ABSTRACT
China has experienced very rapid house price inflation in recent years – by some 10% per annum relative to consumer price inflation. Existing house-owners have benefited from capital gain and have been able to climb the housing ladder. Young household heads – wanting to own a house and facing rising house prices relative to their incomes – have found it increasingly difficult to get onto the housing ladder. However, their difficulty is eased by the strength of family support and the developing market for housing loans. The China Household Income Project (CHIP) surveys of 2002 and 2013 are analysed to test the hypothesis that the age distributions of house ownership and of housing wealth have moved against the young. Probits, tobits, and difference-in-difference analyses measure these effects. There is indeed evidence indicating a receding housing ladder, and evidence that waiting longer to own a house is positively associated with house price inflation.
1. Introduction
Arguably the most important event in the world economy in recent years is the rise of China in little more than forty years, from being on a par with black Africa at the start of its economic reforms to becoming an economic superpower. Associated with China’s rise has been a remarkable housing boom and relative house price inflation – due to a combination of the privatisation of urban housing, the strengthening of market forces, the creation of a mortgage market, ‘the greatest migration in human history’, tardy release of land for housing, and speculation that the inflation would continue. House price inflation has had important implications for the inequality among households of total wealth and of housing wealth (Knight, Shi, and Haiyuan Citation2022). However, a neglected implication is its effect on the intergenerational inequality of housing wealth, and on the housing ladder.
The term ‘housing ladder’ or ‘property ladder’ is sometimes used to denote the first purchase of a house, i.e. getting onto the housing ladder, and thereafter the purchase of a better house or improvement of the existing house, i.e. climbing the housing ladder. The housing ladder might recede in the sense that young people must wait longer to first own a house; that is, the average age at which a householder first acquires a house rises over time. It is a phenomenon that might be found in any country which has experienced rapid house price inflation relative to incomes. For example, in England, where housing prices have outstripped incomes in the twenty-first century, the mean age of first-time buyers increased from 30 years in 1995/6 to 31 in 2005/6 and to 32 in 2015/6, and the figures for house-ownership among those aged 35 and under were 19%, 13% and 9% respectively (English Housing Survey 2016–2017, online).
The complaint heard in some advanced economies is often heard in the cities of China, both by the media and by government. Official concern is evident in various statements of the State Council: the difficulties of house purchase ‘especially for young people in big cities’ were discussed in March 2007, similarly in June 2016, and again in July 2021.
The probable reason is that the price of houses has risen relative to incomes. All householders could in principle benefit from their rising incomes and improved access to housing loans in recent years, However, whereas many older householders already had a house and had the benefit of capital gain on their house, many young householders had neither, and needed to raise funds for their first house purchase. Young people struggled to buy houses given the high prices in relation to their incomes and their capacity to get housing loans or parental support. Our hypothesis is that the average length of time a householder waited before first owning a house increased. We wish to measure this average wait and its change over time. Two national surveys, for the years 2002 and 2013, are examined. We do indeed find support for this hypothesis. More rigorously, by means of intergenerational analyses we examine whether there was a decrease in the conditional probability of house ownership among young people and a decrease in the conditional relative housing wealth held by them.
Our analytical framework is one of demand and supply: the demand for finance to fund house purchase and the supply of funds for house purchase, which is influenced by age-specific differences in access, by the extent of the mortgage market, and by the degree of family support.
In section 2 we describe the data sets to be used, and in section 3 their previous use in the study of China’s household wealth. The next two sections explain the forces at work, drawing where possible on the available literature. Section 4 considers reasons why the housing ladder might recede. It discusses the ways that young people find to cope with the increasing difficulty of house purchase. In section 5 we consider possible reasons why the housing ladder might advance rather than recede, including intra-family transfers. We then test the hypothesis that rapid house price inflation caused young people to become increasingly disadvantaged over the eleven-year period, analysing the average wait to own a house and the age distribution of house ownership (section 6), and the age distribution of housing wealth (section 7). The hypothesis tests involve probit, tobit, and difference-in-difference analyses. The hypothesis is supported. Section 8 summarises and concludes.
To the best of our knowledge, this is a first attempt to examine the age distribution of house ownership and of housing wealth over time in China, and to argue that house price inflation placed young people at an increasing disadvantage, requiring them to wait longer to own a house. We expect these results to make an original contribution for China and more widely.
2. The data
Our data are from the 2002 and 2013 surveys of the Chinese Household Income Project (CHIP). These are household sample surveys, being sub-samples of the national household surveys of the National Bureau of Statistics. We opted for 2002 as the base year rather than an earlier or later year because the 2002 CHIP survey is the first comprehensive data source on housing wealth. The period 2002–2013 was one of very rapid house price inflation. We need the data to be as comparable as possible. Fortunately, the variables relating to housing wealth are very similar in the two surveys. Thus the estimates of housing wealth distributions can be compared given appropriate weighting. The weights used were effectively the same as those applied generally in the CHIP 2002 and 2013 surveys to achieve national representativeness.
