Cohort Trajectories in Hong Kong's Housing System: 1981–2001

Abstract

A household's progression through a housing system depends not just on the current social, economic and policy context but also on past housing experiences and circumstances. Different birth cohorts, because of their location in a specific socio-economic and policy environment, may have distinct residential trajectories. Five waves of Hong Kong Census data are used to trace tenure change of different birth cohort groups from 1981 to 2001. The paper shows that it was the new and expanding middle class entering the labour market in the early 1980s, which were the most likely to enter, and benefit from a booming private housing market. However, the public homeownership sector, which encompasses schemes whereby the supply and price are controlled by the government, offered more uniform assistance to other birth cohorts. Moreover, this state-assisted sector played a particularly important role in sustaining the growth of homeownership when housing was least unaffordable and also during the recession following the Asian Financial Crisis.

Keywords:

• Cohort analysis
• Hong Kong
• homeownership
• census analysis

Introduction

Changes in the nature and availability of employment, demographic structures and behaviour, income distributions and policy regimes all combine to affect people's life chances. In any society different cohorts enter the housing and labour markets under highly varied circumstances. Their progression over the life course is shaped by the particular set of conditions they encounter at that point of entry and by the subsequent economic, social and policy environment. This is not to suggest a uniformity of experience among different groups over time but a distinctiveness of experience, producing notable contrasts between cohorts and specific intra cohort variations. In relation to policy, major changes in, for example, health, educational and housing provision, can have significant distributional impacts. Housing can be a particularly critical element given the influence of well located, good quality accommodation on health outcomes, educational and employment opportunities and public service provision generally. Moreover, housing represents the major cost in most household budgets and for most owners it is likely to be their main asset.

Across a wide range of industrialized societies, the last half-century saw enormous transformations in housing systems. A significant feature was the development of social housing in various forms. For many households, the availability of low-cost housing of reasonable quality had a major impact on their quality of life and on their subsequent route through the housing market. Hong Kong was no exception and was notable for the high proportion of its population in the public housing sector—a sector that continued to grow until the early 1990s. It is with particular reference to the important role of public housing in Hong Kong that this paper examines the housing experiences of different birth cohorts over the last two decades. The study explores, inter alia, patterns of homeownership attainment among different groups, the housing positions of different income groups within each cohort and the changing tenurial distribution.

Pseudo-longitudinal data are created by combining five waves of the Hong Kong Population Census and By-census conducted in 1981, 1986, 1991, 1996 and 2001. These are the only five waves of census data available in electronic format. Analysis of the micro data was conducted via the Self-help Tabulation Service offered by the Census and Statistics Department. Birth cohorts were formed by grouping heads of household born within a five-year period. Essentially, the study starts with cohorts born from 1971 to 1975 (aged 25 to 30 in 2001).

The paper begins with a discussion of the general context within which different cohorts have embarked on their trajectories through Hong Kong's housing system. This discussion is followed by a review of the literature on the use of cohort analysis in tracking, analyzing such changes over a longer perspective, and by an overview of economic change and housing policy development in Hong Kong during the last few decades. Results of the empirical analysis are presented in the fourth section in which tenure change of the various birth cohorts is traced from 1981 to 2001. The paper concludes with a discussion of the impact of wider social and policy changes on these tenure patterns.

Some clarification of terminology is also appropriate at this point. There are various references to ‘public homeownership’ or ‘assisted homeownership’ in the paper and this may be unfamiliar terminology, although it is widely used in Hong Kong. Public or assisted homeownership encompasses various government schemes in which price, supply, eligibility and methods of payment may all be controlled by the government. This includes the Home Ownership Scheme (HOS), which is essentially ‘build for sale’, and the Tenant Purchase Scheme (TPS).

Cohort Trajectories and Housing

Generalizations about shifts in the structure and shape of housing systems at an international level are fraught with difficulties. Moreover, even apparently similar developments in relation to housing provision can produce highly varied consequences with differences in economic cycles, demographies, labour markets and income levels. Nevertheless, there has been a degree of commonality in the direction of change and in the changing portfolios of housing opportunities that have confronted different cohorts at different times. Harloe (1995), for example, has provided the most comprehensive and detailed account of the shifting fortunes and roles of public housing in Europe and North America. While he points to important differences in national patterns and experiences, he refers to a ‘golden age’ of direct provision spanning the period 1945 to 1970. After this the role of public housing declines to a more residual role with a renewed market orientation in provision. One interpretation of these developments is of public or social housing in its varied forms performing a kind of holding operation between a declining private rental sector and an ascendant mortgage based individual homeownership.

