, or of breach of privacy scandals. However, foreign systems cannot be imported piecemeal—the UK may not yet be prepared for all the benevolent bureaucratic intrusions of Scandinavian state feminism. And the largest of the countries discussed here—the Netherlands—has 16 million people compared with the UK's 60 million.

CONCLUSIONIn the short run it is important to implement the improvements to the existing systems proposed by ONS,

some of which are noted the Appendix 1. Other improvements are also suggested. But the systems on which these are based are obsolete and reaching the limit of improvement. Some, such as the IPS, were designed for other purposes and have been press-ganged into functions for which they were always ill-suited; with their inadequacy sharply revealed under modern pressures. Others, like the Census and mid-year estimates, may no longer be “fit for purpose".

More generally, appropriate new systems for determining the number and condition of the people, especially in respect of local authority needs and knowledge of internal migration, have never been developed at all. With the continuous raft of legislation and policy initiatives being implemented, with considerable thought, time and expense going into their formulation, it is regrettable that they cannot be properly evaluated through lack of reliable basic population data. That is what the population register and the person number would provide. The investment—and political will to make that investment—has been lacking.

Thanks to their obsolescence, some of the fog of the 18th century population controversy is descending again on 21st century Britain to cloud our knowledge of our population and immigration. In the longer run it is essential to set aside demographic tinkering and ingenious but inadequate estimation and take the opportunity offered by modern IT to restore the opportunity lost in two centuries ago. Otherwise we will deserve once again the pity for our inadequate statistics recorded early in the 19th century by the distinguished Belgian statistician Adolphe Quetelet:

POSTSCRIPT“Lately at the [British Association] meetings at Cambridge . . . . . .. I heard from several distinguished

persons that there was a general complain to the imperfection of elementary population documents in this country . . . It is indeed a subject of wonder to every intelligent stranger, that in a country so intelligent as England, with so many illustrious persons occupied in statistical enquiries, and where the state of the population is the constant subject of public interest, that the very basis on which all good legislation must be grounded had never been prepared; foreigners can hardly believe that such a state of things could exist in a country so wealthy, wise and great. (Adolphe Quetelet, cited in the Report of the Fourth Meeting of the British Association, London 1835, xxxix, in Glass (1973) p127.)

REFERENCES

Agar, J. (no date) “Identity cards in Britain: past experience and policy implications." History and Policy. Policy Paper 33.pp. 13

An Act for Taking and Registering an Annual Account of the People and the Total Number of Marriages, Births and Deaths; and also the Total Number of Poor receiving alms from every Parish and Extraparochial Place in Great Britain. Geo III 1753. Hansard Parliamentary History of England XIV 1747-53) cols 1317 1365. In Glass 1973. pp 17-20.

Benton, P. and I. White (2003). “Looking Beyond the 2001 Census." Population Trends(113): 7-10.

Birch, D. (2007). “Making national identity cards work." Prospect (October 2007): 52-54.

Bowley, G. (2003). “The last census?" Prospect(92, November 2003).

Boyle, P. and D. Dorling (2004). “Editorial: the 2001 census: remarkable resource or bygone legacy of the `pencil and paper era'?" Area 36(2): 101- 110.

Bruhn, A. (2001). The 2005 population and Housing Census in Sweden will be totally register-based. Symposium on Global Review of 2000 Round of Population and Housing Censuses: Mid-decade Assessment and Future Prospects, New York, United Nations.

Chappell, R., L. Vickers, et al. (2000). “The use of patient registers to estimate migration." Population Trends(101): 19-24.

Diamond, I. (1993). Who and where are the missing million? British Society for Population Studies Annual Conference. Newcastle.

Diamond, I., O. Abbot, et al. (2003). “Key issues in the quality assurance of the one number census." Population Trends(113): 11-19.

Dumont, J.-C. and G. Lema–®tre (2005). Counting Immigrants and Expatriates in OECD countries: a new perspective. OECD Social, Employment and Migration Working Papers No. 25. Paris, OECD.

Duncan, C., R. Chappell, et al. (2002). “Rebasing the annual mid-year population estimates for England and Wales." Population Trends(109): 9-14.

Enterprise Privacy Group (2005). The Identity Project: an assessment of the UK Identity Card and its implications. London, London School of Economics Department of Information Systems.

Evans, H. Vickers, L. and E. Wright (2007) Using administrative data sources in the estimation of emigration. Population Trends 128, 33-40.

Fullerton, P. J. (2003). “A demographic statistics service for the 21st century." Population Trends(113): 32-39.

Glass, D. V. (1973). Numbering the People. Farnborough, Saxon House.

Glover, S., C. Gott, et al. (2001). Migration: an economic and social analysis. Home Office Research and Development Directorate Occasional Paper No 67. London, Home Office.

Home Office (2006). Control of Immigration Statistics United Kingdom 2005 Cm 6904. London, The Stationery Office.

Home Office (2007). Accession Monitoring Report May 2004—September 2007 A8 Countries. London, Home Office Border and Immigration Agency.

Hornsey, D. (1993). “The effect of computerisation of the NHS Central Register on internal migration statistics." Population Trends(74): 34-41.

Mallet, S. B. (1929). “Reform of Vital Statistics: outline of a system of national registration." Eugenics Review 21(87-94).

National Center for Health Statistics (1980). The Person-Number Systems of Sweden, Norway, Denmark, and Israel. Vital and Health Statistics. Data Evaluation and Methods Research, Series 2, Number 84.. Hyattsville, MD, National Center for Health Statistics.

ONS (no date) Implications of 2001 Census—why census shows fewer men. http://www.statistics.gov.uk/census2001/implications.asp

ONS (2003). Impacts and Implications of the Revision to 2001 population Estimates. London, Office for National Statistics.

ONS (2003). Discussion paper: Proposals for an Integrated Population Statistics System. London, Office for National Statistics. http://www.statistics.gov.uk/downloads/theme_population/ipss.pdf

ONS (2003). “Report: Internal migration estimates for local and former health authorities in England and Wales 2002." Population Trends(113): 69-73.

ONS (2004). Results of 2001 census-based local authority studies. Press release 8 July 2004. London, ONS. pp. 4

ONS 2006 A One-Number Census: Census 2001 was the most accurate count of the UK population ever achieved. http://www.statistics.gov.uk/census2001/onc.asp

ONS (2006). Report of the Inter-departmental Task Force into International Migration statistics. London, ONS.

ONS (2007). Improved Methods for Population Statistics Revisions in 2007. London, ONS. http://www.statistics.gov.uk/statbase/Product.asp?vlnk=14834

ONS (2007). International Migration. Migrants entering or leaving the United Kingdom and England and Wales. Series MN No.32. London, Office for National Statistics.

Pereira, R. (2002). “The Census Coverage Survey—the key element of a one-number census." Population Trends (108): 16-31.

Poulain, M., N. Perrin, et al. (2006). Towards Harmonised European Statistics on European Migration (THESIM). Louvain la Neuve, Universitaires de Louvain Presses.

Redfern, P. (1990). “A Population Register or Identity Cards for 1992?." Public Administration 68(4): 505-515.

Redfern, P. (2004). “An alternative view of the 2001 census and future census taking." Journal of the Royal Statistical Society A 167(2): 209-228.

Redfern, P. (2007). Memorandum submitted by Philip Redfern. Written Evidence submitted to the Treasury sub-Committee Report on Preparations for the 2011 Census. House of Commons Treasury sub-Committee: Preparations for the 2011 Census. HC 326-i. London, The Stationery Office: 4.

Registrars-General England and Wales, Northern Ireland and Scotland. (1939). National Register: United Kingdom and Isle of Man. Statistics of Population on 29 September 1939 by Sex, age and marital condition. London, HMSO.

Rose, N. S. (1999). Powers of Freedom: reframing political thought. Cambridge, Cambridge University Press.

Salt, J. (2007). International Migration and the United Kingdom. Report of the United Kingdom SOPMI Correspondent to the OECD 2006. London, Migration Research Unit, University College London.

Salt, J. (2007). International Migration Statistics: Surfeit or shortage? Royal Statistical Society Annual Conference. York, Migration Research Unit, University College London.

Salt, J., A. Singleton, et al. (1994). Europe's International Migrants. Data Sources, Patterns and Trends. London, HMSO.

Simpson, L. (2007). “Fixing the population: from census to population estimate." Environment and Planning A 39(5): 1045-1057.

Smith, L. and A. Sharfman (2007). “Assessing the feasibility of making short-term migration estimates." Population Trends(127): 21-29.

Watner, C. and W. McElroy, Eds. (2003). National Identification Systems: Essays in Opposition, McFarland.

Woodbridge, J. (2005). Sizing the Illegal Immigrant Population in the UK. Home Office Online Report 29 / 05. London, Home Office.

@ANAME@APPENDIX 1

SOME COMMENTS ON CURRENT UK SOURCES OF POPULATION AND MIGRATION DATA

THE CENSUSThe UK has not has a satisfactory census since 1981. Censuses totals are compared with the annual

population estimates for the same year derived from the updating of the previous census by the intervening total of birth and deaths and net migration. In theory the two totals should agree when the census total (usually April) is adjusted to fit the population estimate (usually mid-year). It is, however, a moot point to decide which of the two should be regarded as the benchmark, and neither will be exactly accurate. Both the censuses of 1991 and 2001 deviated considerably from the corresponding population estimate and, despite every effort, each has turned out to be in error, or at least remain controversial. In 1991 the estimates rolled forward from 1981 were preferred (erroneously) over the census. In 2001, the census was preferred (only partly erroneously) over the estimates rolled forwards from 1991.

The census of 1991 generated a population (49,890,000 in England and Wales) over one million fewer than that expected from the 1991 population estimates updated from the 1981 census (51,105,000); the so-called “missing million". For various reasons, particularly its finding of an implausible low sex ratio of 94.1 males per 100 females, the census, not the estimates, was deemed to be in error, and various upward adjustments were made (Diamond 1993).

In order to avoid the repetition of these problems and the difficulties caused, notably to local authorities, of a number of estimates of population for the same year, exceptional efforts were made to create a “one number" census for 2001. All the statistical expertise that money could buy was assembled to this end; no less than 50 statistical working papers were produced in preparation of the census methodology. That included an exceptionally large post –census enumeration survey, the Census Coverage Survey, of 320,000 households, using a sampling methodology separate from that of the census (Pereira 2002). That was intended to avoid the defects in the much smaller Census Validation Survey from 1991 (6,000 households) whose methodology followed that of the census and therefore tended to duplicate its errors, those missing in the census also being missed by the survey.

Accordingly, not without reason, the 2001 census was deemed to be the most accurate ever held (Diamond et al. 2003, ONS 2006). Despite this confidence, it remains the case that the kind of households that are likely to escape or evade the census are also likely to escape or evade any surveys for the same reasons. No serious attempt has been made, or perhaps could be made, to consider the illegally resident population, for which 430,000 is an officially accepted estimate (Woodbridge 2005). Some of them might be caught by the census, others not. ONS (2006) has continued to insist upon the infallibility of the census, a view supported by some outside experts (Boyle and Dorling 2004). Its circumstances, however certainly invite scepticism: with postal delivery, with 30,000 fewer field staff than in 1991, among other factors, the actual response rate was only 94% of the expected number of households in the frame. The missing households and individuals and their characteristics had to be imputed from the CVS (response rate 91%) and other information. Once again the eventual “one number" for 2001, like that of 1991, was about one million persons fewer (the original discrepancy was 1.2 million) than that expected from the population estimates rolled forward from 1991. And like the original results of that survey, the sex ratio was surprisingly low, 94.8 males per 100 females, with a particular deficit of young males compared with the population estimates. Because of the faith placed in the reliability of the census, this time the intercensal estimates were deemed to be in error. For good reasons the registrations of births and of deaths are believed to be as complete as can be. By a process of elimination suspicion therefore fell on only other arithmetic components of the estimates, the migration statistics and the base population of the 1991 adjusted census totals. Migration statistics are the most fragile of all the three components of demographic change.

In the UK they depend primarily on the voluntary International Passenger Survey (IPS) of international travellers. This generates annual estimates of gross immigrant inflow, gross emigrant outflow and the net migration that comprises the consequent annual increase or decrease of the UK population from migration. International migrants are considered to be those entering the UK intending to stay for at least 12 months, and those leaving the UK intending to be away for at least 12 months. That is the internationally accepted definition of an international migrant, as defined by the United Nations. The IPS gross and net totals must be adjusted for movement from the Irish Republic and for visitor switchers (those entering the UK allegedly for shorter periods than 12 months who change their minds) and migrant switchers (those leaving the UK for short periods who stay abroad longer than 12 months) and asylum seekers. The small samples of intended immigrants (2,965 interviews in 2005) and the even smaller sample of intended emigrants (781 in 2005) are then grossed up by the sample fraction to produce national-level estimates. Suspicion fell particularly on the emigration data, backed up by some evidence of over-staying by some

young male UK citizens in Australia and other favoured destinations, provoking much irreverent media hilarity about hundreds of thousands of missing British males disporting themselves on Bondi beach (Bowley 2003).

