Land-based interests and the spatial distribution of affordable housing development: The case of Beijing, China

Habitat International 44 (2014) 137e145

Contents lists available at ScienceDirect

Habitat International journal homepage: www.elsevier.com/locate/habitatint

Land-based interests and the spatial distribution of affordable housing development: The case of Beijing, China Yunxiao Dang a, b, Zhilin Liu c, *, Wenzhong Zhang a a

Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Science, Beijing, China University of Chinese Academy of Sciences, Beijing, China c School of Public Policy and Management, Tsinghua University, Beijing, China b

a r t i c l e i n f o

a b s t r a c t

Article history: Available online

This paper provides an institutional explanation for the spatial distribution of affordable housing units biased toward the urban fringe in Chinese cities. We argue that, given China's inter-governmental relationship, the discriminatory site selection practice is a result of strategic policy implementation by city governments, who strive to balance the top-down political pressure with the local fiscal interests. The empirical analysis focuses on the implementation of the Economical and Comfortable Housing (ECH) Program in Beijing in 1999e2009. Binary logistic regression reveals that, holding local housing needs and socio-demographical attributes constant, sub-districts with a higher land price, a high ratio of old houses, and greater subway accessibility have a lower probability of being designated for affordable housing development. The empirical findings confirm the significance of land-based interests of city governments that contribute to many unintended consequences of affordable housing programs at the local level. © 2014 Elsevier Ltd. All rights reserved.

Keywords: Affordable housing policy Land-based interests Policy implementation Inter-governmental relationship Beijing

Introduction China is at a critical stage in restructuring its urban affordable housing policy to provide decent homes for its growing urban population in the next decade. As the central government rediscovered its responsibility to ensure equitable and affordable housing, unprecedented efforts have been channeled into affordable housing programs. In 2010, affordable housing was officially included in the Twelfth Five-Year Plan as a critical component of the social welfare and public service system. The central government even has sought to enhance political incentives for developing affordable housing by designating affordable housing policy implementation as a critical responsibility for local officials (General Office of State Council, 2011). Despite the heightened political will expressed by the central government, the local implementation of affordable housing programs remains highly contentious and unsatisfactory. The media have consistently reported on the “unintended consequences” associated with affordable housing programs, such as low building quality, poor design (Youth.cn, 2010), inferior locations (Zheng & Zhang, 2010; Zou, 2014), and a lack of transparency and fairness

* Corresponding author. Room 416, School of Public Policy and Management, Tsinghua University, Beijing 100084, PR China. Tel./fax: þ86 10 62785649. E-mail addresses: [email protected], [email protected] (Z. Liu). http://dx.doi.org/10.1016/j.habitatint.2014.05.012 0197-3975/© 2014 Elsevier Ltd. All rights reserved.

in the allocation of affordable housing units (Wang & Murie, 2011; Zou, 2014). Among these problems, the concentration of affordable housing units in inferior locations deserves particular attention from scholars and policy makers. As widely discussed in the international literature, discriminatory location choices for public housing projects have led to poverty concentration and residential segregation in western countries (Massey & Kanaiaupuni, 1993; Schwartz, 2006; Stoloff, 2003). Chinese scholars also have warned about the potential for similar phenomena to emerge in Chinese cities because affordable housing projects often are located in areas without sufficient access to employment opportunities and public services (Chai, Zhang, & Liu, 2011). Through a case study of Beijing, this paper examines the effects of land-based interests as a fiscal incentive mechanism that led to the concentration of affordable housing projects located in the urban fringe. We specifically focus on the case of the Economical and Comfortable Housing Program (jingji shiyong zhufang zhengce, hereafter, ECH) in Beijing from 1999 to 2009. Initiated by the central government in 1998, ECH is a home ownership-oriented program that until recently had been the core pillar of affordable urban housing policy in China. We argue that site selection for ECH projects should be considered a political decision of city governments, i.e., the main implementers of this national program, who strive to balance the top-down political pressure with the local fiscal

138

Y. Dang et al. / Habitat International 44 (2014) 137e145

Table 1 National policy framework for the ECH program: 1998, 2004, and 2007. Time period

1998－2003

2004－2006

2007- present

Stated Policy Goal Target Beneficiaries

Government-regulated commodity housing Low- to middle-income families

Eligibility Assessment

Local criteria is up to city governments; rules unspecified and not enforced

Policy-oriented commodity housing Low- to middle-income families with housing difficulties Local criteria is up to city governments; rules more specified

Upper limit of Building Space Allocation Tenure/Property Rights

No rules Sold by developers Restricted ownership, i.e., resale prohibited within five years Administrative allocation: exempted from land-leasing fees Exempted or reduced Regulated price with restricted profit (3%); Negotiated between government and developers

80 m2 Sold by developers Partial property rights; resale prohibited within a certain number of years Administrative allocation: exempted from land-leasing fees Exempted or reduced Regulated price with little or no profit (less than 3%);

