33
Estimation Algorithm for Predicting the Performance of Private Apartment Buildings in Hong Kong by Yung YAU 1 * Daniel Chi-wing HO 2 Kwong-wing CHAU 2 Wai-yip LAU 2 1 Department of Public and Social Administration, City University of Hong Kong 2 Department of Real Estate and Construction, The University of Hong Kong Paper submitted to Structural Survey * Please send comments to: Dr. Yung Yau Assistant Professor Department of Public and Social Administration City University of Hong Kong 83 Tat Chee Avenue Kowloon Hong Kong Tel. No.: (852) 2788 8958 Fax No.: (852) 2788 8926 E-mail: [email protected]

Estimation algorithm for predicting the performance of private apartment buildings in Hong Kong

Embed Size (px)

Citation preview

Estimation Algorithm for Predicting the Performance

of Private Apartment Buildings in Hong Kong

by

Yung YAU 1*

Daniel Chi-wing HO 2

Kwong-wing CHAU 2

Wai-yip LAU 2

1 Department of Public and Social Administration, City University of Hong Kong

2 Department of Real Estate and Construction, The University of Hong Kong

Paper submitted to

Structural Survey

* Please send comments to:

Dr. Yung Yau

Assistant Professor

Department of Public and Social Administration

City University of Hong Kong

83 Tat Chee Avenue

Kowloon

Hong Kong

Tel. No.: (852) 2788 8958

Fax No.: (852) 2788 8926

E-mail: [email protected]

Estimation Algorithm for Predicti ng the Performance of Private

Apartment Buildings in Hong Kong

[Abstract]

Purpose – For the sake of public health and safety, a territory-wide evaluation of the

quality of buildings in Hong Kong is crucial. However, it is a lengthy process to

assess the performance of the whole stock of buildings in the city. To get around this

predicament, this study aims to propose a statistical approach for a fast and reliable

building evaluation algorithm using the Building Quality Index (BQI) developed by The

University of Hong Kong.

Design/methodology/approach – Using the BQI assessment framework, the conditions

of 133 and 160 private apartment buildings in Yau Tsim Mong and the Eastern District

respectively were assessed and rated. The data of the Yau Tsim Mong buildings were

used to estimate a regression model associating the relationships between building

performance, measured by the BQI, and other exogenous factors. The resulting model

was then employed to predict the performance of the surveyed buildings in the Eastern

District.

Findings – The regression analyses on the Yau Tsim Mong data indicated that building

age, development scale, building management mode were significant determinants of

the existing conditions of the sampled buildings, echoing with the findings in previous

studies. BQI scores of buildings in the Eastern District were estimated using the

resulting regression model, and we found that there was a highly positive relationship

between the predicted BQI and in-situ BQI scores.

Research limitations/implications – In spite of the high correlation between the

predicted BQI and in-situ BQI scores, the estimation algorithm may not be able to

generate reliable condition prediction at a building level. We opine that the estimation

algorithm would be more appropriate for applying to the street-block or even district

levels, rather than individual building level.

Practical implications – The findings of this study imply that a statistical approach can

be adopted to give a broad-brush indication of the performance of all buildings in the

city in a quick and inexpensive manner. Without the need for site visit and detailed

desk study, it takes minimal time and resources to obtain the BQI of each individual

apartment building using the proposed approach. This facilitates the resource

allocation for the urban renewal process.

Originality/value – Our study is the first in the literature to provide an algorithm for

estimating building conditions in a densely developed high-rise urban area.

Keywords – Building Quality Index, performance prediction, building dilapidation,

Hong Kong

Paper type – Research paper

1. Introduction

Due to its high development density, Hong Kong has been faced with great

challenges in urban planning. Being a tiny place of around 1,100 sq.km. in area, Hong

Kong accommodates approximately 6.8 million people (Census and Statistics

Department, 2007). In this light, high-rise and high-density development pattern is a

predestined choice for this concrete jungle. From the positive side, this urban setting

is an economically efficient approach because it maximizes the use of communal

facilities and services (Tong and Wong, 1997). In addition, the problems associated

with urban sprawl can be minimized and efficiency in the use of land can be enhanced

for the concerns of environmental friendliness and sustainability (Freeman, 1993). Yet,

such high-density urban setting potentially provokes health and safety risks to the

community because there is a close linkage between the built environment and people’s

state of health. Evidence for this linkage has been well documented in the literature

(Schmit, et al., 1978; Tanaka, et al., 1996).

As a matter of fact, density-driven problems in the urban areas of Hong Kong

are palpable. Buildings in dilapidation can be spotted across the territory, particularly

in those old districts. The lack of a comprehensive government policy on building

management and maintenance, together with the unawareness of building care among

building owners, attribute to the problems. These derelict buildings do not only affect

the well-beings of their occupants, but also pose threats on the health and safety of the

whole community. To a wider extent, this situation is conducive to the growth of

urban decay in the city. Perhaps, one of the most painful lessons learnt by the local

community is the outbreak of the Severe Acute Respiratory Syndrome (SARS) in 2003.