That conclusion, that emigration, mostly of UK males, had been under-estimated annually by about 350,000 over the decade, for which there was claimed to be some, albeit rather circumstantial supportive evidence from Australia and other destinations, the details of which remain unpublished. That in turn required the annual estimates of emigration and therefore net migration and mid-year populations for previous years to be adjusted downwards in proportion and the 1991 total to be adjusted downwards by 375,000, somewhat closer to its original total.

These adjustments of the national totals required local authority totals to be revised downwards as well, in some areas much more than in others. That inevitably provoked protests from local authorities which saw the basis for rate support grant being cut at a time when their own population estimates, based on rates, school enrolment, electoral rolls and other evidence contradicted the downward revision imposed by ONS. Those protests and other uncertainties provoked further enquiry that began to make the “one number" census into a two –number census at least in some areas. Calculation of the 2002 mid-year population estimates (MYE) for England and Wales revealed the need to revise upwards the 2001 mid-year estimates which had been based on the census, by 193,000. That only amounted to 0.36% of the total but the great majority were males aged 20-50; those whom the census had found to be missing. In fact the census had missed 187,000 males. Their restoration to the 2001 MYE restored the sex-ratios to a more normal level except for the 20-24 and 35-39 age-group (ONS 2003). Further revisions in response to local authority protests restored a further 0.1% of the total (ONS 2004). Some of the LA complaints were vindicated: the 2001 population of Westminster, previously reduced by 25%, was raised by 10% and of Manchester by 7%—three years after the event. The gap between the one-year census and the original mid-year estimates was gradually being nibbled away. A gap of 290,000 (later reduced to 208,000) remained between the 1991-based estimates to 2001 and the 2001 census, however, which defied clear explanation.

This deficit was assigned to a demographic innovation entitled “unattributed demographic change" which for a while was a regular element of population estimates, of 27,000 per year. It provoked some unkind satire in some quarters as a “new miracle ONS ingredient" for shrinking migration estimates, banishing awkward inconsistencies and keeping the 2001 census (nearly) infallible. This product was withdrawn from the market in 2004. These upward revisions in population estimates naturally required a re-revision of the migration estimates, partly restoring the former cuts. Population estimates for 1992-2001 took some time to stabilise.

Finally the 2001 census was criticised, from a more fundamental viewpoint, by Philip Redfern, former Deputy Director, OPCS (now ONS) (Redfern 2004). Demographic analysis was deployed, using sex-ratio evidence, to cast serious doubt on the correctness of the upward adjustments made to the 1991 census and showing that acceptance of the 2001 census results imply a highly unusual sex ratio among British emigrants abroad which is not supported by censuses in the receiving countries, Restoring “normality" would require the addition of 500,000 males to the British population in 2001 (Redfern 2007). That would certainly lay to rest the “unattributed demographic change" discussed above.

BIRTH AND DEATH DATAThese are generally held to be the most complete. Improvements could be made not so much in the

recording of the total number of vital events but in the information that is collected. Major defects here are the lack of information directly relating to true (biological) birth order. That information, important for estimating completed family size and its distribution and the extent to which current period measures of fertility my be giving a misleadingly low picture of future birth rates, has to be estimated indirectly from surveys. With nearly half of all children born outside marriage, the 1938 Population (Statistics) Act needs to be revised to require parents to record all previous live-born children, not just those in the current marriage. There may also be a case for asking for nationality to be collected.

For some time there has been pressure within ONS for information on ethnic origin of mother and father and / or child to be collected on birth registration. With the advent of questions on religion in the census and the LFS and other surveys, pressure to include that variable may also arise. That would undoubtedly be convenient for those (such as this author) who are interested in the demography of ethnic minorities and the projection of their populations. As it stands, it is argued elsewhere, on general grounds, that this should be resisted. The fundamental reason advanced here is that (although accepted in the census and LFS by the majority of respondents) it is an intrusion on privacy, requiring self-definition which some may be unwilling to make. In practical terms this difficulty will become more pronounced in future as an increasing number of parents have multiple ancestry and therefore an increasing number of babies have really complicated ancestry, a problem already emerging in the US census and which our won will follow as open questions on ethnicity and multiple answers become general. Furthermore the

labelling of very young children with an ethnic or religious label at birth would categorise them in ways that they may wish later in life to change or reject. It would open the door to the further subdivision of the British population on ethnic tramlines, from birth onwards.

Marriage and Divorce records need to be altered to include the dates of births of the (ex) partners. That would permit them to be linked with the ONS Longitudinal Study.

MIGRATION STATISTICSGenerally these are the most troublesome aspect of demographic data for various rather intractable reasons.

Migration data are satisfactory in very few countries, Australia being one exception. The basic problems are well known, the very large scale of flows in each direction, of which only a small proportion are intended migrants, the uncertainty of duration of stay or absence, the relative ease in many countries of evasion of control and over-staying.UK sources measure both flow of migrants in and out of the country and the “stock" of immigrants living in the country. None is satisfactory.

Flow data are given by the International Passenger Survey and the Home Office Control of Immigration Statistics.

The manifold problems arising from the voluntary nature, incomplete coverage of ports and small sample size for the International Passenger Survey have been noted above. It does have the advantage, unusual in the developed world, of capturing a small sample of intended emigrants as well as immigrants, thereby permitting annual gross and net population change to be recorded according to the international definition of international migration. Furthermore the IPS apples to all intended migrants irrespective of citizenship, while the Home Office Control of Immigration Statistics data only refer, in any detail, to inflow only of non-EEA citizens only. The latter purport to be a complete description of all legal entries to the United Kingdom, with particular emphasis on acceptances of non-EU citizens according to a time limit (eg under work permit, or provisionally as spouses), or acceptances for settlement: spouses with entry clearance certificates or persons who entered earlier subject to a time limit now lifted.

Difficulties with the International Passenger Survey include the following:

@UL@Response is voluntary, although the response rate is quite high (83% in 2005).@/UL@

@UL@Only 90% of estimated flows are covered (although that is being improved)@/UL@

@UL@Data rest on stated intentions, not actual out-turn. @/UL@

Basic IPS data need to be supplemented with information from separate sources on movement from the Irish Republic, data from the Home Office on visitor switchers (persons given extended leave to remain which puts them into the “international migrant" category, migrant switchers, relating to those who stay abroad for longer than intended, and asylum seekers and dependants adjusted for those removed, or who leave, before 12 months. All that together generates “Total International Migration" (TIM) data.

Those components are published, but not the details about how they are put together. ONS should do that, in the interests of transparent statistics and independent scrutiny.

The small sample size precludes detailed analysis as to country of origin / destination or citizenship etc, purpose of journey, qualifications, or information relevant to legal status of entry of the kind obtained from Home Office statistics. Overall standard error is 3.7% on gross inflows, 4.8% on gross outflows and much larger when subdivided. A8 migration estimates, for example, were based on 78 interviews in 2005. In an attempt to keep standard error within acceptable limits, omnibus categories are lumped together such that they that they preclude effective analysis, other categories are to say the least uninformative. In 2005, of net migration of 185,000. 24,000 (13%) was recoded as “other" and 54,000 (29%) as “no reason stated", and a high proportion of countries of origin were tabulated as “other foreign" (for all details see ONS 2007 and Table 1 below).

“Other foreign" category in the 2005 ONS international migration statistics.

“Other foreign" immigrants (thousands) sourceGross inflow Net inflow table no.

Citizenship 140 62 2.1Country of last or next 162 58 2.2

residenceEU and other foreign country of

89 52 2.4

last or next residenceCountry of birth 147 84 2.5

Note: All tables except table 2.1 include all citizenships including UK.Source: ONS (2007) International Migration Series MN No. 32. London, ONS.

No distinction is made between fiancé(e)s or others moving in order to marry and those coming to join established spouses.

Numbers of those entering and leaving for purposes of work have been surprisingly similar in many years until recently, suggesting a zero net gain for the workforce. That point is sharpened by the OECD finding that the UK is the only major OECD country to lose almost as many tertiary-educated person to emigration as it has gained from immigration (Dumont and. Lemaitre 2005). However without information on whether such emigrants are leaving from an employment in the UK to work abroad or not, it is impossible to tell what proportion are young foreigners who entered as students and are leaving to take up their first job in their country of origin.

In the last few years, the number of countries for which data are returned has been diminished, and tabulated data (in thousands) are no longer presented to one decimal point but as integers.

ONS staff can provide can provide on request special tabulations for some specified countries, necessarily amalgamated over several years, a service which is greatly appreciated. But in the nature of things along a few countries can be so analysed before sample size precludes any statistically respectable comment.

Short-term migration (more than three months but less than 12) is not captured by current IPS procedures, although that is in hand. Short-term migration may now be an important component of additions to the population following the freedom of movement for citizens of A8 countries from May 2004. If half a million entered staying for six months and on departure were replaced by another half million who stayed for six months, the population would be increased by a constant half-million (though different people) although net migration according to INS would be zero. The circumstances of the A8 migration also make switching from short to long-term stay likely, as opinion polls indicate. The Accession Monitoring Report for September 2007 (Home Office 2007) states that a cumulative total of 715,000 approved applications have been made since May 2004, not including the self-employed. ONS estimates of net migration from the A8 countries is 60,000 per year for the last three years. That figure is implausibly low.

All these problems are worse with respect to emigrant flows because of the smaller sample size, making it more difficult to analyse the recent upsurge in emigration.

Except for some categories of visitor switchers, and those granted entry for family reasons, there is little useful connection or correlation with the data published by the Home Office Control of Immigration Statistics.

Although IPS data are available for analysis, that dataset excludes the raw IPS or TIM migration data which can only be obtained conditionally by special request. No Excel files or tables are linked to the web versions of the annual MN International Migration volume (unlike the FM1 series on births).

Home Office Control of Immigration Statistics

These notionally complete data on lawful entries to the UK apply to inward migration only and, in detail, to non-EEA citizens only except that, for the time being, data are also given in respect of citizens of A8 Accession states.

While notionally “flow" data, almost all persons given leave to enter on a long-term basis have already been resident in the country for up to four years. Only a small, although growing proportion (6090 of 179,120 granted settlement in 2005) are accepted for settlement on arrival. It is not possible, at least with the data published, to reconstruct the numbers actually entering the country for particular purposes or other categories in any specified calendar year.

The categories by which data are tabulated refer to the category of the Immigration Rules under which the person is admitted to the UK and citizenship. In respect of the latter they are fully detailed, unlike the IPS data.. They are not primarily demographic. As there are no Rules governing exit from the UK, no data are collected from departures although the overall count that was dropped in 1993 has been re-instated.

Although the categories by which passengers are given leave to enter by purpose of journey are quite detailed, they do not enable some important elements of migration flows to be distinguished. For example in the total of persons granted settlement as a spouse in 2005 (45,970, 26% of all granted settlement in that year) those entering a year or two earlier as fiancé(e)s, mostly through arranged marriage, cannot be distinguished from established spouses joining their husbands or wives. It is therefore impossible to distinguish between family re-union migration and family formation migration. Until 1996 data were published (table 2.16, 2.17), which made it possible to determine the number of years between marriage, or arrival or settlement of spouse, or arrival the year of marriage, or settlement of spouse, and the year of acceptance for settlement for marriage from the Indian-sub-continent that those entering had been married, which permitted such an estimate. That has been discontinued, with no comparable data to replace it. Up to that year, the very large and growing preponderance were of very recently married couples: family formation migration was replacing family re-union migration in the family migration category, as it has in other countries such as the Netherlands, where data are superior in this regard.

It is difficult to understand how 4565 persons in 2005 (2.5%) could be granted settlement as “category unknown" (Table 5.1) or how there could be 290,000 given leave to enter the UK in 2005 under the category “Others given leave to enter"—a much greater number than some of the categories for which detailed are given. Furthermore, when analysed further the 196,000 under this heading in 2004 are revealed numerically to be dominated by 105,000 “Other passengers given limited leave to enter" and 74,700 who were given leave to enter for “unknown" reasons (Home Office Statistical Bulletin 2004 14/05 table 1.2 and related Excel file). How can people be admitted under the Immigration Rules for “unknown" reasons?

All non-EEA international travellers are required to fill in a Landing Card on arrival. Some material which would be very useful in understanding migration flows and helping comparability with IPS data, such as country of birth, is not even extracted from it, never mind tabulated.

Independent analysis of this material though a suitably anonymised sample of records is not possible, although some Excel files can be obtained on request.

Some other annual flow data are generated by various administrative processes; the issue of work permits, of national insurance numbers to overseas citizens, and more recently the data from the Worker's Registration Scheme mentioned above. Labour migration estimates from the national insurance numbers do not correspond to the number of work permits or IPS data on the number entering for work, but there are good reasons why this should be so (see Salt 2007).

STOCK DATA ON MIGRATIONFor obvious reasons no stock data are available for emigration, although some countries (eg Republic of

Ireland) ask questions in sample surveys on intentions to emigrate. Censuses and immigration data of other countries are used for this. Among various administrative sources of data (driving licences, NHS data) only the information on pensions seems a likely contributor to the picture of emigration from the UK (Evans et al 2007). On immigration, a variety of sources give information on the number of persons born in specified countries in considerable detail; the decennial census, a number of surveys including the Labour Force and General Household Surveys and the forthcoming Integrated Household Survey. Some, including the census and the LFS, also ask questions on the residence 12 months earlier. Responses to that count as international migrants, but this is a measure of gross migration inflows, minus those who returned less than 12 months before the census or who had died. Unfortunately, in some ethnic minority populations, the number of persons not answering that particular question in the LFS is about the same as the number who recorded an address overseas twelve months ago given. That makes the data difficult to use as an indication of migration flows.

