Does homeownership impair employment? Evidence from migrant data in transitional China

Abstract

China now has a higher rate of homeownership than most country counterparts in the world. How might have China’s high homeownership rate affected the employment outcome and efficiency of labor markets? In this study, ‌we challenge the theoretical premise that homeownership reduces labor mobility and employment flexibility by examining China’s migrant data‌. Our key empirical finding is that, at the individual level, compared to renting option, homeownership in destination city does lead to a lower employment probability. Older workers, urban hukou holders, and migrants in destination cities with more than 5 million population are more impaired in employment. The possible mechanisms underlying lower employment propensity lie in that homeownership impedes labor mobility, resulting in lower wage prospect from constrained labor market area which further discourages job search effort. Furthermore, purchase costs associated with homeownership may also lower household disposable income, making it harder to venture into self-employment business. Our study sheds new lights on the importance of incorporating employment creation element within current housing policies, particularly in rapid urbanization era of a transition economy.

Introduction

This article aims to examine the relationship between homeownership rate and employment rate and its underlying mechanism. For the past several decades, researchers have paid increasing attention to the relationship between housing market and labor market, particularly after the onset of the 2007-2008 economic crisis (Laamanen 2017). Since embarking on marketization and housing reform in the late 1970s, China has witnessed a striking surge in homeownership (stood at 89 percent in 2017 China Migrants Dynamic Survey (CMDS)). This rate is much higher than most country counterparts.¹ Wang (2011) argues that the housing reform beginning in 1994 set the momentum to transform China into a country with one of the highest rates of homeownership in the world. The house-buying frenzy was fueled by the government policy encouraging urban residents and rural migrants to purchase house in order to boost real-estate industries and promote urbanization. Reasons for China’s high homeownership are various, ranging from institutional, cultural to economic one.² Our focus is on exploring how homeownership affects labor market efficiency in developing China. Specifically, high house ownership rate is detrimental in raising unemployment rate (Oswald, 1996; Brunet and Yves Lesueur 2009; Wolf and Caruana-Galizia 2015), or beneficial in reducing unemployment probability (Van Leuvensteijn and Koning 2004; Munch et al. 2006; Mikula and Montag 2023)? This question is relatively less studied in the existing literature.³ Using individual migrant data (CMDS 2014 and 2017), we examine the relationship between homeownership and labor market outcome, in particular, how homeownership affects individual employment profile. We further explore the underlying mechanisms, such as the impact of homeownership on labor mobility, personal wage earnings and different employment activities (wage work and self-employment) in Chinese context.

The reason for selecting migrant group for this study lies in that, in addition to data availability, unlike local urban Hukou residents’ employment closely relating with house tenure, this group was seldom influenced by urban housing system reform occurred during the past four decades,⁴ which to some extent reduces the degree of endogeneity of housing tenure in employment function estimation. In addition, they are more sensitive to changes in labor markets than urban native workers and hence fits well into the sample of studying employment impact of homeownership in China.

Our study contributes to the literature on labor market and housing market in a number of aspects. First, to our knowledge, this is the first study concerning the impact of homeownership on employment in transitioning China using migrant data. Some previous studies have analyzed the impact of house wealth on labor market (Fu et al. 2016; Li et al. 2020). While the homeownership variable in our analysis includes house wealth aspect, it encompasses broader attributes (such as distance to workplace, mortgage usage for self-employment activity) and can make a comparison with renting house option. In addition, while other studies have focused on labor force participation (Fu et al. 2016; Liu and Yang 2020), our work advances the literature by interrogating how employment status interacts with wage disparities and welfare gaps—a critical yet understudied dimension in China’s migrant labor market. Furthermore, we explore the pathway mechanisms of homeownership impacting on employment via the channels of labor mobility, wage and employment activities. Finally, we address the endogeneity estimation problem of housing ownership using novel instrumental variables (regional homeownership rate for migrants and gender of first child), and conduct several robustness tests, which could be helpful for similar studies in other developing countries.

We find that for migrants in China, compared to renting option, homeownership does lead to lower employment rate. The possible mechanisms causing lower employment may lie in declining labor mobility, lower wage due to narrower search boundary discouraging employment propensity, and dampened entrepreneurship activity due to higher expenditure on housing purchase. Our study sheds new lights on labor market policy (reducing unemployment rate by means of appropriate housing policy), housing policy (providing more renting property or rent subsidies for workers if homeownership hampers job finding) and urbanization policy (reconsidering if homeownership is a necessary condition for urbanization and acquiring permanent residency status in cities for large number of rural to urban migrants in China.⁵). To reduce unemployment and foster urbanization, findings from our study imply that we should encourage freer labor mobility across areas and reduce barriers, which would suggest adopting an alternative housing policy. Rather than encouraging house buying as being strongly advocated in current policies,⁶ a sensible policy might be encouraging renting option, especially for migrants.

The rest of this study is organized as follows. The next section reviews the literature concerning homeownership and employment. Section 3 briefly characterizes the background of housing reform, real-estate market and labor market establishment in China. Section 4 describes data and presents empirical specification. Section 5 discusses estimation results. Section 6 provides conclusions and policy implications.

Literature review and alternative hypotheses

The pioneering work by Oswald (1996) establishes a correlation between high homeownership rates and high levels of unemployment. Henceforth a large number of studies have tested the “Oswald hypothesis” and the empirical evidences in the literature concerning the impact of homeownership on labor market outcome are mixed. Aggregate level studies generally suggest a positive correlation between homeownership and unemployment, both within and across countries (Oswald 1996; Green and Hendershott 2001; Blanchflower and Oswald, 2013; Tohmo and Viinikainen 2023). However, at the individual level, the results are blurred: many studies find that the incidence of unemployment is lower for homeowners than for renters (Van Leuvensteijn and Koning, 2004; Munch et al. 2006; Coulson and Fisher, 2009; Rouwendal and Nijkamp 2010), while still some studies show lower transition into employment with a distant move for homeowners (Battu et al. 2008; Taşkın and Yaman 2019), hampering employment. However, by now most of the empirical studies pertain to developed countries.