We concentrate on the most important data issues.Footnote1 Valuing wealth – in particular housing and land wealth – inevitably encountered problems given China’s marketising, but still semi-marketised, economy. Net housing is housing value minus housing loan. This is based on respondents’ reported values (of both owner-occupied and other houses) in each year, despite the weakness of the housing market in rural China. Missing values had to be interpolated. For instance, where a housing value is missing, its imputation is on the basis of price per square metre at the local (county, city, or city district) level. Real wealth is obtained by reflating 2002 nominal wealth by the NBS’s consumer price indexes, so as to express the 2002 values in constant 2013 prices. We use province-level consumer price indexes, distinguishing also between urban and rural indexes. Throughout the paper our discussion of wealth is real wealth, i.e. measured at constant cpi-adjusted prices. When we use the term housing wealth, we mean household real housing wealth per capita. As the rural and urban surveys are separate, it is possible to examine housing wealth in each of them as well as in the weighted national survey. There is an analytical case for doing so because the rural sector is much poorer and subject to sharply different economic policies and institutional arrangements: relevant trends might differ. There is no direct information on the year in which a householder first owned a house, nor on the means by which the house was funded. This lack is important for our testing strategy.
It will become evident below that the growth of housing wealth and of house prices is central to our story. Analysis is complicated by the fact that our two data sets do not constitute a panel: a pseudo-panel must be created. Two approaches were tried. One was to use data published by the Ministry of Housing and Construction, which show the value of sales of commercialised buildings, and the corresponding sold floor space, at district and county level. From this information it was possible to construct a housing inflation index. The other approach was to calculate house prices from the CHIP surveys for each of the included urban and rural areas within each included province. Districts within cities were used in the case of metropolitan areas. Reflecting the data available, each subsample was divided into ranked subgroups based on average house value per square metre, and these subgroups were compared in 2002 and 2013. If an area was not included in both years another location with very similar housing price was substituted. The resultant house price inflation index was then applied to all households in each area. Robustness tests were passed. The results obtained by the two approaches were fairly similar. Our estimates of house price inflation are based on the second approach. Our interest is in the relative house price inflation, measured as house price inflation in relation to consumer price inflation. We shall refer to this measure as relative house price inflation or real capital gain from house ownership.
3. Previous wealth research on the CHIP surveys
The CHIP 2002 and 2013 household wealth data were examined in the descriptive chapter (Knight, Shi, and Haiyuan Citation2020) and the analytical paper (Knight, Shi, and Haiyuan Citation2022); both explain the sources and methods in some detail. The main evidence from those sources that provides background for this paper is as follows:
At the national level, household overall net (real) wealth per capita increased by 16.6% per annum, and net housing was the asset type that increased fastest (19.9%). Similar patterns were found in urban and rural China, the corresponding figures being 16.8% and 19.4% respectively (urban) and 14.1% and 17.9% respectively (rural). The share of net housing rose from 53% to 73% of China’s total wealth. Housing clearly plays a central role in China’s accumulation of wealth.
The share of the richest wealth per capita decile rose from 37% to 48%, a rise of 11% points. In fact, only the top decile experienced an increase in share over the period. The Gini coefficient of household wealth per capita increased over the eleven years, from 0.50 in 2002 to 0.61 in 2013. The Gini coefficient for housing wealth increased correspondingly from 0.64 to 0.72. The contribution of net housing to the inequality of wealth rose, being 64% in 2002 and a remarkable 79% in 2013. Again, housing plays a central role in China’s growing inequality of wealth.
It is possible to divide the increase in housing wealth into that part which is due to relative house price inflation and that part due to a real increase in housing. However, our measure of relative house price inflation necessarily includes the value of house improvements per square metre: it is not a pure price effect. Thus the real increase (the increase in housing quantity) represents an increase in the average number of square metres reported. Insofar as part of the increase in house values is due to housing improvements, these improvements add to the difficulty faced by young people wanting to own a house. Relative house price inflation, so measured, accounts for most of the increase in housing wealth.
The annual rate of return on housing wealth (excluding real capital gain) was estimated to be 4.9% in 2002 and 4.6% in 2013. The rate of return on relative house price inflation over the period was estimated to be 14.9% per annum. We have to assume that this rate of return applied in the year 2002 and in the year 2013. The overall rate of return on housing wealth was therefore 19.8% in 2002 and 19.5% in 2013. This high return illustrates why there is a strong incentive for people to try to climb onto the housing ladder. There is in principle a large demand for house purchase, but in practice that demand might be limited by the availability of funds.
4. Reasons why the housing ladder recedes
Before testing the receding ladder hypothesis (Section 6) we provide reasons why the housing ladder might recede (Section 4) and reasons why it might instead advance (Section 5). The outcome will be the net product of both sets of forces.
4.1 The increase in house prices
produces three results. One is the rapid annual increase in relative house prices in each geographic area. That increase exceeds 9% in all but the other urban areas, and is highest in China as a whole. This is because the proportionate weight of urban areas, with their much higher house prices, increased over the period. Had no reweighting occurred, the national house price inflation rate would have been 9.4% per annum.
Table 1. The level and percentage increase in average house prices by area.
Table 2. The ratio of household average house value to household average income, by area: national, rural, urban, large cities, other urban; all household heads and household heads aged 20–34.
Table 3. The development of the housing loan market in China, 1998–2013.
Table 4. Probit analysis: coefficient and marginal effects of inheritance and of parental support explaining house ownership by male household heads aged 20–34.
Table 5. Average age, and average years waiting to own a house, of household heads in the age group 20–34 who do not own a house, 2002, 2013, and increase, by area.