Owing to the poor drainage conditions in a few buildings, a local widespread of the

disease was first triggered. Later, the high-density living environment in Hong Kong

served as a catalyst for a community epidemic of the disease across the territory. In

the end, 299 lives were taken away in the plague of SARS. Apart from environmental

health and hygiene, personal safety is also at stake because of the dilapidated built

environment. For instance, falling building fabrics like aluminium windows and

concrete pieces have created numerous deaths and injuries over the past few years (Ho

and Yau, 2004). All these episodes have signified the gravity of the problems of

building dilapidation in Hong Kong.

In light of this urban crisis, it is essential to dig out those buildings which are

poorly-performing. In the opinions of the government and scholars, a building

classification system may work to serve this purpose (Chau, et al., 2004; Ho, 2004;

Housing, Planning and Lands Bureau, 2005). Nonetheless, the effects of the

classification system will not be apparent unless a vast majority of buildings have been

classified using the system so that their performance is comparable. It is envisaged

that this can take a long time and huge amount of resources to complete. Yet, we

cannot afford waiting further more to tackle the building problems in the city. In this

regard, a transitional approach which makes use of the statistical analysis to predict the

performance of buildings in Hong Kong is proposed in this paper.

This paper is organized as follows: the need for a territory-wide classification

scheme on building performance is discussed in the next section. It is then followed

by the introduction of the Building Quality Index (BQI) which is thought to be fit for

the territory-wide building classification. Afterwards, the difficulties encountered in

the assessment of all buildings in Hong Kong are presented and a statistical approach to

solve these difficulties transitionally is proposed. An empirical illustration is then used

to demonstrate how this approach works. Discussions and conclusions of this paper

come at the end.

2. The Need for a Territory-wide Building Performance Assessment

2.1 Proposals to Address Building Problems

Undoubtedly, the responsibility to keep a building in good conditions vests in

the owners of the building. However, non-cooperation among homeowners, inactive

participation and agency problems often hinder the management of buildings in Hong

Kong, leading to building disrepairs of various degrees consequentially (Walter and

Kent, 2000; Walter, 2002; Lai and Chan, 2004). Urban decay in the city is then

inevitably aggravated. In this light, selective incentives, which can be in form of

financial assistance or coercion, should be installed by the authority to tackle the

building management and maintenance problems (Bengtsson, 2000). Therefore, the

Hong Kong government should intervene and take some steps to arrest the urban decay.

In response to the public voice, the government have launched a series of campaigns to

ensure proper building upkeep and public consultations regarding the measures to

promote proper management and management of buildings in the territory since the late

1990s.

In fact, there are a number of possible choices available for the government.

For instance, building-related legislations can be enforced more stringently, say by

increasing the frequency of inspection cycle. Also, the redevelopment process in old

districts can be speeded up. Yet, the government is always subject to a very tight

budget constraint so a sensible allocation of the resources is required (Ho, et al., 2005c).

In other words, it seems that the government cannot take care of all the buildings in

Hong Kong at the same time. Priorities have to be accorded to the most problematic

buildings. To this end, it is necessary to know which buildings are good and which are

bad in their performance or quality.

2.2 Building Performance Assessment Is Not New

To distinguish the good from the bad concerning building quality, building

performance assessment is the most suitable resort. In order to facilitate the

benchmarking of the results of building performance assessment, building classification

tool is generally employed. In general, building classification is a protocol to rank

buildings into different classes according to the performance of various attributes of a

building with reference to certain pre-defined objectives (Ho, 2004).

In reality, building performance assessment or classification is not a novel

concept. Perhaps, the first building classification system was the Use and Condition of

Buildings developed in the UK in the 40s (Duncan, 1971). The system was used to

grade buildings into three classes according to their structural conditions. Later,

countless systems of building performance assessment have been developed and

adopted for various purposes. Well-known examples of these schemes include the

Leadership in Energy and Environmental Design and the Building Research

Establishment Environmental Assessment Method, which provide labelling for

buildings in the US and the UK respectively (Baldwin, et al., 1998; US Green Building

Council, 2000). In Hong Kong, the Built Environment Assessment Method has been

used to assess buildings since 1996 and the government commissioned a consultancy

study in 2001 to develop the Comprehensive Environmental Performance Assessment

Scheme (HK-BEAM Society, 2004; Hui, 2004).

Nevertheless, most of the building performance assessment systems developed

so far focus mainly on environmental issues such as energy saving and sustainable uses

of materials. Other essential aspects of building quality such as health and safety have

been largely ignored in these systems. This results in a niche for the development of

an assessment scheme for benchmarking buildings in respect of their health and safety

performance. The call for such a building classification system was first accorded in

Hong Kong in 2000 when the former Planning and Lands Bureau suggested classifying

private buildings in Hong Kong by their standards of safety, management and

maintenance (Ho and Yau, 2004). A task force was then set up to study the feasibility

of the proposal. Yet, no concrete implementation has been materialized so far even

though the task force considered a voluntary building classification scheme viable and

valuable for the situation of Hong Kong (Ho and Yau, 2004).

2.3 Benefits of Implementing Building Classification

As a matter of fact, implementation of building classification in Hong Kong

can bring about a lot of benefits. To reiterate, a building classification system aims to

distinguish the well-performing buildings from the poorly-performing ones. It is

commonly believed that a well-publicized and well-received classification scheme will

benefit all parties who are interested in how well a building performs, such as building

owners, potential property purchasers, real estate agents, surveyors, property managers,

mortgagees, insurers and the government (Chau, et al., 2004; Ho and Yau, 2004).