INDIVIDUAL DATA ON FOREIGN RESIDENTSAlthough the Home Office gives leave to enter the UK it has no general or central way of knowing where

foreign citizens are living once they have entered the United Kingdom. Foreign citizens are obliged to register with the police locally (are they still?) but the Home office maintains no up to date address list, in fact no address list at all except for asylum claimants. As the Home Office appears to be unclear how many asylum seekers there are in the country by some hundreds of thousand or so, this system does not seem to be very efficient.

THE NEW EMPHASIS ON DATA RELATING TO IMMIGRANTSSince the new immigration policy has increased immigration to unprecedented levels and substantially

increased the size of the immigrant and foreign-citizen population. Accordingly there has been heightened increase in the characteristics of immigrant and foreign, as opposed to ethnic minority, populations in various areas of the national life: incomes, housing, workforce participation and employment, health, education and crime. The new policy has been justified primarily on the grounds of the beneficial effects of immigrant, as opposed to broader ethnic minority population, on the workforce. Some of the bases for these justifications, for example the calculations presented by Gott et al. (2002, p. 29) compare favourably the tax paid by immigrant (not the whole ethnic minority) populations with the welfare benefits and rebates that they receive. Those calculations employ relatively orthodox statistics reasonably available from public sources, in conjunction with demographic data on the size and age-structure of the populations concerned. Other costs and potential benefits, outside the tax and welfare accounts for which data are not so readily available, are not included or are explicitly excluded. For example the earlier paper explicitly assumed that costs per head of primary and secondary education, and share of expenditure on health services provision were equal, within each age-group, across all categories in the UK population as a whole. This is understandable given the difficulties of doing otherwise. The authors go on to acknowledge, however, that rates of utilisation may be different for public education and health and that providers might incur different costs for migrants.

In fact they are known not to be equal, although the statistics are more patchy or hard to get at. Were those to be routinely available, a more balanced analysis would be possible of costs and benefits in respect of this highly significant policy. Statistics on immigrant and even more foreign citizen populations in all the areas above are either not collected or are sparse. Instead, UK statistics in these areas have developed along a different route; with the prime focus on ethnic affiliation, irrespective of birthplace and nationality. Ethnic categorisation seems to have taken priority partly for ideological reasons, from a distaste of the category “immigrant"; to promote the notion of equal entitlement or special measures irrespective of immigration status or nationality. In respect of the contentious issue of NHS utilisation, no statistics of any kind appear to be available whether referring to ethnic origin, birthplace or nationality. Doctors argue that it is not their job to collect such data, and some might object to doing so as being unprofessional. But that duty could quite properly devolve on NHS management, where there is hardly a shortage of personnel.

OFFICIAL EFFORTS TO IMPROVE POPULATION AND MIGRATION STATISTICSSince the early 2000s, probably spurred by the problems encountered by even the best –laid plans for the

2001 census, the ONS has begun a programme to improve statistics on population and migration. No doubt the details will be presented in a submission from the ONS submission and need not be repeated here. These included ambitious proposals for an Integrated Population Statistics System (ONS 2003). That was intended to link census, survey and administrative data at individual level to produce, after the 2011 census, a population statistics database covering all England and Wales to be continuously updated from a variety of sources and to be related, in manner unspecified, to any population register of which it would appear to be a precursor. That would then be the source of all future population statistics. At the same time there were quality reviews of birth and migration statistics and, more recently, the creation of demographic oversight bodies that introduced (2007) a revisions policy for demographic statistics. Further steps were taken in response to glaring inadequacies revealed about data on migration and migrants and their effects upon population, some of them noted above. These included setting up the National Statistics Centre for Demography and a superior review committee.

Following an inter-departmental task force reported on migration (ONS 2006), renewed interest was shown in “administrative" resources of demographic data such as pensions information, NHS information and driving licence data from DVLA. These cover a much higher proportion of the population than does any survey). A rolling programme of improvements to population statistics has been set out by the Improving Migration and Population Statistics (IMPS) initiative October 2006 and its 2007 update.

ONS are thus doing their professional best to improve the tools which they have at their disposal.

Migration statistics are to be improved by:

@UL@1. Increase of the IPS Emigration sample from 800 or so.@/UL@

@UL@2. IPS survey cover to be increased in some ports of entry and extended to others for the first time.@/UL@

@UL@3. Steps to measure short-term migration. That has become much more demographically important following the free access to the UK labour market of the A8 countries since May 2004. In recognition of the potentially powerful effects of continuous short-term migration of a duration that does not show up on the IPS radar (ie less than 12 months), new approaches are being developed to measure it (Smith and Scharfman 2007).@/UL@

@UL@4. using “administrative" sources to estimate migration and its effects: Work permit data, Workers Registration Scheme, NHS records and National Insurance numbers (NiNo)@/UL@

@UL@5. Questions in LFS on previous residence 12 months ago. At present this question is only asked in the Q1 version of this quarterly survey. Henceforth it will be asked in all quarters' surveys, quadrupling the sample size.@/UL@

@UL@6. The Home Office is developing its “e-borders" initiative to record electronically the passport details of everyone entering and leaving the UK. But it is not due to become operational for another six years.@/UL@

@UL@7. Various improvement are proposed for the 2011 census, including restoring the question on nationality, dropped since 1961.@/UL@

The most important innovation is the proposal for e-borders- electronic processing of all arrivals and departures from the UK. But e-borders will take a long time—on stream by 2013. The specification yet vague- it is still out to tender, little has been made public. It is not clear, for example, whether individual records of entries will be linked with the same individuals exit records.

While UK citizens will have biometric passports to be scanned; what to do with others may depend on passport type. The entry cards currently used will be dropped. Those coming from the EEA may not be asked purpose of journey. The collection of data on departures seems to be down to the carrier.

Work permit and the statistics that they generate will be on the introduction of the Points System. It is not clear where information on labour migration will replace it.

ADDITIONAL ISSUES AND PROPOSALS SHORT-TERM REVISIONS TO MIGRATION ESTIMATESSome problems are likely to remain insoluble. Because there are three Registrars-General (for England and

Wales, Scotland, and Northern Ireland) population data abut the UK is always available later, and less abundant, than that referring to England and Wales or the other areas. The level of analysis and detail of publication for England and Wales is superior to that in the other areas. Given the current fashion for devolution, this failure to provide uniform and timely data for the whole country may get worse.

A question on emigration intentions, and on children ever-born, should be asked in each quarterly issue of the LFS (each has a sample size of 60,000) and of the Integrated Household Survey (formerly Continuous Population Survey) which may amalgamate the LFS, GHS and other surveys from 2008.

The migration element of the IPS has always suffered from being a minor component of a Board of Trade survey primarily concerned with balance of payments, tourism and the like. In view of the importance of international migration, the migration element in the IPS should be prioritized to ensure more rapid publication, or separated completely from original Board of Trade enquiry. The sample size relating to migration, in and out, should be greatly increased. Unfortunately a really large (expensive) increase would be needed to reduce the standard error substantially and permit more detailed tabulation.

As far as possible Home office Control of Immigration Statistics and ONS data should be harmonised, with as many categories in common as possible. That would require more detailed tabulation in each, and the common used of grouped categories (the useful combination “New Commonwealth" countries, widely employed in the IPS, was dropped in the CIS data after 1996.

All population statistics—including Home Office data—should be brought under the supervision of the Statistics Commission.

A clear distinction should be made in both data sets between family re-union and family formation migration (relevant data dropped 1997)

IPS and HO data should be available in machine readable form as anonymised records for individual tabulation

The IPS immigration questions could be made compulsory on non—UK citizen immigrants.

Attempts should be made to expand of “Other or unknown" categories in all migration data, and publication / tabulation of IPS data is needed on a more refined basis especially in terms of country of origin / birth / and purpose of journey. Larger sample size would be essential, and closer linkage with Home Office data on inflows.

@ANAME@APPENDIX 2

LIST OF VARIABLES FOR WHICH DATA IS MAINTAINED IN THE BELGIAN MUNICIPAL REGISTERS

A unique person number is given to all individuals at birth and on immigration

For Belgian Citizens the following data are available:

@UL@Name and Surname@/UL@

@UL@Sex@/UL@

@UL@Date and place of birth@/UL@

@UL@Address of residence@/UL@

@UL@Person number@/UL@

@UL@Occupation@/UL@

@UL@Composition of household@/UL@

@UL@Place and date of death@/UL@

For non-citizens the following further details are required

@UL@Immigration Service number@/UL@

@UL@Citizenship@/UL@

@UL@Refugee status@/UL@

@UL@Country and place of origin@/UL@

@UL@Limits on duration of stay@/UL@

@UL@Type number and validity period of work permit@/UL@

@UL@Date of recognition as refugee@/UL@

@UL@Right of return to country of origin@/UL@

@UL@Visa information@/UL@

Type of inscription on voting list (ie whether entitled to vote in EU or local elections)

Source: Poulain, M., N. Perrin, et al. (2006). Towards Harmonised European Statistics on European Migration (THESIM). Louvain la Neuve, Universitaires de Louvain Presses.

December 2007

@HR25@

Memorandum from London Borough of Hackney

INTRODUCTIONHackney recently commissioned a detailed population analysis based on administrative data from the

Council's own records and those of the Primary Care Trust.52 This is the same methodology used by several other boroughs and it has provided indicative evidence that the ONS methodology seriously and significantly undercounts the number of people living in the borough. The most compelling evidence to support this is comparison of the ONS mid-year estimate of 207,800 Hackney residents with the 256,000 GP registrations recorded in the borough. While it is accepted that there are, inevitably, variations between the numbers of people living in an area and those registered with a GP, the sheer scale of the difference suggests a significant shortfall in the ONS population count.

Other external population estimates also indicate an undercount in Hackney; the GLA for example, calculate a low estimate of 215,000 people for the borough and a high estimate of 218,144 implying an ONS undercount of between 7,200 and 10,344. Other local data also suggest discrepancy with the ONS population figures.

The 2001 Census returns in Hackney were 72%, the second lowest rate in the country after the Royal Borough of Kensington and Chelsea. This in itself should have been cause for concern over the accuracy of extrapolated population estimates.

Reliable population statistics are important to all local authorities because they are the basis by which central and local government assess levels of funding and resources needed to deliver local services, and also simply to plan the delivery of services. In deprived areas, where people's lives can literally depend on the services provided by councils and their partners, accurate data is crucial. Hackney is an area of severe health inequalities.

By any measure, Hackney is one of the most economically deprived areas in the country. In 2004, Hackney's average area ranks based on Super Output Areas (SOAs) made it the most deprived local authority in England; ahead of Tower Hamlets (2nd), Manchester (3rd), Islington (4th) and Liverpool (5th). 89% of Hackney's areas—the highest proportion out of all 354 local authority areas in England—are amongst the 20% most deprived SOA's while three areas in Hackney are among the 3% most deprived areas in England.

HOW HACKNEY IS ADDRESSING THIS POTENTIAL UNDERCOUNTFor these reasons, the London Borough of Hackney has commissioned a population analysis and, for the sake

of ease of comparison to other authorities with largely similar socio-demographics, this analysis will use the same methodology as that employed by Brent, Enfield, Southwark and Slough Councils. The London Borough of Hackney will be pleased to share the results of this analysis with the committee once it is completed (initial analysis mid-December, full analysis by February 2008).

FUNDING AND PLANNING IMPLICATIONSWhilst population count is only one of a number of factors in calculating the formula grants allocated to

local councils and Primary Care Trusts (PCTs) the undercount does have an impact on funding. An example of the significance of population undercount to funding can be shown by comparing the GLA low estimate of 215,000 with the ONS mid-year estimate of 207,800 giving an undercount of 7,200. Assuming that Hackney Council receives £950 per person funding from central government this translates to a loss of £6.84 million per year. Perhaps more significant is funding to City and Hackney Primary Care Trust. The City and Hackney PCT receives funding of nearly £2,000 per person, meaning there is a shortfall of £14.4 million.53 In total, the underfunding to the Hackney area may be as much as £21.24 million per annum.

If this undercount is compounded over the 10 years between Census evaluations, the funding shortfall accumulates to £168 million in council and PCT funding available to the area. Hackney has 59% worklessness, and the single largest barrier to employment is mental health and behavioural issues. The “missing" funding would certainly have been able to finance a substantial programme to stabilising the lives of many of our most vulnerable citizens as well as many more interventions that would have enhanced the life chances of our citizens.