Welfare-based subsidized housing Low-income families with housing difficulties Local criteria is up to city governments; more rigorous rules and procedures 60 m2 Sold by developers Partial property rights; resale prohibited within five years Administrative allocation: exempted from land-leasing fees Exempted or reduced Regulated price with restricted profit (3%)

Land Taxes and Fees Price setting Mechanism

Sources: Author's own compilation from national policy documents, e.g., MOC et al., 1998; MOC et al., 2004; MOC et al., 2007.

capability to implement the ECH program. City governments are politically accountable for planning and implementing local ECH programs, whereas national policy design requires cities to bear greatest financial burdens associated with ECH construction. Such institutional constraints at the macro level create a micro-incentive structure that drives the strategic decisions of local officials when implementing national housing programs. We aim to contribute to a growing body of literature on China's affordable housing policy by providing empirical evidence on the institutional mechanism that has distorted local policy implementation and caused unintended consequences for affordable housing programs in urban China. Existing literature tends to focus on housing affordability and equity issues in Chinese cities (e.g., Huang & Jiang, 2009; Li, 2012; Logan, Bian, & Bian, 1999; Wang & Murie, 2000) and the evolution and framework of affordable housing policy in China (e.g., Deng, Shen, & Wang, 2011; Huang, 2012; Wang, 2001, 2011; Wu, 1996; Zou, 2014). Few studies have focused on the local implementation of affordable housing programs as determined by the strategic decision making of city governments. Although many scholars have stressed that land-based interests of city governments determine urban development patterns (e.g., Lin & Yi, 2011; Zhu, 2004), empirical evidence on the impact of government's land-based interests on affordable housing policies has rarely been published. This paper is structured as follows. First, we review relevant literature and highlight the need to study the institutional dynamics underlying the implementation of affordable housing policy. We then discuss the policy design of the ECH program and the consequences for local implementation. We derive our hypotheses from an extensive review of policy documents, existing literature, and interviews with government officials and developers. After describing our research design and methodology, we present our empirical findings and conclude with discussion and policy implications. Literature review In countries where the government subsidizes housing for the poor, location is recognized as a critical dimension of successful affordable housing policy (e.g., U.K. Communities and Local Government, 2006). The provision of affordable housing in convenient locations not only ensures decent housing for the poor but also offers economic and social opportunities to ensure self-sufficiency. Unfortunately, historical lessons from public housing programs in the US and other countries have emphasized the danger of placing affordable housing projects in inferior locations where employment

opportunities and public infrastructure are scarce (Gabriel, 1996; Holmans, 2005). Many scholars have noted devastating consequences brought by discriminatory site selection practices (Stoloff, 2003), such as poverty concentration, degradation of the living environment in low-income communities, and long-term effects such as high unemployment and crime rates (DeKeseredy, Schwartz, Alvi, et al. & Tomaszewski, 2003; Massey & Kanaiaupuni, 1993; Schwartz, 2006, & Tomaszewski). Although such location patterns may have been “unintended” in the original policy design, Massey and Kanaiaupuni (1993) noted that local politics, particularly the strategic behaviors of local politicians, have increased the likelihood that public housing projects are located in poor and black neighborhoods. In China, although urban housing policy in the 1990s had emphasized marketization and commercialization (Wang & Murie, 2000), low-income housing issues have recently assumed a higher priority on the national policy agenda. Nonetheless, the local implementation of affordable housing programs has been problematic and sometimes ineffective (Huang, 2012; Zou, 2014). One particular problem that has received wide attention but insufficient systematic analysis is a discriminatory site selection practice for affordable housing projects. Unlike in the US cities, governmentfunded affordable housing projects in Chinese cities are often located in remote suburbs, locating low-income residents far from employment opportunities, public transportation, and public services. Scholars have warned that low-income households in Chinese cities could face a similar long-term deprivation of economic opportunities and decreased quality of life because of the inferior location of affordable housing opportunities (Chai et al., 2011; Zheng & Zhang, 2010). Scholars such as Huang (2012) have attributed the failure of affordable housing policy implementation to the lack of commitment by local governments to building sufficient low-income housing. Zou (2014) further argued that the inter-governmental fiscal arrangement has worked against the center's policy goal of providing affordable housing. City governments are politically mandated but not financially motivated to provide affordable housing because they must bear most financial and administrative costs of local policy implementation. At the same time, cities have the authority to adjust national policy programs according to their local situations (Xu, 2011). Therefore, to interpret the local implementation of affordable housing programs, including patterns for where housing is located, it is necessary to understand how national policy design and fiscal arrangements have together shaped the strategic behaviors of local governments. The next section discusses the incentives of city governments in the ECH program from which we derive our research hypotheses.