Against this background, certain privileged cohorts have gained particular benefits from direct state provision and have typically been the same groups which have received a further boost in their housing trajectories via policy shifts to privatization and marketization. This has been as true of housing experiences and policy impacts in transitional economies such as Russia and China as in more thoroughly commodified Western housing systems (see, for example, Daniella & Struyk, 1994; Davis, 2003). Conversely, the cohorts which have followed have often faced fewer and lower quality housing opportunities. The rise and fall of direct public housing provision has been closely paralleled, and closely connected with, the growth of individual homeownership and the benefits and opportunities on offer have varied significantly over space and time. Again, there are references to a ‘golden age’ in the literature (e.g. Dymski & Isenberg, 1998), referring to periods of rising real house prices, growing job security, expanding state sectors and subsidies, high general inflation and a homeownership sector associated with relatively affluent purchasers. This can be contrasted with a subsequent ‘new age’ with features such as low inflation, greater price volatility, falling nominal house prices, weakened social protection, reduced state subsidies for homeowners and a greater mix of households or dwellings. The way in which these and other factors have combined to influence housing pathways requires detailed examination in different national contexts. The simple point is that some cohorts have been more fortunate than others. This has perhaps been particularly true in societies in which economic growth and social change have been rapid and compressed into a relatively short period as in Hong Kong and many other parts of East Asia. Cohort analysis offers a powerful analytic tool in capturing the complexity between the varied experiences of an aggregate of individuals in geographical and temporal dimensions.

The earliest use of birth cohorts as an analytic tool can be traced back to the mid-19th century (Elder, 2003). However, modern cohort analysis has moved away from the simplistic view of taking birth cohorts as sharing homogeneous historical and ecological experience (Elder, 2003). Ryder (1965), for example, advocated the use of successive birth cohorts to study social change and recent studies have emphasized the link “between social change and life course patterns, between historical time and lifetime” (Elder, 1975, p. 179).

A cohort can be defined as “the aggregate of individuals … who experienced the same event within the same time interval” (Ryder, 1965, p. 845) and cohort analysis typically involves the “comparison of the characteristics of one cohort at two given points of time” (Glenn, 1977, p. 8). The validity of cohort analysis hinges on the assumption that “particular periods in history have distinct effects on members of cohorts undergoing formative experiences that will shape them for life” (Alwin, 1997, p. 166).

Housing is a durable good that changes slowly over time but also involves a forward accumulating effect. Therefore, for housing researchers, birth cohort analysis offers a particularly valuable analytic tool when a longer-term perspective (e.g. housing history) is required. Repeated cross-sectional surveys are the usual sources of data for cohort analysis. Clark & Dieleman (1996), using the 1980 and 1990 US Census, examined the movement into homeownership of different birth cohorts in the US. They found a rapid movement into homeownership for young adults but a slowdown of entry into the sector after the age of 45. They also found that trajectories into homeownership were different for different income groups. Forrest & Leather (1998), combining data from the UK General Household Surveys from 1973–93, focused their analysis on birth cohorts after the age of 40. In contrast to most cohort analyses, which have tended to look at the historic past, their emphasis was on the projection of demand and equity issues in relation to old age homeowners against the background of a restructured welfare state.

However, longitudinal or retrospective surveys are equally common in cohort analysis. Lelievre & Bonvalet (1994) employed data from retrospective housing history surveys to examine residential mobility and other related changes in the birth cohort born between 1926 and 1935 in France. They conceptualized four types of trajectories, and discovered a close link between geographic and social mobility in which the diffusion of homeownership was one of the major characteristics of housing transformation in post-1950 France. Haurin et al. (1996) used a longitudinal survey of youth in the US to analyze how the movement of young people into homeownership was influenced by wealth, family formation and local house price movements. Feijten & Mulder (2002) looked at the timing of moves into long-stay housing (single-family or owner-occupied homes) from pooled data from two retrospective housing and family history surveys in the Netherlands. They found that despite the continual time delay in the departure from parental homes and in household formation, entry into long-stay homes had not been postponed. However, single persons and couples had very different trajectories. While the former group remained outside the long-stay home sector regardless of which cohorts they belonged to, the latter group was increasingly more likely to enter into single-family homes in the rental sector after forming a union. The also observed that a certain level of commitment still appeared to be decisive for the move into owner-occupied housing, albeit the apparent connection between such commitment and childbirth among older cohorts had become less significant for their younger counterparts.