Based on the information provided by the classification system, these parties can make

informed decisions on what actions they should take.

The release of such information about building performance to the public can

ease, if not totally eliminate, the problem of information asymmetry in the second-hand

housing market.1 In addition, the building information so provided can help change

the deeply-rooted mind of people that aged buildings are derelict buildings. Up to

1 As put forward by Akerlof (1970), information asymmetries happen when the sellers know more about the quality or performance of a good than the purchasers do. Chinloy (1978) suggested that the asymmetric information would result in adverse selection in the housing market.

now, an unrealistically high emphasis has been placed on building age in government

policy formulation and property valuation (Wong, et al., 2005). Take the recently

proposed mandatory building inspection as an example. It was suggested by the

government that buildings that are over 30 years old need to be inspected once every ten

years (Housing, Planning and Lands Bureau, 2007). This situation is largely attributed

to the unavailability of a comprehensive picture of the performance profile of all

buildings in Hong Kong.

Besides, building performance assessment helps motivate building owners to

make efforts to improve their buildings. With the assessment results widely publicized,

a higher property values will be associated with those better performing buildings due to

labelling effects (Chau, et al., 2004). This means positive recognition is awarded to

well-managed and maintained buildings by means of value enhancement. In addition,

well-maintained buildings may attract more favourable mortgage terms (Chau, et al.,

2004). If a reassessment mechanism is in place, owners of buildings with an inferior

grading can carry out improvement works for their buildings with a view to potential

economic benefits. It is thus expected that territory-wide implementation of the

building classification or performance assessment can bring market forces into play to

encourage owners to shoulder their responsibilities in building management and

maintenance.

The ultimate goal of the building classification is to foster a culture of building

care by market forces which are essential for countering the depreciation of our building

stock in the city. As suggested by Yiu (2007), timely maintenance of buildings is

indispensible for deferring building depreciation, with a long-term view in maintaining

the economic sustainability of the building stock. On the other hand, due to the

inherent asymmetric information problem associated with the building quality issues,

pre-mature maintenance or over-maintenance may be resulted because property owners

may signal the quality of their buildings through maintenance and repair works

(Ben-Shahar, 2004). Therefore, building information provided by some classification

scheme is conducive to timely building maintenance and repair, reducing unnecessary

wastages.

3. Building Performance Assessment Using BQI

Although there are plenty of benefits associated with the implementation of the

territory-wide building performance assessment, no suitable assessment systems in the

market are suitable for the purpose. The reasons for the unsuitability are two-folds.

First, as aforementioned, most of the existing systems are tailored for environmental

performance assessment, and not for building health and safety evaluation. Second,

the assessment methods (e.g. computerized simulation and laboratory testing) adopted

in these systems are rather detailed and complicated, making the assessments costly and

time-consuming. In other words, it is not favourable to use these systems for the mass

assessment of building performance within a short period of time and at a reasonably

low cost.

3.1 An Overview of the Building Quality Index

To straddle this gap, the Faculty of Architecture of The University of Hong

Kong has been developing a building classification tool called the Building Quality

Index (BQI) since May 2003 (Ho, et al., 2005b). The classification scheme is

essentially a benchmarking tool which grades buildings with reference to the

performance of the buildings on certain pre-defined objectives. Since the performance

of a building can be assessed from various perspectives, the BQI is composed of

different modules to suit various assessment objectives. So far, there are two modules

or sub-indices developed in the BQI scheme, as shown in Figure 1. They are the

Building Health and Hygiene Index (BHHI) and the Building Safety and Conditions

Index (BSCI).

[Figure 1 about here]

Basically, the BHHI measures the level of achievement of individual buildings

in safeguarding occupants against physical and mental health risks, like infectious

diseases and chronic illnesses (Ho, et al., 2004). Likewise, the BSCI gauges the level

of achievement of individual buildings in safeguarding occupants and the public against

the risk of physical injury and death, like fire and falling objects (Ho and Yau, 2004).

The indices are building-specific, enabling the comparison of health and safety

performance across buildings. Four principles underpin the building assessment

framework (Ho, et al., 2004). First, the factors to be assessed should be general

enough to cater to most variations in building designs and management practices

available contemporarily or in the near future. Second, the assessment factors should,

as far as possible, be measurable and verifiable. The assessment process should require

minimum subjective judgments by the assessors to maintain a high degree of

consistency and impartiality of the assessment results. Third, the assessment process

should be practical, and thus information for assessing buildings should be easily

acquired by inexpensive means. Last, the factors to be assessed should be relevant to

the assessment objectives. This generally requires a clear definition of the objectives,

as well as strong theoretical and empirical backings from the literature.