Secondly, the higher the concentration of population in a residential area the higher the average house price: rising from rural to other urban to large cities; urban is of course a weighted average of large cities and other urban, and national is a weighted average of rural and urban. This progression in prices by area reflects the corresponding progression of land values and the rapid pace of urbanisation.
The third result is that, with one exception, the percentage increase in average house prices by area between 2002 and 2013 is similar to the pattern of house price levels in 2002. Other urban areas had the lowest increase, then rural areas, and large cities had the highest increase. The table indicates why the difficulty of climbing onto the housing ladder is likely to be greatest in large cities.
4.2 The house value/income ratio
examines the ability of households to purchase houses given their income. It shows that the ratio of average house value to average income increased in all areas over the eleven-year period. The table also shows that, in both years, the ratio of house value to income is lower for the household head age group 25–34 than for all age groups. The ratio increased dramatically for young household heads between 2002 and 2013. However, the annual percentage increase was generally lower for them than for all heads, with the exception of large cities, where the respective increases were 9.0% and 7.7%.
It might be inferred that, in both years, it was easier for young than older household heads to get onto the housing ladder. However, this misses the point that most household heads aged 20–34 are not yet on the ladder, and the few who are already there are likely to have done so through inheritance or parental help. By contrast, most household heads beyond the age of 34 are likely to be on the ladder. Insofar as they wish to ascend the ladder by improving their housing conditions, they face an easier challenge. House-owners will have had the advantage of capital gain because of the rapid increase in house prices. Their existing houses will have risen in value, so it is only a matter of funding the additional cost of improved housing.
There is considerable recognition in the literature that house prices have risen relative both to the consumer price index and to income per capita in China, particularly in the big cities, and that houses have become less affordable as a result. See, for instance, the special issue of the Journal of Housing Economics in Lixing and Xiaoyu (Citation2019), including Chen, Hu, and Lin (Citation2019) and Lixing and Xiaoyu (Citation2019). Some consequences of this growing unaffordability were examined (such as the increase in co-residence with parents, and the growing relative attractiveness of second-tier cities). However, to the best of our knowledge, our questions have not been explicitly addressed. Chen, Hao, and Stephens (Citation2010) come closest to doing so. Using the necessary mortgage payment/income ratio as an indicator of affordability, they show that the ratio declined as each cohort of market entrants grew older but that the later a cohort entered the market, the higher was the ratio and thus the less affordable house purchase became for each sequential cohort.
4.3 The role of speculation
China’s housing market has been increasingly driven by speculation, to the detriment of young households. The stock market has fluctuated and has not fully reflected the remarkable success of the private sector. The Shanghai share price index, when set at 100 in 2002, reached 156 in 2013, implying an annual growth rate of 4.1%. Moreover, interest rates available to households on deposits and bonds are kept low, in line with government policy. By contrast, the rate of return on housing wealth is estimated to be 20% per annum, including capital gain of 15% per annum (Knight, Shi, and Haiyuan Citation2022). There was a strengthening view that house price inflation will continue and that housing is the best investment for households. According to a report of the China Household Finance Survey (Citation2018), in 2017 no fewer than 22% of house buyers already owned more than one house in urban areas, 66% owned one, and only 12% owned none. First time buyers were being pushed out of the market by house price inflation, itself aggravated by speculative investment in housing.
4.4. Coping with the receding housing ladder
There is evidence that the rapid increase in China’s urban house prices has elicited several behavioural responses in order to cope with the receding housing ladder. Chen, Yang, and Zhong (Citation2020) analysed the causal relationship between the urban household saving rate and home ownership. Using a natural experiment which created an exogenous variation in housing demand and using a difference-in-difference strategy, the authors found that the reform measures over the period 1998–2001, shifting the burden of funding from the state to households, caused a sharp increase in the saving rates of private households. Lixing and Xiaoyu (Citation2019) showed an increasing tendency of adult children to co-reside with their parents during the recent period of rapidly rising house prices. The incentive for house purchase did not diminish: the authors argued that co-residence reflected the incentive to economise on living costs while saving to buy a house.
Behavioural responses to house price inflation have some notable socioeconomic effects. The average age at which men in China first marry has risen by 1.8 years over 15 years, from 24.5 in 2000 to 26.3 in 2015,Footnote2 but it is faster in large cities, by 1.4 years over the seven years 2005–2012.Footnote3 Nie (Citation2020) has argued that this rise is endogenous: an increase in the male/female ratio among young adults has increased marriage competition. The greater need for young men to improve their marriage prospects has resulted in a higher average saving rate among unmarried men and a higher average male age at marriage. However, we have an additional explanation. The increase in relative house prices is likely to have intensified marriage competition via house ownership: the greater need of the unmarried to build up savings has increased the average age at which men marry. Indeed, there is evidence of a causal effect. Using data from the 2005 sample population census on a man’s age at first marriage and average city house prices, Wrenn, Junjian, and Bo (Citation2019) argue that marriage age does indeed rise in response to a rise in house prices.
5. Reasons for an advancing housing ladder
5.1 The expanding housing loan market
There was no private housing market and no residential mortgage market under central planning. There was, however, a distinction between rural and urban China. Rural households were permitted to own, build and inherit housing: they owned their houses but their collective owned their land. Each household could have only one house-building plot and there could be transfers only within the village. Rural housing markets were suppressed and underdeveloped, but rental and sale of housing gradually emerged especially near the cities (Sato, Sicular, and Ximing Citation2013).