A more sophisticated technique is employed by Myers et al. (1998) who used ‘double cohort analysis’ to examine homeownership attainment among immigrants in the US. This technique incorporates both birth cohorts and the immigration cohorts to capture the life course experience of immigrants. They found that in nearly all birth cohort groups, immigrants who arrived in the 1960s achieved a higher homeownership rate than their counterparts who arrived 10 years later. They also found a complicated relationship between age and assimilation effect on immigrants. Homeownership rates grew over the life cycle, but they also increased with length of residence. Thus, the apparent increase in homeownership among successive cohorts of immigrants was masked by the effect of length of residence. Taking the latter into account, the rate of homeownership of successive cohorts of immigrants in fact decreased, however, factors such as level of education and financial resources also played a role. Controlling for such factors, the result was again an upward trend for new immigrants. It was the lower level of such educational and material resources among recent immigrants that caused the apparent reduction in homeownership attainment.

A more ambitious form of cohort analysis was attempted by Chevan (1989). Using six waves of the US Census data from 1900 to 1980, the study explored the impact of process (change in external social factors) and compositional (changes in the composition of the population) changes on homeownership attainment over a period of nearly a century. Unsurprisingly, it found that homeownership attainment was due to the combined effect of both process and compositional factors. But Chevan's (1989) major contribution was to account for the relative impact of such factors over time. He found that before the 1940s, homeownership attainment levels were low and the main driving force was changing demographic composition, whereas the rapid growth of homeownership from 1940 to 1960 was due to the combined effects of both compositional and process changes. However, the dampening of the compositional factor in the period 1960 to 1980 also caused the growth of homeownership to slow down. This was also the period when the homeownership rate amongst the younger cohorts was the highest.

Methodologically, Myers (1999) offers the most comprehensive discussion of cohort longitudinal analysis amongst a variety of other methods (cross-sectional survey and retrospective/panel survey) in looking at housing histories. Whilst acknowledging various methodological difficulties, he concludes that cohort analysis built from successive cross- sectional surveys offers the best value for money option for analysis of housing issues over time. Such an approach “captures the longitudinal qualities of panel analysis and avoids the mis-inferences from cross-sectional data, while at the same time, enabling the use of the most commonly available data and facilitating aggregate-level analysis” (p. 474).

However, findings from cohort analysis have to be interpreted with caution. First, cohort analysis only measures the net difference of attributes within a cohort between two periods of time. The actual change within the cohort cannot be detected (for instance, a ‘no change’ scenario may be the result of equal change in both directions). Second, the grouping and the boundary of a cohort are often arbitrary and this creates problems of identification. Third, cohort effects may only be an exhibition of a period effect or individual difference Firebaugh (1989) for more detailed discussion of the shortcomings of cohort analysis).

Economic Change and Housing Policy Development in Hong Kong

Different cohorts of households have progressed through the shifting contours of Hong Kong's housing system against a background of major social, economic and political transformations over the last 50 years. In that period, Hong Kong has changed from a poor community of refugees from Mainland China, working in sweatshops and informal trades, to a wealthy society. Real per capita income increased more than five times from 1966 (HK$12 030) to 2000 (HK$61 200, both at constant 1980 prices). Hong Kong has shown a talent for reinventing itself, from entrepôt to manufacturing to a higher order services economy; and these economic changes have created new employment patterns and opportunities. A local economy overwhelmingly dominated by manufacturing employment in 1970 has been transformed into one in which almost 90 per cent of jobs are in the service sector. Moreover, Hong Kong has generally experienced low unemployment and at times suffered from a labour shortage. However, this trajectory masks periods of significant economic and social turmoil. The most recent and probably the most severe has been the Asian Financial Crisis of 1997 and its aftermath. House prices in late 2003 had fallen by over 60 per cent from their peak in mid-1997. An uneven and initially uncertain recovery in property values has occurred since then. These undulating fortunes have shaped the city's economic and social landscape.

