北大—林肯研究工作论文分享，详细内容请至北大-林肯研究中心官方网站免费下载全文： 此处为工作论文《Capitalization of Local Taxation and Public Spending on Housing Values: Evidence from China》内容精摘。

Capitalization of Local Taxation and Public Spending on Housing Values: Evidence from China

Joyce Yanyun Man

Peking University – Lincoln Center

College of Urban and Environmental Sciences, Peking University

Siqi Zheng

Institute of Real Estate Studies, Tsinghua University

Peking University – Lincoln Center

 1       Introduction

This paper analyzes the effects of local taxes and public spending on property values using data from Chinese cities in an attempt to test the Tiebout hypothesis.

According to Tiebout (1956) model, consumers   “shop”   among   different communities  offering  varying packages of local public services and selects as a residence the community which offers the tax-expenditure program best suited to their tastes. His model implies at least at a theoretical level there may exists a system that resembles a market solution to the production and consumption of local public goods that leads to the efficient provision of public goods and services. However, this hypothesis is based upon the assumptions that consumers are fully mobile and have perfect knowledge  concerning local revenues and expenditures, a large number of communities  exist for consumers to choose among, there are no community-based employment restrictions, there are no externalities associated with local public service provision and communities  are sized so that they can produce services at the minimum of their average cost curve. In the U.S. there is a vast number of studies examining the  capitalization  of  property  tax,  because it is the mainstay of local  revenue systems in U.S. Unfortunately empirical literature yields contradictory results concerning the extent to which property taxes are capitalized. If a property tax change is fully capitalized, the selling value of the asset is reduced by the present discounted  value of the tax. Hamilton (1976) argues that public goods financed with property tax can be provided efficiently and the property tax is non-distortionary.  Fees for public services. 

However the “new” view of Mieszkowski(1972) and Aaron(1975) argues that capital owners bear the full burden of a uniform property tax and is progressive, while tax differentials between  communities  give rise to “excise”  effects that may shift  forward to higher housing prices or shifted backward to a relatively immobile factor. Mieszkowski and Zodrow(1989) argue that under the new view, higher property tax rate in the metropolitan will reduce the national average rate of return on  capital and there is a tendency  toward under-provision  of local public services, because local jurisdictions are reluctant to tax mobile capital.

The  bid-rent model developed by  Yinger (1982)  suggest  that the  amount  a household is willing to pay for a unit of housing services in a particular jurisdiction, is based on the jurisdiction’s level of services and taxes. Yinger (1985) argues that the  average tax rate in the metropolitan area is distortionary but the variations from this base tax rate will be perfectly  capitalized  into housing values. Relatively  high  property tax rate in one jurisdiction will not repel capital, since the rate will be fully capitalized into immobile factors. He draws conclusions that differences in service levels between jurisdictions will be inexactly capitalized, depending upon taste parameters in the utility functions, but the  differences in tax rates betweencommunities will be fully capitalized, regardless of the tastes of consumers. If the price of housing is different in different jurisdictions, suppliers will have an incentive  to supply houses to jurisdictions with high home values. In the long-run equilibrium, local fiscal variables will be capitalized into house values.  

A large number of capitalization studies have centered upon the estimation of the degree to which property tax differences are capitalized into property values. If the difference in price between two otherwise identical properties is equal to the present discounted value (PDV) of the tax differential, the full capitalization of the property tax differential occurs.

The seminal capitalization study by Oates  (1969) studies 1960 census  data of fifty-three municipalities in northeastern New Jersey. Oates finds that a higher tax rate  depresses house prices and that increased school spending increases the price of housing. Using  a discount rate of 5%  and 40-year time horizon, he finds that  tax differentials are fully capitalized. Under a 3% discount rate and an infinite house life, the capitalization percentage is 61%. In the  subsequent  years, there are many researchers trying to measure the degree of capitalization of the property tax. Due to  the methodological difficulties and data limitations  and other unobservable  factors, the empirical capitalization literature has resulted in widely varying results of the rate of capitalization.

This study aimed to estimate the rate of capitalization of local taxes and public spending using the data for over 200 Chinese cities. In China, all tax rates are determined by central government, and each provincial/local government is assigned a share of revenue collections within its boundaries. Tax administration is a shared responsibilities between central and local  governments. The central  government is exclusively  entitled to impose taxes, and only  provincial level of  government  can make decisions with respect to tax rates  within the limits set by the central government, Unlike Local governments in the U.S. that relies heavily on property tax for local public finance, Chinese local  governments have all land and property tax revenue as well as business tax, income tax from local enterprises, personal income taxes, receipts of land leasing and transfers, estate tax, stamp tax.

