What do young adults on the edges of homeownership look like in big cities in an emerging economy: Evidence from Shanghai

Abstract

Chinese

The characteristics of young households positioned on the edges of homeownership in a rapidly growing economy are investigated. This empirical work is based on a unique survey conducted in 2011 in Shanghai, which enables us to determine how young renters assess their major financial obstacles to becoming homeowners. We distinguish whether young renters attempting to access homeownership are constrained by a lack of sufficient funds for a down-payment, or by inadequate income, or by both. By connecting young renters’ individual characteristics to the various financial constraints they face, we are able to characterise the renters who are situated on the edges of homeownership. Based on these findings, we make policy recommendations on how the government could improve the homeownership prospects of young households close to the boundary between renting and owning in urban China, as opposed to those with minimal chances of attaining homeownership.

Keywords

China financial constraints homeownership edge housing demand young adults

Introduction

The capability of young adults to become homeowners has attracted substantial interest in the literature (Fisher and Gervais, 2011; Haurin et al., 1994; Ortalo-Magné and Rady, 1999; Ying et al., 2013). This issue has become particularly salient given that the worldwide rate of young household formation has declined since the global economic crisis (Clapham et al., 2012). However, young people’s barriers to homeownership remain insufficiently understood. Recent literature has suggested that it is useful to adopt the concept of the ‘edge’ of homeownership, which refers to a status in which a household is very close to entering into homeownership but for some reason is not yet able to achieve that goal (Wood et al., 2013). This group would benefit the most from policies promoting homeownership. Identifying which young adults are on the edges of homeownership could direct policy attention to those who are most sensitive to market volatility. Such study would also shed new insights into the management of property market risks as well as enhancing the design of homeownership affordability policy. However, few studies have explored this area.

China has determined to shift the Chinese economy away from one mainly driven by investment and export towards an economy with greater reliance on domestic consumption (Li, 2012; World Bank, 2013). This is a task that is commonly faced by many other emerging economies. Enabling young migrants to settle in cities and increasing the homeownership rate of those around the edges of homeownership are definitely among the keys to the success of this strategic goal. Further, there is also a need to note that several emerging economies share an important common goal to promote homeownership as a means to drive urban development and economic growth. As discussed by some scholars, homeownership in these economies was promoted as both a means to boost economic growth and to build up ‘asset-based security’ to preserve self-sufficiency and reduce the citizens’ demand for welfare (Doling and Ronald, 2014).

In this study, we examine the characteristics of youths facing various financial barriers to homeownership in the largest city in China, Shanghai. To achieve this aim, we explore the unique advantages of a large survey of young households conducted in 2011 in Shanghai. In this survey, each young renter was asked the main reason that he or she could not afford to buy a home from the market: lack of down-payment, shortage of income, or both. Our empirical work then gauges the marginal effects of personal characteristics on the probability that an individual is on a certain type of the edges of homeownership.

Our work contributes to the literature in two key respects. First, different from most previous literature, we investigate the determinants of homeownership barriers rather than the determinants of homeownership directly. It is then possible to investigate how economic and demographic variables may have different impacts on different financial barriers to homeownership. Second, this paper provides new insights into how financial underdevelopment depresses housing demand in a nascent housing market. To the best of our knowledge, this is the first paper to study what factors impede young peoples’ pathway to homeownership in post-reform urban China.

Literature review

Studies on the general demand for homeownership

It is well believed that homeownership creates both private and social benefits. Compared with renters, homeowners tend to have a higher social enrollment, greater self-esteem and better life satisfaction (Engelhardt et al., 2010; Megbolugbe and Linneman, 1993). Moreover, homeownership creates external benefits for the neighbourhood and broader community (Coulson and Li, 2013; Rossi-Hansberg et al., 2010). Homeownership is further widely regarded as a means of ‘asset-based’ welfare in the context of socioeconomic and demographic transformations (Doling and Ronald, 2014). Promoting home ownership is thus an important issue in housing policy in nearly every country.

Previous literature on young adults’ homeownership demand mostly focused on the refinement of tenure choice estimation for youths. LaFayette et al. (1995) also confirm the simultaneity of independent living and tenure choice, and obtain a demand elasticity of income of 0.274 for owners after correcting for possible selection biases. However, Bourassa et al. (1994) find that the correlation between these two decisions is insignificant in Australia. In Sweden, although Asberg (1999) confirmed the simultaneity of the two decisions, this does not have a major effect on the estimated coefficients. In general, most studies report that demographic factors are quantitatively more important in explaining household formation, whereas economic factors play a larger role in the demand for homeownership (Andrew and Meen, 2003). See a survey of recent literature in this respect in Öst (2011).

A great deal of literature in the field puts special emphasis on the impacts of borrowing constraints (the joint effects of down-payment constraint and income-liquidity constraint) on the demand for homeownership. Studies in the literature have long argued that mortgage qualification requirements significantly constrain household mobility and tenure choices (Zorn, 1989). Owing to the information asymmetries and moral hazard, lenders typically gauge borrower’s repayment capacity based on observable current income and liquid assets rather than unpredictable future income and total wealth (Haurin et al., 1997). Linneman and Wachter (1989) construct income and wealth constraint variables by estimating the optimal home purchase price for a household. They find that wealth constraints had a stronger negative impact on ownership tendency than income constraints.

However, Jones (1995) argues that potential first homeowners face both endogenously and exogenously determined liquidity constraints. The difference between the aspiration to become a homeowner and the desire to consume housing services themselves may affect the accumulation of wealth. Thus, the endogeneity between wealth accumulation and borrowing constraints in the housing choice decision requires that the impact of household wealth should be treated very carefully (Jones, 1995).

Recently, it has been widely noticed that the opportunities for younger adults to become owner-occupiers have substantially decreased in many countries since the recent global financial crisis (Forrest and Yip, 2012). Researchers relate this declining rate of homeownership to education, careers, labour market risk, levels of debt and access to state welfare. See a review of related studies by McKee (2012). These studies highlight the role of state and government intervention in changing the situations of young households, but they give less attention to characterising the young adults on the edges of homeownership.

Studies on the demand for homeownership in China

With an annual production of 10 million new housing units and a market value as high as a trillion US dollars, China has recently surpassed the USA as the world’s largest housing market. Owing to China’s unique political system, institutional environment, cultural background and unprecedented economic growth (Wang and Wang, 2012), the rapidly growing Chinese housing market provides an ideal laboratory to study some fundamental issues in housing research.