Based on the above principles, a list of building factors that fit the institutional

and cultural settings of apartment buildings in Hong Kong was identified for the

development of the BHHI and BSCI assessment schemes. Then, an advisory

committee comprising representatives from various public organizations (including the

government), academic institutions, and professional institutes reviewed the factors to

see whether they are suitable for the BHHI and BSCI assessment frameworks. Those

building factors considered suitable were subsequently arranged into hierarchies of

factors in line with the objectives of the sub-indices.2 In both the BHHI and BSCI

factor hierarchies, briefly speaking, there are two main groups of factors, namely

Design and Management, as shown in Figure 2. The first group represents the

‘hardware’ of a building such as design features and disposition of the building. It

embraces numerous building factors that are relatively difficult to alter, no matter in

technical or economical sense, once the building has been put into occupation (Ho,

2004). Conversely, the ‘software’ of a building is symbolized by the second group of

building factors which mainly concern the daily operations, management and state of

repair of the building. These building factors are seldom static but relatively easier to

change even after a building is occupied (Ho and Yau, 2004). This dichotomy splits

the building factors into two groups that are respectively within and beyond the control

of building owners, facilitating the owners to make more informed decisions on the

actions that should be taken to improve their buildings’ performance in a cost-effective

manner (Ho, et al., 2004).

[Figure 2 about here]

During the assessment of a building, each of the building factors is rated in

accordance with the respective predetermined rating scale (Ho, et al., 2004; Ho and

Yau, 2004). An example is shown in Table 1. Rating scales were set with reference

to the minimum standards stipulated in the current regulations and the best practices in

2 For details, please refer to Ho et al. (2004) and Ho and Yau (2004).

industry. For each building factor, the maximum score awarded is 100 (for the best

scenario) while zero is the minimum (for the worst scenario). Rating using these

scales enhances the consistency of the assessors’ inputs, maintaining the creditability of

the assessment scheme.

[Table 1 about here]

Since numerous building factors are incorporated in the assessment

frameworks, it may not be manageable by the general public to compare the

performance of buildings using the ratings of all these factors separately. For the ease

of consumption, the ratings of the building factors are aggregated into two indices,

namely the BHHI and BSCI. In principle, these indices are essentially aggregate

figures of the ratings (F) and weightings (w) of all building factors that are related to the

health and safety, respectively, of an apartment building. Mathematically:

hhwFBHHI (1)

sswFBSCI (2)

To arrive at the BHHI and BSCI, what we need are the ratings and weightings of

the building factors, as suggested by Equations (1) and (2). The weightings or relative

importance of the building factors for the BQI scheme were determined using the

Analytic Hierarchy Process (AHP) developed by Saaty (1980). In fact, there are a

number of other methods to obtain factor weightings, such as direct weight assignment,

conjoint analysis and multi-attribute utility technique. Yet, these methods are either

too complicated to implement or too vulnerable to theoretical challenges. To strike a

balance between the practicability and academic vigour of the weighting determination

process, the AHP was used (Ho, et al., 2004; 2005a). Panels of experts were invited to

a series of workshops to provide their professional inputs for the weighting

determination.

3.2 Procedures in BQI Assessment

While sets of factor weightings are ready for the BHHI and BSCI computation,

what are left behind are the factor ratings. The building factors are assessed through

four main processes under the BQI scheme (Ho, et al., 2005a; Wong, et al., 2006).

What comes first is the desk study which aims to provide the background and also the

detailed design information of the buildings under investigation. From the websites of

various government departments, the basic information (e.g. age and number of storeys)

of the buildings can be obtained. Also, the approved building plans deposited in the

Buildings Department are studied. However, discrepancies between the situation on

site and the information indicated on plans are common (Hollis, 2000). To verify the

actual health and safety conditions on site, visits to the buildings are indispensable. As

a matter of practicability, all factors to be inspected on site are confined to the common

areas (e.g. podiums, lobbies, staircases and corridors) and the immediate external

environment of the buildings (Wong, et al., 2006). It is because, owing to their

co-ownership nature, these common areas are usually the worst in terms of management

and maintenance. In contrast, it is an owners’ matter to take care of the conditions of

his or her own dwelling unit.

During the site inspections, building management staff and/or residents are

interviewed to acquire information of the management practices of the buildings (e.g.

frequencies of refuse disposal and fire drill). If needed, the interviewees were

requested to provide documentary records for verification. Finally, the data obtained

from the above processes were consolidated for computing the BHHI and BSCI.

Following these procedures, over three hundred apartment buildings in Hong Kong

were assessed in 2004 and 2005. Figures 3-6 show some photos taken in the

assessments.

[Figure 3 about here]

[Figure 4 about here]

[Figure 5 about here]

[Figure 6 about here]

4. Statistical Approach to Building Performance Assessment

4.1 Difficulties Encountered in the Territory-wide Building Classification

As aforementioned, the implementation of the BQI scheme, being a building

classification system, can create a labelling effect to motivate the voluntary building

upkeep in Hong Kong. However, this can only be achieved if a critical mass of

buildings has been assessed using the scheme. Therefore, it is a paradox. Although

the BQI assessment scheme is tailored for an inexpensive and quick mass evaluation of

building performance in Hong Kong and it is adaptable for assessing commercial and

industrial buildings, it is still impossible to have some 40,000 private buildings in the

territory assessed overnight. In this regard, there must be some approaches to

jumpstart the proposal of territory-wide building performance assessment.

4.2 Statistical Approach as a Resort

As an interim measure, we propose to adopt a statistical approach to unriddle the

current hitch. This proposed method is founded on a theoretical framework which is

graphically illustrated in Figure 7. The health and safety performance of a building is

defined by intrinsic building factors which build up the BQI assessment scheme. The

building is then rated with respect to these intrinsic building factors under the BQI

scheme, and an overall BQI score is allotted to the building after the full assessment.

