The Take-Up of Means-Tested Benefits by Working Families with Children
RICHARD DORSETT* and CHRISTOPHER HEADY?
The purpose of this paper is to analyse the take-up of the two major means-tested benefits that are available for working families with children in the UK: family income supplement (HS) and housing benefit (HB). As eligibility for FIS depended on at least one member of the household working 30 hours per week (24 hours for single parents), we are concentrating on benefits to working families. The take-up of these benefits has become a particularly important issue because of the Governments policy of switching resources from child benefit, which is a universal benefit, to the means-tested family credit (FC) that replaced FIS in the 1988 social security reforms. Family credit has been in operation for too short a time to analyse the take-up behaviour of its potential recipients. However, its method of assessment is sufficiently similar to that of FIS for us to believe that the experience of FIS take-up will be a reasonable guide to the take-up of family credit.
Although most of the discussion of targeting benefits for families with children has concentrated on the switch from child benefit to FIS / FC, it is also necessary to take account of housing benefit in assessing the likely success of targeted benefits reaching families with children. There are several reasons for this. First, housing benefit is a means-tested benefit that is available to a wide variety of groups in the
* Economic Development Department of Kent County Council. t School of Social Sciences, University of Bath.
An earlier version of this paper was presented at the Royal Economic Society conference in April 1991. The research was supported by a grant from the Leverhulme Trust and was mainly carried out when both authors were at University College London. The authors would like to thank the Institute for Fiscal Studies for providing access to the data and allowing them to use their Tax-Benefit Model. The authors have benefited from discussions with Richard Blundell, Andrew Dilnot, John Hudson, Stephen Jenkins, Steve Machin, Costas Meghir, Jane Millar, Graham Stark and Steven Webb.
Take-Up of Benefits
population, including families with children. Second, a large majority of households eligible for FIS / FC are also eligible for housing benefit. Third, benefit receipts under FIS / FC are taken into account in calculating entitlement to housing benefit. Fourth, the experience of claiming one means-tested benefit can be expected to alter a family's attitude to claiming another one. This paper therefore makes a first attempt at analysing the interdependence between take-up decisions on the two benefits.
The best-known econometric study of the take-up of means-tested benefits is the study of housing benefit by Blundell, Fry and Walker (1988). The approach taken in the current paper is similar, but with modifications to take account of the interdependence noted above.
We are not aware of any previous attempts to model the take-up of FIS. One of the main reasons for the absence of such studies is that only a small proportion of families were entitled to FIS, so that surveys such as the Family Expenditure Survey (FES) do not include many eligible families in their samples. This means that it is difficult to obtain statistically significant estimates of behavioural models, based on one year's data.
One solution to this problem is to pool data from a number of years and therefore increase the number of eligible families in the sample. This is the approach taken in the present paper. However, there is a limit to the number of years that can be pooled because of the need to observe families responding to essentially similar benefit systems and the tendency for governments to change benefit systems fairly often. Thus, if we are looking for recent data, we must note the change to the housing benefit system in 1983 and the social security reforms of 1988. This means that the longest reasonably stable recent period consists of the four years from 1984 to 1987. This does not allow us as large a sample of eligible families as we would like but it is the best that can be done and, as the results will show, it is large enough to provide useful insights into the determinants of take-up.
The rest of the paper is organised as follows. Section I1 describes the data, including the patterns of eligibility and take-up of FIS and HB. Section I11 discusses the way in which this observed behaviour can be modelled. Section IV presents the results of estimating these models. Finally, Section V draws together the main conclusions of the study.
11. THE DATA
The data are taken from the Family Expenditure Survey for the years 1984, 1985, 1986 and 1987. For the purposes of this study, families are taken to be the same as the FES-defined tax units because the assessment for FIS entitlement is based on tax units. However, it should be noted that housing benefit is based on households, which may include more than one tax unit.' We have followed Blundell, Fry and
' A tax unit consists of a single person or manied couple together with their dependent children. A household will therefore include more than one tax unit if, for example, grandparents, parents and children all live together.