Urban housing was social housing, owned by central and local government and work units. Privatisation of urban housing occurred in the 1990s, through the sale of social houses to their occupants at subsidised prices. The housing provident fund, introduced in 1994 and adopted nationally before 2000, required contributions from both workers and employers and made possible low-interest bank loans for house purchase (Sato, Sicular, and Ximing Citation2013). The sale of urban land rights became an increasingly important revenue source for local governments. There was unprecedented growth in both the housing markets and the mortgage markets. It is likely that the market was initially supply-constrained but in the 2000s became demand-constrained.
Government played a role as the sole supplier of urban land and in the regulation of mortgage markets. It was important in establishing lending standards and in changing the attitudes of banks towards residential lending. The five (previously) state-owned banks dominated this lending, although other financial institutions (such as city banks, joint-equity banks, and rural financial institutions) also grasped the new lending opportunities.
records how rapidly the mortgage market grew over the period from 1998 to 2013, i.e. roughly the period examined in this paper. Housing investment rose from 2.5% to 17.2% of GDP in 15 years. The balance of individual mortgage loans rose from 0.5% to 17.2% of GDP. This showed the great increase in demand for housing, reflected in the rise in house prices and the expansion of urban floor area per capita. The supply of loans responded but only if various conditions were met.
All residential mortgages in China are adjustable rate mortgages, following the base rate set by the central bank. Default rates are kept low by good due-diligence procedures of lenders, e.g. requiring tax returns, bank statements, and proof of other assets. The minimum down-payment is set by the central bank at 20% of the purchase price and the maximum mortgage maturity was 30 years before 2013. Applicants for house loans must therefore pass stringent tests and find 20% of the purchase value from other sources, usually own savings or family savings. The increased availability of mortgage loans should have conditionally advanced the housing ladder between 2002 and 2013.
5.2 Intergenerational transfers
There is survey evidence that parental support is potentially important for house purchase in China. For instance, Bingqin and Ban Shin (Citation2013), in a Tianjin study, find substantial intergenerational housing support, not only for cohabitation, both ways, but also for parents to transfer, or help to purchase, a house for a child. Zequn and Ning (Citation2022), provide evidence that parental housing contributions to the child involve expectations of care in old age. Ling (Citation2018) finds that the income and wealth of the parents has become increasingly important in determining whether their adult child owns a house.
The housing ladder might be advanced through family inheritance or through family transfers. Consider inheritance: it is unusual for people to become household heads while they are young adults. It can happen, however, through the death of a parent who was the household head. The child of the deceased might then become the head. Among heads in the younger age groups there are likely to be some who have inherited a house by this means. Such young household heads were not subject to the difficulties that others are likely to face in acquiring a house. Correction would be simple if there were sample information about the reasons for having acquired the house. It might be possible if there were sample variables which are indicators that a young person has inherited a house. The most likely available proxy is whether the father, or mother, or father and mother of the household head is not alive.
Consider the possibility of intergenerational transfers within the family. Wei, Zhang, and Liu (Citation2017) examine the effect of the male-female sex ratio among 5–19 year-olds, used as a proxy for the strength of future competition for marriage, on house prices in Chinese cities. In support of their hypothesis they cite evidence that 80% of Chinese mothers would object to their daughter marrying a man who does not own a house, and that 70% of unmarried women prefer that their future husbands own a house (pp.169–170, 177). The authors find that possession of a young son raises the house price, and that variation in the sex ratio accounts for 30–48% of the rise in urban house prices in Chinese cities over the period 2003–9. We take this to be evidence that parents are willing to improve their housing so as to strengthen their child’s marriage prospects, implying that the child will himself benefit from ownership of the house in the future.
Intergenerational gifts of housing are plausible features of China’s family-centred society. They are more important in China than in Western countries, partly because of the traditional custom that children, especially sons, are responsible for supporting their parents in old age. Investment in a house is widely perceived to be the best investment open to an individual in China. That norm and that perception create incentives for parents to provide a house for their son. The incentives might well have grown stronger as house price inflation has spiralled and the ratio of house price to household income per capita has risen: parental support might have responded. In rural areas a young couple normally live with the son’s parents after marriage, with the likely house extension producing a direct channel of wealth transfer. In urban areas either the same co-residence applies or the parents of the bridegroom buy a house for the couple or at least cover the downpayment portion of the cost of the new house (Wei, Zhang, and Liu Citation2017, 177). In some large cities, the parents have to purchase a second home in the name of the child, so as to avoid fiscal disincentives or regulations of the city government.
Wei and Wang (Citation2022) examined the role of inheritance in the determination and distribution of household wealth in China, using the China Health and Retirement Longitudinal Survey (CHARLS) – its 2013 survey and its 2014 life history survey which provided information on inheritance. Unfortunately, the survey covered only inheritors aged 45 or over. 18% of households had inherited wealth, and inherited wealth accounted for less than 5% of their net assets; this figure was less than 1% for all households. Real estate was the main form of inheritance. It was shown on various alternative assumptions that household wealth per capita had a slightly higher Gini coefficient when inherited wealth was excluded. A possible reason given for the unimportance of inherited wealth, and for its equalising effect, is that parents, especially wealthy parents, transfer wealth to their children during their lifetimes owing to the strength of family ties in China.
reports a probit analysis of the probability of a male aged 20–34 owning a house. We test two hypotheses – the inheritance hypothesis and the parental support hypothesis – against each other. If young household heads have no parents alive, the probability that they will have inherited the ownership of a house is higher than for other young household heads. If young household heads have one or both parents alive, the probability that they will have received parental support to acquire a house is increased, again relative to other young household heads. The table contains a dummy variable: one or both parents are alive (with neither parent is alive being the omitted category). Coefficients are reported in the top half of the table. The effect of the coefficients generated by a probit equation on the probability depends on the values of the other explanatory variables. The bottom half of the table reports the marginal effects on the probability of owning a house when other explanatory variables are set at their mean values. Owing to the possibility of endogeneity, the marginals indicate associations and do not necessarily measure the causal effects of the explanatory variables on the probabilities of house ownership.