There exist, nevertheless, some other factors that are not directly related to the health

and safety of a building but have effects on the performance of the building. For

example, property management companies assist building owners in managing their

buildings so their presence should be associated with better building performance.

Using these exogenous factors, the BHHI and BSCI of a building can be estimated

without the need of actual assessment of the building.

[Figure 7 about here]

4.3 Statistical Model Explaining Building Performance

From the literature, a number of exogenous factors affecting the conditions of a

building were identified. These determinants include building age, development scale

and building management mode such as the formation of incorporated owners and the

engagement of property management agent (Werczberger and Ginsberg, 1987; Ho, et

al., 2006; Wong, et al., 2006). Founded on these findings, the health and safety

performance of an apartment building (BP), measured by the BQI, can be expressed as a

function (f):

),,,,( PMIOPMIOUNITAGEfBP (3)

where AGE denotes the age of the building, measured in years; UNIT denotes the

number of dwelling units in the building; PMIO is a dummy which equals one if an

incorporated owners and a property management company exist for the building, and

zero if otherwise; IO is a dummy which equals one if the building has an incorporated

owners (IO) but no property management company (PMC), and zero if otherwise; and

PM is a dummy which equals one if the building has a property management company

but no incorporated owners, and zero if otherwise.

The functional form of Equation (3) is not known a priori, and thus it is taken as

a linear function for the sake of simplicity. Then we have

PMIOPMIOUNITAGEBP 543210 (4)

where αi (for i = 0,1,2,…,5) are the parameters to be estimated and ε is the stochastic

term. The BQI of a building is taken as the average of BHHI and BSCI computed for

the building, and it ranges from 0 to 100. Using the assessment results of a small

sample of buildings, the parameters in the regression model in Equation (4) can be

estimated. The estimated model can then be employed to predict the performance of

other buildings.

4.4 Pilot Study of the Statistical Approach

A pilot study using the empirical data was conducted to illustrate the usefulness

of the proposal statistical approach. This was achieved with the availability of the

building records obtained in the BQI surveys regularly carried out by the Faculty of

Architecture of The University of Hong Kong. With the application of the BQI

scheme, 133 apartment buildings in Yau Tsim Mong and 160 buildings in the Eastern

District were assessed in 2004 and 2005 respectively. The descriptive statistics of the

characteristics of these assessed buildings are summarized by district in Tables 2-4.

[Table 2 about here]

[Table 3 about here]

[Table 4 about here]

The data of the 133 assessed buildings in Yau Tsim Mong were first fitted into

Eq.(3) for estimation by way of Ordinary Least Square (OLS) regression. The

adjusted R-squared for this regression was 0.77 and all estimated coefficients were

found to be statistically significant at the 1 percent level. In other words, the five

independent variables incorporated in the regression model had significant relationship

with building performance, measured by the BQI. About 77 percents of the BQI

variation could be explained by the variations in these exogenous variables. As a

result, the OLS regression analysis returned the following estimated equation:

PMIOPMIO

UNITAGEPB

58.645.589.8

02.029.073.53ˆ

(5)

What comes next was to predict the performance of the 160 assessed buildings

in the Eastern District using Equation (5) and compare the predicted BQI and in-situ

BQI scores (i.e. BQI scores obtained through actual building performance assessments).

The predicted BQI scores for the assessed buildings in the Eastern District ranged from

44.34 to 69.40, with a mean of 57.16 and standard deviation of 8.31. The in-situ BQI

score was then regressed against the predicted BQI score using the OLS technique, the

estimated coefficient was found to be 0.95 and statistically significant at the 1 percent

level. The adjusted R-squared for the regression was 0.98.

5. Discussion of the Findings

The findings of the pilot study suggest that there was a highly positive

correlation between the in-situ BQI and predicted BQI scores. The latter can be

regarded as a good proxy for the former. However, due care should be taken when

interpreting the prediction results. In spite of the high R-squared and close-to-unity

coefficient returned from the regression analysis, whether this proposed statistical

approach recommends reliable performance comparison on a building-by-building basis

remains questionable. If the two BQI scores for each individual building were

contrasted, we found that the difference ranged from 0.02 to 20.63, with a mean of 5.01

and standard deviation of 3.73. Based on these findings, we suggest that

benchmarking of building performance using predicted BQI scores should not be made

on a building-by-building basis. Instead, building performance comparisons should be

undertaken only on a more macroscopic level, such as a street-block or even a district

level. It is because statistic error can be reduced when the sample size (i.e. the number

of buildings involved) is larger.

Nevertheless, this limitation does not render the statistical approach useless in

Hong Kong. In fact, similar statistical approaches have been adopted elsewhere to

study the built environment quality. For example, a prediction model was developed

in the US to find out the properties which were at the greatest risk of abandonment

(Hiller, et al., 2003). An extremely high degree of accuracy is not essential for such a

model but the prediction results should be reliable when applied to a large area.

By the same token, the proposed statistical approach provides a broad-brush

prediction of the health and safety performance of buildings, serving as an early

warning system against urban decay on a district basis. This prediction exercise can

provide the government with performance profiles of buildings in different districts so

the government prioritize action areas and better allocate public resources to community

building management, at a relatively low cost. For instance, street-blocks with

comparatively high concentrations of poorly-performing buildings, as indicated by the

predicted BQI scores, should be accorded more attention. The buildings there should

be assessed with priority to confirm the situation. The findings from these actual

assessments can then serve as extra inputs for re-running the explanatory model in

Equation (4). By reiterating these processes, more and more buildings can be actually

assessed, starting from the most problematic areas.