Along with the expanding freedom of Chinese households’ housing choices since the privatisation of housing in the 1990s, a small number of studies have examined which factors determine Chinese individuals’ demand for homeownership in a market context (Chen and Han, 2014). Coulson and Tang (2013) conducted a survey of 5160 homeowners or potential homeowners in ten cities in China and found that the employees of state-owned enterprises are more likely to be homeowners than the employees of domestic private companies. Zhou (2011) found that uncertainty in income, employment and expenditure significantly and negatively affect urban Chinese residents’ demand for homeownership.

Chinese institutional context deepens the role of homeowning for households and the society. Importance of owning a home has been rooted in Chinese culture. Particularly for young adults, housing stands as social status that improves the relative standing in the marriage market (Wei et al., 2012). Reasons such as the single child policy, sex ratio and cultural traditions have been suggested. In addition, following the ‘buy lots nearby school’ policy, only those residents who live within the key school districts can enroll in the key schools (Feng and Lu, 2013).

Since 1998, welfare housing reform promotes homeownership on one hand, and enlarges income and prices mismatches on the other (Yang and Wang, 2011). This, to a certain extent, excludes most young households, particularly those migrant households, from the expensive housing market. They have become one of the vulnerable groups in urban China (Chen et al., 2010). It has been argued that housing wealth has been one significant reason for the developing social inequality in urban China (Yang and Shen, 2008). Homeownership affordability of young households is thus important for China’s social stability. However, it is always challenging to assess a household’s true potential to afford homeownership based on the observable information of income and wealth; this is especially true in an emerging economy where income and wealth information is generally imprecise. For this reason, Ying et al. (2013) adopted the questionnaire approach to directly ask respondents to assess the home value that they can afford. The authors argue that this approach provides a measure of affordability that is insensitive to a researcher’s subjective definition of ‘standard housing’ (Hancock, 1993). Meanwhile, the households’ self-assessed affordable home value takes into account investment motives, income growth potential, income uncertainty, consumer preferences and other factors influencing homeownership that cannot be observed by researchers (Ying et al., 2013). This paper adopts a similar approach, employing respondents’ self-assessments of their financial constraints as indicators of his or her true position in the homeownership market.

Nonetheless, so far most studies have examined the characteristics of existing homeowners rather than studying those individuals on the edges of homeownership and identifying the main financial obstacles for them to access homeownership. This research gap we wish to address in this paper.

The analytic framework and econometric modelling

This paper explores unique sample data in which the respondents indicate what type of financial obstacles they face when attempting to access homeownership. We then establish connections between the individual characteristics of young renters and their financial constraints through econometric models.

The cases of financial constraint

In the survey, the respondents were asked to indicate which financial constraint they are subject to when planning to buy a home. They were asked to pick one of the following choices:

Case (1). No constraint at all: the respondent can buy a desired home whenever he or she wishes. Those are people who can be identified as ‘can buy won’t buy’ in Hancock (1993).

Case (2). Down-payment constraint (wealth constraint): the respondent does not have sufficient funds available for a down-payment to apply to a standard mortgage to finance the home purchase plan,¹ but his or her monthly income is sufficient to pay the minimum amortisation payment if a mortgage can be obtained.

Case (3). Lending constraint (income constraint): the respondent’s income is insufficient to afford monthly repayment for a standard mortgage to finance the home purchase, but his or her wealth would meet the minimum down-payment requirement.

Case (4). Both the down-payment and the lending constraint are binding. That is, the respondent is constrained by both wealth and income.

In this setting, it is reasonable to deem a renter as a person ‘on the edge of homeownership’ if he or she claims to be subject to only one of the following two financial constraints but not both to purchase a home (case 2: down-payment constraint and case 3: lending constraint), and a renter as far from homeownership if he or she claims to be subject to both the down-payment constraint and the income constraint (case 4). We further denote case 2 as the wealth-based homeownership edge and case 3 as the income-based homeownership edge.

Strictly speaking, the situation ‘can buy won’t buy’ (case 1) is also another form of homeownership edge, in terms of either the sustainability challenge to maintain the homeownership status or the fragile trade-off balance between homeownership utility and other utility over the life course (Wood et al., 2013). However, in this paper we wish to focus financial barriers to first-time homeownership, as households in this group usually attract most attention in the literature and policy debate (Clapham et al., 2012; Clark and Mulder, 2000). Thus we skip detailed discussions of ‘can buy won’t buy’ case in the paper.

As discussed above, the questionnaire approach provides an appealing alternative way of assessing an individual’s true position in the homeownership market, especially in an economic environment where the observation’s current values and potential income and wealth cannot be measured satisfactorily (Ying et al., 2013). This is mainly because the household’s own assessment is not sensitive to the researcher’s subjective definition of ‘desired home’ and has taken into account many influencing factors of homeownership that are unobservable to researchers.

The multinomial logistic model

As the response to the questionnaire has four discrete options, the results are analysed using a multinomial category model rather than a binary category model. Further, as the alternatives of case 2 and case 3 could not be meaningfully rank ordered, ordered category models are not appropriate here. The multinomial logistic model, sometimes named the multinomial logit model, is widely used to predict the probabilities of the different possible outcomes of a categorically distributed dependent variable, given a set of independent variables (Greene, 2003). The basic model set up is the same as logistic regression, with the difference that the dependent variables are categorical rather than binary.

Assume that the outcome of observation i (Y_i) can be only one of k+1 possible integer values such as $Y_{i} \in {0, 1, 2, \dots k}$ ; the multinomial logistic regression uses a linear predictor function to predict the probability that observation i has outcome j (j<=k) in the following form:

\begin{matrix} f (j, i) = β_{0, j} + β_{1, j} X_{1, i} + β_{2, j} X_{2, i} \dots + β_{m, j} X_{m, i} \end{matrix}

(1)

where $β_{m, j} X_{m, i}$ is the regression coefficient of the mth explanatory variable specific to the jth outcome. In a compact form:

\Pr (Y_{i} = j) = β_{j} X

(2)

where $β_{j}$ is the set of regression coefficients corresponding to outcome j and X is the set of explanatory variables that are identical for all observations.

After obtaining the set of regression coefficients $β_{1}$ , $β_{2}$ ,…, $β_{k}$ for each outcome, the following identity holds:

\Pr (Y_{i} = j) = \frac{e^{β_{j} X}}{\sum_{s = 0}^{k} e^{β_{s} X}}

(3)

Setting the coefficient set of baseline outcome $β_{1}$ =0, we have:

\Pr (Y_{i} = 0) = \frac{1}{1 + \sum_{s = 1}^{k} e^{β_{s} X}}

(4)

\Pr (Y_{i} = j) = \frac{e^{β_{j} X}}{1 + \sum_{s = 1}^{k} e^{β_{s} X}}

(5)

Thus, the relative risk of outcome j (Y_i = j) as opposed to the baseline outcome (Y_i =0) is (relative risks are also called odds-ratio):

\frac{\Pr (Y_{i} = j)}{\Pr (Y_{i} = 0)} = e^{β_{j} X}

(6)

Taking the log of both sides in the last equation shows that the log odds-ratios of two alternative outcomes is a function of the parameters between the two alternatives but not of those for any other alternatives.