Overall, there has been substantial economic growth and wealth and opportunity have grown at all levels of the social hierarchy. However, the benefits have not been distributed evenly. Older workers shed by the manufacturing sector have experienced a significant decline in a service sector dominated labour market. Younger workers, taking advantage of expanding educational opportunities, have generally been well placed in this new service economy with a widening recruitment into white collar, middle-class positions (Lui, 1997; Lui & Wong, 1994; Tsang, 1994; Wong & Lui, 1992). While wealth undoubtedly trickled down during the decades of rapid economic growth, levels of inequality were high. Between 1976 and 1991 the official Gini coefficient of household income distribution rose from 0.430 to 0.518 and Hong Kong continues to have one of the highest rates of income inequality among high-income economies (Lui, 1997).

These expanding economic opportunities were complemented by the large-scale provision of social services, including housing, public health and education, that allowed low incomes to go further and moderated the worst effects of these stark income inequalities. Since the mid-1950s the government has been committed to the large-scale provision of public rental housing. In 1976 the Housing Authority introduced a relatively small-scale build-for-sale scheme, the Home Ownership Scheme (HOS). This scheme swiftly expanded to become a significant component of publicly developed housing, both in terms of its contribution to the overall housing stock as well as to the finance of the Housing Authority (the Housing Authority in the mid-1990s generated a surplus of over US$2.5b a year from the sale of HOS). By January 2002 the Housing Authority had provided some 1.09 million public flats or nearly half (46 per cent) of total housing stock in Hong Kong (693 000 public rental flats and 392 000 subsidized homeownership flats) and over 49 000 housing loans or mortgage subsidies for home purchase (HKHA, 2005).

However, the Hong Kong Government has been ambivalent towards the role of public housing over the last decade. Despite the endorsement of a private sector led strategy in the Long Term Housing Strategy in 1987, the public sector housing continued to expand. The first HKSAR Administration even embraced public housing as an instrument in the reinventing of housing policy when an annual housing supply target of 85 000 units was proclaimed in 1997, following the handover from the British Colonial Administration. The plan was for the Housing Authority to build 50 000 new units a year, of which half were to be assisted homeownership flats. The Housing Authority also revitalized a previously aborted plan to sell public housing to sitting tenants (the Tenant Purchase Scheme—TPS) and about 75 000 TPS flats have been sold (HKHA, 2002).

However, the favourable climate for public housing in the late 1990s was short-lived. First, there was pressure from the real estate industry to curb housing supply amidst the market slump in the aftermath of the Asian Financial Crisis and in 2002 the HKHA was forced to permanently scrap the Home Ownership Scheme (Fung & Forrest, 2002). Second, the Review of the Institutional Framework for Public Housing in Hong Kong (RIF Report) proposed a more flexible mixed mode of public rental housing in which in-kind provision would give way to in-cash provision (HKSAR, 2002, p. 26). Third, a disinvestment strategy was adopted in which the commercial assets of the HKHA would be sold off to compensate for the loss of development capital associated with the curtailment of the Home Ownership Scheme (HKHA, 2003). Therefore, the prospect is of a very different pattern of housing provision in which the public sector plays a much reduced role in the direct provision of homeownership opportunities and in the overall shaping of life chances.

Cohort Trajectories in Hong Kong's Housing System

Tenure Change

Hong Kong has witnessed a dramatic increase in homeownership over the last two decades. In 1981 less than a quarter (23 per cent) of households were homeowners but by 2001 this figure had more than doubled to 50 per cent (Figure 1). Although the pace of increase appears fairly steady throughout the two decades, the momentum was fuelled significantly by the mass construction of public ownership flats, especially in the 1990s. The private homeownership rate almost halted in the early 1990s. It dropped half of a percentage point from 1991 to 1996 (during the period when house prices in Hong Kong were rising rapidly) and only increased by two percentage points from 1996 to 2001. On the other hand, HOS flats were developed at an average annual rate of over 12 500 units over the 25 years history of the scheme and in the 1990s accounted for 55 per cent of the units constructed. The sale of over 75 000 units under TPS after 1988 further boosted the growth of homeownership.