Walker in assuming that all housing costs are borne by the first tax unit in the household, and therefore that it is the first tax unit that is entitled to any HB. This means that second (and subsequent) tax units are regarded as ineligible for HB.
Families containing self-employed workers were removed from the sample because their FIS and HB entitlements are determined by the authorities using different criteria from those applied to employed workers. Moreover, these criteria are not known with sufficient accuracy to model with any confidence.
The entitlements to FIS and HB were calculated using the Tax-Benefit Model developed at the Institute for Fiscal Studies (IFS) and described in Johnson, Stark and Webb (1990). In fact, the FES collects data on benefit receipts but these cannot be used in a consistent manner to predict take-up because, by definition, they do not apply to eligible families that do not claim the benefit.
Because FIS receipts are counted as income for the purposes of calculating HB entitlement, these calculations were made under two alternative assumptions: (1) that FIS was taken up and was paid at a rate equal to the calculated entitlement, and (2) that FIS was not taken up. In what follows, a family is regarded as eligible for HB if its entitlement under assumption (2) is positive.
For the interpretation of the results, it is important to realise that these calculations are subject to error. In addition to the standard problems of inaccuracies in survey data, there is the problem that the FES collects income data for a period immediately preceding the interview while the calculation of benefit entitlement will have been made at the time the claim was made, which might have been several months earlier. HB recipients are supposed to report changes in their circumstances but this may not always happen immediately. Also, FIS entitlement is meant to be constant for a period of a year even if income does change. Thus, in some caseR we will be calculating benefit on the basis of an income that is different from the one the authorities used when the claim was made. We may also be using benefit rates that relate to a later period than the one in force at the time a claim was made.
The calculations of entitlement were made in current values using the benefit rates that were in force at the time of the FES interview. The resultant entitlements were then converted to 1984 prices, using the Consumer Price Index. Similar deflations were applied to all the explanatory variables expressed in monetary terms. All explanatory variables also came from the FES.
The IFS Tax-Benefit Model also calculates entitlement to supplementary benefit, and it turns out that some families are entitled to both, although they cannot both be claimed. In order to simplify the analysis, these families were dropped from the sample, as were families who were in receipt of supplementary benefit but did not have a calculated entitlement.
The remaining sample consists of families with a calculated entitlement to either FIS or HB or both. The pattern of this entitlement is given in Table 1. It is clear that almost all families entitled to FIS are also entitled to HB, while the reverse is definitely not true. In fact, the families entitled to FIS but not to HB are often secondary tax units (for example, single parents living with their parents) that might
Take-Up of Benefits
Yes ENTITLEDTOFIS No
be expected to behave differently from primary tax units in their take-up of FIS, and this is confirmed in the statistical analysis of Section IV. We will mainly concentrate, therefore, on those families that are entitled to both FIS and HB.
ENTITLED TO HB Yes No Total
134 38 172 634 - 634 768 38 806
Entitlements to FIS and HB
Table 2 presents the pattern of take-up by tax units with a calculated entitlement to either benefit. A comparison of these figures with those in Table 1 reveals a take-up rate for HB of 50 per cent and a take-up rate for FIS of 59 per cent.
Take-Up of FIS and HB by Whole Sample __ ~~
TAKE UP HB Yes No Total
Yes TAKE UP FIS No
71 31 102 316 388 704 387 419 806
The take-up by families entitled to both benefits is shown in Table 3. Here, the HB take-up rate is 69 per cent and the FIS take-up rate is 65 per cent. If the take-up of the two benefits were independent, we would expect the proportion taking up both benefits to be 45 per cent (= 0.69~0.65~100). In fact, the proportion taking up both is 53 per cent. This provides some support for the suggestion that the two take-up decisions are interrelated. However, it could simply be due to families with high FIS entitlements (and therefore high take-up probabilities) having high HB entitlements. This can only be resolved by the econometric results presented below.