In this study, we regress the median home value of Chinese cities on the local taxes and public spending as well as other factors identified to influence the housing price of a jurisdiction. The results from this study  aims to make a significant contribution to capitalization literature. First, previous research on the capitalization theory still have not reached a consensus. Empirical evidence from China shall add to the debate upon the incidence of the local taxes such as property tax and the extent of capitalization of local fiscal policies. It may provide empirical evidence on the validity of  Tiebout Hypothesis in a non-U.S. environment. Second, the  empirical studies using U.S. data suffer from the endogeneity problem that is fundamental to this literature because the median home value and the tax rate term are simultaneously determined due to the residual property tax system and the definition of the effective tax rate often used by researchers in their capitalization study including Oates(1969).

It is difficult to determine  the  direction  of causation between tax rates and home values. However, in China the tax rate is exogenous because it is set by the Central government and selected by the provincial  government, The local  government in China may have varying efforts in administrating and collecting the tax revenues. Our sample of data is for 220 cities in 31 provinces, allowing for a large amount of sample variation in local tax rate and public spending. Third, the local governments in China also administer taxes other than the land and property taxes such as value added tax and business tax and corporate and personal income tax. The impact of these taxes on housing value may lead to a broad discussion of the degree of substitution between the various inputs at different sectors (industries).

According to Gravele (1994) when there are other taxes on capital, most notably the corporate income tax, housing, especially owner-occupied housing, is taxed at a very low rate compared to other forms of capital, The property  tax actually help to correct the misallocation of resources resulting from the corporate income tax.

2      Models and Specification Issues

This study explores the effects of  various taxes, other revenues and public expenditures on residential property values in Chinese cities.  The current official statistical system does not report city-level median or mean  home values. Our advantage is that we have access to a large-scale Urban Household  Survey (UHS2007) conducted in 2007, which contains the current market values of the households’  homes at the end of 2006. Median home value for every  city is calculated based on this micro dataset (See Section III for details). Since we only have the home value data for 2007, we collect the fiscal variables for 2006 and estimate the following cross-section regression equation:

Where HViis median home value in city i. Tiis the vector of tax structure variables. E is the vector of local public spending variables. Ziis the set of other variables that are believed to affect home value variations across cities. i is the error term.  

To mitigate the possible endogeneity problem between home value and taxes and expenditures, we include lagged indicators of local taxes and expenditures on the right-hand side of Eq. (1). Home value (HV) is of year 2007 (t=2007), and the tax and expending indicators are of year 2006 or 2005 (lag_period = 1 or 2).

3       Data and Hypotheses

Urban Household Survey (UHS2007) conducted by National Bureau of Statistics of China (NBSC), China Urban Yearbooks, and China Fiscal Yearbooks. Table 1 shows the descriptive statistics of the variables used in this study.

Urban Household Survey (UHS) is  conducted  annually  by the  Urban  Survey Department of the National Statistic Bureau of China (NSBC). The sample size is about 50 thousand. In 2007, the Bureau conducted a large-scale survey, which covers all of the 255 prefecture-level cities in China (shown in Map 1) and has a sample size of 300 thousand households. Chinese cities have a 3-tier sub-municipal administrative structure: the first tier is district, or Qu, the second tier is street block, or Jiedao (JD), and the third tier is street neighbourhood, or Juweihui (JWH). Beijing, for example has 18 Qu’s, 130 JDs and 2,625 JWHs in 2006. The 2007 UHS employed the 3-stage stratified sampling method. First, JDs in each city are sorted by their identification (ID) numbers and sampled at  fixed distances; Next, JWHs in each selected JD are sorted by their  ID number and are  sampled at  fixed distances; Finally, 20-40 households are  randomly  sampled in each selected JWH. The selected household reported its annual household income, household head’s education attainment, the size and the current market  value of the  housing unit. The current market value was estimated by the interviewer using a simple market comparison approach.

The dependent variable is measured as city median home value (per housing unit) in logarithm in Equation (1). The median home in the average city has a value of 179 thousand RMB. The most expensive city has the median home value of 690 thousand RMB. The explanatory variables fall into three categories: (1) taxes and other revenues; (2) public expenditures; and (3) Economic and demographic characteristics, industrial composition, housing market attributes and city/region dummies.  The total fiscal revenue is disaggregated into VAT, business tax, corporate income tax, personal income tax, and other taxes/revenues. Local public  expenditure  is disaggregated into education expenditure, infrastructure expenditure and other expenditures. Tax variables are measured as per capita tax revenue or the ratio of the tax revenue to GDP as a measure of average tax rate. Public expenditure variables are  measured as per capita expenditure. In 2006, the average city yields the fiscal revenue of 4072 RMB, and expends 5920 RMB. Regarding the tax structure, business tax takes the largest share (23.0%), and individual income tax has the smallest share(4.1%). On average, 20.3% of the total fiscal expenditure is spent in education.