FUTURE COUNTS AND REMEDIAL ACTIONWe are pleased that the ONS is carrying research and pilot studies to improve the methodology upon which

the 2011 census will be carried out. Considering the impact of inaccurate population counts on the funding of local

52 Hackney has engaged Mayhew Associates.53 This is based on per capita grant figures provided by the CHPCT.

authorities and PCTs and the delivery of efficient and appropriate services, we believe that the ONS (or its successor organisation) should put in place a mechanism for reviewing anomalous counts in partnership with key stakeholders, including the local authority. These reviews should take place early in the 10-year census cycle, perhaps as early as within two years of the census taking place. Local authorities should be allowed to call for a review. It is in the review process that we believe it is most appropriate for the wide variety of administrative data sets may be drawn upon in order to cross-reference with the census data.

For the London Borough of Hackney, it is vital that the 2011 census count is accurate. The 1991 census count was poor, and the 2001 appears to have been nearly as poor. Hackney is eager to work with the ONS and other bodies to assist in making the count more robust.

RECOMMENDATIONConsideration should be given to the creation of a dispute resolution mechanism that allows local authorities

to challenge the ONS findings and result in resolution within two years of the census, eg by 2013. This mechanism should involve the local authority as a key stakeholder and could trigger a process by which the local administrative data is used to enhance the accuracy of the count.

December 2007

@HR25@

Memorandum from the Office for National Statistics

SUMMARYThis submission provides evidence on the statistical issues relevant to the Treasury Sub-Committee's inquiry

into counting the population. The issues covered are as follows:

Uses of population statistics (Section 1)

@UL@— With increasing population mobility, it is essential to have relevant, accurate and timely population statistics. They provide the evidence base for managing the economy, developing policies and allocating resources for service delivery.@/UL@

@UL@— These statistics have assumed increased importance for macro-economic policy. Population estimates are crucial for judging the size of the potential labour supply and therefore the extent to which the economy can grow in a non-inflationary way.@/UL@

Current methods and sources (Section 2)

@UL@— National Statistics on population are focused on usual residents in each geographic area. They are produced by combining the best possible information currently available, to high quality standards. They are based on internationally recognised and transparent methods that are peer reviewed by external experts.@/UL@

@UL@— By combining sources they have real strengths compared with those statistics that, for example, are based on National Insurance registers or the Worker Registration Scheme that only record migrants coming in to the UK but not those going out.@/UL@

@UL@— Most countries produce annual population estimates and regular projections to similar levels of quality and coverage. Population statistics in England and Wales compare well with those in other countries.@/UL@

@UL@— Census estimates are essential as these provide a benchmark and anchor for population estimates and projections in the years between censuses.@/UL@

Issues in producing population statistics (Section 3)

@UL@— In a period of significant population change and individual mobility, users are seeking:@/UL@

@UL@— Estimates of a broader range of population statistics and indicators (eg short term migrants, levels of population turnover);@/UL@

@UL@— More accurate counts of the numbers of migrants entering and leaving the UK; and @/UL@

@UL@— More accurate counts of local populations.@/UL@

@UL@— Population estimates and projections rely on information about births, deaths and migration. Registration of births and deaths occurring in the UK is recognised to be of high quality. The largest area of uncertainty surrounds migration, for which there is no registration system. For this reason, there is no comprehensive count of the numbers nationally or locally.@/UL@

@UL@— Estimates of international migration rely, to a substantial extent, on the International Passenger Survey. National figures have relatively small levels of uncertainty. At local levels, individual estimates are subject to greater levels of uncertainty.@/UL@

@UL@— Levels of internal migration are substantial. At the level of geographic detail at which population estimates are required the use of administrative registers is essential. The timeliness and coverage of existing registers is variable. Although NHS patient registers provide the best fit to usually resident populations, they have recognised limitations (eg list inflation and delays in identifying moves among young, fit population groups).@/UL@

Alternative sources (Section 4)

@UL@— There are a number of alternative sources of information that have the potential to improve estimates of numbers of both internal and international migrants. These include administrative sources such as National Insurance records, Patient Register data, the School Census, information on students in higher education, e-Borders and the Worker Registration Scheme.@/UL@

@UL@— To fulfil this potential will require: @/UL@

@UL@— agreement with the Government Department responsible; @/UL@

@UL@— investment to ensure that migrant information can be identified accurately; and@/UL@

@UL@— Parliamentary approval, where it does not already exist, for data sharing under the Statistics and Registration Services Act 2007.@/UL@

@UL@— ONS is also investigating the improvements needed to surveys and looking at alternatives where necessary (eg port surveys, communal establishment surveys)@/UL@

Planned work for 2008-12 (Section 5)

@UL@— Planning for 2011 Census is in train, focusing on the need for accurate population figures.@/UL@

@UL@— In May 2006, the Inter-Departmental Task Force on Migration Statistics was set up by the National Statistician to recommend timely improvements that could be made to estimates of migration and migrant populations in the United Kingdom, both nationally and at local level. The Task Force report was published in December 2006. Some of the recommended improvements have already been made. Delivery of further, more substantive improvements is scheduled in each of the years 2008-12. However, the problems identified cannot be solved by ONS alone:@/UL@

@UL@— Support is being sought across Government for improving access to and quality of administrative data and for funding of new survey collections and methods.@/UL@

Population Statistics in 2013 and beyond (Section 6)

@UL@— ONS has set out a vision for an integrated population statistics system that would bring together survey and administrative sources. The need for future Censuses will be assessed within this framework. ONS will next year be initiating a project to review the needs for a 2016 census and alternatives to a census for 2021.@/UL@1. USES OF POPULATION STATISTICS1.1 Main uses and users

Understanding the size and characteristics of the population and how it is changing is important to our understanding of society and the economy. It is essential to have relevant, accurate and timely migration and population statistics to provide the evidence base for managing the economy, developing policies and allocating resources for service delivery.

Population estimates and projections are used for planning, resource allocation, business decisions and a broad range of public policy purposes. They provide essential contextual information for research and analysis and in calculating rates for key demographic measures (such as fertility rates and dependency ratios) and economic indicators such as employment rates. With increasing mobility population and migration statistics have assumed increased importance for macroeconomic policy. They are crucial for judging the size of the potential labour supply and therefore the extent to which the economy can grow in a non-inflationary way.

A key use of the population estimates and projections is in funding formulae. The Department for Communities and Local Government (for England), the Scottish Government and Welsh Assembly Government (for Wales) use these data in local government settlements. Population projections are similarly used in the resource allocation formula for the NHS in England by the Department of Health. Health allocations in Scotland make use of population estimates and those in Wales make use of the Census.

Population projections play a key role in the Government's Long-term Public Finance Report, published annually by HM Treasury, and in the development of pension policies.

1.2 How well user needs are met

Current population statistics relate to the usually resident population—that is to say they count people where they usually live. For this purpose, internal and international migrants are counted if they move address for a year or more. Population and migration statistics are produced on this basis by combining the best possible information currently available. They are produced to high quality standard, using internationally recognised and transparent methods that have been peer reviewed by external experts. By combining sources they have real strengths compared with those statistics that cover only people coming in to the UK, but not those going out.

In a period of significant population change and individual mobility, meeting the requirements of users has become more challenging. Having a single definitional base does not meet the needs of all users (eg the Bank of England and increasing numbers of local authorities). The amount of population turnover, both nationally and locally, also makes it more difficult to be certain about the numbers of people in an area and on what basis they are there.

Figure 1

COMPONENTS OF CHANGE IN UK POPULATION ESTIMATES, 1996-2006

/home/PKB/DATA/386539/207.EPS>

Information about births and deaths is based on registration and considered to be reliable. However migration is the now the largest driver of population change (Figure 1), due to increases in both immigration and emigration (Figure 2). It is the also the component that is subject to the greatest degree of uncertainty, as there is no registration system and, for this reason, no comprehensive count of the numbers.

Figure 2

MIGRATION TO AND FROM THE UK, 1996-2006

/home/PKB/DATA/386539/208.EPS>

A substantial part of the headline estimates of international migration rely on the International Passenger Survey. National figures have relatively small levels of uncertainty. At local levels, it is necessary to combine years and distribute figures using other sources. At this level, individual estimates are subject to greater levels of uncertainty.

Levels of internal migration are substantial (Table 1). At the level of geographic detail at which population estimates are required the use of administrative registers is essential. The timeliness and coverage of existing registers is variable. Although NHS patient registers provide the best fit to usually resident populations, they have recognised limitations (see Section 2.2.4).

Table 1

AREAS WITH THE HIGHEST VOLUME OF INTERNAL AND INTERNATIONAL MIGRATION PER 1,000 POPULATION

Mid-2001 to mid-2006Internal Migration International

Migration Volume of migration per 1,000 population1

Area In Out In Out All migration

International migration

1 Cambridge

54,100 57,900 28,500 18,200 278 82

2 City of London2

3,900 3,600 1,500 1,500 276 78

3 Westminster

88,700 95,200 65,700 37,700 259 93

4 Oxford 66,000 70,800 31,900 17,000 258 685 Wandswo 123,400 142,300 40,800 27,300 243 50

rth6 Camden 86,200 96,900 53,400 25,800 243 737 Hammers

mith and Fulham

66,900 83,200 33,200 22,900 242 66

8 Isles of Scilly2

1,200 1,300 - - 238 6

9 Islington 77,300 90,200 29,000 17,200 234 5110 Lambeth 109,300 142,000 32,800 16,400 222 36Notes

1. Volume of migration per 1,000 population is calculated as (in migration + out migration)/population*1000 (separately for all migration and international migration).2. Rates for the City of London and Isles of Scilly are based on less than 10,000 population.- Nil or less than half the final digit shown

Both ONS and users recognise that further improvements to population and migration statistics are desirable, given the current high levels of population change.

There are three specific areas of improvement that users are seeking:

@UL@— estimates of a broader range of population statistics and indicators (eg short term migrants, levels of population turnover);@/UL@

@UL@— more accurate counts of the numbers of migrants entering and leaving the UK; and@/UL@

@UL@— more accurate counts of local populations.@/UL@

In 2003, ONS initiated a substantial programme of work to improve migration and population statistics. It is focussed on making improvements to the methods and data sources used to estimate the population at national and local levels during the inter-censal period. These improvements are needed to minimise the risk of divergence between the rolled forward mid-year population estimates and the 2011 Census-based population estimates, and to better understand the differences that remain.

www.statistics.gov.uk/about/data/methodology/specific/population/future/imps/updates/default.asp

The key strands of this work are:

@UL@— migration research (both international and internal within the UK);@/UL@

@UL@— collaborative work with local authorities to investigate the potential for making greater use of administrative data sources to improve local estimates; and@/UL@

@UL@— population definitions.@/UL@

1.3 Effects of inaccuracies or uncertainties

Population estimates and projections contribute to decisions on the distribution of significant resources. For example, around £100 billion of central government funds each year are allocated to local areas in England through the capitation element of NHS resource allocation and the CLG Formula Grant Settlement. This is equivalent to approximately £2,000 per head. Inaccuracies at local level may lead to imprecision in the allocation formulae. This, in turn, can affect the amount of central finance provided to the various authorities and agencies responsible for

delivering services locally. However, the relationship between headline population numbers and funding provision is complex. For example, for many services the amounts allocated are heavily dependent on the age structure of the population and the calculations incorporate factors other than population numbers.

At a national level, uncertainties about the size and age structure of the population now and in the future have been identified as a concern by a number of organisations, for the Bank of England (in relation to macro-economic policy) and the Pensions Commission (in relation to future pensions provision).

1.4 Engaging with stakeholders

ONS is committed to engaging with stakeholders, and makes significant efforts to seek their views and participation in developments related to population statistics. All significant decisions on statistical developments are accompanied by public consultation. Regular liaison meetings are held with central and local government representatives to discuss population statistics issues. ONS population experts also regularly attend, and present information to, conferences with a population and demography theme.

Recent methodology improvements to international migration statistics were presented at liaison meetings held with central and local government, as well as at the annual conference of the British Society for Population Studies. A series of regional seminars were also held to present these changes to users, supported by the publication of a wide range of methodological material on the National Statistics website. A panel of expert users, from local government and academia, quality assured the improved methods.

ONS has recently carried out a number of studies in collaboration with local authorities to identify how local sources could be used on a nationally comparable basis to improve population statistics. Findings from these studies were discussed at a series of participative workshops with representatives from a wide range of local authorities and other interested parties. Over 70 organisations sent representatives. Papers from each case study were made available on the National Statistics website:

www.statistics.gov.uk/about/data/methodology/specific/population/future/imps/updates/default.asp

A final report on the studies will be published in the near future, incorporating discussion from the workshops.

Stakeholder engagement is a key element in planning the 2011 Census. A Census Stakeholder Engagement Strategy is in development which identifies and classifies the various stakeholders and stakeholder organisations to ensure that they are fully engaged and managed effectively and that any concerns are addressed in advance of the Census so as to maximise response rates and confidence in the Census.

Discussions have also been held with various community and interest groups and organisations. In addition, a number of specific Census Advisory Groups covering the interests of a wide range of census user communities and other stakeholders have been established. These cover local authorities, academics, business and Central government stakeholders in England and Wales. Future engagement is being planned with other key stakeholder groups.

The success of the 2011 Census will be influenced by how well stakeholders are engaged and managed, and this is therefore a key priority for ONS.