Y. Dang et al. / Habitat International 44 (2014) 137e145

National policy design and micro-incentives for city governments Several studies (e.g., Deng et al., 2011; Huang, 2012; Wang, 2011; Zou, 2014) have described a national policy framework for the ECH program and other affordable housing programs such as the LowRent Housing Program (lianzu zhufang) and the Public Rental Housing Program (gongong zulin zhufang). Unlike the other two programs, ECH is a homeownership-oriented affordable housing program that aims to provide incentives for developers to construct low-cost housing to be sold to qualified households at governmentcontrolled prices (State Council, 1998). Since its adoption in 1998, the ECH program's national policy framework has undergone several rounds of adjustments. Table 1 compares key elements of national policy documents concerning the ECH program. The most significant change is a priority shift that emphasizes ECH's “welfare” purpose rather than its nature as a “commodity”. In 1998, ECH was designated “a special type of commercial housing” targeted at low to middle-income households, which are the majority of urban households (MOC et al., 1998). The ECH program was considered the main channel for providing housing for low- to middle-income households and a core pillar of a segmented housing provision system in Chinese cities (MOC et al., 1998). By 2004, ECH was viewed as “policy-oriented commercial housing” (MOC et al., 2004), and by 2007, it was considered “welfare-based subsidized housing” targeted only at low-income households (MOC et al., 2007). Following the change in the targeted beneficiary group, the central government modified and specified many rules regarding building standards, eligibility, and allocation of ECH housing to improve the programs targeting and discourage unqualified families from obtaining ECH housing through rent seeking. For example, these changes included specification of the upper limit of the building space for ECH units, local criteria for assessing household eligibility, and procedural rules for housing application and allocation (see Table 1). However, these policy changes had little effect on most rules and regulations that influence the incentives for policy implementers. ECH is essentially an indirect supply-side subsidy program in which city governments forfeit profits from land development to allow cost reduction and thereby lower prices for low-income housing (Wang, 2011). National policy design has focused on providing incentives for housing developers to participate in the program because the state aimed to increase their role (Table 1). The most important rule is most likely the exemption of ECH projects from land-leasing fees, a lump-sum rent paid by developers to city governments when leasing the use rights for urban land (see Table 1). For housing developers, administrative land allocation is a key feature that distinguishes an ECH project from a typical commercial housing project.1 ECH projects also enjoy a waiver or reduction of taxes and fees, a source of municipal revenues that are usually levied on real estate development activities. Thus, although the state claimed that the ECH program had the potential to incorporate market mechanisms into the provision of affordable housing, city governments shoulder the financial and administrative costs involved in the program. Although city governments are not required to directly fund housing construction, they still face a substantial loss of local revenues, including landleasing fees, other fees, and taxes, which they would otherwise collect from commercial development projects. Unfortunately, land-based revenues, particularly the landleasing fees that the government collects by leasing land-use rights to private developers, have become a dominant source of

municipal revenues in China (Cao, Feng, & Tao, 2008; Tao, Su, Liu, & Cao, 2010; Yeh & Wu, 1996). The sale of land-use rights can account for 25 to 50 percent of a city's revenue (Yang & Wu, 1996). The income generated from the expropriation and commodification of land has become a main source of local revenue and the key to financing urban economic growth and urbanization (Lin & Yi, 2011). The phenomenon of “land-based public finance” (tudi caizheng) is blamed for the rapid real estate development that is promoted by city governments (Tao et al., 2010; Zhu, 2004). The importance of land-based revenues thus creates disincentives for city governments to implement a national affordable housing program such as ECH (Zou, 2014). Because national policy design emphasizes free land allocation and other incentives for developers, city governments must “give up part of [their] revenue from land and administrative fees to bring down the housing price”.2 Affordable housing projects have much less potential than commercial housing development to contribute to local revenues. Conversely, local leaders are politically responsible for implementing the ECH program, which recently has involved a political accountability measure (xingzheng wenze) that holds local officials accountable for failing to fulfill top-down political mandates (GOOSC, 2011). Therefore, a rational strategy for city governments is to take advantage of the decentralized authority of policy implementation in a way that minimizes their fiscal and administrative costs. As the developer of an ECH project commented in an interview, “[Local governments] have no reason to put large ECH projects in the central area of the city, particularly places near the Central Business District because these projects are exempt from many taxes and feesdat least 50% reductiondas well as land rent.” (Interview with developers, August 9, 2005) If locating a new ECH project “in those fast growing places”, city governments “would be concerned about its effects on their revenue income”.3 However, city governments can reduce revenue loss simply by locating ECH projects in less-developed suburban areas where land is much cheaper (Zou, 2014). Furthermore, when city governments designate a site for ECH projects, they are responsible for acquiring the land from the original property owners, as well as the compensation costs incurred if land acquisition involves any demolition or relocation. The compensation costs are particularly high in certain neighborhoods with many old and dilapidated dwellings, which often have a high population density. Although developers eventually must pay compensation costs, in many cases, land acquisition involves a long process of negotiation with residents. It is not unusual to observe many residents refusing to be relocated or demanding favorable terms, placing tremendous pressure on local officials. As reported in some studies, demolition and relocation compensation to original land users could account for as much as half of total development costs (Wang, 2001). Therefore, city governments must face high financial costs and political risks if they choose a site for an ECH project where previous residents strongly object relocation. In summary, anecdotal evidence has indicated that city governments, as the implementer of the ECH program, strive to reduce local fiscal and administrative costs when realizing ECH projects. First, such costs may involve opportunity costs such as land-leasing fees that city governments might collect if the site was used for commercial housing development instead of ECH projects. Thus, they are less willing to place ECH projects in areas that contribute a large amount of land-leasing fees to local budgets. In other words,

2 1

Interview with developers by authors, 10 October and 21 November, 2005.

