Housing and Welfare Regimes: Examining the Changing Role of Public Housing in China

Abstract Following decades of deep marketization, Chinese housing policy has recently revived public housing provision, stimulating debate over whether China is moving towards a more socialized approach to housing. While such a move represents a remarkable shift in the role of housing as a welfare good or service, little research has considered recent developments in housing from a welfare regime perspective. Indeed, while the salience of housing systems to welfare regimes has been widely recognized in comparative research, this link has rarely been explored in the case of China. This paper applies a welfare regime framework to the historical development of Chinese housing system, and specifically, to empirically examine recent transformations in the housing system through contrasting examples from two Chinese cities: Beijing and Chongqing. The analysis indicates the emergence of a hybrid housing system combining socialist, social-democratic and productivist elements, focused on the chimeric needs of urban populations that are balanced between indigenous city residents and urban migrants.


Introduction
Esping-Andersen's "Worlds of Welfare Capitalism" (1990,1999) has provided a powerful framework for comparative welfare state analysis. Building on this, Holliday (2000) introduced a fourth "productivist" welfare regime, which has subsequently become dominant in explanations of social policy and welfare system development in the more recently industrialized East Asian economies (see Aspalter 2006;Choi 2012;Peng and Wong 2010). Despite the pre-eminence of China in this region, it has, nonetheless, largely been neglected in comparative analyses. This omission may in large part derive from China's late arrival and particular manifestation as an industrialized welfare regime in Asia. Nonetheless, over the last few decades, Chinese socio-economic development has adoptedunder an authoritarian communist political regimefeatures common to East Asian capitalist states. There has thus been a growing interest in placing China's welfare system within a more global welfare regime classification (Hudson, Kühner, and Yang 2014). Moreover, housing policy has taken on a particular importance in realignment, reflecting both distinguishing features of Chinese urban and industrial development, and common aspects of productivist and other welfare regime types (Doling 1999;Doling and Ronald 2014).
Indeed, while the position of housing in policy was undeveloped in Esping-Andersen's original theory, and incidental in Holliday's conceptualization, researchers have increasingly considered state approaches to housing as a welfare regime pillar (Barlow and Duncan 1994;Castles and Ferrera 1996;Groves, Murie, and Watson 2007;Harloe 1995;Hoekstra 2003;Hulse 2003;Kemeny 2001;Ronald and Kyung 2013). This paper thus takes housing as a focus for understanding broader shifts in Chinese welfare capitalism. Through applying a framework which links housing systems to welfare regimes, we assert that China is not so unique, with the Communist Party adopting an approach to both market and social policy reforms reminiscent of other capitalist welfare regimes. A specific objective then is to advance regime classification as a basis for understanding China's emerging housing system.
In the period of market-oriented industrialization, the Chinese government has tried to reshape patterns of housing provision to sustain rapid urban growth and economic expansion. For decades, local governments were reluctant to implement public housing programmes, leaving the vast majority of local residents, especially migrants, unprotected 1 (Wu and Webster 2010). The vulnerability of migrants has been highlighted since the recent global financial crisis, with unemployment becoming more concentrated in context of limited social security support. As such, in 2008, the Communist Party set out a core objective 2 to achieve a "Harmonious society", by targeting housing policy and establishing housing as a fundamental human right. Since then, there has been resurgence in public housing provision, albeit with considerable unevenness across cities and provinces. Early analyses have focused on empirical descriptions of recent changes in housing policy and supply (see Chen, Yang, and Wang 2014;Wang and Murie 2011). However, few studies have approached recent housing reforms in China from the perspective of an emerging welfare regime, despite the links identified in other societies. This issue is specifically redressed in this article.
In light of the absence of China from broader discussions, this paper seeks to connect the four welfare regime types to past and present transformations in the Chinese housing system, focusing on variations in, and recent implementations of public rental housing (PRH) provision. Essentially, from establishing a theoretical basis, we go on to apply it. This is done in the first half of the paper by examining different stages in the development of housing and welfare from 1949 to the contemporary period.
In the second half we apply the framework comparatively, focusing on key differences in two of the largest metropolises to illustrate important variegation within the political unity of China. Indeed, cities are a critical scale for understanding housing systems and welfare arrangements, and this paper, in contrast to the domination of country level studies, advances this analytical focus.
The paper proceeds in a number of subsequent parts. The following section addresses relationships between housing systems and welfare regimes, developing a framework for understanding how housing is differentiated in different regime contexts. Section 3 then approaches Chinese housing system reforms before 2008 in terms of three stages, applying our framework to the interpretation of housing policy changes at each stage. Based on the same framework, Section 4 explores contemporary housing in China, and in particular the revival of mass PRH, by comparing distinct processes at work in two cities: Beijing and Chongqing. Section 5 then explains divergence between Beijing and Chongqing housing systems from three perspectives: economic growth, demographic changes and institutional structures. The final section considers how our empirical findings and analysis of China contributes to theoretical debates on welfare capitalism and housing systems, as well as to a broader understanding of how housing in china is responding to social, economic and political realignments.