2. CURRENT METHODS AND SOURCES2.1 Methods used

2.1.1 Estimating populationONS produces usually resident population estimates for England and Wales and combines these with

estimates for Scotland and Northern Ireland to produce figures for the UK as a whole. Details of how estimates are made in England and Wales can be found at:

www.statistics.gov.uk/downloads/theme_population/Making_PopulationEstimate.pdf

Population estimates for Scotland are produced by the General Register Office for Scotland, GRO(S):

www.groscotland.gov.uk/statistics/library/poptest/index.html

The Northern Ireland Statistics and Research Agency (NISRA) produces population estimates for Northern Ireland:

www.nisra.gov.uk/statistics/financeandpersonnel/DMB/datavault.html

The population from the 2001 Census underpins the subsequent mid-year population estimates. The definition of the resident basis of the 2001 Census is given in Box 1.

Box 1: Resident basis of the 2001 Census The 2001 Census was conducted on a resident basis. The means the statistics relate to where

people usually live, as opposed to where they are on Census night. Students and schoolchildren studying away from home are counted as resident at their term-time address. Residents absent from home on Census night were required to be included on the Census form at their usual/resident address. Wholly absent households were legally required to complete a Census form on their return. No information is provided on people present but not usually resident.

In the inter-censal period, estimates of the resident population are made using what is known as the “cohort component method". This updates the previous mid-year estimate, by “ageing" the population by one year, allowing for natural change due to births and deaths in the past year and adding on net migration (that is people moving between areas or countries for a year or more). The process of adding and subtracting components is shown in Figure 3. This is a standard demographic method and is used by most other national statistics institutions in Census taking countries.

Adjustments are also made for some special population groups that are not captured adequately by data sources currently used to estimate internal and international migration. These are home and foreign armed forces, dependants of foreign armed forces, prisoners and pupils in boarding schools.

The sources used in producing population statistics are discussed in Section 2.2.

Figure 3

MID-YEAR POPULATION ESTIMATES COMPONENT METHODOLOGY

/MILES/PKB/DATA/386539/PageEv211.eps>

2.1.2 National population projectionsNational population projections are prepared by the Office for National Statistics on behalf of the Registrars

General for England & Wales, Scotland, and Northern Ireland. National population projections by age and sex are produced for the United Kingdom and constituent countries every two years. The production of these national projections was a responsibility of the Government Actuary's Department (GAD), in consultation with the statistical offices of the constituent countries, until 31 January 2006 when the responsibility moved to ONS.

The 2006-based projections were published on the 23 October 2007, see

www.statistics.gov.uk/cci/nugget.asp?id=1352

They are based on a full scale review of the underlying assumptions about future fertility, mortality and migration. The assumptions are agreed in liaison with the devolved administrations, following consultation with key users of projections in each country and advice from an expert academic advisory panel. The assumptions are based on analysis of demographic trends. They are not forecasts and do not attempt to predict the impact that future government policies, changing economic circumstances or other factors (whether in the UK or overseas) might have on demographic behaviour. They simply provide the population levels and age structure that would result if the assumptions about fertility, mortality and migration levels were to be realised in practice.

For the first time in the last forty years of national projections, the long-term birth rate assumption has been increased for the principal 2006-based projection, on the basis of on analysis of factors contributing to rising fertility in all four UK countries over the past four years. Migration assumptions and life expectancy assumptions have also

been raised. This combination of increases in all three components of population change has caused a significant increase in projected future population size, compared with the 2004-based principal projection.

Variant projections are also produced as an illustration of how the population size and structure would change under various scenarios, and to give users an indication of the sensitivity of the projections to changing underlying assumptions.

2.1.3 Sub-national population projectionsONS produces sub-national projections for England that are fully consistent with the national projections

produced every two years. Across all sub-national areas they sum to the national projections in terms of population (by single year of age and sex) and components of change (births, deaths and international migration). The sub-national projections include numbers of projected population by age (five year age groups) and sex for a twenty-five year period to 2029. They give projections down to Government Office Region, Strategic Health Authority, County, Unitary authority, London borough, and metropolitan and non-metropolitan district level.

Sub-national population projections use local data on population, births and deaths data to calculate local fertility and mortality rates. Internal migration estimates are based on Patient Register Data System (PRDS) data which is available at LA level. International migration consists of a combination of data from the PRDS, International Passenger Survey and Labour Force Survey data and Home Office data on asylum seekers. Where necessary, census data is used to disaggregate this data down to the geographic level needed for the projections.

Details of the projections can be found at:

www.statistics.gov.uk/statbase/Product.asp?vlnk=997

The next set of sub-national projections will be published in summer 2008. These will be based on the recently published 2006 National Projections (see Section 2.3).

2.1.4 MigrationMigration is the most difficult component of population to estimate. Figures are required on moves into and

out of the United Kingdom (international migration), those between its component countries and between areas within each country (internal migration).

International migration

ONS uses the United Nations recommended definition of a long term international migrant in compiling population estimates. This is someone who changes their country of residence for a period of at least 12 months. Those who move or travel for shorter periods are excluded from these estimates. Sources used include:

@UL@— International Passenger Survey (IPS);@/UL@

@UL@— Labour Force Survey data on the location of recent migrants;@/UL@

@UL@— Home Office data estimating the number of asylum seekers and dependants who are not counted by the IPS;@/UL@

@UL@— National Asylum Seeker Support Service data on numbers of asylum seekers by LA;@/UL@

@UL@— Irish Quarterly National Household Survey and National Health Service Central Register (NHSCR) data for estimates of flows between the UK and the Republic of Ireland;@/UL@

@UL@— 2001 Census data; and @/UL@

@UL@— A wide range of other data sources used for out-migration modelling. @/UL@

Estimates are largely based on IPS, supplemented by the information on flows to and from the Irish Republic. Adjustments are then made for asylum seekers and their dependants, people whose intended stay, as reported to the IPS, is less than 12 months, but who subsequently stay for a year or more (known as “visitor switchers"), and, conversely, for those who intend to stay for a year or more but do not do so (“migrant switchers").

Improvements to the methodology implemented in 2007 mean that:

@UL@— By using an ONS major household survey (the Labour Force Survey) in combination with the International Passenger Survey (IPS), better estimates can be made of where migrants go within the country than using the IPS alone.@/UL@

@UL@— The changes recognise that, for example, many migrants who initially arrive in London move on.@/UL@

Details of the improvements ONS is making to population and migration statistics are at:

www.statistics.gov.uk/statbase/Product.asp?vlnk=14834

Internal migration

Internal migration estimates are based on the registration of patients for NHS General Practitioner Services. In this context, an internal migrant is defined as a person who changes their postcode of residence between one year and the next. Address changes are currently identified in two ways:

@UL@— Re-registration with a GP following a move between health areas, as recorded on the NHS Central Register (NHSCR); and@/UL@

@UL@— Notification to GPs of a change of address by patients. ONS are supplied with an annual extract of the information held on local health area systems that relates to postcode of residence.@/UL@

Details of the methods used for estimating internal migration are at:

www.statistics.gov.uk/statbase/Product.asp?vlnk=7070

2.2 Sources currently used

2.2.1 Uses and timeliness of Census IformationThe 2001 Census data has proved to be a rich and uniquely valuable source of information of the population

of England and Wales.

ONS uses include providing the basis for all mid 2001 population estimates (national, local, small area, ethnic group). Census distributions are also used in subsequent years to provide more detail than can reliably be obtained from sample surveys or annual rolled-forward estimates. ONS fulfils its requirements to report on social change by producing analyses that combine information from census with that from surveys, registration of life events and other administrative sources. Publications include Social Trends, the “Focus On" series and Population Trends.

An immense amount of tabular information was made available to users from the 2001 Census. In all about two billion counts were produced. These were issued in 335 standard table formats for various types of geographic areas. These ranged from national level tables down to Local Authorities, wards and Super Output Areas, Health Areas, National Parks, Parliamentary Constituencies, Postal Sectors. They cover all the topics of the Census—demography, ethnicity, country of birth, religion, health, carers, labour market, qualifications, household and family composition, housing types and tenure, availability of cars.

Some 30 Origin-destination tables provide information about flows of people between local authorities, wards, or output areas either due to migration in the year before Census or travel to work.

To date 1,338 ad-hoc tables have been produced for customers ranging from government departments, local authorities, academics, charities and other divisions of ONS. Interest focused, in particular, on having more detail on ethnicity, religion and families than was provided in the standard tables.

To facilitate in-depth research using the Census, ONS provided ESRC with three samples of anonymised records. These are made available to users by Manchester University. A further two, more detailed, samples are available for analysis within the safe setting of the Virtual Microdata Laboratory at ONS sites.

Tabular material is now made available through web based tools that simplify extraction, analysis and comparison with other data for small areas (in particular the Neighbourhood Statistics System and NOMIS).

2.2.2 Port surveysThe IPS interviews passengers travelling to or from the UK through airports, by sea and through the Channel

Tunnel. The survey collects data primarily for tourism purposes, to inform the travel account of the balance of payments, but is also used in estimating international migration.

Interview teams routinely identify every nth person (“n" varies by port and route, taking account of traffic flows) in the flow of passengers past a specified point. Information is collected from any migrants identified through these routine samples. However, for selected ports and routes, additional passengers are selected for a short interview. The questions asked are designed to establish whether the people selected are migrants. If this is the case, more detailed questions are then asked. Prior to 2007, these additional interviews were carried out only for arrivals (to identify immigrants only). Since January, they have been extended to departures (to boost the number of emigrants interviewed). At present the only reliable source of information on emigration is the IPS.

IPS sampling is carried out at all airports with more than 1 million international passengers travelling through them. This currently includes five London airports, Manchester and 10 regional airports. The IPS uses Civil Aviation Authority (CAA) figures to weight up the sample figures to total flows, to ensure that all people are accounted for. In addition the Channel Tunnel and 10 sea ports are covered and again the survey data are weighted to total flows supplied by Eurotunnel, Eurostar and Department for Transport.

Following the recommendation of the Inter-departmental Task Force on Migration Statistics to improve information collected on migrants at ports, ONS initiated a review of port surveys. An interim report of the review was published on the National Statistics website in October 2007.

www.statistics.gov.uk/about/data/methodology/specific/population/future/imps/updates/default.asp

This made a recommendation, taking into account the most recent travel patterns, that additional shifts should be introduced—including at Manchester, Stansted and Luton—from 2008 to better optimise the sample for migration purposes. This recommendation will be implemented. Further changes will be recommended for 2009 in the final report. Improved information from a port survey will be used in combination with other sources, such as the new enlarged household survey starting in 2008 and the 2011 Census.

2.2.3 Population surveysThe Labour Force Survey (LFS) is a sample survey based on postcode address registers that include all types

of private households including multi-occupied dwellings and households in multiple occupation (HMOs). The survey includes questions on address one year earlier, which can be used to identify migrants. This is used to provide important information on where migrants settle, their origins (eg country of birth) and what they do (eg employment, study). The LFS excludes most communal establishments although it does cover most staff resident in National Health Service accommodation. Students in halls of residence are covered by proxy through their parents and thus foreign students living in halls of residence are not likely to be covered—those living in private households will be.

Improvements have been made to the international migration methodology by combining International Passenger Survey (IPS) data with LFS data. This has resulted in:

@UL@— Improving the distribution of international in-migrants (nationally between England and Wales and regionally within England) by using the LFS in combination with the IPS; and@/UL@

@UL@— Refining the existing two-stage process for distributing international in-migrants to local areas within each region in England and Wales, by removing reliance on the smallest IPS sample numbers for some local areas and making adjustments where there is evidence of a tendency to over-report other areas.@/UL@

The method for estimating international migrants uses geographical distribution of in-migrants obtained from the LFS to improve on the regional information available from the IPS. IPS information is based on respondents' stated intended area of residence on their arrival in the country. However, these intentions may not be an accurate reflection of where people actually settle. As the LFS is a household survey, it measures where migrants are actually living.

ONS has initiated a study into the feasibility of conducting a communal establishment survey, asking similar questions to those used in the LFS. This would have multiple benefits—for migration and disability statistics, in establishing a sampling frame of establishments and in improving the range of comparisons that are possible between census counts and surveys information.

2.2.4 NHS registersThe use of patient registers in estimating internal migration was described in Section 2.1.4. When a person

re-registers with a GP a record is made and is used to measure moves between local authority areas. ONS assumes that the average delay between moving and an individual re-registering with a GP is 30 days. This is taken into account in making estimates. However re-registration is known to be particularly slow for particular groups, for example young men.

Information from patient registers is not currently used in estimating international migration. Although the first registration of those allocated a new NHS number on arrival in the country is separately identifiable, this identification is not retained when the patient registers with a second or subsequent GP. Consequently, by the time an annual snapshot is obtained, many new migrants will not be identified. This results in a systematic undercount.

These registers are not appropriate for measuring international emigration as there is currently no incentive to de-register when leaving the country and very few people do so.

Patient register data is also used to produce population estimates below local authority level, for example at electoral ward level. The data were also used for validation of the 2001 Census results. It is likely that they will be used in a similar way following the 2011 Census.

