Is There a Housing Crisis? The Incidence and Persistence of Housing Stress 2001–2006

Prominent on the economic and political agenda in 2007 were claims that Australia was suffering a housing crisis. It was argued that housing had become so expensive that the Australian dream of home ownership was well beyond the reach of the young and other Australians who not already had a foothold in the housing market. Increasing interest rates meant that many low and middle income households found it increasingly difficult to maintain their loan repayments, with increased risk of foreclosure. In addition, rental accommodation was in short supply, so renters began to pay considerably more for their housing, leaving less to save for a deposit and further excluding them from home ownership. During the 2007 election campaign housing affordability was a major issue and a variety of policies were proposed to reduce the incidence of housing stress among both renters and purchasers. House prices in Australia have been trending upwards in real terms for several decades.1 Over the decades mortgage interest rates fluctuated considerably. They averaged 8.65 per cent in the 1970s, 13.21 per cent in the 1980s (although the peak of 17 per cent occurred at the end of the period), 9.95 per cent in the 1990s and 7.21 per cent so far this decade (that is, until end-2007).2 Generally lower interest rates since the early 1990s have helped, somewhat, to cushion the impact of rising house prices on household budgets. However, by the end of 2007 the standard variable mortgage rate had risen to 8.55 per cent which, while low by the standards of most of the previous 25 years, is well above the average for this decade so far and in combination with sharply higher housing prices in several capital cities, has produced a decline in affordability for home ownership. Several measures of housing affordability have been developed. These frequently inform the popular debate. Typically these relate mortgage repayments to household income or calculate the implied ‘deposit gap’ faced by certain households. The Productivity Commission (2004, pp. 26–32) provides a useful discussion of the relative merits of these measures. A feature they share, however, is that they are constructed measures that attempt to capture the circumstances of a ‘typical’ or ‘average’ household purchasing a ‘representative’ dwelling. Data presented at the Melbourne Institute's Public Economics Forum on Housing Affordability in November 2007 by Braddick (2007, slide 13) of the ANZ Banking Group showed that on one such measure housing affordability improved significantly throughout most of the 1990s but has generally deteriorated subsequently and is projected to continue to deteriorate for some years. An alternative measure prepared by the Housing Industry Association and reported in this forum by Lamont (2008) suggests that housing affordability for first homebuyers in the December quarter of 2007 was the lowest since the series began in 1984. The picture in respect of the rental market is more complex. Productivity Commission (2004, pp. 22–4) states that median rents represented a relatively stable proportion of household income in most parts of Australia in the decade ending 2003.3Braddick (2007, slide 13) also presented more recent data to the Melbourne Institute's forum in November regarding the rental market. Real rents either declined or hardly increased for some years after 2000 but very recently have begun to rise sharply, raising questions about the financial situation faced by renters in the period ahead. The purpose of this paper is to assess whether the evidence supports claims that there is a housing crisis in Australia and, if so, to establish its incidence and characteristics over recent years. Using the standard measure of housing stress—a household is defined as being in housing stress if housing costs exceed 30 per cent of gross household income—we examine the incidence of, and trends in, housing stress for all households and adults, for owners and renters separately and by household income quartiles, in the latter case using equivalised4 disposable income. In addition, we provide estimates of the persistence of housing stress, one to five years after the first experience of housing stress. Most previous estimates of housing stress have used static measures of housing stress from cross-sectional data which may be taken to imply that households experience housing stress for considerable lengths of time. Since the standard definition of housing stress is essentially arbitrary, we also examine if there is a housing crisis according to other indicators. These compromise difficulties in paying the rent or mortgage is one such indicator; whether or not households are behind in their mortgage repayments; and movements to and from home ownership. Substantive increases in the proportions: having difficulty in paying their housing costs, behind in their repayments or moving from ownership, would be indicative of housing crisis. The paper utilises longitudinal data from the Household, Income and Labour Dynamics in Australia (HILDA) study using survey data collected between 2001 and 2006. Each year the HILDA study collects comprehensive measures of household income, housing costs and other indicators of the housing situation. Details on the sampling, data, response rates and weights are provided in the HILDA User Manual (Watson 2008). All estimates presented here are from weighted data and the 95 per cent confidence intervals surrounding the estimates are used to assess if differences in pairs of estimates are statistically significant. Table 1 shows the incidence of housing stress, according to the conventional definition. On this indicator the proportion of all households that were in housing stress rose from almost 11 per cent in 2001 to more than 12 per cent in 2006. The rise was from 9 to 10 per