139

3

Interview notes, 20 December, 2005 and 10 October, 2005. Interview Notes, September 17, 2005.

140

Y. Dang et al. / Habitat International 44 (2014) 137e145

Fig. 1. Construction of ECH projects from 1999 to 2009 in Beijing. Source: Beijing Statistical Yearbook, various years.

areas with higher land prices are less likely to be designated for affordable housing development. Second, in addition to attempting to minimize land-based revenue losses, city governments also try to minimize other financial or administrative costs incurred during the land acquisition process. City governments are more likely to avoid sites with many old houses because of concerns over demolition costs. Research design Case city This paper uses Beijing, the capital city of China, as the case city and focuses on the implementation of the ECH program from 1999 to 2009. Beijing was a pioneer in providing low-cost affordable housing under the ECH program. In 1998, the municipal government issued its own policy document for the ECH program and subsequently initiated the first 19 ECH projects with a total of 5.6 million square meters of affordable housing to be constructed (Wang, 2011). Similar to many other Chinese cities, Beijing's progress in ECH development has fluctuated. The first stage, from 1999 to 2004, was a period of rapid growth in ECH construction. Total affordable housing floor space under construction increased from

approximately 3 million m2 in 1999 to nearly 8 million m2 in 2004, an increase of over 160% (see Fig. 1A). However, housing developed under the ECH program was never more than 20% of the total housing constructed or provided in the market. Moreover, its percentage continued to decline after peaking in 2000 (see Fig. 1B) because market-rate housing also experienced an unprecedented boom and grew at a much faster pace than did affordable housing development during this time period. By 2004, the ECH program had attracted significant criticism for failure to meet its stated policy goals, which eventually led to a period of reflection and policy adjustments. As a result, affordable housing construction under the ECH program stagnated from 2003 to 2005 and started to decline after 2005 (Fig. 1A). By 2007, the total floor space of ECH units under construction had declined to 4.4 million m2, 44.5% lower than the 2004 level. Although the amount of floor space under construction started increasing after 2007, the total completed ECH unit space steadily decreased to 0.98 million m2 in 2009 (Fig. 1A). Model specification We used a binary logistic regression model to test the extent to which the potential loss of land-based revenues in an urban area and compensation costs incurred during site demolition and

Y. Dang et al. / Habitat International 44 (2014) 137e145

141

Table 2 Descriptive statistics of key variables in the binary logistic regression model (based on sub-district). Variables

Description

Mean (SD)

S.E. mean

ECH LANDPRICE OLDHOUSE SUBWAY RENTER PUBRENTER SMALLHSG EDU

Whether a sub-district has at least one ECH project (Yes ¼ 1; No ¼ 0) Average land-leasing fee of all the leased land parcels in a sub-district from 1998 to 2008 (1000 Yuan/m2) Percentage of housing built before 1969 in the sub-district Number of subway stations existing or planned in 2003 within 3 km from the center of a sub-district Percentage of families living in rental housing Percentage of families living in old public rental housing (chengzu gongfang) Percentage of families in a sub-district with living space smaller than 8 m2 per capita Percentage of residents in a sub-district with a college degree or higher education

0.24 (0.43) 37,110 (5421) 0.18 (0.21) 8.71 (6.29) 0.42 (0.16) 0.33 (0.18) 0.21 (0.14) 0.22 (0.13)

0.04 468 0.02 0.54 0.01 0.02 0.01 0.01

relocation predict the probability of ECH projects being located in that area, controlling for local socio-demographical conditions. The unit of analysis for the regression model is the sub-district (jiedao), the smallest spatial unit for which most socio-economic indicators are available in Beijing. The regression model is as follows:

  P ¼ a þ b1  LANDPRICE þ b12  OLDHOUSE þ b3 Ln 1p  RENTER þ b4  SMALLHSG þ b5  EDU; Where the dependent variable is a binary coded measure of the presence or absence of ECH development in a sub-district, with 1 indicating the presence of at least one ECH project and 0 indicating the absence of such development project in the sub-district. We compiled a database of all ECH projects, including information such as the year of construction, location, and size, based on public notices obtained from the Beijing Real Estate Transaction Management Network4 of the Beijing Municipal Commission of Housing and Urban-Rural Development. All projects were then geo-coded to allow for the spatial analysis of the pattern of location for ECH projects and to calculate the binary dependent variable for each sub-district. LANDPRICE is the average land price of a sub-district which, as a proxy, captures the potential loss of land-leasing fees that the government would collect if the land was designated for commercial development. We calculated the average land price for each sub-district from a database of land parcels leased by the Beijing municipal government from 1998 to 2008. We compiled and geocoded the land parcel data based on all public notices of land leasing accessed via the Beijing Municipal Bureau of Land and Resources website.5 Based on the hypothesis that city governments attempt to implement local ECH programs in a way that minimizes the loss of land-leasing revenues, sub-districts with higher average land prices have a lower probability of being designated for ECH development. OLDHOUSE is the percentage of housing units built before 1969. We use this variable to measure the potential costs of demolition and relocation compensation if the site is designated for ECH development. Because city governments are more likely to avoid sub-districts with many old houses when determining ECH project locations, we hypothesize that there is a negative relationship between the ratio of old houses and the odds of ECH development in a sub-district. SUBWAY (i.e., subway station accessibility) is measured by the number of subway stations within 3 km of the center of a subdistrict. We calculated this index for each sub-district based on information regarding location of subway stations, retrieved from a