Figure 1 Tenure distribution 1981–2001. Source: Authors' analysis of Census 1981, 1986, 1991, 1996 and 2001.

The rental sectors inevitably lost out to the expansion of homeownership. Whilst in the 1980s, the share of the public rental sector increased from 31 per cent to 36 per cent, it was the private rental sector (and to a lesser extent temporary housing) which declined sharply. The proportion of households in private tenancies halved from 30 per cent to 19 per cent between 1981 and 1991. The private rental sector subsequently stabilized and from the early 1990s it was the public rental sector that contracted. The decline was particularly marked towards the latter part of the 1990s following the introduction of the Tenant Purchase Scheme (TPS). The share of public tenancies decreased from 36 per cent in 1996 to 31 per cent in 2001.

Age Group Breakdown of Tenure

Against this overall background of tenurial change it is not difficult to understand the distribution of homeownership amongst age groups in 1981. With some exceptions (for example, in the 30–34 group), homeownership rates generally increased with age (Table 1). The high house-price income ratios in the 1970s, which necessitated longer periods of saving among prospective homeowners, were a major factor here. However, from the early 1980s to mid-1990s, the share of householders in their late twenties and early thirties in the owner-occupied sector grew substantially. This was in sharp contrast with the pattern in some other mature economies with a longer history of homeownership. In such societies the peak for homeownership is later in the life cycle. For example, Clark & Dieleman (1996) found that the homeownership rate peaked in the US in the 1980s for those householders aged 50 or more then declined when they passed the retirement age. In the UK, the peak rates for homeownership are projected to rise substantially with age as the baby boom generation moves into retirement (Forrest & Leather, 1998).

Table 1. Owner occupation and public renting by age 1981–2001

	Owner occupation (%)					Public renting (%)
Age group	1981	1986	1991	1996	2001	1981	1986	1991	1996	2001
15–19	13	17	21	19	12	35	33	24	20	11
20–24	13	24	32	29	25	22	20	17	11	9
25–29	20	35	51	47	50	18	18	18	13	9
30–34	25	37	49	54	57	21	25	24	17	11
35–39	22	35	46	51	59	28	31	31	27	17
40–44	23	30	43	47	55	33	37	34	33	26
45–49	23	29	38	45	53	37	39	40	35	31
50–54	24	29	38	40	51	39	43	41	42	34
55–59	25	29	36	40	46	39	44	44	44	40
60–64	26	30	34	38	44	37	45	47	48	43
65–69	26	29	34	34	42	35	43	48	53	48
70–74	26	28	33	33	39	33	42	48	54	51
75 or above	26	29	33	33	36	30	37	43	54	52
Source: Authors' analysis of Census 1981, 1986, 1991, 1996 and 2001.

It is thus not surprising to find that because younger householders have increasingly opted for homeownership over the last two decades, their presence in the public rental sector has been greatly reduced. The age distribution of public tenants over this period shows that, with the cut-off age of 55 to 59, the proportion of younger public tenants has gradually decreased while that of the older tenants increased. It is also evident that the younger the tenants, the greater the reduction of their share in the public rental sector with the converse being the case for older householders. The period 1996 to 2001 saw a general reduction in the share of public renting affecting all age groups nearly equally with the exception of the very old.

Cohort Difference in Homeownership Attainment

However, comparisons of age group experience that are based solely on cross-sectional surveys are prone to cross-sectional or age-cohort fallacy (Myers, 1999; Pitkin & Myers, 1994). Because of the quasi-cumulative nature of tenure, tenure achievement between age groups across different years only reflects what has been brought forward from earlier years as people seldom move out of owner occupation until the very late stage in the life cycle (Chevan, 1989). Hence, it is essential to differentiate the age and cohort effect in order to achieve a fuller understanding of tenure change over time.

This is achieved through the creation of birth cohorts that combine Census information from 1981 to 2001. Figure 2 shows the trend in homeownership attainment of different birth cohorts over the two decades and several trends are apparent. First, all birth cohorts show an increasing upward surge in homeownership attainment and there is no sign of a downturn. It indicates an age (and indirectly family life cycle) effect and meshes with the general trend towards an increased desire for a long-stay home when a household becomes more settled. The absence of a downturn also means that there is a continual net addition of owner-occupied stock to each cohort group. Unlike the situation in some other industrialized societies (for example, the US), more old people opt out of homeownership (probably to institutional homes) than their counterparts who move from renting to ownership.