The strengths of these data relate particularly to coverage:

@UL@— the data include all people requiring access to NHS services through a GP regardless of age or reason for visit; and@/UL@

@UL@— individuals staying in the UK for longer than three months can register with a GP and will be included.@/UL@

However, the data excludes:

@UL@— those not wishing to access NHS services from a GP;@/UL@

@UL@— those staying in the UK for under three months; and@/UL@

@UL@— information on the country of origin of new registrations.@/UL@

2.2.5 Fertility and mortalityInformation on births and deaths registered in this country are used in the compilation of mid-year

population estimates. These are obtained from the General Register Office through the compulsory registration of all births and deaths. They are considered to provide a reliable indication of these events and include both information needed for registration and additional statistical items.

The annual number of births in the UK has increased from 669,000 in 2001 to 748,600 in 2006. The recent increase in the number of births in the UK is due to changes in fertility rates and changes in the size and age structure of the female population. In part this reflects net inward international migration of women in the fertile age groups in recent years.

2.3 UK comparisons

Population estimates and projections are produced using comparable methods across the United Kingdom. From the figures supplied to ONS by the constituent countries, figures for the United Kingdom are made available on a consistent basis.

A slightly different methodology is used in Scotland and Northern Ireland to estimate international migration for the purposes of population estimation from that used in England and Wales. Given the small size of the IPS sample for those countries, adjustments are made to migration estimates using administrative sources. For this reason,

figures published by ONS on Total International Migration (which are based primarily on the IPS) differ slightly from those used in estimating the UK population.

A UK-wide work programme on population statistics and demographic analysis is co-ordinated by the UK Population Committee, on which ONS and constituent countries are represented. An Advisory Board, which includes external experts from the local government, academic and business sectors, advises it in that role and helps identify gaps in provision.

2.4 International comparisons

Population statistics in England and Wales, and the UK more widely, compare well with other countries.

Most countries produce annual population estimates and regular projections to similar or lower levels of quality and coverage. The estimates and projections both use internationally recognised and accepted methodology:

http://epp.eurostat.cec.eu.int/cache/ITY_OFFPUB/KS-CC-03-005/EN/KS-CC-03-005-EN.PDF

Unlike some countries though, the UK does not have a comprehensive system of recording international migrants into and out of the country, nor does it have an established and linked system of administrative sources that allows more precise measurement of where people are in the country between Censuses. These are acknowledged weaknesses with our population statistics. However, it is recognised that, with increased mobility, even countries with established population registers are finding it increasingly difficult to ensure comprehensive coverage of international migration through these registers.

www.unece.org/stats/documents/ece/ces/2006/32.e.pdf

In some areas, we are noticeably “ahead of the game" internationally, for example most countries do not have separate estimates of short term migrants (see Section 5.2.1).

3. ISSUES IN PRODUCING POPULATION STATISTICSSome of the key issues in producing population statistics are described in this section under the following

broad headings:

@UL@— concepts and definitions;@/UL@

@UL@— methods of calculation;@/UL@

@UL@— reliability and accuracy of sources for population statistics purposes; and@/UL@

@UL@— revisions policy.@/UL@

3.1 Concepts and definitions

As indicated in Section 1.2, population statistics are based on the concept of usual residence. As an increasing number of individuals are more mobile or spend time at more than one address, this becomes increasingly difficult to operationalise and the statistics are less likely to cover the full range of user needs. The usually resident population does not always coincide with the number of persons to be found in an area at a particular time of the day or year. The daytime populations of cities and the summertime populations of holiday resorts will normally be larger than their usually resident populations. Some areas now experience a constant turnover of short term migrants.

Recognising these issues, work is in hand to develop estimates of short term migrants (Section 5.1.2). Consideration is being given to what population bases are required from the 2011 Census, to underpin statistics in the years that follow. This is described in Section 5.2.1. The increasing sensitivity of population counts to the base that is used has significant implications for the relevance of administrative sources as an aid to counting population. As indicated in the discussion of NHS patient registers (Section 2.2.5), the effective base for a register is determined by the rules governing the administrative system rather than statistical requirements. This is considered further in looking at alternative sources (Section 4).

3.2 Is the estimation methodology appropriate?

ONS population estimates represent the best population estimates that are currently available. ONS has made recent improvements to the way the international migration component of population estimates is calculated both nationally and at local authority level.

The published estimates are produced using consistent methods for all local authorities and strategic health authorities in England & Wales. Population estimates are also available for Primary Care Organisation areas. Detailed information about the components of change is made available to all local authorities, to aid understanding of population changes within each area.

ONS is open and transparent in the methods used to produce population estimates, following international standards and having what we do peer reviewed by external experts. A detailed methodology guide is published, which is open about any limitations. This will be updated in the near future to include the recent revisions to the migration methodology. The methodology guide can be found at:

www.statistics.gov.uk/downloads/theme_population/Making_PopulationEstimate.pdf

3.3 Reliability of available sources

The accuracy of the mid year population estimates is dependent on the quality of data available to measure components of population change (births, deaths and migration). The use of consistent data for measuring these components of change is key to ensuring a common level of accuracy in mid year population estimates for different areas. Migration, both internal and external, is the hardest component to measure.

Of the data sources used to estimate population estimates:

@UL@— The Census provides a reliable base and set of distributions, that worked well in most areas (as is discussed in Section 5.2.7);@/UL@

@UL@— Birth and Death registrations are considered to accurately reflect numbers of events occurring in this country;@/UL@

@UL@— Internal migration data are reliant on people registering change of address with their doctors promptly after a move (as discussed in Section 2.1.4); and@/UL@

@UL@— International migration is difficult to estimate, though good use is made of available sources, but estimates are subject to a margin of confidence (as is discussed below).@/UL@

In addition, there are other data sources that are important in estimating population in some local areas:

@UL@— Counts of “long term" prisoners are highly reliable and accurate;@/UL@

@UL@— Counts of boarding school pupils are generally reliable and accurate though require significant quality assurance and are reliant on accurate form completion by individual schools;@/UL@

@UL@— Estimates of UK armed forces numbers are accurate, but require significant modelling to meet population estimate purposes; and@/UL@

@UL@— Estimates of foreign (American) armed forces are dependent on the continuing goodwill of the USAF, but are reliable and relatively accurate, though significant data manipulation is required to convert these data into data that meet the purposes of population estimation.@/UL@

The principal source of international migration data is the International Passenger Survey (IPS). As with all surveys, the IPS is subject to sampling variability. Standard errors, a measure of how much a sample estimate differs from the true value because of random effects, can be calculated from IPS estimates.

As a guide, the standard error for an estimate of 40,000 migrants from the IPS is typically around 10%. In this example, this would mean that there is a one in twenty chance of the true value being less than 32,000 or more than 48,000 (that is to say, outside a range calculated as the estimated value plus or minus twice the standard error). In general, as the estimated number of migrants decreases, uncertainty becomes proportionately larger. For example,

when the estimate of migration is 1,000 the standard error increases to around 40%. However this can vary for certain routes on which passengers are sampled.

For England and Wales as a whole, the IPS component of the estimate of international in-migration during 2006 was 489,000. This was associated with a standard error of 4%. The comparable figure for out-migration was 345,000, with a standard error of 5%. As the estimated numbers of international migrants varied between regions of England, so too did the standard errors. The largest numbers of migrants went to and from London (151,000 in-migrants, with a standard error of 5%, and 111,000 out-migrants, with a standard error of 9%). The North East had the fewest migrants, 13,000 in each direction (with standard errors of 27% for in-migrants and 30% for out-migrants). Figures for Wales were 13,000 international in-migrants (standard error of 22%) and 8,000 out-migrants (standard error of 25%).

By using three year averages in estimating the migration component of population estimates at regional levels, the effects of sampling variation are significantly reduced.

As a survey, the IPS is subject also to non-sampling errors including non-response. The questions also focus on intentions (about length of stay and where migrants will live). These are often not realised, or may subsequently change. To estimate the amount by which actual and intended length of stay differ, new IPS question were introduced in 2004, to be asked of those interviewed at the end of a stay. Based on the answers to these questions, adjustments have now been made to national estimates of long term migrants. For 2006, this added 28,000 to the estimate of net migration. These adjustments are, of course, themselves subject to sampling variation.

3.4 Accuracy of population projections

ONS's national and sub national projections are trend based. They project what is likely to happen in the future if current demographic trends continue. They do not provide a forecast or prediction of what might happen to future trends. If current trends change, then the future population will almost certainly differ from the projection.

For a number of reasons gauging the accuracy of past projections is complex. ONS has published an article in Population Trends looking at the accuracy of national projections:

www.statistics.gov.uk/downloads/theme_population/PT129.pdf

The national total population has historically been projected reasonably accurately but this is largely a result of compensating differences. Differences between assumptions and outturns for future mortality, migration or fertility would affect the accuracy of the population projection, and each element can be—and has been in the past—subject to significant change.

A simple comparison of mid-2004 based projections for mid-2006 with mid-2006 population estimates indicates that over 95% of English LAs differed by less than 2% from the projection, with the biggest differences at +3 and -5%.

Recent analyses of past UK projections provide figures on the accuracy of national projections and the component assumptions.54 55In general, migration assumptions have the greatest uncertainty. One study found that the mean absolute error in assumptions of net migration to the UK has been around 60,000 10 years ahead and 100,000 after 20 years. 54 A separate analysis55 suggests that the accuracy of UK migration projections has been around the European average.

These analyses showed that, in the short to medium-term, projections of the working age population are far more accurate than those for the very youngest and oldest age groups. For example, errors in projections of the population aged 20-64 10 years ahead have averaged around 1-2% compared with errors of around 10% for the 0-4 or 85+ age-groups.

3.5 Revisions policy

54 Shaw C. Fifty years of United Kingdom national population projections: how accurate have they been? Population Trends 12855 Keilman N. UK national population projections in perspective: how successful compared to those in other European countries? Population Trends 129

A formal revisions policy came into effect on 5 July 2007. The policy, based on the National Statistics Revision Protocol, relates to the population statistics produced under the National Statistics Centre for Demography work programme by the Office for National Statistics, the General Register Office for Scotland, the Northern Ireland Statistics and Research Agency and the Statistical Directorate of the Welsh Assembly Government.

ONS will revise population statistics following a new Census, deriving a new population base. Between Censuses, ONS continues to develop its methods and sources. In doing this the key principles are:

@UL@— To revise outputs when significant effects of methodological change are established and quantified. There is an element of judgement in deciding what is significant in this context, taking account of user needs.@/UL@

@UL@— To announce substantial methodological changes before the release of statistics based on the new methods, together with an indication of the likely effects.@/UL@

@UL@— To seek to avoid frequent revisions and attempt, as far as is possible, to ensure that any revisions are in place prior to the production of the biennial population projections rounds.@/UL@

In addition, from time to time, there will be a need to make revisions outside of the planned schedule, for example because of unforeseen errors. Such unplanned revisions will be monitored and the possibility of such revisions minimised. All revisions, both planned and unplanned, will be pre-announced.

4. ALTERNATIVE SOURCESNational Insurance Number (NINo), Worker Registration System (WRS) and NHS Patient Registration data

are commonly cited as alternative measures of population change. A comparison of these sources was published recently:

www.statistics.gov.uk/about/data/methodology/specific/population/future/imps/updates/downloads/admin.pdf

ONS is investigating how aggregate level counts from these and other alternative sources might be used in migration estimation and in producing additional indicators of population change at local level. The uses of and access to individual records is also being reviewed, following the recommendations of the Inter-Departmental Taskforce on Migration Statistics (see Section 5.1). Use of individual records from administrative sources would, in particular, make it possible to undertake linkage between information held on different sources. However such data are potentially disclosive and so are subject to limitations on access.

Work has recently been undertaken within Departments, such as DWP, to investigate how existing linked data can be further exploited to improve the identification of migrant histories, including improved identification of emigration from the UK. This is included in the statistical evidence presented to the House of Lords Economic affairs Committee:

www.parliament.uk/parliamentary_committees/lords_economic_affairs/eaffwrevid.cfm

4.1 National insurance registers

National Insurance Numbers (NINo) are issued to individuals when they reach age 16 and are used to record a person's national insurance contributions and social security benefit claims. New numbers are issued to the following migrants:

@UL@— All non-UK born nationals aged 16 or over working, planning to work or claim benefits legally in the UK, regardless of how long individuals intend to stay.@/UL@

The following are excluded:

@UL@— Dependants of NINo applicants, unless they work or claim benefits;@/UL@

@UL@— Individuals from overseas not working, planning to work, or claim benefits—for example, this will include many students;@/UL@

@UL@— Migrants who are not of working age if not claiming benefits; and@/UL@

@UL@— Those with an existing national insurance number, for example returning UK nationals.@/UL@

This source provides numbers registering for a NINo in order to work or claim benefits and tax credits. Others who come to the UK are excluded. There is no requirement to de-register on leaving the country or. For this reason, the figures do not show the number of foreign nationals working or claiming benefit at any given point nor do they distinguish between long and short term migrants.

4.2 Pupil and student information

The School Census provides, every term, a snapshot of all school pupils in state education in England. It is collected by DCFS and is used for monitoring the effectiveness of policies and school/local authority funding. Similar systems are in place in the devolved administrations. There are a number of definitional issues which limit the value of comparing aggregate School Census figures with population estimates (children in independent schools, short term migrants, visitors and children with more than one address). However, these data may be useful if linked to other sources to provide more information about migrants.