Ideal Types of Welfare Regimes and Housing Systems
The welfare regime literature has largely preoccupied itself with the three criteria by which Esping-Andersen (1990 originally differentiated welfare regimes: de-commodification, stratification and the market-state-family relationships. Based on these, three ideal-typical welfare state regimesassociated with societies in North America and Western Europecan be identified: social-democratic, conservative and liberal. To sum up, in social-democratic welfare states, universal social rights are given to a large proportion of the population based on citizenship, with the state dominating provision. The regime is thus highly de-commodified with low stratification. The market is often crowed out and the cost of raising a family is also socialized. In conservative welfare states meanwhile, the distribution of social rights is often based on class and status, consolidating divisions among wage earners. Social policies characteristically maintain social differentiation with individual welfare conditions modified by non-state providers: faith-based communities, trade unions, kinship networks, etc. Conservative regimes thus ensure both a measure of de-commodification and a high level of stratification. It is often shaped by the church and committed to the traditional role of family. Thirdly, in liberal welfare states, governments usually ensure limited well-being for the very poor, providing some welfare services based on means testing. The policy regime typically seeks to maximize the function of the market while minimizing the state's involvement. There is thus a low degree of de-commodification and high levels of stratification. Debates on welfare regimes have increasingly become more considered and diversified as they have been applied to a wider range of societies. A particularly early criticism was the neglect of Southern European (or Mediterranean), approaches to welfare that are dominated by family provision. Latin America has also provided an illustrative case of welfare regime variation with system features that ensure both low-level de-commodification and high-level stratification, with the family again playing a critical role (Gough and Wood 2004). South Africa, meanwhile, has also been considered a hybrid system with features of social democratic, corporatist and liberal welfare state ideologies, shaping a particular approach to housing (Venter et al. 2015).
In light of variations, Holliday (2000), Holliday and Paul (2003) extended Esping-Andersen's typology to the welfare arrangements of five East Asian welfare states, Hong Kong, Japan, Singapore, South Korea and Taiwan. This "productivist" welfare regime, he suggests, constitutes a distinct fourth regime type 3 and provides a challenge to Esping-Andersen's assumption that the universe of welfare capitalism contains only those advanced capitalist orders that take social policy so seriously as to be identifiable as fully-fledged welfare states (Holliday 2005, 1-5). Holliday argues that this position rules out examination of other developed capitalist states that engage in social policies but are also subject to other policy objectives, like economic growth and political legitimation. He suggests that in productivist welfare states, which are also advanced economies, economic objectives largely define social policies, with the application of social security as a means to target economically important interest groups. Priority in policy formulation is thus given to enhancing economic and social development, with state, family and market relationships aligned around these objectives. In this sense, some distinction can be made between developed societies where the focus of the welfare regime is "productivist" rather than "protectionist", with, in the former, measures of de-commodification being far less meaningful for understanding welfare conditions (Hudson, Kühner, and Yang 2014).
Apart from the overriding economic goal, Holliday (2000) points out that features of the productivist regime are variegated across countries. First, although large-scale public expenditure on welfare services is resisted, the state can be more or less active in extending the cover and quantity of social assistance. Through interventions, the state may either encourage the market to provide extensive social programmes or support families as providers, and by doing so promote legitimacy and diminish labour unrest. This means that levels of social protection can be fairly high in the productivist world, although the degree of de-commodification is low. Second, the degree of stratification is not always necessarily high, depending on the relationship between privileged occupational groups and the rest of the population. When the labour market favours a large working class, social policies may become more universalistic. Alternatively, when the labour market favours professional elites, social policies may become more particularistic, with the state directing limited supplies of welfare benefits. The productivist welfare regime thus focuses on human capital development in particular. Third, although market and family oriented social provision encourages or rewards economic participation, governments regard welfare regulation as their duty, resulting in differentiated patterns of state-market integrations in housing provision.
The family remains central to the actual delivery of welfare in productivist arrangements. In contrast to corporate or Mediterranean regimes however, relationships between the family, state and market are more distinct, with the state taking on considerable responsibility for sustaining the family as a welfare provider. Education and housing policies have played particularly important roles in this regard (Doling and Ronald 2014), with families expected to take advantage of opportunities provided by the stateusing educational infrastructure to improve employment and income, and applying subsidies (like home ownership schemes) to accrue housing market assetsto reinforce family welfare self-reliance.
While expanding the application of welfare regimes more comprehensively to economically developed societies, we should be cautious about the addition of a fourth, productivist type of welfare capitalism. Firstly, some studies point out that increasing social inequality, unemployment, immigration and the ageing of society in context of increasing democratic competition, may be undermining productivism, leading to enlargement of public social security provision in East Asia (Aspalter 2006;Peng and Wong 2010). Furthermore, with industrial upgrading, divergence between productivist and Western welfare regimes has diminished (Choi 2012;Hudson, Kühner, and Yang 2014). Nonetheless, although the primacy of high-speed economic development may have weakened in recent decades, researchers agree that economic growth continues to determine East Asian social policy-making, meaning that productivism still represents an effective framework for understanding welfare-systems and transformations, and may have particular relevance, especially in historic context, for China.