Homeownership hampers employment

Homeownership may impair labor market outcome in several ways. The most prominent reason may lie in declining labor mobility due to homeownership. The long time involved and high transactions costs entailed in selling and buying a house makes job change difficult and less likely. Workers are often trapped in local labor market due to transaction costs associated with buying and selling their properties. According to Catte et al. (2004), the transaction costs for selling medium-sized houses in the U.S. are around 9% of the sale price, whereas in the European countries the transaction costs are higher, above 10%. Similarly, renting public housing also constitutes a barrier to mobility as job choice of renters is restricted by the location of available public housing. If not moving to where the job is, high commuting cost to the thriving region also makes some jobs untenable. Blanchflower and Oswald (2013) find that high homeownership areas have longer commute-to-work times, which are expected to raise transport costs for both employers and employees. Increased commuting time also reduce job satisfaction through deprivation of leisure and family time. Even though some homeowners can rent an apartment near the work place⁷ to save commuting time and cost, the cost of renting plus up-keeping another owned house is also considerable. Manning and Petrongolo (2017) find that the attractiveness of jobs to applicants sharply declines with distance, suggesting the existence of locally confined labor markets. As a result, homeownership leads to long job tenure (less quitting) and harder to find a nearby job, hampering labor market efficiency and employment outcome. Lisi (2024) studies the steady-state equilibrium of the labour marke and find that an increase in homeownership increases unemployment in the long run.

Regarding macro evidence of homeownership’s impact on labor mobility, Hamalainen and Bockerman (2004) find that net migration (moving in minus moving out) to a region in Finland appears to be depressed, ceteris paribus, by a greater level of homeownership in that region.⁸ Modestino and Dennett (2013) find that one standard deviation increase in the share of homeownership in the origin state reduces the outflow of migrants from the origin to destination state by 2.9%. Hence homeownership in both origin (moving out) and destination (moving in) areas will inhibit labor mobility. In this study we intend to explore the impact of destination homeownership at individual level on migrant’s residency aspiration and employment in China, as origin homeownership rate in the countryside in China is very high and it did not prevent millions of rural workers from migrating to cities to seek jobs.

In connection with reduced mobility, homeowners would need a higher wage to cover the house transaction costs when changing job. A higher reservation or asking wage in turn will make job-hunting more difficult and presumably lead to lower employment rate. On the other side, the lower wage available in the narrower market place also dampens labor force participation rate and employment probability.

Moreover, with money being spent on house purchasing rather than on business venture and innovation, homeownership may have negative impact on innovation and entrepreneurship, which in turn reduce employment creation. Bracke et al. (2012) find that while homeownership and entrepreneurship appear to be positively associated, a more careful fixed effects estimation suggests that homeownership reduces the likelihood of entrepreneurship by 25%. Similarly, some researchers have found that high housing price reduced the likelihood of entrepreneurship and business venture in China (Li and Wu 2014). We will examine the channels of wage and self-employment activity in the following mechanism analysis.

House owning has investment as well as consumption attributes. Those who acquire more than one house are often driven by investment motives. Facing rapidly appreciating housing market, the owner with extra houses would rather stay out of the labor force and make a living on renting property owned and speculating in the real estate market (in a sense, taking up the “job” of property management and speculation). In addition, a higher home equity mitigates the need to quickly find employment if being displaced, which, in a sense, entails a wealth effect so that it is affordable not having to work to make a living. This aspect implies a negative influence of household property income on labor supply. Indeed, Fu et al. (2016) find that a 100,000 Yuan increase in housing value leads to a 1.37% decrease in female homeowners’ probability of participating in the labor force. On the other hand, not being able to sell a house (or cannot sell it at a sufficiently high price to cover a deposit needed for purchasing the next house) may cause homeowners to be ‘locked’ (or “trapped”) into a house, even though an appropriate job is available elsewhere (Head and Lloyd-Ellis 2012). Furthermore, high loan-to-value ratios may require more financial security, which may inhibit mobility if expected income gains are uncertain.

Homeownership raises employment probability

Some studies have explored the labor supply and demand implications of homeownership, and found that homeownership raises employment (or lowers unemployment), especially at the micro-level studies. They hypothesize that homeowners, in their attempt to stay in the residence region, may have lower reservation wages for accepting jobs in the local labor market. Welters et al.(2024) confirm that homeownership reduces residential mobility and increases neighborhood social capital but find no effect on reservation wages of the unemployed. There is evidence that homeowners tend to accept lower wage offer due to restricted mobility, which in turn leads to their high probability of employment (Munch et al., 2006). We can test whether homeowner’s reservation wage is higher or lower if reservation wage information is obtainable. On the other hand, in a booming region, surges in housing prices may make house-selling profitable to compensate the moving costs of homeowners, thus making them more mobile and reducing unemployment rate. Further, homeownership can facilitate business set-up, as the house can be used as business office and as collateral for loan application. All these contribute to business start-ups and self-employment activities. Harding and Rosenthal (2017) find that increase in home value raises the likelihood of entry into self-employment, particularly when housing prices are rising rapidly.

Furthermore, Morescalchi (2016) stresses that the direct contrast between homeowners and renters could be misleading as mortgage-bearing homeowners, unlike mortgage-free homeowners, may share some similarities with renters in that they have to make monthly payments to avoid foreclosure. Mortgage loan pressure induces workers to earn more money to repay the debt, thus contributes to active job search, work participation and lower probability of voluntary unemployment. As Arulampalam et al. (2000), Flatau et al. (2003), Rouwendal and Nijkamp (2010) and Mikula and Montag (2023) suggest, homeownership could boost exit rates off unemployment through higher debt pressure to return to work. Even for mortgage-free homeowners, Morescalchi (2016) also finds that they tend to select more effective search methods so that they return to work faster despite of lower search intensity.

Considering the labor demand side, if employers are more likely to hire home-owning workers, the employment rate of homeowners will be higher than otherwise. There are ample reasons for this to happen. First, employers may save hiring cost because they do not need to provide lodging for employees, especially in Chinese migrant context. Second, employers may perceive home-owning workers more deep-rooted and less likely to quit, as they have higher investments in local capital (DiPasquale and Glaeser 1999). Therefore, Coulson and Fisher (2009) conclude that consideration of firm behavior is important for understanding the labor market impacts of homeownership.

