Professional Documents
Culture Documents
David K. Tse1
and
Nan Zhou1
Restoring demand for commodity housing is crucial for the success of housing
reform in previously centrally planned economies; yet weaning workers away from
publicly subsidized housing is a difficult task. Taking advantage of a unique survey,
we show that the intention to buy commodity housing by Chinese urban workers is
sensitive to various incentives, including housing mismatch, liquidity constraints,
risk attitudes, access to publicly subsidized housing, and commodity housing prices.
Our probit estimates indicate that the access to publicly subsidized housing is at
least as important as the affordability of commodity housing in discouraging private
home ownership. 2000 Academic Press
Key Words: housing choice; housing reform; economic transition; China.
I. INTRODUCTION
In a centrally planned economy, as China was until recently, housing was
not a commodity, but merely as a social obligation to be met by adminis-
trative organizations, state enterprises or municipal governments Re-
1
We thank the seminar participants at the University of Southern California and two
anonymous referees for helpful comments. Zhou and Tse acknowledge the financial support
provided by the Research Grant Council CERG CityU 1176r97H.. Xiao Ming provided
excellent reasearch assistance. All authors contributed equally.
61
0094-1190r00 $35.00
Copyright 2000 by Academic Press
All rights of reproduction in any form reserved.
62 FU, TSE, AND ZHOU
naud w18, pp. 12471248x.. Before the current housing reform, urban
households in China had little choice in their housing, which was provided
by the state for a nominal charge. Housing reforms started in 1979 and
were initially aimed at alleviating the extreme shortage of housing by
decentralizing housing investment decisions and funding. Local experi-
ments in housing privatization were encouraged since the mid-1980s as
fiscal expenditure on urban housing subsidies soared from 2.03% of GDP
in 1980 to 4.21% in 1988. with the rapid expansion in publicly subsidized
housing built by urban work units World Bank w22x; Wong, Heady, and
Woo w23x.. In addition to gradual adjustment in public housing rents and
selling of public housing stocks, commodity housinghomes built for sale
at market priceshas also been promoted as part of the privatization.
Although significant progress has been made in privatizing housing re-
sponsibilities, weaning urban workers away from publicly subsidized hous-
ing remains a major challenge for China Yatsko w24x..
Whereas Chinese housing reforms have followed gradual steps in transi-
tion to a housing market, scholars like Buckley, Hendershott, and Villani
w4x consider rapid housing privatization to be essential for restoring con-
sumer demand in a housing market. The Chinese experience, however,
offers a valuable opportunity to study consumer behavior in a housing
market when a socialist housing legacy remains. The purpose of this paper
is to present an empirical study of housing choice behavior of urban
workers in Chinas transition to a housing market. In particular, we show
supporting evidence that the continuation of publicly subsidized housing
holds back the demand for commodity housing. We do so with the benefit
of a unique data set collected from a recent consumer survey carried out
at factories in a number of major Chinese cities. The data allow us to study
individual workers intentions to buy commodity housing conditional on
their personal and family circumstances that may affect their housing
choices, as well as on their access to subsidized housing.
There have been many studies on various aspects of Chinese housing
reforms. Among the studies based on survey data, Fleisher, Yin, and Hills
w8x show a lack of labor compensation differential in relation to the
housing cost born by Chinese workers. Pudney and Wang w17x examine the
distribution of housing resources, using the household survey data from
two Chinese provinces collected between 1986 and 1990; they find a
positive relationship between the access to housing resources and house-
hold income. Zax w25x studies the relationships between family characteris-
tics and dwelling sizes in the household-built and state-owned sectors and
finds that the relationship differs dramatically between the two sectors.
Lim and Lee w13x and Chen and Coulson w5x both use aggregate data and
find housing expenditure to be positively related to regional income level;
HOUSING CHOICE AND CHINAS TRANSITION 63
2
Tolley w20x reports on the challenges in introducing market housing in Chinese cities.
64 FU, TSE, AND ZHOU
With respect to housing investment, urban work units were given more
discretion to allocate their capital budget and to raise extra capital funding
from their employees for housing development. Small work units were able
to pool their funds and participate in housing projects organized by
municipal governments. In addition, private housing construction was also
allowed in cities. The share of housing investment financed by the state
budget thus fell sharply, from over 90% before reforms to 16% in 1988.