Connecting Housing Systems to Welfare State Typologies
Historically, welfare regime approaches have not considered housing as a core feature of the welfare state. Since Kemeny's (1995Kemeny's ( , 2001 application of Esping-Andersen's approach to housing systems, however, attention to housing as a social dimension and an intersection of the welfare mix has advanced (Allen 2006;Lennartz 2011;Kurz and Blossfeld 2004;Stephens and Fitzpatrick 2007). Specifically, Hoekstra (2003) has proposed, a more comprehensive framework, linking elements of welfare regimes with aspects of housing systems. Stamsø (2009) andVenter et al. (2015) have applied a similar framework to Norway and South Africa, respectively. In the East Asian context, Ronald and Doling (2010) have drawn connections between housing systems and the operation of welfare regimes, but have stopped short of integrating China into this analysis. Thus, while research has pointed to the centrality of housing systems in East Asian welfare practices, writers have only hinted at the connection in China (see Ronald 2013;Ronald and Kyung 2013;Wang and Murie 2011).
One of Esping-Andersen's core welfare regime concepts is de-commodification. Housing research has thus largely focused on housing tenure and its role in de-commodifying households, 4 assuming that homeownership and private rental housing commodify social relationsmaking the household more dependent on market provision and their position in the labour market to sustain their homewhile social rental housing has been understood to de-commodify (Barlow and Duncan 1994;Harloe 1995;Kemeny 1995). Doling (1999), nonetheless, points out the limitations of tenure in understanding de-commodification as, for example, owner-occupied housing might be dominated by state subsidy, may not necessarily circulate as a free market commodity and access may not necessarily be a function of ability to pay. Hoekstra (2003) therefore defines de-commodification as the extent to which households can access housing independent of income, and suggests measuring it by the scale of housing subsidies (both object and subject subsidies) and price regulation.
Another process by which housing systems impact welfare regimes is stratification: the way the welfare state more or less equitably distributes well-being among different social groups. Within housing studies, stratification has been largely translated as housing segregation and inequality. Although Musterd and Ostendorf (1998) have pointed out that welfare states have significant influences on housing segregation, most other studies focus on the dynamics of specific housing outcomes. Kemeny (1995) is more emphatic on this issue, suggesting that states may adopt either a tenure neutral or dualist approach, with the latter being more unequal, supporting deeper housing stratification. In making the link between welfare regimes and housing more directly, Hoekstra (2003) applies the "rules of housing allocation" in explaining how certain groups are favoured through regulatory privileges in the Dutch case. Hoekstra borrows a number of indicators from Esping-Andersen's 1990s model: the scope and organization of welfare provision; the level of the well-being quality and rules for accessing welfare benefits.
Building on the above, especially Hoekstra's approach to the welfare housing nexus and Holliday's definition of productivist welfare regime, we extend the link between housing system and welfare state regime to the Fourth (East Asian) World, translating criteria of de-commodification, stratification and state-market-family relationship into 10 aspects of the housing system (Table 1). Below we apply this framework in understanding development of the Chinese housing system in terms of different stages.

The Development of the Chinese Housing System and (Four) Welfare Regime Types
Bearing in mind that the four types of welfare regimes identified above are ideal types, and no single country or case purely corresponds to any of the types, in this section we attempt to position Chinese experiences.

Before 1978: A Socialist System with Conservative Elements
Between 1949 and 1977, public housing provision became increasingly dominant (Wang and Murie 1999). Under Communism, home ownership declined steadily and private housing was gradually transferred to public ownership via local governments, especially after 1956 (Wu 1996). While the state was ideologically communist, the idea that the public ownership of housing would help to maximize the use of all housing sources, complimented the rhetorical claim that state ownership would help eliminate social inequality and class exploitation (Wang and Murie 1999). Consequently, the Chinese housing system came to resemble a social-democratic regime but also reflected some conservative corporatist traits, especially in terms of the links between housing and employment.
Although transformations in tenure and ownership appeared to promote public welfare, the early communist housing system also critically embodied socialist state ambitions to nurture industrialization and working class formation (Selden and You 1997). Indeed, although the state claimed that it took responsibility to provide housing welfare, most of housing was combined with the employee programmes carried out by work units (danwei), 5 the financiers and providers of public housing. It also became evident that, while inspired by the Soviet Union, Chinese Communism demonstrated many social-democratic features (in Esping-Andersen's terms). For instance, China was committed to full employment, and by the 1980s the housing system was able to cover more than 90% of the urban labour force (and dependents) through public housing (Gu 2001). Moreover, in principle, the distribution of housing to employees was guided by more non-financial criteria. The council of housing distribution (established by both managers and employees) carried out the distribution work, attaching importance to people who needed housing the most, like those who had or were going to have bigger families living together or those who had been living in dormitories for many years. It also considered the needs of people who had ostensibly contributed more to society, like military veterans. Specifically it offered a 'universal' lifetime housing guarantee, which represented a large-scale de-commodification of the housing system (Wang and Murie 1999; Zhao and Bourassa 2003). Object subsidies were maximized at every stage of housing construction meaning rents could be set at very low levels, often almost zero. The formal market played a minimal role in housing provision and families were not responsible for accommodating themselves (Wu 1996;Zhu 2000). This is quite similar to other socialist economies where low-cost housing also functioned as a form of social wage (Huber and Stephens 2001). However, unlike western social democratic societies, there were few alternatives to public housing, especially in cities (Wu 1996).
With respect to work unit policies and provision, early housing interventions resulted in high levels of stratification, echoing the features of the corporatist system. First, the state only provided minimum basic housing assistance to urban residents who could not receive housing welfare either from the danwei or from their family: like disabled and unemployed people. On the other hand, work units provided better housing service to their employees. Second, as work units were the sole providers of public housing, the quality housing depended upon the financial capacity of work units. Thus, housing provision varied along the occupational hierarchy, with employees in collective units with permanent contracts prioritized. Third, within the same work unit, although public housing allocation was not wage-based, the shortage of public housing caused further stratification among employees in terms of corrupt allocation. Last, since rents were kept very low and government revenues were limited, the development of public housing fell behind housing demand, and consequently, housing quality declined rapidly (Wang and Murie 1999).