The control variables indicate that having more education (possibly a proxy for parental wealth or own income) raises the probability of owning a house and that higher income does so in large cities and other urban areas. So also does membership of the Communist Party, except in rural areas, and increasingly so over time. It is understandable that rural China is exceptional because of the ease with which all households can acquire a village house. The negative coefficients and marginals at the national level are due to the much lower income but – institutionally determined – higher house ownership in rural than in urban areas.
The coefficient on having one or both parents alive is expected to be positive if lifetime family transfers of wealth are more important than inherited wealth, and negative if the reverse is the case. The coefficient is positive in every case and significantly so in five of the eight cases. The marginal effect is indeed positive and large in large cities (0.27*** and 0.12***), in other urban areas (0.10* and 0.12***), and in China as a whole (0.11*** and 0.17***). It suggests that intra-family gifts are more important than inheritance. This disadvantage of orphanhood is consistent with the finding that inheritance is unimportant in explaining household wealth, referred to above (Wei and Wang Citation2022). It is possible, moreover, that becoming an orphan at an early age reduces the parental wealth available to be inherited. When a distinction is made between householders with one parent alive and both alive (not reported in the table), the coefficient on both parents is consistently positive but not significantly so. Having two parents alive might help to strengthen family support.
6. The age distribution of house ownership, 2002 and 2013
We examine the age distribution of house ownership in four ways: first, by means of a descriptive figure, showing how house ownership varies by age group; second, by measuring the change in average age of young household heads who did not own a house; third, by explaining this change in terms of a proxy for increasing difficulty of acquiring a house; fourth, by conducting a probit analysis of the determinants of house ownership including age-group of the household head;. In all four cases it is the difference in outcomes between 2002 and 2013 that is important to our argument.
shows the percentage of each five-year age group who own a house in 2002 and in 2013. It enables us to conduct a difference-in-difference analysis. For clarity it is divided into . The former distinguishes rural and urban China and the latter large cities and other urban areas plus the national picture. We see in that rural house ownership is extremely high, being generally over 90% in both years. This reflects the institutional arrangements by which each household is allocated land for a house and reflects also the strength of village family ties that assists young household heads to build a house. The drop in house ownership among the elderly suggests that they have passed ownership to the next generation. As transfers of house ownership are permitted only among villagers, house prices are little influenced by market competition. It is understandable, therefore, that there is little sign of a receding housing ladder in rural China.
Figure 1A. The proportion of household heads owning a house by five-year age group, by area (rural, urban), 2002 and 2013.
Figure 1B. The proportion of household heads owning a house by five-year age- groups, by area (national, large cities, other urban cities), 2002 and 2014.
The urban curves provide some evidence for the existence of a receding housing ladder. We might expect the later curve to be above the earlier one because incomes rose and housing loans became more freely available. However, this is true only of the older household heads. The two curves are very similar until age 40, beyond which the percentage owning a house in 2013 is well above that in 2002. Whereas house ownership of older household heads rose over the eleven years, younger household heads were unable to make progress.
provides clearer evidence that the housing ladder has receded. However, there is a sharp difference between large cities and other urban areas. In the latter, the 2002 and 2013 curves follow each other closely with ageing. By contrast, in large cities the percentage of owners in 2013 is well below that in 2002 up to age 45 and well above it beyond that age. For instance, the gap in favour of 2002 exceeds 15% points in the age-group 30–34 and falls short by nearly 15% points in the age-group 45–49. All householders could in principle benefit from their rising incomes and improved access to housing loans over these years. However, many old householders already had a house and had the benefit of capital gain on their existing house, whereas many young householders had neither, and needed to raise funds for their first house purchase. The large cities dominated the picture for China as a whole. Up to age 45 the national percentage of owners was lower in 2013 than in 2002 – by over 15% points in the age-group 20–34 - but beyond that age the two curves were closely aligned.
It is possible to measure the average age, and thus the average years waiting to own a house, of household heads in the age-group 20–34 who do not own a house. The same exercise conducted in 2002 and in 2013 provides a measure of the change in average age over time (). By subtracting 20 years we obtain the average number of years a twenty-year old had to wait to own a house. The final row of the table shows that, at the national level, the average wait increased by 3.6 years between 2002 and 2013. It increased least, by 0.7 years, in rural China but by 4.2 years in urban China. The greatest increase was in large cities, by a remarkable 6.0 years. The housing ladder did indeed recede rapidly over that eleven-year period.