Figure 2 Cohort trajectories of home ownership (public and private) 1981–2001. Source: Authors' analysis of Census 1981, 1986, 1991, 1996 and 2001.

Second, most cohorts occupy a higher level contour line than their immediate senior cohort and homeownership attainment increases at a higher rate over the decades. This is a gradual cohort effect. Not only do younger cohorts start to buy at an earlier age, more of them are able to buy when they move along the family life cycle. For instance, while householders born between 1946 and 1950 could only achieve a 25 per cent homeownership rate at the age of 30–34, the comparable figure for their counterparts born 20 years later (1966 to 1970) was 57 per cent. The youngest cohorts achieved a fourfold jump in the rate of homeownership in two decades whilst their older cohorts could only achieve a 50 per cent increase over the same period. Chevan (1989) finds a very similar pattern in the US for the period 1940 and 1980 when rapid social change also occurred in that society. Younger birth cohorts generally have higher education attainment, higher occupational status and higher incomes (Census 2001). It is also relevant that free education was introduced in the late 1970s within the context of escalating manufacturing exports. Higher education expanded in the late 1980s against the backdrop of a restructured economy with the rapid growth of service sector employment. These developments signal a strong cohort effect, which reflects the differential benefit of expanding education opportunity and social mobility on different groups in the population.

Third, this trend appears to falter with the youngest two cohorts whose trajectories overlap with that of the 1961–65 birth cohorts. This suggests that the upsurge of homeownership seen in successive earlier cohorts can no longer be sustained for these younger cohorts. This may reflect the stagnation in the expansion of higher education and social mobility closure in the 1990s (Wong & Lui, 1992). However, such observations have to be interpreted with caution since data for these younger cohorts are not complete (they have fewer observations compared with their older counterparts), and the impact of such factors would require further confirmative analysis.

Between Private and Assisted Homeownership

Analysis of the attainment of homeownership in Hong Kong requires particularly detailed attention. Although over 50 per cent of households were in homeownership in 2001, only two-thirds were in the market sector. The remaining one-third were in assisted homeownership (HOS and TPS) with supply and price controlled by the government. Hence, we would expect the cohort experiences in these two sub-sectors of homeownership to be rather different.

A breakdown of cohort trajectories between private and public homeowners does, indeed, show a sharp contrast. The homeownership trends in both the market and assisted sector for all cohorts in the first decade from 1981 to 1991 follow a pattern similar to that depicted in Figure 2. However, in the second decade the homeownership rate for nearly all birth cohorts in the market sector stabilizes at between 35 per cent and 40 per cent (Figure 3). This indicates the period effect of macro-economic factors that dominate over the cohort effect. The first half of the 1990s saw rapid house price inflation, which created serious affordability problems among home buyers and suppressed further advances in homeownership attainment in the market sector. However, after 1997, around one-third of the market value of housing evaporated with the Asian Financial Crisis and created the deepest depression in the housing market in recent decades. Advancement in homeownership was largely halted.

Figure 3 Cohort trajectories of homeownership (private) 1981–2001. Source: Authors' analysis of Census 1981, 1986, 1991, 1996 and 2001.

The exceptions to this pattern are the birth cohorts at the two extremes. The youngest two birth cohorts (born after 1966) show a spectacular increase in home buying in the late 1990s. Unlike their older counterparts, the depressed prices in the late 1990s did not impose a disincentive to home buying but opened up an opportunity. The smaller households in this birth cohort whose incomes were largely unaffected by market conditions (for example, public sector workers) suddenly found that small flats had become very affordable. The second group that stands out is the birth cohort born before 1910. This group shows a continual increase in recruitment into homeownership—in contrast with their counterparts in many Western societies where the reduction of homeownership among the very old is often as a result of movement into institutional care.