1978-1997: A Gradual Transfer to a More Liberal System
As the state excluded the market and maintained low rents, a heavy financial burden was brought upon work units, ultimately undermining provision (Wang and Murie 1999). Thus, pressure came to bear on the Chinese government to reform housing and to relieve the severe financial burden. Consequently, from 1978, after which the Chinese economy was transformed in line with a new vision of the role of the market in socio-economic development, social-democratic elements of the old housing system were transformed in terms of two core shifts in the housing system: decentralization and liberalization (Wu 1996;Zhu 2000). In terms of the former, work units gradually became independent enterprises focused on the market. Thus, from 1978 to 1997, the share of direct state investment in construction was drawn down from 78 to 6% (NBSC 1998). Instead, the state assigned responsibilities to work units and local governments for the production of new apartments, and assisted them by establishing preferential policies on land and taxes. Private homeownership began to be promoted and was the focus on new building, and existing PRH was increasingly sold-off to sitting tenants at very low prices (close to the construction costs) under various local housing privatization schemes.
Work unit housing, thus, did not disappear, but was provided in new ways that were not exactly liberal, but not entirely social democratic either. Essentially work units became more important intermediaries, and housing provision became more stratified in terms of advantages for work unit employees. The initial shift may thus be considered more corporatist. In terms of subsidies, the state established two new systems: The Housing Allowance System, requiring work units to disburse housing benefits to their employees as rental housing, and the Housing Provident Fund, requiring work units to contribute to a provident account for their employees that could be drawn upon to purchase a home (see Zhao and Bourassa 2003). As the construction and allocation of housing was occupationally based, housing stratification was intense. Additionally, as state subsidies were also given based on work unit rankings, small and private firms were less able to supply housing or improve housing quality for their employees. Stratification thus occurred among employees of different work unit ranks. Meanwhile, privileges for political and other elites continued.
The emerging system also reflected some liberal featureswith access to housing based on the market rather than citizenshipwhich were later extended. First, the state gradually increased rents to cover the costs of building and managing housing property. Second, financial criteria (i.e. income) were increasingly applied in housing allocation and distribution. Higher income residents were given the freedom to rent better and relatively more expensive dwellings in both public and private sectors. Thirdly, the state encouraged households to take greater responsibility for their own housing by facilitating saving in the Housing Provident Fund. Fourthly, the state diversified arrangements by which residents could purchase their own homes, and in addition to buying existing PRH, tenants could also buy economic affordable housing (a new type of subsidized owner-occupied housing), from their employers (Wang and Murie 1999;Wu 1996Wu , 2015.

From 1998 and 2008: A Liberal Housing System with Productivist Elements
Housing privatization and rent liberalization alleviated the public housing cost burden and, by the late-1990s, the state was in a position to deflect the objectives of housing reform towards a more radical marketization that also aligned with strategies to stimulate urban and economic growth (Wu 2015). A new era of liberalism really kicked off in 1998 with the termination of the lifelong housing subsidy and work unit housing. Instead the state developed a dualist housing supply system in which owner-occupied housing (including low-profit, limited price housing and economic affordable housing) dominated housing provision, while at the same time a small amount of rental housing (including old and unsalable work unit housing) was designated for the poor and disabled. Housing de-commodification was limited to the minimum.
The state not only reduced subsidies on low-profit housing, it also loosened its role in price setting. This left house prices to the whim of the market, which, in the emerging economic landscape of the 2000s, took a primary role in financing, constructing and allocating homes. In this context, the rental sector withered and owneroccupied house prices inflated rapidly, stimulating extreme housing stratification. Since housing allocation had shifted to a costumer-based system, a small group of richer people could increasingly monopolize supply, transforming housing into a speculative commodity (Wu 2015). This undermined the market housing choices of low-to-middle income households who were increasingly priced out, especially in cities. In the rental sector, meanwhile, housing allowances for the poor were limited and low-to-middle income households struggled independently to meet rents (Shi, Sato, and Sicular 2013).
While the emerging housing system framework was characteristically liberal, elements of the productivist regime were also evident. Firstly, with the high speed of urban growth and the wealth generated by it, the real estate industry became a central pillar of the Chinese economy, with the interests of government and the real estate sector increasingly aligned around growth. Secondly, providing suitable labour for economic growth became a concern of housing policy. During rapid urbanization and industrialization, the flow of migrants to cities increased annually (Wu and Webster 2010). As they were largely excluded from public sector housing, however, migrant housing access and affordability problems began to represent a threat to rapid industrial growth, putting pressure on local governments to act. In some places, experimental housing reforms were carried out to extend public housing provision to limited numbers of migrants in key labour sectors, in the interests of sustained economic expansion (Li and Zhang 2011). While migrant housing measures opened up the housing system and encouraged social solidarityechoing an earlier socialistic erathey primarily served productivist objectives with migrants essential to economic and industrial upgrading.
Although housing provided to migrant-employees is reminiscent of earlier work unit housing, it differed from the corporatist housing system in that the scale of provision was far more limited and housing provided to employees did not represent any social rights. The new system also reinforced a high level of housing stratification. First, as the number of beneficiaries was limited, most migrants were marginalized, and became concentrated in low-quality housing. Second, for those migrants given access, since benefits varied based on occupational contribution, high skilled and professional migrants could often acquire better-quality family housing, with others often allocated to cramped dormitories at their work place (Huang 2012;Wu 2002).

The Contemporary Urban Housing System in China
So far we have attempted to tease out different regime traits in the ongoing development and restructuring of the Chinese housing system up until 2008. Each period illustrates a dynamic transformation in which housing reforms and de/re-regulation reflect and reinforce wider welfare regime realignments and shifting responsibilities of the state, market and family households for providing or finding secure and adequate housing. The home appears to have transitioned from being a universal welfare service, to a market commodity, to, more recently, some combination of both. At each stage housing has played a differentiated but critical role in social stratification. Features of all four ideal welfare regime types have been evident, with the balance shifting during different periods of policy realignment.