4 Beijing Municipal Commission of Housing and Urban-Rural Development http://breaa.cn/publicbjjs/house/project_list_jjsyf.asp. 5 Beijing Municipal Bureau of Land and Resources network. http://www.bjgtj.gov. cn/tabid/3259/Default.aspx.

digitized map of the Beijing Subway System Plan issued in 2003. We include this variable to test whether city governments give sufficient consideration to transportation accessibility for lower-income residents living in ECH neighborhoods. If that is the case, we should observe a positive relationship between subway station accessibility and the probability of ECH projects being located in a sub-district. However, we also noticed that it was not until 2007 when the Ministry of Construction for the first time stated that local governments should “fully take into account the need of low-income families for public transportation so as to locate ECH project appropriately” (MOC, 2007). Yet such guideline was not fully implemented at the local level. Instead, the government prefers not to place affordable housing developments near subway stations, where land is more profitable. In that case, we may instead find an opposite pattern that the more accessible to public transit an area is, the less likely the area is designated for affordable housing development. We also include two variables (RENTER and SMALLHSG) as measures of the living conditions of a sub-district to test whether greater housing needs may predict a greater likelihood of ECH projects being located within a sub-district. RENTER is the percentage of families in a sub-district living in rental housing. We also calculated the ratio of families living in old public rental housing (gongfang; PUBRENTER) to capture the effects of different rental housing sectors. SMALLHSG is the ratio of families in a sub-district with a per capita living space smaller than 8 square meters. All variables are calculated based on the Fifth Population Census aggregated to the sub-district level. Finally, because the census in China does not provide income information, we used the percentage of people with a college degree or higher education (EDU), also aggregated from the Fifth Population Census to indicate the socio-economic status of a sub-district. See Table 2 for a description of the variables.

Empirical findings Spatial distribution of ECH projects in Beijing, 1999e2009 We identified 88 projects constructed under the ECH program in Beijing from 1999 to 2009, with 51 constructed from 1999 to 2004 and 37 constructed from 2005 to 2009. Fig. 2 illustrates the spatial distribution of all ECH projects constructed during this time period, as well as the total floor space under construction by sub-district. Only three of 51 ECH projects constructed from 1999 to 2004 were located in the convenient central city (i.e., within the Third Ring Road), whereas eight projects were located in the outer suburban districts (e.g., Changping and Tongzhou). Despite policy adjustment after 2004, the spatial distribution of ECH projects remained similar or was even farther from the city center. Only two projects were located inside the Third Ring Road, whereas 18 projects (48.7%) were placed outside the Fifth Ring Road (see Fig. 2). Moreover, larger projects tended to be located farther from the city center. Using living space (i.e., floor space) as a measure,

142

Y. Dang et al. / Habitat International 44 (2014) 137e145

Fig. 2. Spatial distribution of ECH projects in Beijing: 1999e2004 and 2005e2009.

approximately 70% of all ECH units were built outside the Fifth Ring Road from 1999 to 2004, whereas almost 90% were built in this area from 2005 to 2009. We further compared the location pattern of ECH projects with the spatial distribution of land parcels leased for commercial housing development during 1999e2009 (see Fig. 3). Apparently, as shown in Fig. 3, a significantly higher proportion of commercial housing developments are located in the central areas of the city, whereas ECH projects are more likely to be located in the fringe area. Specifically, from 1999 to 2009, 65.9% of ECH projects are located outside the Fourth Ring Road and 40.9% located outside the Fifth Ring Road. By contrast, 42.4% of all land parcels leased for commercial housing development are outside the Fourth Ring Road and only 9.9% outside the Fifth Ring Road. As discussed earlier, the land-leasing fee charged for commercial real estate development is an important source of local revenues. Thus, city governments are keener to designate land parcels at central and more profitable locations for commercial housing development as opposed to policy-oriented affordable housing development. We conducted a multivariate spatial autocorrelation analysis to examine the spatial relationship between the location of ECH projects and land prices across all 132 sub-districts for which data were available. The global spatial autocorrelation analysis revealed a significant negative correlation between the location of ECH developments and average land-leasing prices (Moran's I ¼ 0.1176, P value ¼ 0.005). This result indicates that the average land-leasing prices are lower in sub-districts where ECH developments are present than in sub-districts without ECH projects.