Was the lengthening wait due to the growing difficulty that young people faced to get onto the housing ladder? Our hypothesis is that, at the sampled locality level, the increase in average waiting time between 2002 and 2013 was determined by the rate of house price inflation over that period.Footnote4 That gives the appropriate specification. However, because the data set is not a panel, the change in average waiting time is not available at sampled locality level.Instead, we use the average waiting time in 2013 as the dependent variable. reports a regression analysis in which the unit of analysis is the sampled locality, i.e. each village or city. A positive association would imply that the greater the increase in difficulty to acquire a house, the longer the average wait. The table shows the coefficients in every geographical area to be positive and significantly so.Footnote5 An increase in the annual rate of house price inflation is associated with an increase in waiting time of 0.5 years in large cities.
Table 6. Regression analysis of the association, at the sampled locality level, between the average wait of household heads to own a house in 2013 and the average house price inflation rate over the period 2002–2013.
We cannot provide a formal test of causality because the data set lacks valid instruments, but the positive coefficients suggest that the relationship is causal. The literature evidence in sections 4 and 5 also implies causality: the various behavioural changes are generally interpreted as responses by young people to the growing difficulty of acquiring a house.
Like conducts a probit analysis of the probability of house-ownership, but this time uses all households and employs age-groups as the test variables. There are five age groups, with the youngest, ages 20–34, being the omitted category in the dummy variable analysis. Consider the control variables. We include all the potential explanatory variables that are available in the data setFootnote6. The coefficients and marginal effects for years of education are positive and significant in all cases. The coefficients and marginals for the income variable are positive and significant in large cities and other urban areas, but at national level are negative, reflecting the institutional divide between rural and urban China: rural areas are poorer but have a higher incidence of house-ownership. Except in rural China, Communist Party membership raises house-ownership. However, except in large cities, the marginal effects are lower in 2013 than in 2002, suggesting that party membership has become less important.
Table 7. Probit analysis: coefficients and marginal effects of age groups explaining house ownership by household heads, 2002 and 2013.
Our hypothesis is that young household-heads became relatively disadvantaged over the period. The marginals show the increase in probability of owning a house of the higher age-groups over the (omitted) group 20–34. The test is whether the disadvantage of the youngest age-group increased between 2002 and 2013. We therefore expect the marginals for age-group 35–44, and for subsequent age-groups, to increase between the years. This is indeed true of all marginals, and for all coefficients as well. For instance, in the age-group 35–44 the relative probability increases by 0.03 (and the coefficient by 0.22) in large cities and by 0.18 in China as a whole, and in the age group 45–54 the corresponding rise in probability is 0.27 and 0.31 respectively.The hypothesis test should be not only that the disadvantage of the younger age group is greater in 2013 but that it is significantly greater. A significance test is provided by pooling the 2002 and 2013 data in each area and introducing 2013 × age-group interaction terms: the coefficient on the 2013 × age-group 35–44 should be positive and significantly so. We see in the final row of that the coefficient is indeed positive, and significant except in other urban areas. We take the results as evidence that young household-heads have experienced much more difficulty in accessing the housing ladder.
7. The age distribution of housing wealth, 2002 and 2013
and 2B show household real housing wealth by five-year age group. We choose household wealth rather than household wealth per capita because sharp changes in household size complicate the picture. For instance, in large cities the average household size for heads aged 25–29 in 2002 was 3.4 in 2002 but fell to 2.0 in 2013. Indeed, this fall might reflect the decisions of young household heads to postpone having children as the housing ladder receded.
Figure 2A. Household real housing wealth, urban and rural, 2002 and 2013.
Figure 2B. Household real housing wealth, large cities, other urban, and national, 2002 and 2013.
shows that urban real housing wealth did not rise between the age-groups 20–4 and 35–9 in 2002 but rose sharply in 2013. Young urban household heads in 2013 were at a huge housing wealth disadvantage relative to older heads. It is plausible that their accumulation of housing wealth was postponed by the increasing difficulty that they faced. We see in that the contrast between 2013 and 2002 is particularly marked in large cities. Young household heads were at little disadvantage in 2002 but were at a huge disadvantage relative to older heads in 2013. The same pattern can be observed at the national level and, more weakly, in other urban areas. The figures provide descriptive evidence of a receding housing ladder.
reports the determinants of household real housing wealth in different residential areas of China. Tobit analysis is employed, reflecting the fact that some households have zero housing wealth. The table reports both coefficients and marginal effects. The coefficients reflect the effect of each explanatory variable on the housing wealth of house-owning households. The marginal effects, taking account of the fact that a marginal increase in a variable might have no effect on the wealth of households which do not own a house, is relevant for the sample of all households.Footnote7
Table 8. Tobit analysis: coefficients and marginal effects explaining household real housing wealth, 2002 and 2013.
The control variables have marginal effects that differ somewhat from those in . Years of education, household income per capita and (except for rural China in 2013) Communist Party membership raise household housing wealth, and increasingly so. Our test concerns the different age groups. Again, 20–34 is the omitted category in the dummy variable analysis. The coefficients on the older age-groups are expected to be positive, and indeed all but one of the 32 are; young household heads have less wealth. Our hypothesis is that the young became increasingly disadvantaged over time as they faced greater difficulties in acquiring housing wealth. Thus, we expect the coefficients on the older age-groups to be larger in 2013 than in 2002. Indeed, this is true of all 16 cases.