\ln \frac{\Pr (Y_{i} = j)}{\Pr (Y_{i} = 0)} = β_{j} X

(7)

Thus, the interpretation of regression coefficients in the multinomial logistic model must be specific to two alternative outcomes. For example, $β_{m, j}$ is interpreted as the marginal effect of the mth explanatory variable (X_m) on the log odds-ratio of alternative j to the baseline alternative, 0. Usually the property of IIA (Independence of Irrelevant Alternative) is assumed, which states that the odds-ratio of one outcome over another does not depend on the presence or absence of other ‘irrelevant’ alternatives.

The marginal effect of the mth explanatory variable (X_m) on choosing alternative j is based on the derivative of the probability of outcome j $\Pr (Y_{i} = j)$ with respect to X_m.

\begin{matrix} \frac{\partial \Pr (Y_{i} = j)}{\partial X_{m}} = \Pr (Y_{i} = j) \\ [β_{m, j} - \sum_{s = 0}^{k} \Pr (Y_{i} = s) β_{m, s}] \end{matrix}

(8)

As in the logit model, this equation involves not only the parameters of k but also the parameters of all other alternatives. Thus, the marginal effect of $β_{m, j}$ does not need to have the same sign as $β_{m, j}$ .

Institutional background and data description

Institutional background

In China, until the early 1990s more than 90% of urban residents were living in public housing (World Bank, 1993). Through the massive privatisation of the public housing stock in the late 1990s, China was quickly transformed into one of the nations with the highest homeownership rates within one decade (Mak et al., 2007). This reform stimulated homeownership in urban China, reaching a rate of 89.3% in 2011. Only 38% homeownership was achieved through the market, with the rest resulting from either privatisation (40.1%) or inheritance (11.2%) (NBSC (National Bureau of Statistics of China), 2011). In recent years, China’s major cities have experienced significant increases in house prices in parallel with sustained economic growth and an unprecedented rate of urbanisation. Young Chinese people who have no chance to benefit from housing privatisation face continued increasing housing prices and are in a highly vulnerable position.

Shanghai is the largest city in China and operates the most developed real estate market. Similar to other Chinese cities, neoliberal housing policy has transformed the housing sector in Shanghai from a public-owned system to a full-fledged market-oriented system. In the early 1980s, 80% of urban housing stock in Shanghai were owned by the state (including working units) (Shanghai Statistics Office, 1983). However, at the end of 2013, the state ownership has decreased to no more than 15% and the rest was primarily owned by private persons (SSO, 2014).

Since the 1998 housing reform, the housing market in Shanghai has experienced a sustained boom and is now one of the most expensive markets in the world. According to official data, in 2014 the mean housing price in Shanghai was approximately 16,415 RMB/sqm,² while in the same year the average annual disposable income per capita in Shanghai was only 48,841 RMB, implying that the PIR (price to income ratio) of the cost of a 90 sqm apartment or a three-person household would be an intolerable 9.92 times.

A major interest in this paper is the difference in financing barriers between natives and migrants when accessing homeownership. Under the Chinese hukou system, migrants are largely excluded from the welfare package reserved for local residents, including unemployment insurance, healthcare, pension, housing benefits and even the right for children to enter local schools (PFPC (Population and Family Planning Commission), 2012). The hukou system has been constantly referred to as the main constraint on the housing consumption of migrants (Sato, 2006; Wu, 2004). At the national level, an official survey indicated that the rate of homeownership among migrants was around only 10% in 2010 (PFPC, 2012), in sharp contrast to the 89% rate among permanent urban residents (NBSC, 2011). The exclusion of migrants from the local housing welfare system further exacerbates migrants’ difficulties in the Chinese urban housing market (Sato, 2006).

Nonetheless, because of the massive privatisation of the public housing stock at the end of the 1990s (Chen and Han, 2014; Yang and Shen, 2008), the homeownership rate in urban China is generally very high, and Shanghai is no exception. Restricting our analysis to local households, Shanghai is currently a city of homeowners: in 2010, 80% of local households owned their homes (SSO, 2012). However, as the homeownership rate among migrants is as low as 20%, the overall homeownership ratio in Shanghai was just approximately 58% in 2010 (SSO, 2012).

According to the sixth population census (2010), the Shanghai metropolis accommodated 23.02 million residents (8.25 million households) in 2010 (SSO, 2012). Among all residents, local households (with permanent registration status, or hukou) accounted for 61% of households, with migrants (without hukou status) accounting for 39%.

The mortgage business in China operates in a way very similar to those in advanced economies. In 2011, the terms of mortgage loans in China were regulated as follows: the maximum loan-to-value ratio is 70% (of the lower of the appraisal value or purchase price), and the maximum mortgage term is 30 years. Mortgage interest rates are controlled and set to track central bank rates. In addition to commercial banks, financing is available to homebuyers in China from the Housing Provident Fund (HPF), which is a compulsory saving scheme aiming to support workers’ homeownership (Chen and Deng, 2014).

Further details on the structure of the housing market and housing finance system in Shanghai can be found in earlier studies on this topic (Chen and Deng, 2014; Chen et al., 2010; Han, 2010; Zhang and Chen, 2014).

Data description

Our data set is based on a household survey conducted by a survey team commissioned by the Shanghai Municipal Consulting Committee in 2011. Researchers from Fudan University and the staff of the Horizon Research Consultancy Group jointly designed and implemented this survey. As the survey was supported by the municipal government, the process of data collection was smooth and the survey obtained high quality data. The survey is limited to persons aged between 20 and 40. A total of 3200 observations can be used in the analysis after eliminating those with missing information on key variables.

As the survey only covers formally employed youths, migrants are underrepresented in the sample at 31.5% (cf. Table 1), considerably lower than the city-wide share of 53.6% for the 20–40 age group in 2010 (SSO, 2012). The sample is also slightly skewed towards the younger end of the youth age bracket: 61.4% of respondents were younger than age 30 (cf. Table 1) as opposed to 52.8% city-wide. Nonetheless, these sample biases may not have serious consequences for our research purpose because we are interested in the possibility for young middle-class individuals to face borrowing constraints, which is the dominant priority of Chinese housing policy (Chen and Deng, 2014).

Table 1.

Living types by hukou, age group and family type (%).