Two socially and culturally contingent factors may have been at work here. First, the majority of elderly people in Hong Kong are still living with their family and it is not particularly uncommon to regard the most senior male as the head of the household. Thus, such elderly heads of household would be counted as the homeowner regardless of whether the home is actually owned by him. It should be noted that homeownership as a joint family venture is not uncommon in other cultural contexts (Forrest & Murie, 1995). Second, it is also possible that asset and income-rich baby boomers jointly purchased properties for their parents as an expression of filial piety as well as a contribution to their retirement care.

In sharp contrast with the stagnation in homeownership attainment for most birth cohorts in the latter half of the 1990s, those in the assisted homeownership sector show an even steeper upward surge in homeownership during this period (Figure 4). This indicates the combined impact of a cohort and period effect. The cohort effect in the first decade (described in the last section) also applies to homeowners in the public sector. However, the changing economic factors which produced a strong period effect and restricted homeownership growth in the market sector in the second decade have been mediated by policy initiatives. This produced a markedly different impact on homeowners in the assisted homeownership sector. When the market sector was hampered by affordability problems in the first half of the 1990s, the Hong Kong Housing Authority increased its share of equity in the assisted homeownership scheme (which is a shared-equity scheme) and effectively reduced the downpayment and subsequent repayment burden for hard pressed home buyers. When the housing market bottomed out in the latter half of the 1990s, a revamped Tenant Purchase Scheme (sale of public rental housing to sitting tenants), which offered deep discounts (up to 85 per cent) to sitting tenants, sustained the demand for assisted homeownership.

Figure 4 Cohort trajectories of homeownership (public) 1981–2001. Source: Authors' analysis of Census 1981, 1986, 1991, 1996 and 2001.

The Rental Sectors

Figure 1 shows that the proportion of households in the public rental sector was stagnant at around 35 per cent from 1986 to 1996. It then shows a sharp downturn to 31 per cent in 2001. However, a disaggregated picture of the trajectories of different birth cohorts shows a distinct cohort effect. Three distinct public renting trajectories can be identified (Figure 5). The first group is the older birth cohort (those who were born before 1920 and at the age of 60 when the first records are available for analysis) whose trajectories within public rental housing have shown a continual increase. This matches shifting patterns of demand and public housing allocation policy over the last few decades. Older applicants received higher priority in the 1980s as a consequence of long queues when the average waiting time was around 10 years (even longer for urban area estates). In the early 1990s this group also benefited from a preferential allocation policy for families that included elderly relatives as well as accelerated construction of dedicated housing for single elderly people.

Figure 5 Cohort trajectories of public renting 1981–2001. Source: Authors' analysis of Census 1981, 1986, 1991, 1996 and 2001.

The second group, the birth cohorts in the middle who largely follow the general trajectory of public rental housing attainment, shows a slight departure for its 1986 to 1996 experience. For those cohorts, public renting trajectories still trend upwards (in contrast to the stagnation in the general population) and the increase among some birth cohorts is quite substantial (for example, among the older, middle aged in the 1921–25 and 1926–30 cohorts). However, for such birth cohorts, their experience in public housing shows a strong age effect in which the proportion of public renters in each cohort generally increases with age. This pattern is partly explained by the long waiting time for public housing in the 1980s. Despite their surge in public renting attainment in the previous decade, such cohorts also experienced downward trajectories in public renting after 1996, a pattern which is very similar to the rest of the cohorts. As the general contraction of public renting in the latter half of the 1990s can be attributed to the Tenant Purchase Scheme, it indicates that these middle-aged public tenants were equally receptive to the opportunities to purchase.

The third group comprises the youngest birth cohorts born after 1961 and who show a downward trajectory over the last two decades. The notable change occurs in the birth cohort born between 1956 and 1960. Unlike their older cohorts, their representation in the public rental sector initially decreases from 1981 to 1996 when they were aged between 20 and 24. It then increases. This may be explained by family life cycle factors. Most younger householders in the public sector inherited their tenancies from their parents when they moved out of the sector at marriage (the typical marriage age in the early 1980s was around the mid-twenties). Subsequently, more families were formed within this birth cohort and some of them went into public housing after years of waiting (at this time single persons were not eligible for public housing). Their senior cohorts may have had similar experiences but there are no available data on these groups. Such a pattern would represent a departure from the experience of their younger cohorts. The 1960 to 1965 birth cohort, for example, does not show an upward trajectory at the point when most of them should have formed their own family (only 10 per cent of people in this birth cohort were never-married in 2001 (Census, 2001)). However, such observations match the trajectories of homeownership presented earlier in which younger cohorts show a conspicuous surge in homeownership attainment. Therefore, it is not surprising to find public renting as the major complementary tenure, exhibiting a movement in the opposite direction.