The strongest evidence for our hypothesis is to be found in large cities. Comparing those aged 35–44 with those aged 20–34, the coefficient rises from 9.53*** in 2002 to 49.36*** in 2013, a five-fold gain of 39.83. The marginal effects are 4.54*** and 29.27***, a gain of 24.73. The same comparison in other urban areas yields a gain in coefficient of 7.91 and in marginal of 6.07. At the national level, the gains are 18.83 and 12.05 respectively. However, there is no support for the hypothesis in rural China.Footnote8 The final row of follows in testing whether the increase in the coefficient is not only positive but also significantly so. The increase is significant in three cases but not in the rural case,
The relative housing wealth of households aged 20–34 fell dramatically between 2002 and 2013. Our explanation is that the fall was due to the greater difficulty they faced to get onto the housing ladder. Can there be a different interpretation? If the same fall is found in non-housing wealth, the explanation is less likely to be found in the housing market. When was equivalently estimated but with real non-housing wealth as the dependent variable (not reported), there was a relative fall in the non-housing wealth of the young but the fall was only minor. As an example of the pattern, in the case of large cities, the non-housing coefficient for the age-group 20–34 relative to the age group 35–44 fell by 11.4 between 2002 and 2013 whereas their housing coefficient fell by 39.83. Most of the relative fall in the wealth of young households was specific to housing. Moreover, the fall was limited to house owners (who were likely to have used their savings for house purchase); there was no significant fall in the case of those who did not own a house.
Again, there is further evidence to suggest that house price inflation was responsible for the greater relative wealth disadvantage of the youngest age group in 2013 than in 2002. The literature review contained in sections 4 and 5 argues that house price inflation has elicited several behavioural responses in order to cope with the receding housing ladder. For instance, Wrenn, Junjian, and Bo (Citation2019) find that higher house prices caused marriage age to rise, and Ling (Citation2018) finds that, during the period of rapid house price inflation, the income and wealth of parents became a more important determinant of whether their adult child owns a house.
In all but two (Chongqing and Changsha) of the large cities in our sample, by 2013 there were recently introduced regulations that forbid or penalise ownership of multiple houses. In such circumstances parents have an incentive to transfer ownership of a second house to a child. The wealth disadvantage of young adults might be reduced in that way. In we test this hypothesis by adding to the large cities equation of a dummy variable denoting that a city had such regulations in 2013, and interaction terms combining regulated city with the different age-groups. In 2013 the coefficient and the marginal on regulated city are significantly positive, and larger than in 2002. House prices are higher and have risen more in the regulated cities. The coefficients and marginals on the interaction terms in 2002 are small and generally not significantly different from that on age-group 20–34. In 2013 they are negative and significantly so for all age-groups within the range 35–64. Young household heads aged 20–34 are at a lesser wealth disadvantage relative to older household heads in the regulated cities. The test is not conclusive because failure to regulate might have been endogenous. Nevertheless, the evidence is consistent with our hypothesis.
Table 9. Tobit analysis: coefficients and marginal effects explaining household real housing wealth in large cities, distinguishing regulated cities from others, 2002 and 2013.
In summary, we have adduced evidence that the age group 20–24 in all but rural China experienced a relative fall in housing wealth over the period. However, the most powerful evidence of a receding housing ladder was to be found in the large cities, where house prices rose rapidly and the young suffered a substantial disadvantage in housing wealth by comparison with all older age-groups.
8. Conclusion
We have examined the consequences of China’s rapid house price inflation in recent years, analysing evidence from the CHIP national household surveys relating to 2002 and 2013. There is a contrast between those households which were already house-owners – and which therefore benefited from capital gain – and those which did not own a house. The latter faced increasing difficulty in getting onto the housing ladder as the ratio of house prices to their incomes rose. This difficulty was to some extent eased by the expanding market for housing loans, previously very weak, and by the strength of family ties in China, providing inter-generational sources of funding for house purchase.
Our estimates indicated that the time young household heads spent waiting to acquire a house increased substantially in the eleven years between 2002 and 2013, being by 6.0 years in large cities and 3.6 years nationally. Evidence was adduced from probit, tobit and difference-in-difference analyses that younger households became relatively disadvantaged, both in house-ownership and in the value of their housing wealth. For instance, in large cities the probability of house ownership in the age-groups 35–64 over that of the age-group 20–34 almost doubles between 2002 and 2013. Similarly, in large cities the average premium on the real wealth of the age group 35 = 44 over the wealth of the younger age-group increases nearly four-fold over the eleven years. We explained why the phenomenon is little found in rural China and why it is particularly important in large cities.
Young households faced greater difficulty in the housing market in 2013 than in 2002. The rate of house price inflation, although still high, has slowed since 2013. Only the growing possibility of a disastrous housing crash or broader government restrictions on multiple house ownership might bring our housing ladder story to an end.
The limitations of the data set require that our conclusions be qualified. In the tables reporting equations, the coefficients or marginals indicate associations and might not measure the causal effects of the explanatory variables. For instance, the lack of valid instruments for house price inflation prevents us from establishing that it has a causal effect on the length of time that young householders wait to acquire a house. Similarly, the lack of measures of other plausible explanatory variables means that alternative or additional explanations cannot be ruled out. Nevertheless, the originality and plausibility of our explanation for the receding housing ladder might make it a baseline for further research.
Disclosure statement
No potential conflict of interest was reported by the authors.