Model results Before the model analysis, we conducted collinearity diagnosis analysis for all independent variables and found no serious collinearity problem in the model except the correlation between PUBRENTER and RENTER.6 Nevertheless, since RENTER and PUBRENTER are only included in the model separately, their collinearity is neglectable in our analysis. We ran five models to further test the effects of various sets of independent variables. Model 1 is the base model, which includes all variables except LANDPRICE and OLDHOUSE. Models 2 and 3 include these two variables, although subway accessibility is excluded in Model 3 due to concerns regarding a possible correlation between land price and subway accessibility. Models 4 and 5 are similar to Models 2 and 3, respectively, except that the ratio of rental housing is replaced with the ratio of old public rental housing. All variables were standardized before being introduced into the models, and all five models are significant (see Table 3 for model results).

6 The tolerance values for all independent variables are larger than the critical value of 0.1 and the VIF values smaller than the critical value of 10 with the only exceptions of PUBRENTER and RENTER6, which are highly correlated with each other. Further collinearity diagnostics results indicate that, for all independent variables except RENTER and PUBRENTER, the eigenvalues are greater than the critical value of 0 and the condition indexes are smaller than 10.

Y. Dang et al. / Habitat International 44 (2014) 137e145

143

Fig. 3. Spatial distribution of ECH projects and land parcels for commercial housing: 1999e2009.

Land-leasing revenue

Ratio of old houses

As predicted, the coefficients of LANDPRICE are negative and significant at the 0.01 level in Models 3 and 5 when subway accessibility is excluded and are significant at the 0.1 level in Models 2 and 4. These results confirm our hypothesis that sub-districts with less potential to contribute land leasing revenues to local coffers are more likely to be designated as sites for ECH projects, thereby reducing the loss of fiscal revenues for local governments.

All coefficients of OLDHOUSE are negative and significant in all five models (p < 0.05 in Models 1, 4, and 5; see Table 3). The negative coefficient indicates that sub-districts with a higher ratio of old housesdand consequently, high demolition costsdare less likely to be designated by city governments as sites for ECH projects. This confirms our hypothesis that although city governments are politically responsible for implementing the ECH program, they

Table 3 Empirical results of the binary logistic regression models. Variables

Model 1

b LANDPRICE OLDHOUSE SUBWAY RENTER PUBRENTER SMALLHSG EDU Constant 2 Log likelihood Cox & Snell R2 Nagelkerke R2

Model 2 S.E.

0.642 0.631 0.748

0.038** 0.017** 0.022**

0.433 0.052 1.362 130.965 0.115 0.172

0.260 0.844 0.000***

b

Model 3 S.E.

b

Model 4 S.E.

2.135 0.540 0.076 0.678

0.054* 0.082* 0.837 0.035**

2.287 0.539

0.436 0.125 1.690 125.066 0.153 0.229

0.240 0.654 0.000***

0.438 0.117 1.711 125.108 0.153 0.229

0.670

Note: * e significant at 0.1 level; ** e significant at 0.05 level; *** e significant at 0.01 level.

0.007*** 0.083* 0.036** 0.237 0.674 0.000***

b

Model 5 S.E.

b

S.E.

1.992 0.991 0.171

0.069* 0.016** 0.649

2.327 0.969

0.007*** 0.017**

0.984 0.074 0.066 1.663 122.832 0.167 0.250

0.011** 0.785 0.807 0.000***

0.949 0.089 0.048 1.709 123.038 0.166 0.249

0.012** 0.742 0.858 0.000***

144

Y. Dang et al. / Habitat International 44 (2014) 137e145

may intentionally avoid sub-districts where land acquisition is costly and time consuming. Subway accessibility Surprisingly, all coefficients of subway accessibility are negative, although the results are only significant in Model 1. In other words, a sub-district with existing or planned subway stations within 3 km from the center of the sub-district are not more or less likely to be designated as sites for ECH projects. As can be seen in Fig. 2, most of the ECH projects constructed during 1999e2009 are located outside the Fourth Ring Road where subway stations are apparently scarce. This result contradicts our expectation that governments locate ECH projects near subway stations to reduce transportation costs for low-income residents. In fact, the model result suggests the opposite, i.e., that public transit accessibility is not a significant consideration for the city government when it plans an ECH project. Aware of this problem of insufficient public transit services, the central government reiterated its guideline in the 2011 policy document, mandating that local governments should “improve the public transportation system and construct on-site public services and facilities” near affordable housing neighborhoods (GOOSC, 2011). Prior to this regulation, instead of locating new ECH developments near a subway station, city governments often started building public transportation facilities near an existing ECH community only after residents had moved in and begun complaining about the traffic conditions. For instance, the Hui-Long-Guan community is one of the first 19 ECH projects in Beijing and the largest ECH project that had been developed so far. Although families had moved into Hui-Long-Guan by 2000, the first subway stations began operating after 2002 and they were located in the far south end of the community. Until 2006, only a few bus routes provided services between the stations with the community. Likewise, in the TianTong-Yuan community, another mega-project started in 1998 that was planned to house 180,000 people, there was no subway service connecting the community with the city center until 2007. Local housing needs The results for the three variables regarding local housing needs are complicated. First, the ratio of families with housing difficulties has no significant explanatory power across all five models, although the coefficients are negative in all models. One possible reason is that most sub-districts with serious overcrowding issues are usually located in Beijing's city center, where land prices are too high for the areas to be used for ECH projects for free. Second, the ratio of rental housing (or old public housing) is significantly and positively related to the odds ratio of a sub-district where ECH projects are located. That is, ECH projects are more likely to be located in sub-districts with a higher ratio of families who are renters, particularly those where most residents rent old public housing. Conclusions and policy implications Through the case study of the ECH program in Beijing, this paper empirically investigated the extent to which land-based fiscal interests of city governments may have jeopardized local implementation of affordable housing programs in China. Several factors have collectively contributed to the concentration of ECH projects in the fringe areas: the concern over losses of the land-transfer fees by using a site for ECH projects, the concern over additional costs e financial and administrative e occurred during the land acquisition process, and the inadequate attention to transit accessibility for low-income households. Most importantly, we demonstrated that a