The mechanisms of homeownership’s impact on employment

The above review reveals two conflicting hypotheses concerning the impact of homeownership on employment. Positive impact mechanisms are manifested as:(1) High transaction costs associated with homeownership hinder workers from pursuing job opportunities across regions, locking them into local labor markets and reducing employment matching efficiency (Catte et al. 2004; Oswald 1996); (2) Homeowners are more likely to accept long-distance commuting to retain their housing, which confines employment opportunities to local markets and further suppresses employment rates (Blanchflower and Oswald 2013; Battu et al. 2008); (3) Homeowners demand higher reservation wages to offset housing transaction costs when switching jobs, leading to reduced probability of successful employment matching (Van Leuvensteijn and Koning 2004; Taşkın and Yaman 2019). Negative impact mechanisms are manifested as:(1) Homeowners tend to maintain existing employment and reduce resignation rates, thereby lowering short-term unemployment (Van Leuvensteijn and Koning 2004; Munch et al. 2006); (2) Local social networks developed through prolonged residency enhance access to informal job information, improving employment opportunities (Coulson and Fisher 2009; Blanchflower and Oswald 2013); (3) Wealth effects from housing appreciation may incentivize homeowners to pursue entrepreneurship‌, thereby generating new employment opportunities (Green and Hendershott 2001; Bracke et al. 2012). We will apply the migrant data to ascertain which hypothesis is more applicable in Chinese context.

Background of Hukou system and housing system reform in China

Hukou is a document issued by the local government to certify a person’s residency and eligibility for local welfare. Since 1958 the government has implemented a Hukou (household registration) system to differentiate rural residents from urban ones, and local residents from non-local ones. Without Hukou transfer (which is strictly controlled by government), workers couldn’t migrate to work and live in another place. Since the inception of economic reform in the late 1970s, control over migrant population has been gradually relaxed and migration has become more common, although some migration barriers and disparities still persist, especially in large cities. To obtain urban Hukou, rural migrant should reach quite a few requirements such as education attainment, living house and social security record depending on different city’s different policy. In this study, migrants refer to those living out of their home Hukou county or city for more than 6 months.

Urban Hukou holders enjoy favorable benefits and treatments in employment, social security and housing. In the planned economy time, work units usually provided housing for their urban Hukou employees to rent. Following the onset of housing reform in later part of 1990s, working unit ceased to provide housing for urban Hukou workers. Consequently, employment status would no longer closely affect homeownership and pose serious endogeneity problem. However, some weak connections may still exist, as part of housing reform permitted employees to buy house from their working-units with deeply discounted prices, although they only obtained partial ownership from the transaction. Unless a worker quit or was laid off from former working unit and found another new job, the connection between employment and homeownership would exist in the long run. Even after housing reform, working-units continue to subsidize their urban employees’ purchase of “commodity house” (which grants complete deed) in real-estate market in the form of “housing fund”.⁹

As opposed to urban Hukou holders, migrant workers face a quite different situation. There is no cheap house built by work units or “housing fund” available and they have to self-finance the purchase of “commodity house” when planning to settle down in destination cities. Therefore, our migrant sample was not influenced by housing reform occurred during the past two decades, which to some extent reduces endogeneity of housing tenure in employment function estimation. However, work units often provide dormitory for migrant workers. For migrants, unit-provided dorms play an important role in reducing family housing expenditures. As a result, migrants have lower demand for house purchase in cities (Fan et al. 2015).¹⁰ We therefore exclude from the sample those renting from work units or living in unit-provided free housing to reduce the degree of endogeneity further. Due to the policy available, renting cheaper public house is seldom and most migrants have to rent private house from urban residents. In addition, during the process of rapid urbanization and privatization of public housing in China, urban residents often become landlords renting out their additional apartments to migrants, which to some extent lowers the employment rate of urban Hukou holders.

What accounts for high homeownership rate in China? Is it endogenous, closely connected with employment, or also due to un-observed personal preference? It could be the case that some households are inherently less mobile than others, and owning house provides a higher sense of stability and safety. In this case, homeownership could reflect some element of selection bias. Furthermore, individuals with more unstable career paths or expecting more frequent job changes are less likely to invest in home-buying. This pertains to migrant group and young people. Unemployment risk has a significantly negative effect on homeownership and prolongs the timing of home purchase (Moriizumi and Naoi 2011). These potential endogeneity issues need to be addressed using appropriate instrumental variables.

Following Coulson and Fisher (2009), we use regional homeownership rate for migrants as an instrumental variable. The rationale of using this IV is that regional homeownership rate is more closely connected with personal homeownership probability. Meanwhile, it is argued that the macro homeownership variable is not directly correlated with labor market outcome of individual worker, which seems plausible and reasonable. Furthermore, we also apply Propensity Score Match (PSM) approach to reduce the heterogeneity between homeowner and renter.

Traditional Chinese culture puts great emphasis on deep-rooted family life and homeownership. Currently in China there is a new phrase called Mother-in-law Economy, which emphasizes homeownership as a prerequisite for approving a marriage proposal. Given this background, we propose to use the first child’s gender as another suitable instrumental variable (IV), as families having a son would be more likely to buy house, while child’s gender rarely affects parents’ employment decision.¹¹ The shortcoming of using this IV is sample size reduction as only families with children are included in this case.

With economic reform deepening and labor markets getting established, workers in China can more freely move across regions, especially for rural migrants. As Hukou system is relaxed in smaller cities, migrants are allowed to purchase house and settle down in cities. The Hukou regulation is still stringent and migrants cannot purchase house without local Hukou in some larger cities, hence in the following heterogeneity analysis (Section 5) we divide the sample into two groups according to city scale to reflect difference in Hukou restriction and possibility of obtaining homeownership.