The BQI prediction process can be further facilitated through the application of

geographic information system (GIS). With the GIS, not only the results of the

building performance prediction can be shown graphically on maps, but also analyses

on the clustering phenomenon can be done. This may be constructive for the further

exploration of the factors determining urban decay in Hong Kong.

6. Conclusions

The extremely high-density high-rise development pattern in Hong Kong lays

down a close linkage between the quality of the built environment and the well-beings

of the community. Nonetheless, the prolonged neglect of building care among the

building owners has resulted in a serious problem of urban decay and posed threats to

the public. Given the tight budget constraint of the government, it is practically

impossible to work on all the buildings in the city simultaneously. Identification of the

most problematic buildings in the concrete jungle can definitely help ease, if not

completely eliminate, the predicament at this moment. For that reason, a system for

classifying buildings in Hong Kong should be put in place.

In view of this niche, The University of Hong Kong has carried out a project to

develop the BQI since 2003. Using the BQI scheme, the health and safety

performance of private apartment buildings in the territory can be assessed and

compared. The uses of this assessment tool are versatile. Although the BQI

assessment framework is tailored for a first-tier screening of building performance, it

still takes a very long time and consumes a lot of resources to complete the assessment

of the whole building stock in Hong Kong. In view of the urgent need of urban

renewal, a short-cut approach to the mass evaluation of building performance is

necessary.

Based on the findings from previous empirical studies, a statistical approach of

building performance assessment is proposed in this paper. It serves as an alternative

to having all buildings in Hong Kong actually assessed. Using the estimation results

of a regression model which associates the relationships between building age,

development scale, building management mode, and the health and safety performance

of the assessed apartment buildings, the performance of other non-assessed buildings

can be predicted.

A pilot study on the building assessment results of 133 and 160 apartment

buildings located in Yau Tsim Mong and the Eastern District respectively was carried

out. The data of the Yau Tsim Mong buildings was used to estimate the regression

model and the resulted equation was then employed to predict the performance of the

buildings, measured by the BQI, in the Eastern District. We found that there was a

highly positive relationship between the predicted BQI and in-situ BQI scores. These

findings indicate that the statistical approach can be used to predict the performance of

other buildings in Hong Kong quite accurately though we opine that the estimation

algorithm would be more appropriate for applying to the street-block or even district

levels, rather than individual building level.

Without the need for site visit and detailed desk study, it takes minimal time and

resources to obtain the BQI of each individual apartment building using the proposed

approach. A broad-brush indication of the health and safety performance of all

buildings in Hong Kong can be obtained in a quick and inexpensive manner. This

facilitates the resource allocation for the urban renewal process. For example, it helps

the interested organizations prioritize the stock of buildings for redevelopment and

rehabilitation.

Acknowledgements

We gratefully acknowledge the financial support provided by the Research

Grant Council of the Hong Kong Special Administrative Region (HKU 7107/04E and

HKU 7131/05E), the Small Project Funding of The University of Hong Kong and the

HKU Research Group on Sustainable Cities Seed Grant. We would also to

acknowledge the Buildings Department and Home Affairs Department of the

Government of the Hong Kong Special Administrative Region for their kindly provision

of information and support for the study. Last but not least, we also thank the

insightful comments made by the participants of the World Sustainable Building

Conference held in Hong Kong in December 2007.

References

Akerlof, G.A. (1970) “The market for lemons: quality uncertainty and the market

mechanism”, Quarterly Journal of Economics, Vol. 84 No. 3, pp. 488-500.

Baldwin, R., Yates, A., Howard, N. and Rao, S. (1998) BREEAM 98 for Offices: An

Environmental Assessment Method for Office Buildings, Building Research

Establishment, Garston.

Ben-Shahar, D. (2004) “Productive signaling equilibria and over-maintenance: an

application to real estate markets”, The Journal of Real Estate Finance and

Economics, Vol. 28 No. 2/3, pp. 255-271.

Bengtsson, B. (2000) “Solving the tenants? Dilemma: Collective action and norms of

co-operation in housing”, Housing, Theory and Society, Vol. 17 No. 4, pp.

175-187.

Chau, K.W., Ho, D.C.W., Leung, H.F., Wong, S.K. and Cheung, A.K.C. (2004)

“Improving the living environment in Hong Kong through the use of a building

classification system”, CIOB(HK) Quarterly Journal, October, pp. 14-15.

Census and Statistics Department (2007) Hong Kong Annual Digest of Statistics,

Census and Statistics Department, Hong Kong.

Chinloy, P. (1978) “Depreciation, adverse selection and housing markets”, Journal of

Urban Economics, Vol. 5 No. 2, pp. 172-187.

Duncan, T.L.C. (1971) Measuring Housing Quality – A Study of Methods, Occasional

Paper No. 20, Centre for Urban and Regional Studies, The University of

Birmingham.