State-owned enterprises financed 52% of housing investment in 1988,
urban collective enterprises and local governments financed 6% each, and
private sources financed the remaining 20% World Bank w22x.. The
decentralization of housing investment quickly led to a housing construc-
tion boom. Housing investment as a proportion of GNP rose from an
average of 1.5% before reforms to over 7% during the 1980s World Bank
w22x.. As a result, per capita housing floor space rose to 5.2 square meters
by 1985 and to 8.5 square meters by 1996 State Statistical Bureau w19x..
In 1988, the Chinese constitution endorsed private property rights in
urban land, and long-term land leases were granted for private real estate
development. The real estate industry has grown rapidly since then. In
1996, sales of buildings in real estate developments amounted to 145
billion yuan, or 5.9% of total retail sales China Economic Yearbook w6x..
Housing accounted for 86% of the building floor area sold.
Table 1 lists the urban housing completion in work-unit-funded housing
projects and commodity housing sales from real estate development. Sales
of commodity housing columns 7 and 8. accounted for no more than 21%
of total annual housing completion in cities column 1. before 1993 and
rose to around 25% in recent years. Urban work units, however, purchased
most of the commodity housing sold, although individuals share of the
commodity housing market columns 9 and 10. increased from 14% in
1986 to slightly over 50% in recent years. Accumulation in the unsold
commodity housing inventory is apparent, given persistently low sales
relative to commodity housing completion columns 5 and 6. in recent
years. Table 1 reveals that, although housing investment and commodity
housing supply has risen significantly since the early 1980s, the commodity
housing market has remained a minor channel for urban residents to
acquire new homes. On the other hand, publicly owned3 urban work units
have continued to play a major role in providing housing as an employ-
ment benefit to urban residents, either through housing construction or
through acquisition in the market.
In a survey of Beijing residents in 1997 China Commercial Times,
October 8, 1997.,4 76.8% of respondents reported living in public rental
3
Owned by the state or by urban collectives.
4
Zhong Hua Gong Shang Shi Bao, a Beijing-based newspaper.
HOUSING CHOICE AND CHINAS TRANSITION 65
TABLE 1
Housing Completion and Sales in Urban Areas
Year 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.
1. Total housing completion in urban areas million sq m.. Source a. for 19921995 and
c. for 1991 and 1996. The number may exceed the sum of column 2., 3., and 5., the
residual being housing completion in the informal housing sector.
2. Housing completion for state-owned work units million sq m.. Source b..
3. Housing completion for collective-owned work units million sq m.. Source b..
4. Housing completion for publicly owned work units as proportion of total completion:
w2. q 3.xr1..
5. Commodity housing completion million sq m.. Source a..
6. Commodity housing completion as proportion of total completion: 5.r1..
7. Commodity housing sold million sq m.. Source b..
8. Commodity housing sold as proportion of total completion: 7.r1..
9. Commodity housing sold to individuals. Source b..
10. Commodity housing sold to individuals as proportion of total sales: 9.r7..
Sources: a. China Economic Yearbook w6x, various years; b. State Statistical Bureau w19x;
c. Liu w14x, Table 2.
TABLE 2
Ratio of Housing Expense to Total Living Expenditure and Real Urban Wage Index
5
According to China Economic Yearbook 1995, p. 311. w6x, in Zhejiang province one-third
of the public housing stock had been sold to tenants by the middle of 1994. The average price
in Hangzhou, the capital city of the province, was 220 yuan per square meter, compared with
the average building cost of 702 yuan per square meter at the time.
HOUSING CHOICE AND CHINAS TRANSITION 67
6
Liu w14x also discusses various institutional factors preventing housing market clearing.
7
The three semi-private housing options are public housing with private user rights, partial
property rights in work-unit provided housing, and employeeemployer jointly funded hous-
ing development. These options are cheaper than commodity housing but are subject to
restrictive and ambiguous private property rights.
68 FU, TSE, AND ZHOU
TABLE 3
List of the Cities Surveyed and Their Economic Conditions
TABLE 4
Equity Structure of Different Ownership Types
Ownership type
no. of reporting cases. Ownership of equity capital Average share
TABLE 5
Work-Unit Size Number of Workers. and Housing Benefits
units. Foreign joint ventures are largely funded by nonpublic capital. Table
4 lists the equity structure of these four types of work units; it shows that
all types of work units, except for foreign joint venture enterprises, are
largely publicly owned in nature.
Table 5 shows that the average size of joint-stock enterprises is the
largest, followed by state-owned work units, and that the average size of
the collective-owned units is the smallest. Table 5 also shows the percent-
age of each type of work units that indicated being responsible for
employee housing.8 State-owned units and joint-stock units were more
likely to be responsible for housing. For state-owned and foreign joint
venture units, the average size of those providing housing is much larger
than that of those not providing housing.