Type (of sample total, %)	Type 1	Type 2	Type 3	Type 4
Local (68.47%)	4.15	10.95	42.36	42.54
Migrant (31.53%)	19.82	60.06	14.67	5.45
Age 20–24 (15.50%)	15.12	33.47	9.07	42.34
Age 25–29 (45.41%)	10.69	29.34	23.89	36.08
Age 30–34 (27.06%)	5.08	22.98	49.31	22.63
Age 35–40 (11.53%)	4.07	13.55	68.56	13.82
Single (51.59%)	13.2	32.04	13.51	41.25
Married without children (18.66%)	4.86	24.46	48.41	22.28
Married with children (29.75%)	4.62	17.96	59.24	18.17
Total	9.09	26.44	33.63	30.84

Note: Type 1: Work-unit renting; Type 2: Private renting; Type 3: Owner-occupation; Type 4: Living with parents or relatives.

Table 1 reports the distribution of living arrangements according to age, family type and whether the respondent is a local or a migrant. If the respondent does not live with his/her parents or relatives, he or she is defined as living independently. As Table 1 suggests, 70% of respondents live independently. Note that while 42.5% of young locals are still not living independently, only 5.5% of young migrants live with their parents or relatives. However, only 14.67% of migrants own a home in Shanghai, only one-third of the rate of locals (42.36%).³

Not surprisingly, the likelihood of living with parents/relatives decreases dramatically with age. Nonetheless, the effect of age on homeownership is significantly different between locals and migrants; while the homeownership ratio increases sharply with age for locals, this trend is much less clear for migrants (cf. Table 2). Tables 1 and 2 also suggest that the likelihood of homeownership is greater for individuals with higher levels of education attainment and those who are married with children.

Table 2.

Homeownership ratio by age, hukou and educational attainment.

Age	Hukou		Educational attainment			Total
Age	Local	Migrant	College and below	University	Graduate	Total
20	14.12%	3.42%	7.00%	10.50%	20.00%	9.07%
25	29.44%	13.71%	18.78%	25.29%	26.53%	23.89%
30	55.66%	28.57%	35.37%	55.78%	52.51%	49.31%
35	76.83%	20.37%	61.18%	73.10%	75.00%	68.56%
Total	42.36%	14.67%	26.42%	35.29%	42.27%	33.63%

In addition to the rich information on individual characteristics, a key advantage of this survey is that it provides young renter respondents’ self-assessments of the borrowing constraints they face when considering homeownership. Among the 1263 young renters who indicated what constrained them from buying a home they desired, 8.71% responded that they were not subject to any financial constraint (‘can buy won’t buy’), 31.5% answered that they were short of funds for a down-payment, 18.97% indicated that they were constrained by income, and 40.82% replied that they were subject to both down-payment and income constraints.

The central aim of our empirical work is to explore how the set of individual characteristics can be used to predict a young renter’s risk of certain financial constraint of homeownership. The variables that we incorporate in the multinomial logit model are listed and explained below.

Estimated results and discussions

Permanent income

There is a consensus in the literature that permanent income dominates housing decisions rather than current income (Goodman, 1988). Permanent income estimation is typically calculated as the fitted value of the regression of current income on a set of human capital variables.

Table 3 gives the variable definitions, and the estimation results of our permanent income regressions are reported in Table 4. Column 1 in Table 4 is the output for the whole sample. Note that the ownership status itself affects permanent income, suggesting that current household income may be endogenous to housing choice (Jones, 1995). We thus estimate permanent income separately for owners and renters. In all cases, the coefficient magnitudes of educational attainment dummies are statistically positive and substantial, as expected. The effects of industry, firm type, occupation and availability of HPF are also consistent over different subsamples and have the expected signs. However, it is worth noting that after controlling for human capital variables, there is no systematic difference between the income of locals and migrants. In other words, the income gap between locals and migrants can be explained by their differences in human capital, suggesting that there is no strong evidence of migrant discrimination in the Chinese urban labour market.

Table 3.

Variable definitions.

Variables	Definition
Dependent variable of permanent income regressions
Lnhinc	Log of current household income (in 10,000 yuan)
Dependent variable of multinomial logistic models of financial constraints
Afford = 0	The renter is not subject to any financial constraint to buy a home
Afford = 1	The renter is subject to a down-payment constraint to buy a home
Afford = 2	The renter is subject to an income constraint to buy a home
Afford = 3	The renter is subject to both income and wealth constraints
Independent variables
Age	Age of respondent
Workyear	Years of work experience
Age30	= 1 if older than 30, 0 otherwise
Female	= 1 if female, 0 otherwise
Migrant	= 1 if migrant, 0 otherwise
College	= 1 if three-year college education or below (education reference)
University	= 1 if four-year university education, 0 otherwise
Graduate	= 1 if graduate education, 0 otherwise
Manu	= 1 if working in the manufacturing industry (industry reference)
Private service	= 1 if working in the market-based service industry, 0 otherwise
Public service	= 1 if working in the public service sector (school/government, etc.)
SOE	= 1 if working at an SOE firm (firm reference)
Private	= 1 if working at a private or collectively owned enterprise
Public	= 1 if working at a public sector agency, including the government
Foreign	= 1 if working at an enterprise with foreign investment
Manager	= 1 if the respondent holds a managerial/leadership position, 0 if he or she is a worker or technician
FincgrowFast	= 1 if the respondent expects future income growth greater than 15% annually
Plnhinc	Log of predicted permanent household income
Temlnhinc	Log of temporary household income
Uncertaintyhinc	The ratio of the absolute value of temporal household income relative to predicted permanent household income

Note: the reference dummies are listed here to illustrate but are not used in the models.

Table 4.

Permanent income regressions.