Freeman, H. (1993) “Mental health and high-rise housing”, Burridge, R. and Ormandy,

D. (eds.) Unhealthy Housing: Research, Remedies and Reform, E & FN Spon,

London, pp. 168-190.

Hiller, A.E., Culhane, D.P., Smith, T.E. and Tomlin, C.D. (2003) “Predicting housing

abandonment with the Philadelphia Neighbourhood Information System”,

Journal of Urban Affairs, Vol. 25 No. 1, pp. 91-105.

HK-BEAM Society (2004) Hong Kong Building Environmental Assessment Method –

Existing Buildings, HK-BEAM Society, Hong Kong.

Ho, D.C.W. (2004) “Current research on building classification/labelling system”,

Proceedings of the Symposium on Green Building Labelling, Hong Kong, 19

March 2004, pp. 61-68.

Ho, D.C.W., Chau, K.W., Cheung, A.K.C., Yau, Y., Wong, S.K., Leung, H.F., Lau,

S.S.Y. and Wong, W.S. (2008) “A survey of the health and safety conditions of

apartment buildings in Hong Kong”, Building and Environment, Vol. 43 No. 5,

pp. 764-775.

Ho, D.C.W., Chau, K.W., Wong, S.K. Yau, Y. and Cheung, A.K.C. (2005a) “The

Building Quality Index – a tool for building classification”, Proceedings of the

CII-HK Conference 2005 on Healthy Building, Hong Kong, 30 November 2005,

pp. 37-45.

Ho, D.C.W., Chau, K.W., Yau, Y., Cheung, A.K.C. and Wong, S.K. (2005b)

“Comparative study of building performance assessment schemes in Hong

Kong”, The Hong Kong Surveyor, Vol. 16 No. 1, pp. 47-58.

Ho, D.C.W., Leung, H.F., Wong, S.K., Cheung, A.K.C., Lau, S.S.Y., Wong, W.S.,

Lung, D.P.Y. and Chau, K.W. (2004) “Assessing the health and hygiene

performance of apartment buildings”, Facilities, Vol. 22 No. 3/4, pp. 58-69.

Ho. D.C.W., Then, D.S.S. and Yau, Y. (2005c) “Facilitation of urban renewal with

Building Safety and Conditions Index”, Proceedings of the CIB Combining

Forces – Advancing Facilities Management and Construction through

Innovation. Volume on Facilities Business and its Management, 13-16 June,

Helsinki, pp. 475-486.

Ho, D.C.W. and Yau, Y. (2004) “Building safety & condition index: benchmarking tool

for maintenance managers”, Proceedings of the CIB W70 Facilities

Management and Maintenance Symposium 2004, Hong Kong, 7-8 December

2004 in Hong Kong, pp. 49-155.

Ho, D.C.W., Yau, Y., Wong, S.K., Cheung, A.K.C., Chau, K.W. and Leung, H.F.

(2006) “Effects of building management regimes of private apartment buildings

in Hong Kong”, Property Management, Vol. 24 No. 3, pp. 309-321.

Hollis, M. (2000) Surveying Buildings, RICS Books, Coventry.Housing, Planning and

Lands Bureau (2005) Building Management and Maintenance – Public

Consultation on Mandatory Building Inspection, Housing, Planning and Lands

Bureau, Hong Kong.

Housing, Planning and Lands Bureau (2007) Report of the Public Consultation on

Mandatory Building Inspection, Housing, Planning and Lands Bureau, Hong

Kong.

Hui, M.F. (2004) “Comprehensive Environmental Performance Assessment Scheme”,

in Proceedings of Symposium on Green Building Labelling, 19 March, Hong

Kong, pp. 54-60.

Lai, L.W.C. and Chan, P.Y.L. (2004) “The formation of owners' corporations in Hong

Kong's private housing estates: A probit evaluation of Mancur Olson's group

theory”, Property Management, Vol. 22 No. 1, pp. 55-68.

Saaty, T.L. (1980) The Analytical Hierarchy Process, McGraw-Hill, New York.

Schmitt, R.C., Zane, L.Y. and Nishi, S. (1978) “Density, health, and social

disorganization revisited”, Journal of American Institute of Planners, Vol. 44,

pp. 209-221.

Tanaka, A., Takano, T., Nakamura, K. and Etkeuchi, S. (1996) “Health levels

influenced by urban residential conditions in a megacity – Tokyo”, Urban

Studies, Vol. 33 No. 6, pp. 879-894.

Tong, C.O. and Wong, S.C. (1997) “The advantage of a high density, mixed land use,

linear urban development”, Transportation, Vol. 24 No. 3, pp. 295-307.

US Green Building Council (2000) Green Building Rating System™ Version 2.0 –

Leadership in Energy and Environmental Design, US Green Building Council,

Washington, D.C.

Walters, M. (2002) “Transaction costs of collective action in Hong Kong high rise real

estate”, International Journal of Social Economics, Vol. 29 No. 4, pp. 299-314.

Walters, M. and Kent, P. (2000) “Institutional economics and property strata title – a

survey and case study”, Journal of Property Research, Vol. 17 No. 3, pp.

221-240.

Werczberger, E. and Ginsberg, Y. (1987) “Maintenance of shared property in

low-income condominiums”, Housing Studies, Vol. 2 No.3, pp. 192-202.