Data and empirical methodology

Our data source is the 2014 and 2017 China Migrants Dynamic Survey (CMDS) conducted by the National Health and Family Planning Commission of China¹². This is a survey of migrants’ work, living condition and welfare across 31 provinces, with over 200,000 observations in 2014 and over 160, 000 observations in 2017. In China, the term “migrant” (floating population) refers to individuals who reside in a city or region other than their official household registration (hukou) location for over six months. Unlike international migrants, these are internal migrants whose mobility is heavily shaped by the hukou system—a institutional legacy that ties access to public services (e.g., education, healthcare, subsidized housing) to one’s registered hukou area. Rural-to-urban migrants, the focus of this study, often retain rural hukou status even after long-term urban residency, rendering them institutionally excluded from urban welfare systems unless they obtain local hukou through criteria like homeownership. This creates a unique “citizenship gradient” where migrants navigate labor markets and housing choices under structural constraints absent in most other countries. The China Migrants Dynamic Survey explicitly targets this population, capturing their employment, housing, and mobility patterns within this institutional framework. In addition, in some cities, particularly smaller cities, buying house can actually qualify migrants for local urban hukou status. In such case, obtaining homeownership and getting local hukou will also bring forth stable employment for those migrants. However, this consideration is not relevant here, as our sample only contains those migrants without local Hukou, even if they have purchased house in the destination city.¹³

We consider the following specification:

$${E}_{i}=\alpha {H}_{i}+\beta {X}_{i}+{\mu }_{i}$$

(1)

Where ${E}_{i}$ is a binary variable with 1 indicating that individual i is employed, and with 0 otherwise (having no job).${H}_{i}$ is a binary variable with 1 indicating that the individual is a homeowner, and 0 a renter. ${X}_{i}$ is a set of demographic, socioeconomic status and geographic location characteristics. ${\mu }_{i}$ is error term. $\alpha$ is the parameter of primary interest. To ascertain the underlying influencing mechanisms, we also assess the impact of home-ownership on three variables: inclination to reside permanently in current city, monthly wage and self-employment or wage work (employee) probability. In the influencing mechanism analysis, we then replace ${E}_{i}$ with inclination to live permanently in current city variable, the natural log of wage and employment activity dummy variable, and keep the same explanatory variables as those in Eq. (1).

Considering the endogeneity of homeownership in Eq. (1) and its being also a dummy variable, we conduct biprobit model and use regional homeownership rate as the instrumental variable in the first stage of regression for personal homeownership determination.

Table 1 displays the definition of main variables. The dependent variable, employment status, is inferred from the survey question “During the one week before Labor Day this year, did you hold a paid job working more than one hour?”. Answer of “Yes” is defined as 1, otherwise is 0. The most obvious reason for not holding a job is that the individual cannot find a job. Other reasons include job separation related to work issues between employee and work units. Individuals not holding a job due to personal reasons such as illness or childrearing are excluded from the sample as they are not capable of working. In our study, individuals not holding a job may include those choosing not to participate in the labor market and therefore cannot be regarded as unemployed as defined by official unemployment statistics. Since a large proportion of those out of the labor force are discouraged workers and most migrants move to cities for job-related reason¹⁴, we do not distinguish different categories of individuals not holding a job in the baseline analysis, and only apply the stricter unemployment standard (having no job and actively searching for job) in subsequent robustness checks. With individual employment information, we can also calculate the average employment rate at city level for macro analysis in robustness checks.

Table 1 Definition of variables.

Table 2 Summary statistics (Mean and Std. Dev.).

Homeownership variables are constructed to include purchasing “commodity house”, purchasing security housing provided by government, and building house by oneself. Renting options are divided by several categories including renting private house, renting inexpensive public housing provided by government, renting house provided by work unit, rent-free house provided by work unit, borrowed house, living on working site and other informal living places. To reduce the extent of endogeneity (homeownership associated with employment), we delete from sample three house categories provided by work unit. The rationale stems from the key considerations: in contrast to homeownership, work-unit housing provision in China constitutes an administrative welfare benefit tied to non-market determinants such as job tenure and bureaucratic rank (Wang and Zhang 2020). Concurrent analysis of these two distinct populations risks conflating the effects of institutional interventions with those of market-driven homeownership outcomes.Below we also compare different employment impact of renting public house and renting private house. Homeownership is explicitly defined as owning property ‌in the destination city‌ (where migrants currently reside and work), ‌not‌ in their rural hometowns. We exclude rural housing from our definition of homeownership for two reasons: First, Rural housing in China is collectively owned (tied to rural hukou), lacks tradable market value, and cannot be mortgaged or leased freely. Thus, it does not function as urban property in terms of enabling labor market mobility or wealth accumulation. Second,Our analysis centers on how housing tenure ‌in the destination city‌ shapes migrants’ employment behavior. Rural homeownership, while common, does not constrain or incentivize labor market decisions in urban areas, as rural properties are spatially and institutionally decoupled from urban labor markets.

Aggregate regional homeownership rate is the average level of homeownership for migrants in a destination city, and is used as an instrumental variable in empirical analysis of the impact of homeownership rate on employment rate at individual level. One weak point of measuring regional homeownership variable this way is that it covers migrants only, while local residents being excluded.¹⁵ However this shortcoming does not pose a serious problem as this study is not a study on regional homeownership’s impact on individual employment, which would need a more precise regional homeownership rate data, and hence our way of measuring aggregation regional homeownership can work as an appropriate instrumental variable.

As shown in Table 1, other control variables include gender, age, health condition, education attainment (the reference group being primary school or below), Hukou type, family size, marital status, property ownership and wealth in the source location, migration scope the reference group being migration in a given prefecture-level city: Inter-county migration), migration experience (years since last migration in 2014, years since first migration in 2017), city size (small city as the reference group¹⁶), destination provinces and Hukou provinces (we delete Xinjiang Production and Construction Corps from the sample). We use two instrumental variables: regional homeownership rate and gender of first child. Since only one household member was surveyed and only the respondent provided employment information, we can only capture the impact of housing tenure choice on one individual (primary breadwinner) in the family.¹⁷

Table 2 presents descriptive statistics of the main variables. In 2014, the employment rate in our migrant sample is 86.21%. Homeowners appear to have approximately 6 percent lower employment rate than renters, which seems to support the Oswald hypothesis. These preliminary findings from descriptive statistics are examined rigorously in the following section. Only 20.21% migrants own their house, considerably lower than rural or urban residents.¹⁸ The homeownership rate rose to 33.01% in 2017. However, the employment possibility gap between these two group also expanded (over 11 percent). As to personal characteristics, migrant homeowners are older, more educated, more likely to have urban Hukou and be married. They tend to migrate within home province and live in smaller cities. We will control them and conduct PSM model to further reduce the differentials between these two groups in the following empirical analysis.