Variables	All	Natives	Migrants	Owners	Renters
Variables	lninc	lninc	lninc	lninc	lninc
Migrant	0.009			0.092	−0.005
Migrant	(0.034)			(0.081)	(0.037)
Age	0.152***	0.121***	0.176**	0.133**	0.164***
Age	(0.032)	(0.041)	(0.073)	(0.061)	(0.047)
Age square	−0.002***	−0.002**	−0.003**	−0.002*	−0.002***
Age square	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)
Workyear	0.010	0.007	0.019	0.003	0.015*
Workyear	(0.006)	(0.007)	(0.013)	(0.009)	(0.008)
Female	−0.088***	−0.080***	−0.112**	−0.036	0.015
Female	(0.025)	(0.030)	(0.048)	(0.072)	(0.037)
College	0.381***	0.388***	0.376***	0.395***	0.377***
College	(0.035)	(0.042)	(0.068)	(0.059)	(0.044)
University	0.590***	0.624***	0.485***	0.614***	0.590***
University	(0.049)	(0.055)	(0.105)	(0.084)	(0.059)
Private Service	0.085***	0.071*	0.114**	0.082	0.074*
Private Service	(0.031)	(0.039)	(0.052)	(0.052)	(0.040)
Public Service	0.032	−0.021	0.141*	0.062	0.006
Public Service	(0.039)	(0.046)	(0.077)	(0.073)	(0.046)
Private	−0.068*	−0.063	−0.072	−0.110*	−0.051
Private	(0.035)	(0.044)	(0.057)	(0.061)	(0.043)
Foreign	0.186***	0.178***	0.194***	0.098	0.221***
Foreign	(0.038)	(0.044)	(0.070)	(0.065)	(0.048)
Public	−0.239***	−0.195***	−0.373***	−0.367***	−0.169***
Public	(0.053)	(0.060)	(0.130)	(0.098)	(0.063)
Manager	0.083***	0.072**	0.108*	0.122**	0.038
Manager	(0.028)	(0.032)	(0.060)	(0.048)	(0.034)
HPF Dummy	0.095**	0.167***	0.060	0.038	0.108**
HPF Dummy	(0.039)	(0.056)	(0.063)	(0.103)	(0.043)
Owner dummy	0.154***	0.135***	0.258***
Owner dummy	(0.031)	(0.033)	(0.082)
Constant	−1.389***	−0.990	−1.728*	−0.985	−1.596**
Constant	(0.476)	(0.631)	(1.004)	(0.976)	(0.671)
Observations	3021	2067	954	1026	1995
R-squared	0.197	0.181	0.217	0.135	0.171

Notes: Robust standard errors in parentheses; ***p < 0.01, **p < 0.05, *p < 0.1.

The reference for education dummies is high school or lower; the reference for industry dummies is manufacture industry; the reference for firm dummies is state-owned enterprise (SOE); the reference for occupation dummy (manager) is worker.

Financial constraints and the edges of homeownership

Owing to the possible endogeneity between household income and ownership status, we use the estimation results of the renter sample (column 3 in Table 4) to predict renters’ permanent income.

Through the multinomial logit model, we explore the connections between a renter’s permanent income and his or her self-assessed financial constraint to access homeownership. Following the standard practice in the literature (Goodman, 1988), the renter’s predicted temporal income, computed from the residuals of permanent income regressions, is also included as a control variable in the model to check whether income liquidity is a major issue. Previous studies in the literature have noted that income uncertainty is high in urban China and that this significantly affects a household’s housing demand (Zhou, 2011). Therefore, the model includes a measure of income uncertainty, which is calculated as the ratio of temporal income relative to permanent income. In addition, we also test how income growth potential may affect the self-assessed financial constraints by including a dummy indicator which is 1 if the renter predicts that his or her income growth in the next few years will be more than 15% on an annual basis, and 0 otherwise.

Table 5 reports the multinomial logit regression results. The interpretation of multinomial logit coefficients is not intuitively straightforward, but in spirit it is similar to the interpretation of logit coefficients (see explanations in section ‘The multinomial logistic model’). For example, the coefficient for migrants is positive in all three equations for all renters (columns 1–3), suggesting that a migrant has a higher log odds-ratio than locals to face these types of financial constraints as opposed to being free from any constraint. The coefficient estimate of migrants is 0.579 in the equation where both income and wealth are constrained (column 3), and the exponential of this figure is 1.784.⁴ Thus, the relative risk of simultaneously facing two constraints rather than being free from any constraint is 78.4% higher for migrants than for locals, everything else equal. However, it is interesting to note that, while the relative risk of migrants to be allocated on the down-payment-based homeownership edge as opposed to no constraints is 97.5% higher than comparable locals (the coefficient is 0.681 in column 1 and the associated odds-ratio is 1.975), the relative risk of being on the income-based homeownership edge is rival (the coefficient is 0.039 and the associated odds-ratio is 1.040, they are statically insignificant). This finding suggests that, compared with locals, the major financing obstacle of migrants to access homeownership is the inadequacy of down-payment rather than income shortage.

Table 5.

Multinomial logit model results of renter’s financial constraints (income-based). (0: no constraint, baseline and coefficients omitted; 1: down-payment constraint-edge 1; 2: income constraint-edge 2; 3. both).

Variables	All			Locals			Migrants
Variables	1	2	3	1	2	3	1	2	3
Migrant	0.681***	0.039	0.579***
Migrant	(0.192)	(0.210)	(0.191)
Age30	0.250	0.634**	0.466**	0.297	0.904***	0.601**	0.094	0.062	0.193
Age30	(0.236)	(0.254)	(0.232)	(0.295)	(0.308)	(0.291)	(0.409)	(0.474)	(0.403)
Female-headed	−0.190	−0.154	−0.126	−0.394	−0.303	−0.215	0.233	0.201	0.136
Female-headed	(0.203)	(0.218)	(0.202)	(0.243)	(0.258)	(0.238)	(0.402)	(0.429)	(0.400)
Lnphinc	−0.216	−1.066***	−0.792**	−0.251	−1.390***	−0.979**	−0.103	−0.585	−0.556
Lnphinc	(0.319)	(0.340)	(0.317)	(0.425)	(0.440)	(0.425)	(0.501)	(0.552)	(0.494)
Lntemhinc	−0.695	−0.713	−1.881***	−0.705	−0.948*	−2.231***	−0.618	−0.338	−1.360
Lntemhinc	(0.429)	(0.452)	(0.433)	(0.487)	(0.504)	(0.495)	(1.010)	(1.083)	(1.005)
Uncertaintyhinc	0.117	0.149	0.169	0.119	0.162*	0.189	0.151	0.190	0.186
Uncertaintyhinc	(0.099)	(0.097)	(0.118)	(0.095)	(0.095)	(0.116)	(0.318)	(0.329)	(0.397)
FincgrowFast	−0.552***	−0.709***	−0.800***	−0.614***	−0.784***	−0.753***	−0.473	−0.577	−0.819**
FincgrowFast	(0.185)	(0.201)	(0.185)	(0.228)	(0.246)	(0.228)	(0.327)	(0.361)	(0.327)
Constant	3.418***	4.569***	7.706***	3.591***	5.760***	8.935***	3.557	2.779	6.516**
Constant	(1.159)	(1.218)	(1.184)	(1.373)	(1.412)	(1.415)	(2.611)	(2.753)	(2.596)
Observations	1949	1949	1949	1160	1160	1160	789	789	789

Note: *** if significant at 1%; ** if significant at 5%; * if significant at 10%. Robust standard errors in parentheses.