With regard to cohort trajectories in private renting, there appears to be no strong cohort differentiation. Instead, a period effect is evidenced with most cohorts following a general downward trend in the private renter population over the two decades. However, the effect of age is also apparent. Again the evident surge in private renting among younger cohorts may reflect the incompleteness of the data in which potentially similar trajectories among their older counterparts cannot be explored due to the lack of relevant data prior to 1980.

Concluding Comments

Cohort analysis offers a powerful analytic instrument with which to explore longer-term social change. The two decades of Census data utilized in this analysis provide a window through which some of the interconnections between age, period and cohort effect on tenure change in Hong Kong can be examined. As with Chevan's (1989) analysis it is evident that the impact of process and compositional changes on tenure development in Hong Kong are both, if not equally, important. Hong Kong has undergone rapid, if uneven, changes in many aspects of its social and economic environment. These broader transformations have combined with people's unique lifetime experiences to produce very different housing histories within the various social, income and age groups.

During the last two decades the growth of homeownership in Hong Kong has been uneven with different factors fuelling recruitment of different cohorts at different times. It is the new and expanding middle class that has shown the highest propensity to move into the private homeownership market. They are also the group that has probably derived the greatest material benefits in terms of asset accumulation—given their pattern of recruitment during periods of price inflation. However, the development of public homeownership appears to have provided relatively uniform assistance across the other birth cohorts and has been of significant strategic importance in ‘democratizing’ access to the tenure. The Home Ownership Scheme was also a key factor in maintaining the further growth of homeownership at the time when escalating house prices imposed serious affordability hurdles for most home buyers before the Asian Financial Crisis as well as in its aftermath when the market slump seriously undermined market confidence. There is also some initial indication of an end to the expansion of homeownership among young adults—a trend common across a number of societies at present. This may be attributed to the general economic difficulties which Hong Kong has faced over the last few years or to the specific impact of a changing socio-economic environment in the last decade in which economic restructuring and social closure has put younger cohorts in a less favourable position to buy than their immediate older cohorts.

The other strong pattern to emerge from this analysis is the ageing of the public rental sector over the last two decades, with the diminishing recruitment of younger cohorts. Demographic ageing accelerated sharply even during the period when the public rental sector was still marginally expanding. Again, this is a pattern that is consistent with trends in other countries. The impact of privatization via the Tenant Purchase Scheme also begins to take effect after the Asian Financial Crisis indicated by declining levels of public renting across all birth cohorts.

Changes in the housing sector have both reflected and acted upon transformations in the wider economic and social profile of Hong Kong. The growing aspiration for homeownership has been fuelled by rising real incomes, demographic shifts, employment change, higher participation rates of women in the labour market and by government housing policies. Broad shifts in the pattern of housing tenure are well documented both in Hong Kong and across a wide range of (post) industrialized societies. However, this analysis offers a more detailed picture of the timing and rate of change in the housing sphere among different birth cohorts. As has been stressed earlier, however, it is important to acknowledge that the conditions of entry to particular housing tenures can involve variations in quality, marketability and asset appreciation. One cannot simply ‘read off’ social mobility from tenurial change. Nevertheless, in relation to homeownership, the fivefold increase in property values between 1987 and 1997 has been an important factor in the transformation of the vast majority of those households who gained entry to the tenure at the most propitious moments. Earlier and subsequent cohorts did not, and will not, derive the same financial springboard. Equally, a ‘golden age’ of homeownership is intimately intertwined with a period of pervasive state provision in relation to higher quality rental opportunities and state assisted entry to homeownership. When and where households joined the policy convoy has shaped their subsequent housing market trajectory.

Acknowledgement

The authors acknowledge the financial support of the Research Grant Council of Hong Kong to the Research Project ‘Public Housing, Economic Change and the Shaping of Family Housing Histories in Hong Kong’ (CityU 1165/01H).