2008 and Beyond: Ongoing Transformations Under Debate
As housing commodification in China in the early 2000s resulted in a highly volatile and over-heated real estate market, distorting housing supply and creating significant affordability problems (Wu 2015), since 2008, the state has adopted various new regulatory housing strategies to curb speculative investment in commercial housing, constrain house price inflation and prioritize the development of social housing to a certain degree. PRH has been revived and has become a core means of serving the housing demands of urban low-to-middle income residents, including both local residents and migrants. Housing policy and provision has thus lost much of its tenure bias, moving towards a more balanced regime featuring both pro-homeownership and pro-PRH schemes. Indeed, housing reforms did not purely target the development of social welfare, and they also dealt with the economic shocks of the global financial crisis. The assumption was that the mass construction of public housing would have two main benefits: sustain GDP growth and reduce housing shortages (Wang and Murie 1999). In light of this "double" purpose, the state has begun to abolish, through hukou 6 reform, restrictions on migrant access to PRH, and focused on a more universal allocation of housing and housing benefits since the 11th National People's Congress in 2009.
Even though housing reforms have increased public housing provision, there has been some debate about whether reforms really aim to provide a universal housing service for all residents (Zhu 2014). Some researchers regard recent reforms in terms of a transition towards a more just and equitable society (see Stephens 2010). Chen, Yang, and Wang (2014), for example, argue that in recent years the Chinese public housing model has developed in terms of the overall transformation of development ideology from "productivist welfare" to "developmental welfare". Nonetheless, some scholars argue that the housing system retains key productivist traits. In general, they suggest that housing in China continues to be understood as an economic issue rather than a project to enhance social harmony (Huang 2012;Zou 2014). Wang and Murie (2011) have recently suggested that the Chinese approach to housing policy is becoming a distinctly hybrid one. A core objective of this paper then is to contribute to this analysis by applying the framework set out earlier in this paper. While Wang and Murie's study mainly focuses on the other two types of public housing in China, affordable housing (to buy) (jingji shiyong fang) and government assisted rental housing (lianzu fang), our analysis focuses on public rental housing (gongzu fang) (PRH), one of the more common types of public housing with more universal applicability. PRH is the only type of housing that is in principle open to both local residents and people without local hukou registration (migrants). Moreover, since the end of 2013, local governments have been required to merge lianzu fang programme into the PRH programme, and have also started to merge other types of housing into the PRH programme. To a large extent therefore, our focus on PRH provides a wider perspective on the broader and ongoing development of the contemporary Chinese housing system.
We pursue our goal empirically by considering recent PRH system transformations in two Chinese cities: Beijing and Chongqing. These contexts characterize two different approaches within China to PRH provision. Historically, analyses of either housing systems or welfare regimes have not dealt particularly well with territorial scale and have tended to focus on countries as units of analysis (see Hoekstra 2003). This critique is particularly salient in the case of China due to scale and diversity in urban development. As the policies of administrative and financial decentralization have intensified since the 1980s, in the contemporary period, public housing provision responsibilities have mostly rested with local governments, but under strong central government directives (Chen, Yang, and Wang 2014;Zhu 2014). Local governments plan and set up rules for housing provision based on individual situations (Aspalter 2006;Peng and Wong 2010). Therefore, the following city level analysis provides a more meaningful scale and a comparative scope from which to assess the complex role of PRH in housing and welfare regime restructuring in China.