8
It is not necessarily the case that the work units reporting not being responsible for
employee housing never provided housing to employees. About a quarter of the workers in
our sample at such work units were living in housing provided by their own work unit.
HOUSING CHOICE AND CHINAS TRANSITION 73
TABLE 6
Number of Observations by Housing Tenure and the Type of Work Unit
Ownership types
State- Collective- Joint- Foreign- Subtotal % of
Housing tenure types owned owned stock joint-venture grand total.
Table 6 shows the housing tenure distribution among the 3964 workers.
About 6% lived in commodity housing over four-fifths of them were
owners.. About 12% had invested, but were not living, in commodity
housing. About 78% occupied public housing provided by work units or
local governments. Eleven percent lived in traditional private housing or
household-built housing., largely inherited from previous generations.
About 3.5% rented their homes privately. Finally, a few interviewees,
accounting for less that half a percent of our sample, purchased work-unit
housing.
The workers participating in our survey were asked to indicate their
intention to purchase commodity housing in the next 3 years,9 ranked on a
five-category scale from 1 impossible. to 5 definitely possible.. Existing
research on consumer behavior largely confirms that purchasing intention
can be regarded as a valid measure of action tendency see Bagozzi and
Baumgarten w1x, for example., especially for high-involvement purchase
decisions characteristic of home purchasing McQuarrie w16x..
9
Ideally we would like to know the respondents reason for purchasing commodity housing.
Unfortunately, such information was not collected in our survey. Nevertheless, we believe the
majority of the respondents would consider buying commodity housing as their home rather
than as an income property. Our empirical results appear to be consistent with our belief.
74 FU, TSE, AND ZHOU
TABLE 7
Sample Statistics and Ordered-Probit Estimates
Definitely
Categories Impossible Unlikely Uncertain Possible possible
Equally 0.183 0.478 0.153 0.159 0.027
weighted
Weighted* 0.256 0.268 0.215 0.223 0.038
Sample mean and probit estimates dependent variable is BUY , number of observations is 3749 .
2 . 5 .
1 . Weighted 3 . 4 . Weighted probit
Independent Sample mean sample mean Probit estimates Probit estimates estimates
variables Std Dev. Std Dev. t value . t value . t value .
INCOME 3.431 1.726 . 3.470 1.751 . 0.0300 2.37 . 0.0344 2.80 . 0.0362 2.99 .
NRMPP 0.796 0.363 . 0.795 0.363 . y0.1516 2.92 . y0.1296 2.57 . y0.1525 3.06 .
CHNGINC 4.285 1.695 . 4.325 1.718 . 0.0347 2.90 . 0.0337 2.82 . 0.0394 3.34 .
LUXUR 0.225 0.417 . 0.234 0.423 . 0.1560 3.46 . 0.1536 3.42 . 0.1642 3.69 .
CHNGJB 2.017 0.932 . 2.027 0.956 . 0.1372 7.13 . 0.1364 7.13 . 0.1491 7.91 .
NEWHM 0.297 0.457 . 0.296 0.457 . y0.0759 1.85 . y0.0805 1.97 . y0.0801 1.97 .
CHNGSVN 3.002 0.880 . 3.026 0.887 . 0.0940 4.17 . 0.0877 3.91 . 0.0972 4.35 .
HINV 0.125 0.331 . 0.120 0.325 . y0.1193 0.57 . y0.1618 0.77 . y0.1237 0.61 .
STKINV = 0.224 0.417 . 0.233 0.423 . 0.1175 2.65 . 0.1258 2.89 . 0.1519 3.54 .
1-HINV .
JOBSEC = 1.685 1.015 . 1.715 1.027 . 0.0829 3.60 . 0.0756 3.31 . 0.0749 3.32 .
1-HINV .
STKINV = 0.043 0.203 . 0.044 0.205 . 0.0268 0.26 . 0.0369 0.37 . 0.0181 0.18 .
HINV
JOBSEC = 0.247 0.725 . 0.239 0.717 . y0.0906 1.59 . y0.0887 1.56 . y0.1078 1.95 .
HINV
EPCITY = 0.298 0.835 . 0.286 0.822 . 0.1435 2.28 . 0.1543 2.46 . 0.1600 2.64 .