Additional information
Notes on contributors
John Knight
John Knight is emeritus professor in the University of Oxford and emeritus fellow of St Edmund Hall, Oxford. He is also academic director of the Oxford Chinese Economy Programme. He has conducted research on the Chinese economy for thirty years. Recent books include Towards a Labour Market in China, 2005, and China’s Remarkable Economic Growth, 2012 (both OUP).
Haiyuan Wan
Haiyuan Wan, at the Beijing Normal University Business School, is a researcher in the China Institute for Income Distibution, and a member of the CHIP team.
Notes
1. More detailed accounts of the data are available in the text and data appendices of Knight, Shi, and Haiyuan (Citation2020, 2022).
2. National Population Census, 2000, 2015.
3. China General Social Survey.
4. The other possible proxy for the increasing difficulty of young people to acquire a house is the increase in the house value/income ratio over our period. However, because the data set is not a panel, the change in the ratio is not available at sampled locality level.
5. Various other specifications were attempted but this one produced the strongest positive association. House price in 2013 and the ratio of house price to income in 2013 also produced positive coefficients but each was dominated when included with house price inflation.
6. We lack some of the potential explanatory variables that are discussed in sections 4 and 5, such as extent of family support and access to loans. Marriage and number of children are excluded because they might be consequences rather than causes of house ownership.
7. We choose household real housing wealth as the dependent variable because the logarithm of household real housing wealth would eliminate households with zero housing wealth, and because household housing wealth per capita results would be confounded by sharp changes of household size in some age-groups.
8. As a further robustness test, we confined the sample to house-owners and compared the OLS results with housing wealth or ln. housing wealth as the dependent variable. In both estimations and in each sample except the rural, the coefficient on age-group 35–44 was positive, significant and greater in 2013 than in 2002. The choice of ln. housing wealth would make no difference to the story.
References
- Bingqin, L., and H. Ban Shin. 2013. “Intergenerational Housing Support Between Retired Old Parents and Their Children in Urban China.” Urban Studies 50 (16, May): 3225–3242. doi:10.1177/0042098013483602.
- Chen, J., Q. Hao, and M. Stephens. 2010. “Assessing Housing Affordability in Post-Reform China: A Case Study of Shanghai.” Housing Studies 25 (6): 877–901. doi:10.1080/02673037.2010.511153.
- Chen, J., M. Hu, and Z. Lin. 2019. “Does Housing Unaffordability Crowd Out Elites in China’s Superstar Cities?” Journal of Housing Economics 45: 1–14. September. doi:10.1016/j.jhe.2018.03.003.
- Chen, B., X. Yang, and N. Zhong. 2020. “Housing Demand and Household Saving Rates in China: Evidence from a Housing reform‘.” Journal of Housing Economics 46: 1–17. September. doi:10.1016/j.jhe.2020.101693.
- China Household Finance Survey 2018. ‘2017 China Urban Housing Vacancy Report’, December, online.
- Knight, J., L. Shi, and W. Haiyuan. 2020. “‘The Increasing Inequality of Wealth in China’, Ch. 4.” In Changing Trends in China’s Inequality. Evidence, Analysis and Prospects, edited by T. Sicular, S. Li, X. Yue, and H. Sato, 109–43. New York: Oxford University Press.
- Knight, J., L. Shi, and W. Haiyuan. 2022. “Why Has China’s Inequality of Wealth Increased Rapidly in the Twenty-First Century?” Review of Income and Wealth 68 (1, March): 109–138. doi:10.1111/roiw.12507.
- Ling, Z. 2018. “Intergenerational Housing Asset Transfer and the Reproduction of Housing Inequality in China.” Chinese Journal of Sociology 4 (4): 453–480. doi:10.1177/2057150X18792835.
- Lixing, L., and W. Xiaoyu. 2019. “Housing Prices and Intergenerational Co-Residence in Urban China.” Journal of Housing Economics 45: 1–9. September. doi:10.1016/j.jhe.2018.08.001.
- Nie, G. 2020. “Marriage Squeeze, Marriage Age and the Household Savings Rate in China.” Journal of Development Economics 147: 1–26. November. doi:10.1016/j.jdeveco.2020.102558.
- Sato, H., T. Sicular, and Y. Ximing. 2013. “‘Housing Ownership, Incomes, and Inequality in China, 2002-2007’, Ch. 3.” In Rising Inequality in China. Challenges to a Harmonious Society, edited by L. Shi, H. Sato, and T. Sicular, 85–142. New York: Cambridge University Press.
- Wei, H., and Z. Wang. 2022. “The Impact of Inheritance on the Distribution of Wealth: Evidence from China.” Review of Income and Wealth 68 (1, March): 1. doi:10.1111/roiw.12513.
- Wei, S.J., X. Zhang, and Y. Liu. 2017. “Home Ownership as Status Competition: Some Theory and Evidence.” Journal of Development Economics 127: 169–186. doi:10.1016/j.jdeveco.2016.12.001.
- Wrenn, D. H., Y. Junjian, and Z. Bo. 2019. “House Prices and Marriage Entry in China.” Regional Science and Urban Economics 45 (1): 47–71. January.
- Zequn, T., and W. Ning. 2022. “Will Downward Intergenerational Housing Support Increase parents’ Expectations for Old-Age Care from Adult Children?” Journal of Applied Gerontology 10: 1–12.
- Zhou, Z. 2015. “The Development of Mortgage Finance in China: The Implications of International Experience.” China Perspectives 2015 (4): 51–61. doi:10.4000/chinaperspectives.6862.