As shown in Table 5, renters older than 30, especially those local renters, have greater odds of being subject to income constraint as opposed to no constraint, but their relative odds of being subject to wealth constraint are not significantly higher. This finding may have several interesting messages: first, it suggests that there is a large cohort difference of housing demand among urban residents in China. While young people can accept smaller homes when realising their first-home dreams, older people generally may demand better homes; second, work experience does not help much in enhancing one’s labour income in post-reform urban China; third, work experience however helps a lot in accumulating one’s wealth.

Table 5 also suggests that the relative odds of facing income constraint as opposed to no constraint decreases as permanent income increases. Nonetheless, the growth of permanent income does not help much to reduce the relative odds of down-payment constraint. This could be because housing in Shanghai is too expensive and the downpayment threshold is so high that marginal growth in permanent income has limited help in alleviating the wealth constraint. It could also be attributed to the fact that households with higher permanent income may demand higher-value homes and then the down-payment thresholds of their desired homes are higher than that of the average person. Interestingly, a growth in temporal income could only reduce the relative chance of being far from homeownership (subject to two constraints simultaneously) but have limited help on relaxing either down-payment constraint or income constraint. This suggests that one-shot cash subsidies may have little effect in boosting homeownership among Chinese young renters. However, except for locals, greater income uncertainty will not lead to higher odds of financial constraint relative to no constraint. Meanwhile, if a renter has an optimistic expectation of his or her income growth, the relative chance of assessing himself or herself as being subject to a financial constraint would decrease.

The average marginal effect estimates of all control variables with respect to the probability of being free from any financial constraint are reported in Table 6 (cf. equation 8 in section ‘The multinomial logistic model’ for details of such calculations). These results show that migrants on average have a 3.74% lower probability of being free from any financial constraint than comparable locals. Meanwhile, one percentage increase in permanent income would increase the probability of being free from any financial constraint by 4.87%, and higher expectations of future income growth shift the probability by 5.8%.

Table 6.

The average marginal effect estimate on the probability of no constraint.

Expression	Pr(Afford==0), predict(outcome(0)), Delta-method
Variable	dy/dx	Std. err.	z	P>\|z\|	(95% confidence interval)
Migrant	−0.0374	0.0126	−2.9600	0.0030	−0.0621	−0.0126
Female-headed	0.0123	0.0153	0.8000	0.4220	−0.0177	0.0422
Age30	−0.0301	0.0143	−2.1000	0.0360	−0.0582	−0.0020
Lnphinc	0.0487	0.0230	2.1200	0.0340	0.0037	0.0937
Lntemhinc	0.0906	0.0312	2.9000	0.0040	0.0294	0.1518
Uncertaintyhinc	−0.0111	0.0075	−1.4800	0.1390	−0.0259	0.0036
Fincgrowfast	0.0580	0.0158	3.6700	0.0000	0.0270	0.0890

Note: based on the results of the multinomial logit model for all sample reported in columns 1–3 of Table 5.

Based on the predictions from the multinomial logit model, Figure 1 reveals how the gap in the probability of being free from financial constraint between renters with different expectations of future income growth varies with permanent income. It suggests that higher expectations of future income growth decrease the probability of being free from financial constraints when the permanent income is sufficiently high. This could be because higher expectations of future income growth are usually associated with greater income uncertainty, which prevents renters from accessing homeownership.

Figure 1.

Probability of no constraint on homeownership as a function of permanent income at different levels of expected income growth.

Figure 2 describes how the effects of firm type on the relative chance of accessing homeownership vary with permanent income. For example, compared with employees of SOEs, private companies and the public sector with the same permanent income, employees of foreign-owned firms have the greatest relative chance of accessing homeownership when their permanent income is low, but they become the furthest from the homeownership boundary when their permanent income is very high. There are two possible explanations for this interesting observation. One is that employees of foreign-owned firms are more familiar with the mortgage market, affecting constraints for employees with lower salaries, but higher salaried employees of these firms may face more intense job competition, decreasing their relative likelihood of attaining homeownership. Another reason is that employees of foreign-owned firms with high incomes may have higher quality demands.

Figure 2.

Probability of no constraint as a function of predicted permanent income for different firm types.

Robustness check: Individual characteristics and the edges of homeownership

So far, we have tested how permanent income affects the relative risk of being subject to a certain type of financial constraint or the relative chance of being on the different edges of homeownership. Nonetheless, the precision of permanent income is sensitive to the quality of reported income, which can contain measurement errors with unknown extents. Meanwhile, we would like to know in more detail how the relative chance of accessing homeownership varies with the values of the individual characteristics. Thus, we run a multinomial logit model where the control variables are a set of individual characteristics.

Table 7 reports the outputs of this model for all renters and separately for migrants and locals. The average marginal effects of control variables on the probability of being free from any constraint are reported in Table 8. The results in Tables 7–8 fit well with the results in Tables 5 and 6. One item in Table 8 merits further attention: the employees of the public service sector have a higher relative probability of being close to homeownership, while the employees of private firms tend to be the furthest from the boundary between renting and homeownership.

Table 7.

Multinomial logit model results of renter’s financial constraints (individual characteristics-based). (0: no constraint, baseline and coefficients omitted; 1: down-payment constraint-edge 1; 2: income constraint-edge 2; 3. both).