HINV
EPFIRM 2.144 0.901 . 2.168 0.908 . 0.0786 3.57 . 0.0964 4.52 . 0.1110 5.25 .
EPFMLY 2.250 0.801 . 2.274 0.815 . 0.0976 3.71 . 0.1026 3.91 . 0.1181 4.52 .
FMLYSDL 1.988 0.613 . 2.003 0.622 . 0.1722 5.19 . 0.1702 5.15 . 0.1952 5.93 .
CLGEDU 0.288 0.453 . 0.301 0.459 . 0.1077 2.54 . 0.0938 2.24 . 0.1146 2.77 .
MANAGER 0.088 0.283 . 0.089 0.284 . 0.1479 2.25 . 0.1446 2.21 . 0.1673 2.59 .
AGE 37.834 9.192 . 37.499 9.215 . y0.0054 2.48 . y0.0064 3.00 . y0.0061 2.87 .
PRVRENT 0.037 0.190 . 0.041 0.197 . 0.3212 3.39 . 0.3466 3.67 . 0.3558 3.82 .
COMHSNG 0.012 0.110 . 0.015 0.122 . 0.4099 2.64 . 0.4080 2.66 . 0.3199 2.17 .
PVTPLAN 0.102 0.303 . 0.103 0.304 . 0.2176 3.26 . 0.2275 3.48 . 0.2481 3.80 .
PVTPAST 0.237 0.425 . 0.227 0.419 . y0.0100 0.21 . 0.0256 0.56 . 0.0421 0.93 .
LRGSOE 0.068 0.251 . 0.067 0.250 . y0.1936 2.66 . y0.1673 2.40 . y0.2096 3.06 .
PUSPV 0.041 0.199 . 0.043 0.202 . 0.0005 0.01 . 0.0215 0.25 . 0.0246 0.29 .
PVSPU 0.147 0.354 . 0.145 0.352 . 0.0608 1.14 . 0.0618 1.17 . 0.0753 1.44 .
PVEMPL 0.104 0.305 . 0.107 0.310 . 0.1311 2.07 . 0.1398 2.22 . 0.1620 2.58 .
PRICE 2.587 1.263 . 2.581 1.244 . City fixed effects y0.1306 7.70 .
y0.1506 8.86 .
1 y 0.9097 5.73 . 0.1640 1.10 . 0.6095 4.11 .
2 y 2.3684 14.70 . 1.6146 10.80 . 1.4237 9.57 .
3 y 2.9062 17.77 . 2.1519 14.22 . 2.0906 13.92 .
4 y 4.0608 23.84 . 3.3042 20.93 . 3.3719 21.66 .
LLR chi- 682 35 . 650 28 . 846 28 .
squared d.f..
*The weight for category Unlikely is 2r5 of the equal weights for other categories.
The estimates t value . are 0.3539 4.47 ., 0.3469 4.21 ., 0.2429 3.24 ., 0.5215 6.74 ., 0.6070 6.99 .,
0.5002 6.19 ., 0.6141 7.83 ., 0.6093 7.58 ., respectively, for city 2 through city 9.
78 FU, TSE, AND ZHOU
11
While permanent income may have an ambiguous effect on housing tenure choice, it
positively affects the demand for housing adjustment. Given the lack of housing choices in
the private rental market in Chinese cities, demand for housing adjustment motivates the
decision to buy commodity housing.
80 FU, TSE, AND ZHOU
12
Since the workers in the estimation sample did not report owning their work unit
housing, they appear not to have benefited from past housing privatization efforts.
HOUSING CHOICE AND CHINAS TRANSITION 81
TABLE 8
Observed and Predicted Outcomes
Percentage
Predicted Definitely Total correctly
Observed Impossible unlikely Uncertain Possible possible number predicted
TABLE 9
The Impact of Institutional Factors on the Incentive to Buy Commodity Housing
Based on Equally Weighted Probit Estimation.
Percentage
Predicted probability of Incremental difference in
indicating Possible or probability over probability over
Definitely possible the base case the base case
*Base case: all independent variables are set to their sample mean except i. PRVENT s
COMHSNG s PVTPLAN s PVTPAST s LRGSOE s PVEMPL s 0 and ii. PUSPV s
PVSPU s 1r2.
The predicted probability equals 3 y y X ., where is the cumulative distri-
bution of standard normal, X is the vector of independent variables, and 3 y and are
estimated parameters as reported in column 4. of Table 7.