The Higher Education Statistics Agency (HESA) collects data about students attending all publicly-funded higher education institutions in the UK. The term-time full postcode will be collected from 2007-08, thereby overcoming a limitation in the coverage of current data. This will inform both internal migration estimates of students and international migration of foreign students in higher education. ONS also intends to explore the availability of data on students not covered by HESA. For instance, we will be looking at what data may be available on further education students from the Learning Skills Council.

4.3 Local authority information

ONS has worked with local authorities to investigate local sources of information to assess whether they can be used on a nationally comparable basis to improve population estimates. Studies of four local authorities were conducted, each representing areas with specific issues in estimating population accurately. The sources examined included:

@UL@— The Electoral Register records people resident in each local authority who would be 18 or older during each year beginning 16 February and are eligible to vote in local government, devolved administration and or Parliamentary elections;@/UL@

@UL@— Housing Needs Surveys that are carried out by all local authorities in England; and@/UL@

@UL@— Council Tax Billing and Exemptions list based on a register of dwellings kept at the Valuation Office Agency. Each local authority is responsible for administering its own billing list.@/UL@

ONS published reports on the individual studies earlier in 2007. These can be found at:

www.statistics.gov.uk/about/data/methodology/specific/population/future/imps/updates/default.asp

Workshops were held to discuss these reports, to which all local authorities were invited. ONS will shortly be publishing a final report on this work.

4.4 Migrant registration

Nationals of countries (other than Cyprus and Malta) that joined the EU in May 2004—the A8 countries—who wish to take up employment in the UK for a period of at least a month are generally required to register with the Worker Registration Scheme (WRS). Workers who are self-employed do not need to register. They must register more than once if they are employed by more than one employer and must re-register if they change employer. Each application represents one job, not one applicant.

The population covered on the WRS includes:

@UL@— long-term international migrants from A8 countries working as employees in the UK;@/UL@

@UL@— visitors and short term migrants from A8 countries, staying for over a month, and intending to work as employees in the UK; and@/UL@

@UL@— dependants of WRS applicants. It is likely that there is some double counting as dependants may also be registered in their own right on the WRS.@/UL@

The following are excluded:

@UL@— migrants from A8 countries who are self employed;@/UL@

@UL@— A8 migrants staying for less than a month;@/UL@

@UL@— A8 migrants who migrate or visit the UK for reasons other than work, for example including potentially many students; and@/UL@

@UL@— migrants from non-A8 countries.@/UL@

Figures relate to the address of the applicant's employer rather than their own usual residence and are produced by date of application rather than date of entry into the UK. The data only include those registering when they take up a job, when intended length of stay is recorded. However neither actual duration of employment or if and when the applicant returns home are recorded.

4.5 E-Borders

The Home Office e-borders programme aims to establish a modernised, intelligence-led border control. In the long-term e-Borders will provide the best option for recording individual movements into and out of the UK and length of stay in the UK for all modes of transport. However, it will only provide information about people arriving and leaving the UK not where in the UK they will live. In so far as it is possible to link journeys into the country with journeys out for the same individual, by identifying travel patterns it is in principle possible to separately identify long term migrants, short term migrants (including seasonal workers) and visitors. However detailed analysis will be required to develop rules for categorising people with more complex travel patterns, for example distinguishing those who take up residence in this country for a prolonged period but make frequent trips abroad from those who reside abroad but frequently visit the UK.

The ONS is responsible for statistics on international and internal migration whereas the Home Office is responsible for statistics on immigration control. The ONS has been participating fully with the Home Office in the early stages of the e-Borders procurement process to ensure that full use can be made for analysis purposes of the extensive data that should start to become available from 2009.

4.6 National surveys

As indicated in Section 2.14, data from a survey (the IPS) is used with administrative sources to estimate international migration to and from the UK and is used in combination with the LFS and Census figures to distribute these estimates to regions and local areas. The IPS is currently the only source of information on emigration.

Surveys are an essential tool in estimating international migration because:

@UL@— they provide a cost-effective means of updating Census information each year;@/UL@

@UL@— social and economic questions can be asked and related to migration. These are not available on many administrative sources and provide both context and outcome information (eg employment); and@/UL@

@UL@— they are better suited to probing issues, such as intentions and reason for visit, than Census or administrative sources.@/UL@

However, as indicated in Section 3.3, they do have limitations. Specifically:

@UL@— all estimates are subject to sampling variation;@/UL@

@UL@— numbers of migrants sampled are too small, even in the large population and passenger surveys, to provide robust estimates at local levels; and@/UL@

@UL@— non-response tends to vary across population groups and areas, making accurate estimation more difficult.@/UL@

This analysis suggests that, to get the best estimates of international migration, information from well designed sample surveys needs to be used in combination with that obtained from the alternative sources discussed earlier in this section. The following section describes plans to improve the information obtained from surveys and other sources.

5. PLANNED WORK FOR 2008-125.1 Improving population and migration statistics

In 2006, the National Statistician set up an Inter-departmental Task Force on Migration Statistics to identify timely improvements that could be made to international migration statistics, in advance of those that might flow from long term strategic systems (such as e-borders).

www.statistics.gov.uk/statbase/Product.asp?vlnk=14731

The recommendations of the Task Force are summarised in Annex A. The developments being taken forward as part of this programme and the wider programme of improvements to population and migration statistics are described below.

5.1.1 Short term migration estimatesONS currently produces National Statistics on international migration based on the United Nations (UN)

recommended definition of a long-term migrant, someone who changes his or her country of usual residence for a period of at least a year. It is widely recognised that these estimates do not capture all visits, and ONS have been exploring how to estimate short-term migration.

The UN defines short-term migration as moves made for between three and 12 months for reasons of employment or study. An initial feasibility report, published in January 2007, sought opinions as to whether this definition met all user requirements.

The feedback to the feasibility report revealed that different users had different requirements. Requests were made to broaden both the “length of stay" and the “reason for visit" criteria. In October 2007, ONS published the first set of short-term migration estimates for mid-2004 and mid-2005. National estimates for a number of differing definitions were published for England and Wales, with separate estimates shown for London.

www.statistics.gov.uk/about/data/methodology/specific/population/future/imps/updates/downloads/STM_Research_Report.pdf

There is significant demand for short-term migration figures at local authority level. However, the International Passenger Survey (IPS) does not provide sufficiently robust estimates at this level, and no single alternative data source is currently available to produce such estimates. The availability of local area level estimates is dependent on the development of a suitable methodology and sources. Research work to look at identifying a method for producing local area level short-term migration estimates is being taken forward over the next year.

5.1.2 Access to administrative recordsThe Inter-departmental Taskforce on Migration Statistics recognised the potential for improving statistics

through using records from administrative sources. It recommended that access to a number of such sources was essential to develop record linkage approaches to measuring migration. Sources identified include components of the Work and Pensions Longitudinal Study (WPLS), migration statistics from new NHS IT systems, the School Census, and the Higher Education Statistics Agency (HESA). Taking forward the recommendations of the taskforce, ONS has prioritised negotiating access to these sources. A framework for access will be provided by the Statistics and Registration Services Act 2007, due to come into force on April 1st 2008. The Act, establishing the independence of ONS, contains powers to enable information sharing between public authorities and the newly established Statistics Board for statistical purposes. This is subject to the following safeguards:

@UL@— protection of confidentiality;@/UL@

@UL@— only where there is no existing power/ it is otherwise prohibited by law;@/UL@

@UL@— subject to bilateral Ministerial agreement;@/UL@

@UL@— sharing only for specific purposes and subject to a public interest test and@/UL@

@UL@— Parliamentary approval.@/UL@

Use of administrative records in combination will require development of new tools and techniques, including the ability to evaluate the quality of the information for statistical purposes. First steps have been taken in a pilot project in 2007. Further projects specifically related to population statistics, Census and administrative data will be undertaken in 2008/9 to build on this work.

5.1.3 Survey enhancementsThe Inter-departmental Task Force on Migration Statistics made a number of recommendations on

improvements that could be made to surveys to improve the quality of the migration data they provide.

The task force recommended a review of port surveys. This is underway, and an interim report was published in October 2007. The review proposed a number of enhancements that could be made to the International Passenger Survey (IPS) in 2008 that would increase the number of migrants sampled. More fundamental changes to the IPS are also being considered, including whether separate specific surveys should be run to meet the needs of current customers of the IPS (for example migration, tourism, balance of payments).

The main concerns that users have with the Labour Force Survey as a source of information about migrants are to do with the non-coverage of communal establishments and with the sample size of the LFS. ONS is assessing the feasibility of a communal establishment survey to increase the coverage of migrants.

In 2008 the LFS will include a special module on migration. In 2009 the new, continuous, household survey will replace the LFS and be enhanced to contain migration questions on the core sample every quarter.

5.1.4 Statistical modellingWhile survey data are likely to remain the main source of international migration statistics at national and

regional level they are already enhanced by the use of a model in apportioning the numbers of emigrants to Local Authority Districts. A wide range of other data sources are used in the construction of this model. Work is already under way to refine this model by using small area estimation techniques. Consideration will also be given to the possibility of constructing a similar model for immigration.

A joint ONS/Economic and Social Research Council studentship is researching the modelling of internal migration in the UK. Future research might also investigate the use of time series models for migration estimates.

5.1.5 Coherent reportingONS is working with other Departments to take forward plans to produce better, more coherent reporting on

migration statistics, a key recommendation from the Inter-departmental Task Force on Migration Statistics.

Work is well advanced on a cross-government review of migration reporting and on deciding which statistics should be within scope for coherent reporting. A public consultation on coherent reporting of migration statistics will be carried out in 2008.

In consulting with other departments, ONS is also considering how best to use a panel of migration experts to add value to existing migration reporting (another task force recommendation). It is anticipated that public consultation on the plans for coherent reporting will be published early 2008.

Earlier this year, ONS published figures on the volume of internal and external migration experienced by each local authority (see Table 1 in Section 1.2). This was widely welcomed by users and indicates the need to establish a wider and more relevant set of data about local populations. At the simplest level, these data would include short term migrant figures and population turnover due to migration, but could be widened into a coherent package of indicators.

5.2 2011 Census preparation

The final Census design has yet to be confirmed so the following is dependent on final design decisions, contract costs and Test evaluation. It reflects plans at the time of writing. Latest information about the 2011 Census can be accessed from:

www.statistics.gov.uk/census/2011Census/default.asp

5.2.1 Matching population definitions to user needsChoice of population bases for enumeration of the 2011 Census is critical to the type of data obtained at

output for users. Discussions in 2004 presented four options for the 2011 enumeration:

@UL@(a) population present;@/UL@

@UL@(b) population usually resident;@/UL@

@UL@(c) population present plus usual residents temporarily absent; and@/UL@

@UL@(d) population usually resident plus visitors.@/UL@

In considering the optimal enumeration base for 2011 one issue was continuity of measurement with previous censuses; another was providing continuity with post-2011 population measurement. British censuses in the years 1801-1971 employed a population present enumeration base; subsequently, the 1981 and 1991 censuses, although principally using a population present base, were designed to count BOTH population present and usual residents; the 2001 census employed usual residence alone. Post-2011 developments will probably be founded on administrative sources and based on usual residence.

Extensive consultations in 2004 amongst users (local/health authorities, central government departments and academia) concluded that:

@UL@— usual residence is the key population output base required;@/UL@

@UL@— a population present base would fail to provide accurate data on household composition, migration, resident population, journey to work etc; and@/UL@

@UL@— within the usually resident base, clarification is required of rules to reduce ambiguity in recording hard-to-count groups.@/UL@

Further discussions have endorsed these views, adding that:

@UL@— information is needed about short term migrant workers for labour market purposes; and@/UL@

@UL@— in addition to usual residence data, information on visitors (both from within the UK, and abroad) is required.@/UL@

5.2.2 Census TestBackground

The 2007 Census Test in England and Wales was held on 13 May 2007 in five different local authorities covering about 100,000 households. A report can be found at:

www.statistics.gov.uk/census/2011Census/2011Project/2007test.asp

The main objectives of the Test were to assess:

@UL@— The effect on response of:@/UL@

@UL@— the use of post-out to deliver questionnaires; and@/UL@

@UL@— the inclusion of a question on income.@/UL@

@UL@— The feasibility of major innovations in operational procedures, such as address checking, the outsourcing of recruitment, training and pay, and questionnaire tracking. @/UL@

The overall return rate for the Test was 46%. The Test over-sampled in the hardest to count areas (40% of Test areas compared to 10% of England & Wales). If the sample had been representative of England & Wales, it is estimated that the Test return rate would have been 58%, a comparable rate to that achieved in the 1997 Test (59%).

Key results

The key results from the Test are:

Post-out—A post-out methodology, rather than hand delivery, reduces the significant risk of being able to recruit a large, sufficiently skilled, field force; and, provides cost savings that can be channelled into targeting hard to count populations.