The Cases of Beijing and Chongqing
4.2.1. Socio-economic development in Beijing and Chongqing. Although PRH policies vary between Beijing and Chongqing, both cities converge in a number of important respects. On the one hand, both these mega-cities have populations of over 10 million and have undergone relatively recent and exceptional modernization, industrialization and economic growth (GDP Annual Growth Rate over 16%). They have also been subject to similar housing system pressures with rapid economic expansion resulting in increasing inequalities in income and access to market housing. Both cities have also struggled to accommodate a large and growing urban population driven by the arrival of migrant labour (approximately 50% of the urban population). Migrant populations in Beijing and Chongqing also share a number of common features which indicate that urban migrants have a strong desire to settle down and enjoy the same citizenship rights as local residents: in 2010, more than half have stayed in cities for at least three-years; around 40% are aged 30 years old and under, and over 70% of them have arrived with families (BMBS 2012; CMBS 2012) 4.2.2. Comparing housing system characteristics between Beijing and Chongqing. Table 2 compares housing policies announced in Beijing and Chongqing drawing on the framework provided in Table 1. We find, in both cities, their PRH policy rhetorics for local residents are ostensibly alike and are significantly socialdemocratic-like, and although their approaches to migrant housing contrast significantly, the mechanisms behind them reflect quite similar productivist elements.
In terms of housing policy for local residents, both cities have adopted strategies of maintaining social stability as their overall housing policy objective, representing elements of a social-democratic regime. Thus their degree of housing de-commodification is relatively high, with housing stratification among local residents being relatively low. Most local households can, in principle, access certain housing schemes and subsidiesespecially PRHbased on their local hukou citizenship. Except hukou, other non-financial criteria are adopted to help local residents, especially those who are not homeowners, to access PRH. For example, in Beijing, around 60% of local residents are covered by the pension plan, and have the right to apply for PRH in 2012 (BMBS 2013). Additionally, according to the regulations concerning PRH access in Beijing and Chongqing, comparing the income limits with the average income in both cities in 2012, the scope of PRH provision in Chongqing covers all income groups, while it is more focused in Beijing. Lastly, both city governments strongly regulate housing prices, but at different levels. Whereas PRH rents are 20% below market price in Beijing, in Chongqing the figure is 40% (CAG 2012). Although strong on policy rhetoric, in terms of de facto PRH provision, Beijing is typical of major cities that have done relatively little to provide PRH for local residents. Chongqing meanwhile, has focused on providing public housing for local residents. In 2012, Chongqing authorities built 140 thousand units of PRH, 13 times more than Beijing (BMBS 2013;CMBS 2013). By 2013, the volume of PRH in Chongqing represented more than all the other major cities combined.
Unlike the policy on local residents, regulations on access to PRH for migrants appear to be more strongly different. The degree of housing de-commodification for migrants in Beijing is relatively lower than in Chongqing. Beijing has implemented a series of additional requirements that restricts access. In Beijing, only migrants who have consistently been employed and participated in the pension plan and Housing Provident Fund, as well as paid taxes for at least five years in the same district, are eligible to apply for PRH. Moreover, the Beijing government only allows migrants to apply for PRH through their employers, and gives priority to migrants who work for companies with a higher political and economic status. However, Chongqing has applied relaxed rules on access to PRH to encourage migrants take up. Migrants with a one year work contract or even retired migrants are allowed to apply for PRH independently. Consequently, a report from the Beijing public housing centre shows that in 2012, only 1% of the tenants in PRH were migrants in Beijing, while the proportion rose to over 45% in Chongqing in 2011(Kaifeng Foundation 2011. In terms of affordability, PRH is not really affordable for low to middle income migrants in Beijing, as of 2012, the rent-to-income ratio was around 0.9 even for below-middle income, local residents who unusually have higher incomes than migrants (calculated from BMBS 2013), 7 . Therefore, many migrants who are occupationally and financially impoverished are marginalized by the PRH system in Beijing.
Variations are also revealed in terms of the respective roles of state and market in providing housing. The Chongqing municipality is more active than Beijing in organizing housing construction, although both municipalities play a regulative role and have established preferential land and tax policies to support the construction of PRH. The Chongqing municipality also largely facilitates family access to PRH through providing a housing allowance. In Beijing, the municipal government encourages multiple sectors, like real estate developers, social institutes, industry parks, etc., to take responsibility for the financing and contracting of PRH programmes; while in Chongqing, PRH construction is still controlled by the government, through the management of two specific state-owned-enterprises.

Understanding Divergence Between Beijing and Chongqing
Despite variations between Beijing and Chongqing, we argue they represent similar features of a productivist regime, aiming to reinforce economic objectives: i.e. productive elements of the population who can contribute most to economic growth (either individually or in terms of their contribution to a particular industrial sector). We now pay specific attention on the mechanism behind different PRH policies towards migrants in order to establish whether both cities reflect similar elements of the productivist welfare regime (i.e. housing policy facilitates the economic growth). Holliday (2000) emphasizes the need to understand the emergence of different clusters of welfare regimes from the perspective of social base, economic base and political superstructures. Building on this notion and in context of the analysis above, three factors are particularly relevant in explaining the Chinese housing system: institutional restructuring, demographic changes and economic development. Below we explore how these three factors interact in shaping housing systems, and, more specifically, how these processes help explain divergence between Beijing and Chongqing housing systems.

The General Context: Decentralized Provision of Public Housing
Prior to the 1990s, as the Chinese economic transition had brought about an increase of the urban population, housing demand intensified in cities. Local and national authorities have thus, in principle, had a shared interest in expanding housing welfare. Both central and local governments sought to develop public housing in terms of two core objectives: firstly economicto house the workforce necessary to The municipality input a few public expenditure The municipality grants a few preferential policies on land distribution and tax cuts The municipality grants strong and many preferential policies on land distribution and tax cuts Price setting Strong influence on price setting, based on market value Very strong influence on price setting, based on market value e.g. PRH, 40% below market prices e.g. PRH, 10-20% below market prices sustain urban developmentand secondly socio-politicalto ensure housing security for the urban poor, maintaining social stability. However, in context of the institutional restructuring since 1990slocal government motivations to invest in public housing have become more complex (Chen and Gao 1993). In the interests of national economic growth, the central government has increasingly exercised political authority, instructing local governments on their political and fiscal responsibilities. However, this has produced a division in financial incentives, which has become a particular barrier for achieving national policy goals at the local level. As the central government has taken a larger share of tax revenues and reduced its commitments, local governments have received a smaller share of tax income, but been given greater responsibilities. Local governments receive around 40% of tax revenues, but their share of welfare and infrastructure expenditure is around 70% (Chen and Gao 1993). Local governments thus have different pressures on them and often use the autonomy appointed to them by the national government to resist central state goals of expanding housing welfare provision. Indeed, as the financial benefits of increasing commercial housing have competed with PRH (in context of diminishing local tax revenues), local governments may be more interested in leasing land for building for-profit housing rather than "wasting" land on public housing (Zhu 2014). Indeed, local governments often respond quite passively to the targets and objectives of unfunded state mandates. Although the state evaluates local government achievements in the promotion of PRH, it is of secondary importance to economic growth. Thus, to some extent, although local governments have some incentives to compete in housing provision to see who can provide more and better services, incentives depend upon the economic and political benefits that the PRH programmes potentially offer local officials. Due to the decentralization process, local governments also have become independent in deciding how housing welfare is constructed and which groups qualify. The structure of housing provision provides us a basic context for understanding local governments' policy practices in the two cities. Within this context, below we explain how political and economic incentives of local governments shaped the housing strategies on migrants in Chongqing and Beijing.