HOUSING CHOICE AND CHINAS TRANSITION 83
ing price by 1000 yuan per square meter about 38% off the mean price
level. would increase the demand for commodity housing by only 21%.
Clearly, housing welfare policies, especially the policy of public rental
subsidy, have substantial impact on the individual demand for commodity
housing.
VII. CONCLUSION
Transition to a housing market is one of the most important reforms in
China, with deep economic and social consequences. A difficult task in this
transition is to wean urban residents away from publicly subsidized hous-
ing to become private home owners. We have shown that Chinese urban
workers intention to purchase commodity housing is sensitive to various
incentives, including housing mismatch, liquidity constraints, risk attitudes,
economic confidence, the cost of commodity housing, and housing welfare
policies. In particular, the individuals who are likely to participate in the
housing market are those who have a relatively high income and live in a
crowded home, are mobile in the job market and invest in stocks, are
economically confident and young, and have no access to publicly subsi-
dized housing. An important result emerging from our empirical analysis
regards the role of continued access to publicly subsidized housing in the
transition to a housing market. Our result indicates that the access to
publicly subsidized housing has a stronger negative impact than the afford-
ability of commodity housing in motivating individuals to participate in the
housing market.
The results in this study suggest that there are three major ways to
facilitate the development of Chinas housing market. First, the housing
market and the other related markets need to be efficient. Producers need
to improve the local public infrastructure and lower the cost of commodity
housing so as to stimulate demand. The housing prices need to respond to
the market mechanism so as to clear the inventories. It would be helpful to
establish consumer housing mortgage systems to provide more financial
resources to the housing market. Moreover, job market efficiency and the
social security system must be enhanced to improve the economic confi-
dence of urban workers.
Second, our study suggests that selected groups of urban workers will
have a greater inclination to move into the housing market. Those who
have a higher income, are younger, and live in areas where prices are not
as high showed a higher probability to buy private housing. Accordingly,
these groups may be the prime agents who facilitate the housing market.
Third, decision makers need to consider reversing the age-old subsidized
housing policy. Indeed, the removal of publicly subsidized housing is
inevitable for China to successfully establish a housing market. The
Chinese central governments commitment to ending subsidized housing
84 FU, TSE, AND ZHOU
for urban employees by the end of 1999 is a correct move. This decision
reflects how urgent removing subsidized housing is to Chinas housing
reform. Simultaneously, it shows how determined the government is to
tackle it.
Sampling
A total of 3964 respondents from 180 firms in nine cities participated in
the survey. They were recruited through a multistage sampling procedure.
The sampling frame of the first stage consisted of all Chinese mainland
cities of provincial capital or provincial status. They were grouped into
three tiers based on their per capita income in 1996. Three cities were
selected from each tier; they were Beijing, Shanghai, and Guangzhou high
income.; Nanjing, Wuhan, and Chengdu medium income.; and Changchun,
Guiyang, Xian low income..
In the second stage, the manufacturing firms with 100 or more employ-
ees in each of the nine cities were classified into four types by ownership
structure state-owned, collective-owned, joint-stock, and Chinese-foreign
joint ventureaccording to the 1996 national industrial statistical data
bank supplied by the Chinese Statistical Bureau. Twenty firms were
randomly selected from each type. Only the manufacturing sector was
covered, because in other sectors fewer types of ownership existed. After
eliminating the firms that declined the survey, went out of business,
located too far away from the city center to be reached by our field
interviewers, or did not have enough middle-level managers as required
by our sample criteria ., 78.3% of the randomly selected firms were
included in the final sample. The remaining number of firms in the final
sample were selected by the field interviewing firm, with the balance in
enterprise types, industrial sectors, and location distribution taken into
account.
In the final stage, 22 employees were selected from the employee list
supplied by the personnel department of each firm, excluding the employ-
ees from such side-line departments as firm-run kindergartens or hotels
that had little to do with the firms main line of business. Among the 22
interviewees, one was a personnel manager, one a marketing manager, 10
were randomly selected middle-level managers, and 10 were randomly
selected nonmanagerial employees. There were four extra employees, from
different firms, interviewed in Shanghai.
Data Collection
The data were collected in a 6-week time period from November 17 to
December 26, 1997.
HOUSING CHOICE AND CHINAS TRANSITION 85
Firm-Le el Data
Firm-level data were first obtained from a questionnaire completed by
the personnel manager. They were then verified according to the data
available from the previously mentioned national industrial statistical data
bank. In the few cases where discrepancies were found, we use the data
from the data bank because it had been officially audited.
REFERENCES