The delivery of questionnaires via post-out proved successful in the Test. The same resources were devoted to follow-up of postal and hand delivered questionnaires. As a result, more resources were put into hand-delivery areas overall, with contact being made on the door-step during delivery by hand. Nonetheless, the return rate for questionnaires delivered by post was only 3 percentage points lower than hand delivery (44% post-out, 48% hand delivery). Modelling suggests that this difference could be recovered through additional follow-up of postal delivery and still provide significant cost savings. Therefore, in 2011 the majority of households in England & Wales will receive their questionnaires by post.

Income—The return rate for a questionnaire including income was 3 percentage points lower than those without (45% income, 48% no income). A decision on the inclusion of an income question will be based on further assessment of the quality of responses and the strength of user requirement and questionnaire space considerations.

Address checking—The Test was the first time that a full separate operational address check has been conducted, previously this was undertaken as part of questionnaire delivery.

The Test demonstrated that an address check will be required to correct deficiencies in the currently available address registers, which are mostly associated with converted flats within existing buildings.

Recruitment, training and pay—For the first time recruitment, training and pay of field staff was outsourced together. Based on the success in the Test and additional market and options analysis, the Census intends to outsource these services for 2011.

Questionnaire tracking—Individual questionnaires were successfully tracked from delivery to receipt with the updated information exchanged with the field force to enable efficient and effective follow-up of non-response.

5.2.3 Census design and questionsONS recommendations for the content of the 2011 Census will be based on a number of factors, and user

requirements are a key consideration. These requirements are considered alongside issues of data quality, respondent burden and public acceptability as well as operational and financial considerations.

The current working assumption is for a 24-page household questionnaire with three pages of individual questions per household member as in 2001. However, it has been clear from consultation with users that there is significant demand for more topics than can be accommodated within three pages of individual questions.

ONS is currently prioritising two new topics for inclusion in 2011 which would increase our understanding of the population count, namely, second residences and month/year of entry into the UK. However, this would mean the exclusion of some topics which were included in 2001 (for example, carers, qualifications, industry) unless a fourth page of questions is introduced. ONS is currently seeking funding for an additional page of questions per person so that more information can be collected on the population.

5.2.4 Coordination of the Census across the UKThe UK is unique in the world in having three Census Offices for carrying out the census and responsible to

separate Parliaments and Assemblies. Other countries have a single census office. Devolution has increased the local accountability for census questions and operations, and makes co-ordination harder.

The National Statistician seeks to maximise the extent of UK Census harmonisation, but has little authority to compel it. Some census users want the best information for their country, regardless of consistency with other parts of the UK. However, most uses for policy development and monitoring require consistency across the UK.

The three Census Offices seek to co-ordinate census questions, census operations, and census outputs. An agreement in principle between the Registrars General was signed in 2005, and is regularly reviewed and updated. There is close working-level liaison and joint working to harmonise census questions—whilst allowing reasonable variation in each census office.

The Registrars General have agreed a common statement on disclosure control, with a view to a single method for disclosure control in 2011 (not achieved in 2001).

5.2.5 Improving coveragePost out and hand delivery will be used to deliver Census questionnaires in 2011. The primary method will

be post out, but in areas where ONS are not confident in the completeness of the address register, questionnaires will be hand delivered. In either case, if someone does not receive a questionnaire they can request one through a contact centre or web based help facility. New/replacement questionnaires will be sent out through the post. During the follow up exercise, field staff will also hand out replacement questionnaires.

There needs to be a strong role for local authorities in:

@UL@— Agreeing the address base in advance;@/UL@

@UL@— Providing local knowledge to identify potential enumeration difficulties; and@/UL@

@UL@— Promoting Census and encouraging participation.@/UL@

People will either complete the questionnaire on paper and return it through the post to a central location or use the internet questionnaire. Central post back will enable prompt, cost effective receipting of questionnaires and enable Census management to have up to date information on the rate and patterns of response across the country and take decisions accordingly. Internet capture will offer another route for the public to respond potentially appealing to groups such as students that traditionally are hard to engage in the Census.

Support will be offered through a contact centre and a web based help facility and there will also be supporting literature, including translations of the questions. Blind and partially sighted people will be offered large print questionnaires. Respondents will also be able to request field visits via the contact centre, to help them complete the questionnaire.

Enumerators will follow up questionnaires that have not been returned. When contact is made, the enumerator will be encouraged to offer help to the household to enable them to complete their questionnaire, from writing in the responses to posting it for them.

5.2.6 Improving response ratesThere are a number of strands to this:

@UL@— The field design for 2011 means a greater proportion of resources are being put into the enumeration of hard to count groups and areas, to people who can't or won't fill in their form without enumerator intervention. Where these areas coincide with deprivation, higher pay rates will be offered to ensure that we do attract sufficient staff. Enumerators will be mobile rather than assigned to a pre-determined area so we can move resources during the field operation in response to events/response rates. This means that if an area is proving more difficult to enumerate than previously anticipated, additional staff can be deployed to support the process. The use of central post back and an operational intelligence system will enable us to have accurate up to date information on response patterns at local levels during the Census operation so take prompt remedial action if necessary.@/UL@

@UL@— The use of post out will enable delivery of questionnaires to some properties where we would otherwise struggle for example those with access control and security arrangements.@/UL@

@UL@— A programme of community liaison is being developed that will work with national and sub-national organisations that represent many of the population groups in areas that are typically hard to count.

Information from this will both feed into how we approach different population groups as well as providing a means to encourage participation from the “inside".@/UL@

@UL@— A strong publicity campaign that exploits media channels that are used by these groups and puts forward messages that are meaningful and relevant to the specific groups will also be in place.@/UL@

5.2.7 Response estimationNo census is perfect and some people are missed. This undercount (or non-response) is not uniform across

geographical areas or other sub-groups of the population such as age and sex groups. For resource allocation, this will have a disproportionate effect if the populations that are missed are those that attract higher levels of funding (eg children or the elderly). For this reason, without any adjustment, the allocations relying on uncorrected census counts would result in resources being wrongly allocated. To prevent this happening, census undercount is measured and the outcome disseminated to users.

In the 2001 Census, a methodology and process were put in place to identify and adjust for the number of people and households not counted in the 2001 Census. Both the Statistics Commission and the Local Government Association published reviews that concluded that the methodology used in 2001 was the best available.

ONS outlined its 2011 Census coverage assessment and adjustment strategy in the spring 2007 edition of Population Trends

www.statistics.gov.uk/downloads/theme_population/PopulationTrends127.pdf

The overriding strategy is to build on the methodological framework used in the 2001 Census, using it as a platform to develop an improved methodology with a number of objectives:

@UL@— The strategy will aim to address the lessons from 2001, looking for improvements and taking into account the changes to the census design.@/UL@

@UL@— ONS are looking to build in more sources of data, eg Council Tax, and more time for Quality Assurance and analysis into the processes to reduce the risk of having to make post-census adjustments.@/UL@

@UL@— The methodology should produce results that are robust and of high quality, with no Local Authority having poorer quality results than in 2001.@/UL@

@UL@— Gaining acceptance of the methodology from users is a key objective. Users will not accept their census population estimates if they are not confident about the methodology used to derive them.@/UL@

@UL@— Simple methods should be developed where possible to aid communication of the methodology with all users through appropriate channels and with tailored materials.@/UL@

The development of the methodology that will be used to achieve the above aims and objectives is underway. ONS has been consulting with users through its advisory groups, and has been peer reviewing the methodological proposals developed to date. This engagement and quality assurance will be widened as the detailed methods develop.

5.2.8 Census estimation in 2011For the 2011 Census, ONS will be refining the procedures it introduced to ensure the accuracy of the 2001

Census estimates (see Section 5.2.7). The approach provides an estimate of the level of under enumeration in the main Census, in terms of both households and individuals. Among the refinements for 2011 are improved identification of difficult to count areas and accounting for over enumeration.

In addition, ONS have been reviewing the definitional bases to be used in the 2011 Census. This work includes consideration of population sub-groups, such as short-term migrants, not estimated by the 2001 Census. The definitional bases used are important to ensure consistency between Census and other ONS products, such as the mid-year population estimates.

5.2.9 Census outputsOutput from the 2011 Census falls into three broad categories of information:

@UL@— Estimates of the numbers of population units—people, households and dwellings;@/UL@

@UL@— Population structures—eg family and household relationships, ethnic groups; and@/UL@

@UL@— Population and housing characteristics—eg education and health.@/UL@

A significant volume of pre-planned statistics in tabular form will be devised so as to provide a set of consistent output. This will be supplemented by services to deliver customised output for specific users.

Availability will be from first release of population estimates, scheduled for September 2012, with a rolling programme of releases through to the end of 2013.

Dissemination is expected to be predominantly via the web. Paper reports and other electronic media (eg DVD) are also expected to be available.

We will be providing commentary and analysis of key results subsequent to the initial release of population details.

6. POPULATION STATISTICS IN 2013 AND BEYOND6.1 Future of the Census

ONS has set out a vision for an integrated population statistics system that would bring together survey and administrative sources.

www.statistics.gov.uk/about/Methodology_by_theme/downloads/Demographic_Statistics_Service.pdf

The need for future Censuses will be assessed within this framework. ONS will next year be initiating a project to review the needs for a 2016 census and alternatives to a census for 2021.

To date, the pattern of Census taking in the UK has focused on holding them at ten yearly intervals, with only a few exceptions. The evaluation of the needs for a 2006 census concluded that effort would be better put into improving other data sources. ONS has taken this forward through the development of a continuous population survey—bringing together four existing household surveys to provide a large annual sample core.

The new survey aims to increase the range and quality of statistical outputs while preserving the integrity of key time series. It will provide regular estimates of key socio-demographic variables at sub-regional level between decennial censuses.

The sample will be made up of the total number of addresses sampled by the existing surveys, making it the largest ever continuous survey to be conducted in Great Britain. It would be possible to boost this survey to provide core population information for smaller areas.

A phased timetable for implementation is planned from January 2008, with the LFS survey module being incorporated later in 2008.

6.2 Population registers

The ONS led Citizen Information Project (CIP) that reported in 2006, found that there is significant value to both citizens and the public sector in greater sharing of basic citizen information in a secure way across central and local Government.

www.gro.gov.uk/cip/Definition/FinalReportAnnexes/index.asp

The CIP is now closed and its recommendations are being taken forward by other government departments, ie where the opportunities exist, subject to legislation, to develop systems that have the potential to deliver many of the CIP benefits in the longer term. It is recognised that a statistical imperative alone will not be sufficient to establish and maintain a comprehensive and up to date population register.

In the UK there are several administrative sources of information about the population that are registered for a particular service eg NHS patient registers, NI numbers). None of them are sufficiently comprehensive or up to date

to constitute a population register. In particular, there are significant practical difficulties in identifying those people on these registers who no longer live in the UK.

6.3 Plans to consult stakeholders

As indicated in Section 6.1, next year ONS be initiating a project to review the needs for a 2016 census and alternatives to a census for 2021 within the framework of an integrated population statistics system. Engagement with stakeholders will be central to this review.

@ANAME@Annex A

RECOMMENDATIONS OF THE INTER-DEPARTMENTAL TASK FORCE ON MIGRATION STATISTICS

Recommendation A: Improve information about migrants as they enter or leave the country

@UL@— Establish a port survey to capture an increased sample of migrants, especially emigrants. Bring forward key elements of the e-Borders project, including passport scanning. Improvements to statistics could be delivered in three to five years;@/UL@

@UL@— Provide more data on controlled migrants from landing cards, in particular, duration of stay and destination in the UK. Link data to subsequent information about the migrant. A sample of all travellers would need to complete a landing card on arrival in the UK;@/UL@

@UL@— Collect migration-related information from the points-based information system being developed to manage the flow of migrants coming to the UK to work or study.@/UL@

Recommendation B: Improve information about migrants living in this country

@UL@— Survey people living in communal establishments, who are not currently covered by ONS household surveys. Make more use of information on migrants collected by local authorities, employers or agencies providing work for migrants;@/UL@

@UL@— Include in the 2011 Census questions that identify short-term and long-term migrants;@/UL@

@UL@— ONS and devolved administrations should have access to timely administrative information that potentially identifies migrants.@/UL@

Recommendation C: Link official data sources to relate migrants' intentions at entry to the UK to subsequent events, such as employment, having a child and when they leave the country

@UL@— Explore the potential for using a current survey, the Work and Pensions Longitudinal Study, to provide information on patterns of employment, children and benefits among migrants;@/UL@

@UL@— ONS and the devolved administrations should continue to have access to the population statistics items, available on individual level health registers, that are needed to estimate migration;@/UL@

@UL@— Access should be provided to individual level School Census data for statistical purposes, to allow linking with other sources so as to improve the statistics on migrants and their families.@/UL@

Recommendation D: Improve key indicators of migrant numbers and trends

@UL@— Improve the use of statistical and demographic models to enhance migration and population estimates;@/UL@

@UL@— Convene an expert committee on migration statistics reporting to the National Statistician to provide an up to date interpretation of current UK migration statistics.@/UL@

Recommendation E: Publish all statistics collected across government on migration and migrants in a single UK-wide report

@UL@— Provide a more coherent picture than is available from the wide range of disparate information currently available;@/UL@

@UL@— The National Statistics Centre for Demography, part of ONS, should publish an Annual Report on Migration Statistics.@/UL@

December 2007

@HR25@