Variables	All			Locals			Migrants
Variables	1	2	3	1	2	3	1	2	3
Migrant	0.414**	−0.176	0.333*
Migrant	(0.182)	(0.198)	(0.178)
Female-head	−0.071	0.071	0.126	−0.279	−0.046	0.027	0.314	0.370	0.376
Female-head	(0.193)	(0.207)	(0.188)	(0.238)	(0.250)	(0.229)	(0.360)	(0.390)	(0.352)
Age30	0.129	0.183	0.151	0.107	0.188	0.141	0.219	0.176	0.239
Age30	(0.205)	(0.219)	(0.201)	(0.246)	(0.259)	(0.240)	(0.375)	(0.415)	(0.369)
University	−0.238	−0.404*	−0.317	−0.314	−0.642**	−0.422	−0.075	0.014	−0.136
University	(0.201)	(0.213)	(0.198)	(0.266)	(0.272)	(0.262)	(0.332)	(0.358)	(0.328)
Graduate	−0.283	−1.15***	−0.404	−0.190	−1.48***	−0.259	−0.687	−0.635	−0.915*
Graduate	(0.281)	(0.322)	(0.280)	(0.354)	(0.397)	(0.350)	(0.475)	(0.551)	(0.477)
Private service	−0.166	−0.402*	−0.244	−0.263	−0.405	−0.172	0.028	−0.331	−0.242
Private service	(0.208)	(0.228)	(0.203)	(0.267)	(0.285)	(0.256)	(0.345)	(0.388)	(0.340)
Public service	−0.512**	−0.595**	−0.96***	−0.94***	−0.681**	−1.29***	0.516	−0.250	−0.022
Public service	(0.254)	(0.280)	(0.258)	(0.299)	(0.317)	(0.302)	(0.633)	(0.702)	(0.641)
Private	0.561**	1.158***	0.603**	1.039***	1.369***	0.966***	0.018	1.008**	0.161
Private	(0.238)	(0.257)	(0.235)	(0.341)	(0.355)	(0.337)	(0.380)	(0.440)	(0.379)
Foreign	0.338	0.716***	0.042	0.172	0.629**	−0.005	0.315	0.796*	−0.027
Foreign	(0.231)	(0.251)	(0.229)	(0.286)	(0.295)	(0.279)	(0.409)	(0.475)	(0.407)
Public sector	1.096***	1.423***	1.141***	1.389***	1.477***	1.359***	0.621	1.430	0.811
Public sector	(0.334)	(0.364)	(0.335)	(0.367)	(0.391)	(0.365)	(0.955)	(1.049)	(0.960)
Manager	−0.53***	−0.54***	−0.59***	−0.58***	−0.574**	−0.57***	−0.333	−0.457	−0.517*
Manager	(0.171)	(0.186)	(0.167)	(0.218)	(0.229)	(0.211)	(0.294)	(0.338)	(0.291)
Constant	1.374***	1.024***	1.837***	1.562***	1.279***	1.899***	1.555***	0.418	2.079***
Constant	(0.263)	(0.278)	(0.260)	(0.327)	(0.339)	(0.325)	(0.430)	(0.460)	(0.426)
Observations	2124	2124	2124	1263	1263	1263	861	861	861

Note: *** if significant at 1%; ** if significant at 5%; * if significant at 10%. Robust standard errors in parentheses.

Table 8.

The average marginal effect estimate on the probability of no constraint.

Expression	Pr(Afford=0), predict(outcome(0)), Delta-method
Variable	dy/dx	Std. err.	z	P>\|z\|	(95% conf. Interval)
Migrant	−0.0200	0.0124	−1.6200	0.1060	−0.0443	0.0042
Age30	−0.0113	0.0140	−0.8000	0.4210	−0.0387	0.0162
Female-headed	−0.0036	0.0136	−0.2600	0.7920	−0.0303	0.0231
University	0.0224	0.0130	1.7200	0.0860	−0.0032	0.0479
Graduate	0.0381	0.0221	1.7300	0.0840	−0.0051	0.0814
Private service	0.0175	0.0138	1.2700	0.2040	−0.0095	0.0446
Public service	0.0611	0.0222	2.7600	0.0060	0.0177	0.1045
Private	−0.0575	0.0175	−3.2800	0.0010	−0.0918	−0.0232
Foreign	−0.0277	0.0196	−1.4100	0.1580	−0.0662	0.0107
Public sector	−0.0812	0.0181	−4.4900	0.0000	−0.1166	−0.0458
Manager	0.0450	0.0131	3.4300	0.0010	0.0193	0.0707

Note: based on the results of the multinomial logit model for all samples reported in columns 1–3 of Table 7.

Conclusions

This paper studies what factors impede youths’ pathways to homeownership in post-reform urban China. We distinguish three types of borrowing constraints to access homeownership: constrained by both wealth and income, constrained by wealth but not income, and constrained by income but not wealth. We denote the latter two cases as edges of homeownership.

Using a large set of survey data collected in Shanghai in 2011, we investigate how personal characteristics affect the relative risk of being on a certain type of homeownership edge. Our findings help to identify what young adults on the edges of homeownership look like, which may have significant policy implications for housing market development and mortgage product design, especially for big cities where the down-payment threshold is very high. For example, compared with comparable locals, migrants do not have much higher relative risks to be subject to income constraints but are much more easily subject to down-payment constraints. This implies that, for migrants, the shortage of wealth is the major obstacle to attain homeownership in Shanghai. Then, if the Chinese government really wishes to encourage the migrants to settle down in the big cities,⁵ the provision of mortgage products with lower down-payment requirement may appear more effective than the cash assistance policy.

We also find higher permanent income can reduce the relative risk of income constraints but have little help on alleviating down-payment constraint. Thus, wage growth alone would be insufficient for boosting homeownership among young renters if it occurred without suitable mortgage innovations. Compared with the USA and other mature housing markets, the minimum down-payment requirement of 30% appears too strict. Given the robustness of the growth of household incomes in urban China, there is considerable room to lower the down-payment requirement. Further, we also find that young renters with higher expectations of future income growth are much more likely to enter the ownership market even at the same permanent income level. This suggests maintaining positive expectations of long-term economic prosperity are very important in encouraging homeownership dreams. We also discover that a young renter’s relative risks of different financial constraint vary considerably with the industry and type of firm in which they work. This suggests that the banks should take more detailed occupation information into account when assessing a mortgage borrower’s financing hazards.

Admittedly, as our work relies on the respondent’s subjective assessment of financial constraint, it could have some biases due to either overconfidence or underestimation of one’s true homeownership potentials. Nonetheless, in our approach there is no need of a subjective definition of ‘desired home’ and it is possible to take into account influencing factors of homeownership barriers that are unobservable to researchers. Thus, we suppose that our approach can help to highlight the determinants of homeownership barriers. In this perspective, our work can have direct implications on housing development policies in China as well as in many other emerging economies.

Footnotes

Funding

Funding was received from National Science Foundation of China, 71173045, 71573766, 71461137002; Shanghai Education Development Foundation, 13SG35; The Ministry of Education of PRC, 13JZD009.

Notes

References

Andrew

Meen

(2003) Housing transactions and the changing decisions of young households in Britain: The microeconomic evidence. Real Estate Economics31(1): 117–138.

Asberg

(1999) Housing decisions of young Swedish adults. Journal of Housing Economics143(2): 116–143.

Bourassa

Haurin

. (1994) Independent living and home ownership: An analysis of Australian youth. Australian Economic Review27(3): 29–44.

Chen

Deng

(2014) Financing affordable housing through compulsory saving: The two-decade experience of Housing Provident Fund in China. Housing Studies29(7): 937–958.

Chen

Han

(2014) The evolution of housing market and its socio-economic impacts in post-reform China: A survey of the literature. Journal of Economic Survey28(4): 652–670.

Chen

Hao

Stephens

(2010) Assessing housing affordability in post-reform China: A case study of Shanghai. Housing Studies25(6): 877–901.

Clapham

Buckley

Mackie

. (2012) Housing Options and Solutions for Young People in 2020. York: Joseph Rowntree Foundation.