Estimation results and discussion

Impact of homeownership on employment: baseline results

We first estimate a model assuming exogenous homeownership, then relax this assumption and use the instrument variables to deal with endogeneity problem. As shown in Table 3, probit model (1) indicates that compared to renting, homeownership leads to 4.80% lower possibility of having a job in 2014. This supports the view that homeownership hampers employment prospect.

Table 3 The impact of homeownership on employment.

Model (2) in Table 3 shows that the employment probability of homeowners is 4.83% lower, compared to migrants renting from private parties, while that of renting government provided housing is also 3.31% lower. Although not strongly significant, this result still warrants paying attention to. In fact, the estimation results using 2017 data (not reported here to save space) show that compared to renting private house, renting public house leads to a significant 4.73 percent lower employment probability. This is consistent with studies focusing on advanced countries (Flatau et al., 2003; Battu et al. 2008; Wood et al. 2009). Liu and Xing (2013) also find that residing in public housing reduces worker’s job-search incentive during unemployment period. This relative immobility of public renters may stem from public housing rents being below market rates, restrictive transferability within public housing, long waiting lists and security of tenure. Similar to homeowners, public renters are then ‘locked in’ to the current residence and face higher costs if they were to accept a job that entails a long distance move (Hughes and McCormick 1981; Battu et al. 2008). This immobility carries a policy implication that providing public housing to migrants or the unemployed might not be the most appropriate measure to induce active job search and reduce unemployment. Instead, an alternative measure such as cash subsidy for renters may be more effective to avoid the unintended counter-incentive effect.

The above results may be biased if homeownership is endogenous (employment affecting house buying) and self-selected (those buying house have different unobserved attributes). To cope with these problems, we re-estimate the model replacing individual homeownership with an instrumental variable (IV) in a bivariate probit specification. The IV used here is the regional homeownership rate for migrants by averaging the individual information of a certain city. As shown in model (3), when homeownership rate is added in employment probability regression, its coefficient is small and insignificant, reinforcing that it hardly plays a role in employment determination. Thus, this is an appropriate IV as it is correlated with individual homeownership but orthogonal to individual employment. The first stage result of Bivariate Probit in model (4) shows that, regional average homeownership rate is closely related to individual homeownership possibility, with the coefficient value 3.0705 and significant at the 1 percent level.

The marginal effect of the second stage Bivariate Probit in model (4) indicates that homeownership leads to 6.60% lower probability of having a job, compared to the renting counterpart. This magnitude is larger than that in column 1 (4.80%), which could be attributed to self-selection correction for the unobserved characteristics of homeowners. In addition, the biprobit result of 2017 data in model (5) shows that the negative effect of homeownership is even larger, reaching 15.04 percent, which also reinforces the 2014 regression result and conclusion. The observed increase in homeownership’s negative impact on employment probability likely reflects mechanisms intensified by China’s post-2015 housing de-stocking policies. These policies, through lowering down payments and subsidizing purchases, incentivized homebuying among liquidity-constrained groups (e.g., rural migrants), but simultaneously increased household debt burdens. Elevated mortgage obligations likely reduced labor mobility (e.g., reluctance to relocate for jobs) and incentivized asset-dependent income strategies (e.g., prioritizing property appreciation over wage work). The policy-driven housing market expansion thus amplified housing’s role as a financial constraint, disproportionately suppressing employment opportunities for leveraged homeowners by 2017.The underlying reasons are explored further in below.

Another concern is that if a renter cannot find a job in a city, she/he will leave this city and move to another one for job-hunting. However, this phenomenon (failing to find a job and depart) rarely happened among the homeowners. Since the remained renters in our sample are more likely to hold jobs in the survey city, this is a censored sample and there could be selection bias. It is interesting to note that the possible existence of this selection bias and renter leaving due to unsuccessful job search exactly indicate that renters tend to be more mobile than homeowners and thus homeownership hampers mobility and employment.¹⁹

Different facets and underlying mechanisms of homeownership’s impact on employment

Impact of homeownership on residency propensity (labor immobility)

First, lower employment rate mainly results from lower labor mobility across areas or regions. We evaluate the impact of homeownership on residence spells. Model (1) in Table 4 indicates that homeowners are 41.37% more likely to settle down (more than 5 years) in destination cities, which also imply that their job search range is constrained. The result from 2017 data is 21.31% higher. If labor demand in current city were to fall in the near future, homeowners would be reluctant to migrate to other cities where positive shocks increase labor demand, as the expected benefits from such move do not necessarily outweigh the high transaction costs of moving away from the owner-occupied house. Biprobit model points toward homeowner’s lower propensity of job search outside of current city. This deeper rooting tendency can be used to explain homeowner’s lower employment rate. This also corroborates with the finding of Wolf and Caruana-Galizia (2015) showing that homeownership is associated with lower migration outflows relative to inflows, and hence homeownership inhibits labor mobility.

Table 4 The impact of homeownership on residing permanently, personal earning and different employment activity.

In the same vein, even within the boundary of the city, the homeowners’ job search efforts and results are also constrained by the geographical periphery of their property as they do not like to work far away from their home. Unfortunately, no information is available to test it. We can only test the probability of migrant homeowners moving out of the city, and believe that the moving out of the city decision can be extended to mobility inside the city.