site that potentially contributes more to local revenues has a lower chance to be designated for affordable housing construction. The empirical findings call for more attention to policy implementation by the literature on affordable housing policy, particularly the institutional mechanisms that prevent better policy outcomes. Although this paper only focuses on one particular aspect of a particular housing program (i.e., ECH), our findings have broad implications for the overall affordable housing policy system in China. Despite substantial differences in beneficiary targets and allocation procedures, the LRH, PRH and ECH are more similar than different in terms of the inter-governmental relationship that underlies policy implementation. As a matter of fact, city governments have to shoulder even greater financial and administrative burdens in LRH and PRH because, in addition to free land, city government also have to provide capital funds for constructing low-cost rental units. Since 2007, the central government created special funds to subsidize local LRH or PRH projects. But the total amount of such funds is hardly enough for even half of the construction costs, while city governments are responsible for the allocation, management, and maintenance of low-cost rental units. Nevertheless, findings from this research only reveal part of the story with regards to the complicated dynamics between landbased interests and affordable housing development. Whereas our model results confirm that locations with higher land prices have a lower probability to be used for ECH projects, subsequent impacts that locating an ECH project e or other low-income housing development e may have on land prices entail further research in the future. On one hand, certain attributes of affordable housing communities, such as lower income of affordable housing dwellers, poor quality of housing, insufficient supply of public services, or simply the fear for poverty concentration and social safety risks, may negatively affect the land prices of surrounding areas. On the other hand, our interviews have shown that city governments often combine ECH development with other local policy initiatives such as facilitating large-scale new-town development and stimulating real estate development in the suburban areas. In that case, locating ECH development may well serve as a means to drive up land prices in the urban periphery, which in turn generate higher land-based revenues for city government in the long term. Finally, our research by no means suggests a total failure of affordable housing policy in China. Rather, we contend that the national policy design has greatly improved by learning from past lessons. However, as we suggested before, policy makers have put more efforts to ensuring fair and transparent allocation of lowincome housing units but less on setting up the right incentives for local implementers. Certainly, the central-local fiscal arrangement, and consequently the “land-based public finance” of Chinese cities, is hardly the only institutional factor, albeit an important one, for so many “unintended consequences” in the Chinese affordable housing policy. The decentralized policy implementation has enabled city governments to take advantage of the “fragmented authoritarianism” in China (Xu, 2011) so that they can implement affordable housing programs in ways to advance other local policy agendas. The performance evaluation system over-emphasized the quantity (i.e., total number of units) of housing construction but ignored other aspects such as quality of housing, location, and accessibility. All these factors have played their roles in leading to many of the controversies revolved around the Chinese affordable housing policy, including the discriminatory location of affordable housing units e be them ECH, PRH, or LRH e across Chinese cities. Therefore, for the central government, a pertinent task is not to dictate upon all localities a “perfect” framework of affordable housing, but rather to initiate a comprehensive institutional design process so as to restructure the inter-governmental fiscal and

Y. Dang et al. / Habitat International 44 (2014) 137e145

administrative relationships and to enable self-enforcing implementation at the local level.