Clark

WAV

Mulder

(2000) Leaving home and entering the housing market. Environment and Planning A32(9): 1657–1671.

Coulson

(2013) Measuring the external benefits of homeownership. Journal of Urban Economics77: 57–67.

10.

Coulson

Tang

(2013) Institutional and demographic influences on the presence, scale and geographic scope of individual Chinese real estate investment. Regional Science and Urban Economics43(2): 187–196.

11.

Doling

Ronald

(eds) (2014) Housing East Asia: Socioeconomic and Demographic Challenges. Hampshire: Palgrave Macmillan.

12.

Engelhardt

Eriksen

Gale

. (2010) What are the social benefits of homeownership? Experimental evidence for low-income households. Journal of Urban Economics67(3): 249–258.

13.

Feng

(2013) School quality and housing prices: Empirical evidence from a natural experiment in Shanghai, China. Journal of Housing Economics22(4): 291–307.

14.

Fisher

JDM

Gervais

(2011) Why has home ownership fallen among the young?International Economic Review52(3): 883–912.

15.

Forrest

Yip

(2012) Young People and Housing: Transitions, Trajectories and Generational Fractures. London: Routledge.

16.

Goodman

(1988) An econometric model of housing price, permanent income, tenure choice, and housing demand. Journal of Urban Economics23(3): 327–353.

17.

Greene

(2003) Econometric Analysis. Englewood Cliffs, NJ: Pearson Education.

18.

Han

(2010) Housing demand in Shanghai: A discrete choice approach. China Economic Review21(2): 355–376.

19.

Hancock

(1993) ‘Can pay? Won’t pay?’ or economic principles of ‘affordability.’Urban Studies30(1): 127–45.

20.

Haurin

Hendershott

Kim

(1994) Tenure choice of American youth. Journal of Urban Economics35(1): 28–45.

21.

Haurin

Hendershott

Watchter

(1997) Borrowing constraints and the tenure choice of young households. Journal of Housing Research8(2): 137–154.

22.

Jones

(1995) Testing the central prediction of housing tenure transition models. Journal of Urban Economics38(1): 50–73.

23.

LaFayette

Haurin

Hendershott

(1995) Endogenous Mortgage Choice, Borrowing Constraints and the Tenure Decision. NBER Working Paper No.5074. Cambridge, MA: National Bureau of Economic Research.

24.

(2012) Deepening the strategy of expanding domestic consumption through furthering the reform and openness. Qiu Shi54(4): 3–10 [in Chinese].

25.

Linneman

Wachter

(1989) The impacts of borrowing constraints on homeownership. Real Estate Economics17(4): 389–402.

26.

McKee

(2012) Young people, homeownership and future welfare. Housing Studies27(6): 853–862.

27.

Mak

SWK

Choy

LHT

WKO

(2007) Privatization, housing conditions and affordability in the People’s Republic of China. Habitat International31(2): 177–192.

28.

Megbolugbe

Linneman

(1993) Home ownership. Urban Studies30(4–5): 659–682.

29.

National Bureau of Statistics of China (NBSC) (2011) Series Reports on China’s Economic and Social Development Achievements During the 11th Five-Year Planning Period (2006–2010): No.9. The Official Website of National Bureau of Statistics of China (NBSC). Available at: http://www.stats.gov.cn/ztjc/ztfx/sywcj/201103/t20110307_71321.html (accessed 1 March 2013).

30.

National Population and Family Planning Commission (PFPC) (2012) Report on China’s Migrant Population Development 2012. National Population and Family Planning Commission of P.R. China (PFPC). Beijing: China Population Press.

31.

Ortalo-Magné

Rady

(1999) Boom in, bust out: Young households and the housing price cycle. European Economic Review43(4–6): 755–766.

32.

Öst

(2011) Parental wealth and first-time homeownership: A cohort study of family background and young adults’ housing situation in Sweden. Urban Studies49(10): 2137–2152.

33.

Rossi-Hansberg

Sarte

P-D

Owens

III (2010) Housing externalities. Journal of Political Economy118(3): 485–535.

34.

Sato

(2006) Housing inequality and housing poverty in urban China in the late 1990s. China Economic Review17(1): 37–50.

35.

Shanghai Statistics Office (SSO) (1983) Shanghai Statistics Yearbook 1983. Shanghai: SSO.

36.

Shanghai Statistics Office (SSO) (2012) The Handbook of Key Survey Results of the Sixth Population Census in Shanghai. Shanghai: SSO.

37.

Shanghai Statistics Office (SSO) (2014) Statistical Bulletin of National Economic and Social Development in Shanghai 2013. Shanghai: SSO.

38.

Wang

(2012) What is unique about Chinese real estate markets?Journal of Real Estate Research34(3): 275–289.

39.

Wei

S-J

Zhang

Liu

(2012) Status Competition and Housing Prices: Some Evidence from China. NBER Working Paper 18000. Cambridge, MA: National Bureau of Economic Research.

40.

Wood

Smith

Ong

Cigdem

(2013) The edges of home ownership, AHURI Final Report No. 216. Melbourne: Australian Housing and Urban Research Institute.

41.

World Bank (1993) Housing Indicator Program: Report of the Executive Director. Washington, DC: World Bank.

42.

World Bank (2013) China 2030: Building a Modern, Harmonious, and Creative High-Income Society. Washington, DC: The World Bank.

43.

(2004) Sources of migrant housing disadvantage in urban China. Environment and Planning A36(7): 1285–1304.

44.

Xinhua News (2015) The Public Announcement of the Economic Conference of Central Politburo of the CCP (Communist Party of China) on Dec.14, 2015. Xinhua News. Available at:http://news.xinhuanet.com/politics/2015-12/14/c_1117456577.htm.

45.

Yang

Shen

(2008) The affordability of owner occupied housing in Beijing. Journal of Housing and the Built Environment23(4): 317–335.

46.

Yang

Wang

(2011) The impact of privatization of public housing on housing affordability in Beijing: An assessment using household survey data. Local Economy26(5): 384–400.

47.

Ying

Luo

Chen

(2013) The determinants of homeownership affordability among the ‘sandwich class’: Empirical findings from Guangzhou, China. Urban Studies50(9): 1870–1888.

48.

Zhang

Chen

(2014) The changing prevalence of housing overcrowding in post-reform China: The case of Shanghai, 2000–2010. Habitat International42(2): 214–223.

49.

Zhou

(2011) Uncertainty and housing tenure choice by household types: Evidence from China. China Economic Review22(3): 408–427.

50.

Zorn

(1989) Mobility-tenure decisions and financial credit: Do mortgage qualification requirements constrain homeownership. Real Estate Economics17(1): 1–16.