Impact of homeownership on personal earnings

Low employment rate may also result from the homeowner’s demanding higher wage to cover the relocation cost from moving. We consider the impact of home-ownership on personal reservation wage and in turn on personal wage income. It is possible that homeowners would demand higher reservation wage due to higher home transaction cost and non-labor income stemming from homeownership, making job hunting and employment more difficult. Higher reservation wage from supply side in turn can be converted to higher personal income in the labor market. On the other hand, it is also possible that homeowners would adjust their reservation wage since their job search would be limited to labor market within commutable distance. This implies that homeowners may have a lower reservation wage, which in turn can be converted to lower personal wage income.

As we have no information on reservation wage, we test homeownership’s impact on personal earnings as an alternative. Since earnings will help financing house-buying making homeownership endogenous, we instrumented the homeownership variable with the regional homeownership rate. The IV estimation results in model (2) show that homeownership exerts a negative effect on earnings, (−21.18 percent) significantly. In 2017 data, the wage penalty is −17.81 percent. The biprobit result seems to support the second conjecture alluded above. If lower wage could stand for lower reservation wage, we can observe higher employment probability, just as Munch et al. (2006) suggest. However lower potential wage could also be associated with lower employment by the way of dampening the work incentive.²⁰Therefore, the earnings mechanism of homeownership is not sufficiently manifested. In this case homeowners suffer from both lower employment probability and lower earning.

Impact of homeownership on employment activities

First, we evaluate how home-ownership shapes the employment activity profile: choosing to work for pay, self-employment or employment disengagement. As shown in multinomial logit model (3), home-owning migrants are no different than those renting migrants in the aspect of engaging in self-employment activity, comparing to becoming a wage earner. However, they are more likely to choose to be jobless (employment-disengagement). That is, the homeownership impairment effect on employment is mainly displayed in jobless. Using 2017 data, we also find both self-employed and wage worker are hindered by homeownership, while self-employment suffering heavier.

Considering that there exist more entry barriers to wage work sector for migrants, we compare the choices between self-employment and employment-disengagement seperately in evaluating homeownership’s role. The result of Biprobit model (4) provides a distinct picture: homeownership inflicts self-employment activity, 15.70% lower than the employment-disengagement counterparts. The similar 2017 data result is also consistent with Bracke et al. (2012). House purchase commands a lion share of family income, leaving less funds available for a business venture. Further, if the house was financed with a sizable mortgage, monthly payment may be higher than rental payment so that less money is left to finance self-employment activity.

Summing up the influence of these pathway mechanisms, homeownership constrains labor and geographic mobility, which leads to lower employment propensity. Also, homeowners tend to have lower earnings prospects, which may hamper their future job search incentive and employment profile. Finally, higher expenditure in house buying leads to lower rate in self-employment activity, which further hampers employment prospect in the market, considering the significance of self-employment activity in job creation in developing countries.

Heterogeneity analysis

Heterogeneous effect of homeownership by city size

Given the wide gap in housing prices and employment opportunities between large and small cities, we test whether the negative impact of homeownership on employment differs by city size. Here we examine the respective impact of homeownership in megalopolis vs smaller cities.

As shown in model (1) and (2) of Table 5, the negative impact of homeownership on employment is 8.15% in larger cities with more than 5 million populations, while in smaller cities with less than 5 million populations, the negative impact is 6.61%, suggesting that for migrants buying house in larger cities poses a greater negative effect on employment, which may be due to the relatively lower homeownership rate and higher cost of house purchase in large cities for migrants. This in turn hampers labor mobility and impairs self-employment potential seriously. Another possible reason may be that migrants in large cities are a selected group (compared to other migrants that cannot sustain the expensive stay and exit the large city), and they may care less about employment but pay more attention to pursuing the better social services of megacity, thus the wealth effect of homeownership maybe larger in megacities. The 2017 data result gives a similar conclusion, however the gap between larger and smaller cities is much larger.

Table 5 Heterogeneity analysis—the employment impact in different migrant group (biprobit).

Table 6 presents a heterogeneity analysis based on urban administrative tiers. Defining “large cities” solely by population thresholds exceeding 5 million may overlook the more significant influence of China’s hierarchical urban administrative system (e.g., first-tier, second-tier cities) on housing policies. Column (1) reports regression outcomes for first-tier and new first-tier cities, Column (2) for second-tier cities, and Column (3) for third-tier and smaller cities.

Table 6 Heterogeneity analysis—the employment impact in different migrant group (biprobit).

Heterogeneous effect of homeownership by distance of migration

Migrants moving in the home province may differ from those moving across home province in some un-observed characteristics. As we often observe that local migrants are more likely to purchase house²¹, it is necessary to examine the relationship between homeownership and employment across different migrant groups by distance of migration.

As indicated in column (3) and (4) of Table 5, for local migrants in home province boundary, homeownership leads to 8.03% lower employment probability. As for migrants moving across provinces, individual homeownership’s negative effect is much smaller, only 5.15%, implying personal heterogeneity between local migrants and migrants across province and different employment opportunities between home province and another province. Local migrant homeowners are more likely to be locked in the vicinity of their home due to their hometown link. This lock-in tendency bounded by hometown link exacerbates the negative effect of homeownership on employment. Guler and Taskin (2018) show that homeownership introduces additional frictions into labor market especially when the local market is weak so that as local labor market opportunities deteriorate, homeowners become more likely to stay put and remain unemployed. If the policy intent is to increase employment, their finding casts doubt on the common policy of encouraging rural migrants to purchase house and settle down in small cities of home province with less job opportunities, despite the policy’s seemingly reasonable argument that buying house in small cities is more feasible and affordable for them. However, analysis using 2017 data does not support the view and the real reasons deem further study.

Heterogeneous effect of homeownership by age: are the youths different?

It is commonly held that younger workers are more likely to be unemployed, more mobile towards where the job is and less likely to become homeowners. For older workers, even though they are more likely to own house, the possibility of unemployment could also rise with age. The negative effect of homeownership on employment found in our baseline results may differ between young and old migrant workers.