Differentiated Local Governance and Institutional Incentives
In Beijing, PRH provision is further decentralized to a lower administration level, the local districts, who are required to supply land and finance for the building of PRH. In this process, the conflict mentioned above, between the national and local governments, re-emerges between municipal and district governments. In context of limited funds, local districts try various ways to minimize their obligations for PRH, exaggerate the amount of PRH provided and keep their fiscal subsidies minimal. For instance, they make their specific rules for PRH access, and are very reluctant to supply PRH to residents who are registered in other districts. They often count housing built by the private sector on suburban-rural land, which is identified as illegal according to the Chinese management land law, in the PRH figures. They also rely on the investment from other sectors, and mainly realize their obligations for PRH in cooperation with private housing developers, based on the exchange of a small amounts of PRH (usually 5-10%) for big quotas of commercial housing (95-90%) construction. Critically, the three actors involved -Beijing municipality, local districts and private developersdo not share motiva-tions for providing PRH for migrants leading to conflicts and often resulting in ineffective policy implementations.
By contrast, Chongqing has demonstrated unprecedented enthusiasm towards the PRH programme as it also aligned with objectives in political and economic spheres. Bo Xilai, the secretary of the Communist Party's Chongqing branch between 2007 and 2012, used his mandate in Chongqing as a platform to significantly further his own political ambitions at the national level (Miller 2012). Bo was subsequently convicted of corruption and sentenced to life in prison. Before his downfall, however, the giant PRH programme was one of the five core innovations considered to help Bo win both local support (in Chongqing) as well as muster wider political recognition in order to help him bid for a seat at the top table of Chinese politics at the 18th party congress in 2012 (Miller 2012). In order to provide political incentives, the PRH programme in Chongqing was run in a rather radical way. Instead of decentralizing the provision task to local districts, the main actors of the programme have been the municipality and state-owned-housing enterprises, with the latter being the actual provider of PRH, but also a client of the former. The state-owned housing enterprises were heavily empowered, with the municipality prioritizing them in policies and in the allocation of housing land, as well as guaranteeing bank loans for them in the large-scale production of PRH (40 million square metres of PRH) over three years. Although the government claimed that the PRH programme in Chongqing was an approach which combined public ownership with a market oriented economy for satisfying public needs (Huang 2011), scholars and social media generally agree that it was the coercive power of Bo that efficiently mobilized tax breaks, the supply of land and bank loans in favour of these goals (Cheng 2015). After the downfall of Bo, most of his other programmes were cracked down upon, although the PRH programme has been continued by the incumbent mayor, Mr Huang Qifan. Initially, this was because the programme remained consistent with state goals on housing.
However, we still need to demonstrate some caution in regard to the sustainability of the PRH provision in Chongqing (Zhou and Ronald 2016). First, there are concerns over how state-owned-housing enterprises will deal with the high debt associated with the PRH programme. According to the original plan, from 2016, the government should have started to gradually sell off PRH units to sitting talents in order to begin repaying their huge bank loans. However, by August of 2016, the government had not announced any new relavant regulations. Unless the PRH programe pursues other further financing strategies, it may not be able to repay the loans as the government expected. Moreover, it may make it more diffiuclt for the state-owned housing enterprises to get more loans for building further commercial and residential facilities near the newly built PRH communities. It has also been revealed that some PRH neighbourhoods still have poor access to public transport (Kaifeng Foundation 2011). Second, there have also been questions as to whether the general public has really benefited from the PRH programme? In order to further highlight the achievements of Bo, the municipality has often exaggerated official statistics in regard to the amount of PRH built. For instance, from 2010 to 2011, 18 thousand housing units built by either non-public providers (such as worker's dormitories and student hostels) or in remote counties has been counted in the total amount of urban PRH (110 thousand units) (Chongqing Morning Post 2011). As the state-owned housing enterprises' ability to find finance decreases, the government also has to rely more on the industry parks to build more new PRH in order to maintain a stable supply, and unfortunately, these houses will only be provided to employees working in the industry parks.

Differences in the Urban Economy
Apart from differences in political incentives, differences in urban economies have also contributed to discrepancies in housing policies for migrants between Beijing and Chongqing. Both the state and local governments consider a rapid economic growth as the primary criteria in evaluating the success of local development, and as expanding PRH for migrants has also represented a means to improve the urban economy. Nonetheless, Beijing and Chongqing have established their own PRH provision programmes based on distinctions in economic and industrial bases, with migrants who do not contribute to the local economy in each case more likely to be excluded from the public housing system.
As Beijing has established itself as the centre of the high end production services, it has sought to build a highly professionalized labour force featuring entrepreneurial talent (BMCDR 2010). However, as Figure 1 shows, there are few migrants with or above senior high school education levels in the labour market and few migrants have skills that are recognized by the city of Beijing as contributing to the advancement of industry and economic growth. Consequently, although the hukou barrier for housing access has been eliminated in the 2009 Housing Act, the municipality has established stringent regulations (see Wang and Murie 2011), that only allows the very elite of the migrant population access to public housing, excluding migrants with low educational or occupational status (Huang 2012 In contrast, as a centre of manufacturing in China, Chongqing is at a very different stage of industrial development with an exceptional need for workers and land in comparison with most other cities (Huang 2011). For example, in 2012, around 10% of the Chongqing labour force worked in knowledge intensive industries compared to 30% in Beijing. Meanwhile, over 80% of Chongqing's employees work in labour intensive industries, with average salaries around 60% of those working in similar occupations in the more developed coastal provinces (BMBS 2013;CMBS 2013). Within a broader context of labour shortage among Chinese big cities (Chen and Du 2011), Chongqing has prioritized its competency in secondary and service industries, leveraging its relative strengths in the labour market for achieving a higher GDP growth. Thus, a more inclusive PRH allocation scheme has become crucial for attracting migrants, and has helped maintain a sustainable labour supply for the continued urban development. As Figure 1 shows, Chongqing's migrant population contributes a labour force more suited to the manufacturing base of the city. Additionally, Chongqing estimated that investment in the PRH programme (110 billion yuan) would boost GDP by 400 billion yuan, and would attract 910,000 workers to the region (Xinhua net 2011). To some extent, providing PRH for migrants helped Bo create an 'economic miracle' in Chongqing. In the five years that Bo was in power (2008)(2009)(2010)(2011)(2012), the average GDP growth rate in Chongqing was 1.5 times that of China (NBSC 2009, CMBS 2009. This economic achievement also contributed to Bo's political campaign. Overall, on examining mechanisms behind the distinct migrant housing regimes in Beijing and Chongqing, it appears that it was often more the political and economic incentives of local officials rather than public needs that guided the formation of the PRH programmes in China. Both cities exemplify an emergent productivist welfare regime, but have different features due to their different degrees of socio-economic development and variations in governances. The variation between them confirm, to a large extent, Holliday (2000) theory of productivist welfare, in which market's involvement and the degree of stratification can be either high (i.e. Beijing) or low (i.e. Chongqing). The analysis of the two housing systems then has provided considerably more insight into the divergence around the current housing policy in China, bringing us a broader picture of Chinese urban housing and welfare regimes.