Acknowledgments Research partially sponsored by National Natural Science Foundation of China (No. 41230632) and Beijing Municipal Philosophy & Social Science Research Program (No. 11JGC125). References Cao, G., Feng, C., & Tao, R. (2008). Local “land finance” in China's urban expansion: challenges and solutions. China & World Economy, 16(2), 19e30. Chai, Y., Zhang, Y., & Liu, Z. (2011). Spatial differences of home-work separation and the impacts of housing policy and urban sprawl: evidence from household survey data in Beijing. Acta Geographica Sinica [Dili Xubao], 66(2), 157e166. Dekeseredy, W. S., Schwartz, M. D., Alvi, S., & Tomaszewski, E. A. (2003). Crime victimization, alcohol consumption, and drug use in Canadian public housing. Journal of Criminal Justice, 31, 383e396. Deng, L., Shen, Q., & Wang, L. (2011). The emerging housing policy framework in China. Journal of Planning Literature, 26, 168e183. Gabriel, S. A. (1996). Urban housing policy in the 1990s. Housing Policy Debate, 7(4), 673e693. General Office of the State Council. (2011). Guidelines on affordable housing construction and management (Document No. 45) Accessed 24.02.12 http://www. gov.cn/zwgk/2011-09/30/content_1960086.htm. Holmans, A. E. (2005). Housing and housing policy in England 1975e2002: Chronology and commentary. Report for the Office of the Deputy Prime Minister. Cambridge: Centre for Housing and Planning Research. January 2005. Huang, Y. Q. (2012). Low-income housing in Chinese cities: policies and practices. The China Quartely, 12, 941e964. Huang, Y. Q., & Jiang, L. W. (2009). Housing inequality in transitional Beijing. International Journal of Urban and Regional Research, 33, 936e956. Li, S. M. (2012). Housing inequalities under market deepening: the case of Guangzhou, China. Environment and Planning A, 44, 2852e2866. Lin, G. C. S., & Yi, F. X. (2011). Urbanization of capital or capitalization on urban land: land development and local public finance in urbanizing China. Urban Geography, 32, 50e79. Logan, J. R., Bian, Y., & Bian, F. (1999). Housing inequality in urban China in the 1990s. International Journal of Urban and Regional Research, 23, 7e25. Massey, D. S., & Kanaiaupuni, S. M. (1993). Public housing and the concentration of poverty. Social Science Quarterly, 74, 109e122. Ministry of Construction, State Plan Commission, & Ministry of Land and Resources. (1998). Directions on Promoting the Development of Economical and Comfortable Housing. Document MOC-154, September 14, 1998.

145

Ministry of Construction, National development and Reform commission, ministry of land and resources, & People's Bank of China. (2004). Regulations of the ECH program. Document MOC-77, May 13, 2004. Ministry of Construction, National Development and Reform Commission, Ministry of Supervision, Ministry of Finance, Ministry of Land and Resources, People's Bank of China, et al. (2007). Regulations of management of economical and comfortable housing (Document MOC-258, November 19) Schwartz, A. (2006). Housing policy in the United States: An introduction. New York: Routledge. State Council. (1998). Circular on further reform of urban housing system and speeding up housing development. Document No. 23, July 3rd. Stoloff, J. A. (2003). A brief history of public housing. Report for US Department of Housing and Urban Development, Office of Policy Development and Research. Washington, DC http://reengageinc.org/research/brief_history_public_housing. pdf.  in China's Tao, R., Su, F., Liu, M., & Cao, G. (2010). Land leasing and local public fiance regional development: evidence from prefecture-level cities. Urban Studies, 47(10), 2217e2236. U. K. Communities and Local Government. (2006). Planning Policy Statement 3 (PPS3): Housing. November 2006. Wang, Y. P. (2001). Urban housing reform and finance in China a case study of Beijing. Urban Affairs Review, 36, 620e645. Wang, Y. P. (2011). Recent housing reform practice in chinese cities: Social and spatial implications. In J. Man (Ed.), China's housing reform and outcomes (pp. 19e44). Cambridge, MA: Lincoln Institute of Land Polices. Wang, Y. P., & Murie, A. (2000). Social and spatial implications of housing reform in China. International Journal of Urban and Regional Research, 24, 397e417. Wang, Y. P., & Murie, A. (2011). The new affordable and social housing provision system in China: implications for comparative housing studies. International Journal of Housing Policy, 11, 237e254. Wu, F. L. (1996). Changes in the structure of public housing provision in urban China. Urban Studies, 33, 1601e1627. Xu, C. (2011). The fundamental institutions of China's reforms and development. Journal of Economic Literature, 49(4), 1076e1151. Yang, C., & Wu, C. (1996). Zhong Guo Tu Di Shi Yong Zhi Du Gai Ge Shi Nian (Ten-Year Reform of Land Use System in China). Beijing, China: Zhong Guo Da Di Publisher. Yeh, A. G. O., & Wu, F. L. (1996). The new land development process and urban development in Chinese cities. International Journal of Urban and Regional Research, 20, 330e353. Youth.cn. (2010). Do not let a heart-comforting program turn into a heartbreaking program. April 13, 2013 http://news.youth.cn/pl/201004/t20100413_1200460 Accessed 20.11.13. Zheng, S. Q., & Zhang, Y. J. (2010). Location patterns of social indemnificatory housing: theories, international practices and the reality in China. Modern Urban Research, 9, 18e22. Zhu, J. (2004). Local developmental state and order in China's urban development during transition. International Journal of Urban and Regional Research, 28(2), 424e447. Zou, Y. H. (2014). Contradictions in China's affordable housing policy: goals vs. structure. Habitat International, 41, 8e16.