As shown in model (5) and (6) of Table 5, for young generation born after year 1980, the negative impact of homeownership is smaller (1.74% less likely to have a job, and statistically insignificant), compared to 8.36% less likely for those homeowners 35 years or older. The differences are also distinct and significant in 2017 data results. This is consistent with other scholars’ findings. For instance, using data of U.S. states, Green and Hendershott (2001) find that the negative effect of homeownership on employment is present only among middle-aged group. The larger negative effect of homeownership for older workers is due to their lower mobility linked to higher homeownership, which further hurts employment profile heavier. On the contrary, younger workers are more inclined to move towards where the job lies, even if the relocation entails selling their house, thus making the impairment effect of homeownership on employment insignificant or less.

Heterogeneous effect of homeownership by Hukou types: urban Hukou holders versus rural Hukou holders

Considering that the characteristics such as welfare entitlement may differ between migrants holding urban Hukou and those holding rural Hukou, we examine possible differences in homeownership’s impact on employment between the two groups. Column (7) and (8) show that for urban Hukou migrants the negative impact of homeownership is twice more severe in 2014, suggesting that their employment prospects are more inhibited by homeownership than their rural Hukou counterparts. The reason may lie in that urban Hukou comes with better social welfare benefits which make them less inclined to move simply for better job prospects. This implies that portable social security can facilitate urban Hukou holder mobility, in turn contributing to unemployment reduction. Rural Hukou workers are not entitled to the same welfare provision as urban Hukou workers, and hence are more likely to move, resulting in less homeownership’s barrier effect on employment. This mobility propensity is evidenced by the fact that almost everyone owns a house in his countryside hometown, and that does not seriously restrict them from migrating to cities in the first place. While numerically different, homeownership exerts the same negative impact on employment for both urban Hukou and rural Hukou migrants in 2017, but the difference between urban Hukou holders and rural Hukou holders narrows.

Robustness tests

We checked the robustness of baseline results by replacing the above individual survey data with macro migrant data. The 274 cities estimation results of 2014 further confirm the view that homeownership hampers employment. Controlling for regional dummy variables, population and average income, 1% increase in average regional homeownership rate leads to 3.66% decline in average employment rate. This magnitude of homeownership’s negative impact is smaller than the effect found earlier using individual data.

The attenuated effect of homeownership on employment probability under the strict unemployment definition suggests that housing tenure may disproportionately deter labor force participation among individuals excluded by narrow metrics—notably, the “hidden unemployed”. While our data do not directly observe this subgroup, the magnitude shift implies that homeownership’s financial burdens (e.g., mortgage obligations) or reduced geographic mobility could exacerbate disengagement from the labor market. Policymakers should consider synergies between housing affordability programs (e.g., subsidized rentals for homeowners in economic distress) and active labor market interventions (e.g., skills training, job-matching services) to mitigate such disincentives. Future research should prioritize longitudinal datasets tracking both housing status and job-seeking intent to disentangle these mechanisms, particularly in contexts with high underemployment or spatial mismatches.

We conducted another robustness test using a narrower and more accurate definition of unemployment (actively searching for job in the previous month of survey, defined as 0 in dependent variable). The coefficient of homeownership in 2014 became smaller, but statistically negative (−1.12 percent in Biprobit). As we mentioned earlier, there could be many discouraged workers that have dropped out of the labor force, so that homeownership’s negative effect is underestimated in this regression. In addition, for the unemployed group, we find that compared to unemployed homeowners, the unemployed renters are less likely to stay permanently in destination city (45.95 percent less likely) in 2014, thus illustrating the renters’ higher propensity to move to where job is available.

The weaker aggregate employment effects of homeownership rates, as observed in macroeconomic studies, may stem from countervailing forces that offset micro-level disincentives. First, regional agglomeration effects—such as localized job creation in areas with high homeownership—could dampen the negative labor supply responses identified at the individual level (e.g., reduced mobility or job-seeking). Second, macroeconomic analyses often capture heterogeneous regional dynamics: in areas with flexible labor markets or robust public transit, homeownership’s spatial “lock-in” effects may be mitigated. Conversely, micro-level estimates isolate individual behavioral responses (e.g., reservation wage adjustments, reduced search intensity) that aggregate analyses may obscure. Future work should integrate dynamic spatial models to disentangle how regional labor demand shocks interact with housing tenure to shape this macro-micro gap.

Thirdly, considering that some people migrate for other reasons²²rather than employment and including them in the sample may have blurred our earlier results, we “purified” the sample using only migrants whose sole purpose of migration is for work and business (accounting for 88.13% of total sample) in 2014 and conducted a similar Biprobit regression as was in model (4) of Table 3. The results show that homeownership leads to 6.44% lower employment probability, a slightly smaller magnitude but qualitatively the same negative impact of homeownership.

Fourthly, we experiment with an alternative instrument variable, gender of first child. This reduces the sample size for the study as only the migrants having children are included in the regressions. The gender of child is an appropriate instrument variable for homeownership as it may influence the homeownership decision (explained earlier) but exert little or no impact on employment decision. The result shows that homeownership still impairs employment significantly, and the employment probability is 8.37 percent lower in 2014.

Lastly, we also conduct robustness diagnostics by PSM analysis using 2017 data to deal with the heterogeneity of different group and selection bias. As shown in Table 7, all results of ATT, ATE, and ATU support that homeownership exerts negative effect on employment, although the magnitude differs slightly. In sum, the conclusions of heterogeneity analysis from column (1) to column (8) are similar to those in Table 5.

Table 7 PSM result of homeownership’s impact on employment.

Other considerations regarding the negative impact of homeownership

We need to distinguish between homeowners who have to make mortgage payments and those who are debt-free, and test if those paying mortgage have a lower unemployment rate due to the loan payment pressure. On the other hand, homeowners facing a higher housing debt may need to find a higher wage job to finance the debt payment, which increases job finding difficulties and thus lowers the employment possibility. To explore the labor supply implications of homeownership, we add mortgage variables in the estimation to examine if debt-servicing burden causes homeowners to participate more in labor market, and if workers from a family with housing loan have a higher likelihood of employment than those without. The un-shown estimation result provides evidence of the work incentive from housing debts, one percent increase in mortgage loan raises the possibility of employment 0.9 percent in 2014. This is consistent with findings of Arulampalam et al. (2000), Flatau et al. (2003) and Rouwendal and Nijkamp (2010). This consideration also mitigates the negative impact of homeownership on employment, although it cannot fully change the direction.