Conclusion
By applying the framework developed by Hoekstra (2003), this paper has linked the housing system in China to the welfare regime theories elaborated by Esping-Andersen and Holliday at a more global level. Our review on the development of Chinese housing system indicates that, although China developed from a socialist welfare regime to a more market orientated one, the Chinese government has continued to adjust the housing system in line with social protection and social investment motivations. Housing policies and approaches in China have shared a number of features that reflect both western and eastern welfare regime constellations, with these features varying across different stages of economic growth.
Our city level comparison on contemporary PRH provision points to the salience of urban scale in understanding relationships between housing and welfare practices as well as the significance of diversity within a single national context. While variations have been observed, researchers have tended to use one type of welfare regimeeither liberal, productivist or developmentalto classify the current housing system in China (Chen, Yang, and Wang 2014;Choi 2012;de Haan 2010;Stephens 2010). However, we argue that the cities of Beijing and Chongqing share a broadly similar pattern of public housing supply in the sense that their PRH system flows into two sub-systems, a local citizenship-based and a migrant-based system. The former is very much similar to the universalistic system that Esping-Andersen proposed, while the later appears closer to the idea of the productivist welfare regime, with PRH provision subordinate to the policy objectives of economic growth. On the one hand, the expansion of PRH for local residents represents more protective intentions and movement (back) in the direction of a social-democratic-like regime. Strong political and institutional control is also strongly indicative of a resurging protective concern in the Chinese welfare framework (Zhou and Ronald 2016).
On the other, while also protective, public migrant housing provision in Beijing and Chongqing reflect more productive functions that align with a productivist welfare regime ideal type: the reinforcement of the position of productive elements in society and state-market-family relationships directed towards economic growth. Although the logic of PRH provision for migrants has been similar in Beijing and Chongqing, distinct patterns and discrepant scales of housing provision have been developed. Chongqing has sought to significantly extend an open access public housing system. The direct motivation behind its PRH programme was the unusual political incentives aligned to the Bo Xilai administration. The programme sought to attract a substantial flow of labour to the city in order to serve the continued growth of secondary and service industries. Indeed, PRH output since 2011 represents, notwithstanding some concern with the accuracy of the figures, a large chunk of the global supply. Beijing, meanwhile, has maintained a modest supply of PRH for skilled migrants in order to promote an occupational upgrading within the city's internal economy.
Overtime, the Chinese housing system has transformed from a more socialistic to a more corporatist, to a more liberal one, and is now coming to resemble a hybrid system combining both social-democratic and productivist regime elements. While Holliday (2000) has emphasized the link between economic growth and social policy in East Asia, which is considered stronger than in western countries, it is a link that is particularly evident in China, and especially in the housing system that has helped shape and reshape both the market and welfare conditions and practices.  Since the 1980s, as many as 300 million people have migrated from rural to urban areas. These migrants are identified as non-local residents according to the Chinese household registration system (hukou). Migrants are thereby discriminated against, with access to housing and other benefits restricted.

2.
At the seventeenth Chinese Communist Party congress.

3.
We can consider productivism as a fourth world in light of other claims where Mediterranean and Latin American regimes may also be considered a fourth and fifth type (see Esping-Andersen 1999; Gough and Wood 2004).

4.
Another generality has been concern with the production, rather than consumption of housing, defining de-commodification as a way of suppressing profits from the construction process.

5.
A general term for publicly owned organizations that employed most urban residents, including state owned enterprises, civil associations and governmental organs. 6.
Hukou is the Household Registration system enacted in 1958. Personal hukou status is classified by original residential location (such as "agricultural hukou"/"non-agricultural hukou"). A key outcome is that access to local resources, especially to public housing, are dependent on having a local hukou status. 7.
Based on the calculation from Sina Finance (http://finance.sina.com.cn/g/20111020/ 171210659082.shtml) the monthly rent for the first PRH programme was 1715 yuan in 2012.
According to the Beijing Statistical Yearbook (2013), in 2012, the average income of the lowest two income deciles were 1569 yuan and 2415 yuan, respectively. Rent-to-income ratios of the respective groups are 1.09 and 0.71, and, therefore, 0.9 on average for both.