Ideally, we would like to have a measure of various effective tax rates, per se. However, any attempt to use statutory tax rate schedules in constructing a measure of effective rates  across  Chinese cities is problematic because of the substantial non-uniformity of enforcement practices and variations in rate structures of some taxes across regions in China. As an alternative, tax revenues from various taxes as a share of GDP in each city is used as tax rate proxies. In variable  group (3), we include each city’s economic  and demographic characteristics, such as median annual household income, GDP, population, human capital level (average  years to schooling), the shares of primary  and secondary industries in each city’s GDP. We also include two housing market variables: share of commodity housing in total housing stock, and median housing unit size. The cities with larger shares of commodity housing are expected to have higher home values because commodity housing is newer and more expensive than average housing stock. The cities with larger houses on average should also have higher home values. The regression results are discussed in the following section.

4       Empirical Results

Table 2 presented the regression estimation of the property value equation for the sample of this study. Our explanatory variables can explain  60%+ of  home value variation across cities.

According to regressions results from Equation 6 in Table 3, the variable of our primary interest, the per capita expenditure, has a positive and statistically significant coefficient estimate, indicating that the level and quality of public goods and services are reflected into the home value of a city in China. A 10 percent increase in per  capital public spending leads to nearly 4 percent increase in residential home value, everything else being constant.  

This result suggests that Chinese consumers do take into account of the provision of local public goods and services in their location choices. A City with a higher per capita public expenditure is resulted in a higher value of houses. In China sub-national government spending account for nearly 70 percent of total public spending  and there exists a vast disparity in quality and quantities  of public goods and services including public housing, social security,  public health, education and civil services among others. It may demonstrate that with rapid urbanization and increasing household mobility, the differences in services among cities in China are capitalized into home values. It provides empirical evidence that even if in an environment where the set of assumptions of Tiebout model are  unrealistic or too restrictive, Tiebout hypothesis is still relevant in China. It supports the Tiebout hypothesis by demonstrating that consumers or households in China do consider the  available programs of public services in their  choices of  locality of residence. So other things being  equal (including tax rates), the  communities that

offers more attractive package of public goods and services results in higher gross rents and therefore property values would be higher in a community.

As a result, the outputs of public services  influence  the attraction of a community to potential  residents and thereby affect local property values. This  result  confirms Yinger’s  theory that individual  families  desiring to consume higher levels of public output, would presumably tend to bid up property values in communities with high-quality programs of public services, affecting locational decisions of   consumers   or households.

However, the total local revenue as a share of GDP in a city has a negative sign but statistically not significantly different from zero at 10 percent level. This result may be caused by the possible competing effects of multi tax structures and rates used  by the local  governments  in China. Unlike the U.S. where the local property  tax accounts for roughly 75 percent of revenues on average, the tax on the value of land  and structure as a property tax accounts for a small share of tax revenue. There is no tax on the value of owner-occupied residential housing. As a result, the property tax in China does not fully reflect the average cost of the local public goods and services.

Consequently, a package of local taxes used in the financing of public  goods and services in a Chinese city are included in the estimation equation of the residential housing value equation to better measure the differential impacts of varying taxes. 

The income elasticity of housing demand is about 0.3-0.5, which is consistent with Zheng (2007) who estimates the elasticity using micro household data in Chinese cities. Larger homes have higher values. All else equal, a 1% increase in housing unit size results in a 0.5% increase in home value. Larger cities in terms of non-agriculture population have higher home values. All else equal, a 10% population growth causes 1.5% home value appreciation. Cities with larger shares of tertiary industry also have more expensive homes. Consistent with our intuition, cities with larger shares of commodity housing have significantly higher home prices.

Not surprisingly, the tax on personal income tax reduced the median house value of a community. It may be well likely that the personal income tax burden on  households depresses the  purchasing  power of the consumers, leading to lower property value in the community via the demand side effect.  The other local taxes including those on the value of real estate in the urban areas results in a statistically significant negative effect on property value. This result is consistent with the empirical results of many previous researchers, suggesting that the differences in local taxes are capitalized in the value of residential housing.

（以上内容版权，系文章作者和北京大学林肯研究院城市发展与土地政策研究中心所有。如需转引、转帖或联系相关媒体报道，务请联系中心出版事务干事赵敏女士。)

联系电话：8610—6275 6535 8610—6275 7884（传真）

联系邮件：zhaom@plc.pku.edu.cn

中心主页：http://plc.pku.edu.cn/

联系地址：北京大学廖凯原楼508室 邮编：100871