One consideration concerns rural migrants, who mostly already own houses in their home countryside or nearby small county or city. In our opinion, purchasing house in hometown is the second-best choice under the circumstance of Chinese Hukou system (harsh restriction on house purchase in large city). It is often the case that the migrants’ wife and children live in hometown-county, while they have to work in large city and rent house. The family have to endure this kind of arrangement’s undesirable aspect of family separation and additional living cost (associated with buying a house back home and renting a house in the migrant city). Admittedly, homeownership in home-county in fact becomes a barrier to their finding jobs and moving across county due to limited job openings in the local market confinements. If rural workers want to increase the prospect of landing a job, they have to leave the county and work in larger cities, renting a house there (for persons doing that, hometown house is no longer a barrier to employment). The same case applies to some homeowners who rent a house near their workplace, even for those local urban Hukou holders. All these scenarios render further support for our main result that homeownership impairs employment. There are myriad of reasons for homeownership in destination city including access to social service, child education, marriage, etc. Still, the majority in the sample stated their objective of migration is for work and business, hence it is pertinent and necessary to study the relation between homeownership and employment.

Conclusions and policy implications

Housing market can impact labor market in various ways. In this study, we use China Migrants Dynamic Survey data of 2014 and 2017 to examine the impact of homeownership on employment outcomes in transitioning China. We also explore the underlying mechanisms of homeownership (via lowering labor mobility, raising reservation wage and reduced self-employment activity). We control for endogeneity of homeownership by means of instrumental variables, and further explore the possible heterogeneous effect of homeownership on employment by distinguishing between different migrant groups (by size of city, scope or distance of migration, age of migrants, urban Hukou versus rural Hukou holders). Finally, we conduct several robustness tests using macro data, different micro data and different IV and PSM methods. Our major finding is that, compared to the renting option, homeownership does lead to lower employment rate. Possible mechanisms for lower employment outcome of homeownership may lie in declining labor mobility. Also, large expenditure spent in house purchase crowds out funding source to start a business and hence lowers self-employment activity.

‌Additionally, our study reveals that China’s institutional context as a transition economy fundamentally reshapes the Oswald Hypothesis. The‌ ‌hukou‌ ‌system exacerbates homeownership’s spatial rigidity: rural migrants face institutionalized exclusion from urban welfare unless they purchase housing, creating a perverse tradeoff between securing residency rights (via homeownership) and retaining labor market flexibility. Simultaneously, local governments’ reliance on land finance—a hallmark of China’s urbanization model—artificially inflates housing costs, which not only crowds out entrepreneurial capital but also incentivizes households to prioritize housing speculation over productive labor supply. These dynamics highlight how state-market interdependence in transitional economies amplifies and complicates the housing-employment nexus, diverging from mechanisms observed in mature market economies.

Some policy implications from our study call for rethinking of current urbanization, housing and labor policies. During the urbanization process, policies were typically proposed to encourage migrants to buy house and settle permanently in destination city. Our study cautions the proponents of such policy and stresses that we should not ignore the unintended side-effects (particularly the employment impairment effect) of homeownership on labor market. The renting option may be more applicable or tenable to preserve employment flexibility, particularly for low-income rural to urban migrants. Homeownership is not necessarily the best course of action for urbanization and for large number of rural to urban migrants to permanently settle in cities. However, it is no denying that under current institutional circumstance, purchasing house can help migrants obtain local urban Hukou and other welfare benefits embedded in it. In this regard, reforming current Hukou system and endowing the renter with accessible public service and rights hence become essential.²³ As indicated by our empirical results, urban Hukou migrants are hampered more heavily by homeownership, hence reforming social security system and making the benefits portable can facilitate urban Hukou holder’s mobility and job finding in the labor market. Also, the rental property market is currently underdeveloped, which dampens the renting propensity. Therefore, establishing sound and well-functioning rental housing market becomes crucial to increase employment, which tends to be hampered by homeownership. Since renting public housing also inhibits labor mobility, perhaps a better policy option for the government is to give direct cash subsidy rather than providing public housing for rent, at least from the perspective of promoting full employment in developing countries.

‌Our findings extend the Oswald Hypothesis by demonstrating that in transitional economies like China, institutional barriers (e.g.,‌ ‌hukou‌ ‌) and state-driven market distortions (e.g., land finance) transform homeownership from a mere individual choice into a mechanism of‌ ‌institutionalized immobility‌. ‌While the Hypothesis posits geographic rigidity as the primary channel, our results underscore that rigidity is not merely spatial but embedded in institutional hierarchies. By linking housing tenure to welfare access and fiscal incentives, the state inadvertently deepens labor market fragmentation. This recalibrates the Hypothesis to account for how institutional path dependence in transition economies mediates housing-employment tradeoffs—a critical marginal contribution to debates on urbanization and labor policy.

It is said that soaring house price is attributed to speculative investment demand for housing. From individual family’s perspective, it could be desirable to retreat from labor market and receive more income from collecting rents and from housing value appreciation. Yet from macro perspective for healthy economic development, the resulting employment reduction due to managing extra house and speculating on housing market is not desirable.²⁴

In future research, by acquiring pertinent urban residents’ panel data and adopting better approaches to address the endogeneity issue of homeownership, we intend to test whether the negative effect of homeownership is unique for migrant group, or the negative effect can also be found for a broader group of workers in developing China. Second, working at home under the circumstance of telecommuting, establishing better transport facility to what extent can mitigate the employment barrier effect of homeownership is an interesting problem and deserves deep study. Lastly, future studies should further investigate how institutional reforms—such as decoupling‌ ‌hukou‌ ‌status from housing ownership or curbing local governments’ land-finance dependency—could attenuate homeownership’s adverse employment effects. Comparative analyses across cities with varying degrees of‌ ‌hukou‌ ‌liberalization or fiscal decentralization would help disentangle the role of transitional institutions in shaping labor market outcomes.