Wong, S.K., Cheung, A.K.C., Yau, Y., Chau, K.W. and Ho, D.C.W. (2005)

“Distinguishing the decrepit from the old: Is building age a good proxy of

building performance?”, in Cheung, Y.K. and Chau, K.W. (eds.) Tall Buildings

– From Engineering to Sustainability, World Scientific Publishing, Singapore,

pp. 687-692.

Wong, S.K., Cheung, A.K.C., Yau, Y., Ho, D.C.W. and Chau, K.W. (2006) “Are our

residential buildings healthy and safe? A survey in Hong Kong”, Structural

Survey, Vol. 24 No. 1, pp. 77-86.

Yiu, C.Y. (2007) “Building depreciation and sustainable development”, Journal of

Building Appraisal, Vol. 3 No. 2, pp. 97-101.

Table 1: Rating scale for the conditions of cantilevered structures

Grade Rating Description

Satisfactory 1 No apparent defects in the cantilevered structures are

observed; and no misuse of the cantilevered structures

Above

average

0.75 Infrequent spalling and/or cracks at the slab soffits of

cantilevered structures

Acceptable 0.5 Frequent spalling and/or cracks at the slab soffits of

cantilevered structures; or infrequent spalling and/or cracks

in the tension zones of the cantilevered structures

Deficient 0.25 Frequent spalling and/or cracks in the tension zones of

cantilevered structures

Poor 0 Undue deflection in any cantilevered structure; or spalling

and/or cracks and/or spalling along the supporting edge of

any cantilevered structure; or any misuse of a cantilevered

structure observed

Table 2: Summary statistics of the characteristics of the assessed buildings in Yau

Tsim Mong

Characteristics Maximum Minimum Mean Standard Deviation

Age (Year) 51 4 32.0 1.025

Number of units 420 3 57.3 70.78

BQI 69.7 35.7 50.8 8.478

Table 3: Summary statistics of the characteristics of the assessed buildings in the

Eastern District

Characteristics Maximum Minimum Mean Standard Deviation

Age (Year) 59 3 32.7 1.007

Number of units 440 4 62.0 70.92

BQI 74.5 37.3 54.1 8.540

Table 4: Summary statistics of the management mode of the assessed buildings in the

two districts

Building with Number of Buildings (% of the Sample)

Yau Tsim Mong Eastern District

Both IO and PMC 51 (38.4%) 52 (32.5%)

IO but no PMC 29 (21.8%) 56 (35.0%)

PMC but no IO 12 (9.0%) 13 (8.1%)

No IO and PMC 41 (30.8%) 39 (24.4%)

Building Health & Hygiene Index

(BHHI)

Building Safety & Conditions Index

(BSCI)

Othermodules

Figure 1: The composition of the BQI (Ho, et al., 2005)

Level 1 Level 2 Level 3 Weight* Category Weight* Building factor Weight* Design 53.6% Architecture 18.5% Size 2.5% Plan Shape 3.5% Headroom 2.0% Windows 5.7% Noise Reduction 3.4% Open space 1.4% Building 19.3% Water supply 5.6% services Drainage 6.8% Refuse Disposal 4.7% Lift 2.2% External 15.8% Density 1.9% environment Adjacent use 1.7% BHHI Air Quality 5.2% Aural Quality 2.6% Visual obstruction 1.6% Thermal comfort 2.8% Management 46.4% Operations & 27.1% Cleaning 5.1% maintenance Pest control 3.1% Refuse handling 4.6% Drainage condition 4.6% Unauthorized alteration 4.0% Water quality 5.7% Management 19.3% Owners’ duties 7.9% approaches Documentation 6.8% Emergency preparedness 4.6% Design 47.0% Architecture 22.1% Height and deposition 3.8% Means of escape 9.3% Means of access 6.3% Amenities 2.7% Building 16.6% Fire service installations 8.3% services Electrical installations 4.3% Fuel supply 4.0% External 8.2% Proximity to special hazards 6.4%

BSCI environment Proximity to fire station 1.8%

Management 53.0% Operations & 33.5% Structural condition 8.6% maintenance Building services condition 5.3% Exit routes condition 8.4% Fire compartmentation 4.3% Illegal appendages 6.9% Management 19.5% Owners’ duties 4.3% approaches Documentation 3.5% Emergency preparedness 7.8% Financial arrangement 3.9%

Note: *Weightings were obtained from an expert panel in the form of a workshop using Saaty’s (1980)

Analytic Hierarchy Process (AHP). Details of the workshop were reported in Ho, et al. (2004).

Figure 2 Building factors assessed under the BHHI and BSCI, and their relative

weightings (Ho, et al., 2008, pp. 767)

Figure 3: Debris in the staircase of a building in Yau Tsim Mong assessed in 2004

Figure 4: Vegetation in the drainage system of a building in Yau Tsim Mong assessed

in 2004

Figure 5: Unhygienic conditions in the rear lane of a building in Yau Tsim Mong

assessed in 2004

Figure 6: Exposed reinforcement in a building in the Eastern District assessed in

2005

Figure 7: The conceptual framework of the statistical approach of building

performance assessment

Intrinsic Building Factors

(e.g. hygienic conditions

and structural integrity)

determine Exogenous Building Factors

(e.g. building age and mode

of building management)

define

Conditions of

a Building

evaluate

estimate

benchmark