It should be noted that these take-up proportions are based on those households with a calculated entitlement. Table 4 shows that 77 households without a calculated entitlement to FIS and 154 households without a calculated entitlement to HB took up these benefits. These cases most probably arise from the difficulties in calculating entitlements noted above (difficulties that can also account for the differences between entitlements and receipts for those recipients who we have recognised as entitled).
Entitlements Non-recipients Recipients
Take-Up of FIS and HB by Families Entitled to Both
Receipts Entitled Non-entitled
TAKE UP HB No Total
Yes TAKE UP FIS No
flO.O f11.7 f13.2 f 10.4
f5.2 f12.0 f8.9 4.2 (70) ( 102) ( 102) (77)
(38 1) (387) (387) (154)
71 16 87 21 26 41 92 42 134
It is interesting that the recipients of FIS without a calculated entitlement receive an average benefit that is almost as large as that of recipients with a calculated entitlement. This contrasts with the case of HB, where recipients without acalculated entitlement receive a relatively low benefit, so that the failure to identify entitlement could easily arise from measurement error (of income or housing costs). This difference presumably arises from the fact that the size of FIS is fixed for a year. Thus we are observing people who may have experienced substantial changes in hours of work (19 of the 77 reported hours that were too low to qualify for FIS) and / or income.
111. THE MODELLING OF TAKE-UP
We follow Blundell, Fry and Walker by modelling take-up in terms of a balance between the (expected) financial value of the benefit and the costs (including stigma and hassle) of take-up. The complication here is that we are looking at the take-up of two benefits.
Take-Up of Benefits
This would not be much of a problem if the value of taking up one benefit were independent of the value of the other, and if the costs of taking up one benefit were independent of whether the other was being taken up. However, the values of the benefits are hardly ever independent of each other because FIS receipts are counted as income for the purposes of calculating entitlement to HB. Also, the costs of taking up a second benefit may well be less than those of taking up the first: the stigma might well be less and the family will be more experienced in dealing with the authorities. This means that instead of looking separately at the costs and benefits of the two take-up decisions, they must be modelled simultaneously.
We suppose that the attractiveness of receiving benefits is given by U(FIS, HB, z), where FIS is the amount of FIS received, HB is the amount of HB received, and z is a vector of family characteristics (including income).
The values of the two benefits have not been summed because the family may not regard them as equally attractive. For example, local authority tenants receive their housing benefit as a credit against their rent rather than in cash. This could be less attractive because they cannot spend it on other items while allowing their rent payments to fall into arrears.
It is assumed that the cost of claiming benefits is independent of the size of benefit expected. Therefore it can be represented by a function C(DF, DH, z), where DF has the value 1 if FIS is taken up and 0 otherwise, and DH has the value 1 if HB is taken up and 0 otherwise.
The net benefit will be given by the difference between these two functions:
V(FIS, HB, DF, DH, Z) = U(FIS, HB, Z) - C(DF, DH, z).
Note that, by definition, V(0, 0, 0, 0, z) = 0. We suppose a family will choose the take-up combination which gives the highest
net benefit. Thus, the family will choose to take up both benefits if
V(FISENT, HBENT1, 1, 1, z) > 0 and V(FISENT, HBENT1, 1, 1, z) > V(FISENT, 0, l,O, z) and V(FISENT, HBENT1, 1, 1, z) > V(0, HBENT2,O. 1, z)
where FISENT is the size of FIS entitlement; HBENTl is the size of HB entitlement under assumption (1) above (that FIS is received); and HBENT2 is the size of HB entitlement under assumption (2) above (that FIS is not taken up).
Similar conditions can be written down for the family to choose FIS only or to choose HB only. In each case, three inequalities have to be satisfied: ensuring that the net benefit is positive and that it is higher than the other two benefit-receiving possibilities.
In principle, it should be possible to estimate the net benefit function V(.) from observing the choices made by different families. However, this is difficult to do
because there are three conditions that have to be satisfied for any outcome to be chosen, and each condition could be subject to random error. We have therefore chosen a si...