cent for individuals. These movements are significant—the confidence limits for the estimates in the later years do not overlap with the estimates from earlier years, implying that there has been an increase in the proportion in housing stress over the period.5 The discrepancy between households and individuals arises because larger households tend to have higher incomes. These estimates for households are lower than an earlier estimate of 15 per cent derived from ABS data (Yates 2007). The finding of an increase contrasts with Yates' (2007) finding of no increase in housing stress over the 10-year period from 1993–94 to 2003–04. Table 1 also shows the incidence of housing stress by tenure type. The proportion in stress is much lower among owners (which includes those in the process of paying off their mortgage) than in the total adult population. However, the proportion of owners in housing stress has increased by more than in the total sample—the estimates for 2006 are about 2 to 3 percentage points higher than in 2001 for both individuals and households. The confidence limits for the two estimates do not overlap, indicating that there has been an increase in the incidence of housing stress in the population from which the sample was drawn. In general terms this seems consistent with the rises in house prices and interest rates since 2002. However, the analysis presented in the next section suggests that the explanation is not as simple as this. The incidence of housing stress among renter households ranges from 23 to 25 per cent over the period, that is, about three or more times that of owners on the standard measure. Among individual renters the incidence is about one fifth. Although housing stress is much higher among renters, there is no indication that it increased between 2001 and 2006. Indeed the data suggests there has been some improvement for renters over the period. This finding is generally consistent with data presented at the Melbourne Institute Economics Policy Forum in November 2007 which shows that although real rents have increased sharply very recently, there were either declines or little real change up until 2006 (also see endnote 3). Not unexpectedly, the proportion experiencing housing stress is substantially higher among low income earners. Table 2 shows that more than one fifth of households in the lowest quartile of equivalised disposable income were in housing stress each year from 2001 to 2006. In the next quartile, the incidence was between 8 and 10 per cent, with further progressive declines in the proportion of those in stress in the two highest equivalised disposable income quartiles. Unexpectedly, Table 2 shows that housing stress has not increased in the lowest disposable income group: indeed, the incidence of housing stress for this group is nearly 2 percentage points lower in 2006 than in 2001. In contrast, the trends for housing stress in the other (higher) income quartiles are upwards. In the third income quartile the incidence of housing stress rose from 3.0 per cent to 5.7 per cent and in the highest disposable income quartile it increased from 1.7 per cent to 3.7 per cent. The confidence limits surrounding the estimates suggest that these trends are statistically significant. Among owners there is no upward trend in housing stress among those in the lowest equivalised disposable income group but increases are found among the other income quartiles. Although the incidence of housing stress is low in these owner groups, in percentage terms the increases are fairly substantial. Increased housing payments among higher income households are more likely to reflect rational choices rather than housing stress. Higher income earners may be buying higher quality or better located housing6 or using the flexibility available these days with many mortgages to redraw against a partly repaid mortgage to finance other consumption (consumer durables, holidays etc.). The question asked of respondents to the HILDA Survey should capture payments on any mortgages taken out on the family home, even if the funds were applied to consumption or other purposes. Data about the costs incurred in servicing mortgages on investment properties or second properties is also collected but not used in this analysis. It is frequently argued that Australia's taxation arrangements encourage households to borrow to upgrade their dwellings, especially the favourable treatment of owner occupied housing for capital gains tax. The strong rise in house prices experienced in most capitals over the period and, at least in the initial years, the lower returns available on alternative assets such as equities (see Productivity Commission 2004, p. 52 and box 3.3) may have encouraged such activity.7 Housing stress among low income renters is especially high. Of renters with equivalised disposable household incomes in the lowest quartile nearly half were in housing stress in 2001 and 2002. However, among this group, housing stress declined over the period studied by almost 7 percentage points. Even so, by the end of the period the proportion of the lowest quartile in stress among renters was almost four-fold that of owners. The incidence of housing stress among the second lowest income quartile of renters was less than 20 per cent and also showed a significant decrease over time. The incidence of housing stress among renters with incomes in the third and top income quartiles tended to be similar to, or in several instances lower than, that for the comparable groups of owners and, in contrast to owners, showed no increase. A detailed examination of the idiosyncrasies of the rental housing market is beyond the scope of this paper. As has been previously noted, while house prices rose strongly and rental vacancy rates declined significantly over the period, median rents only just kept pace with inflation over the period and are likely, therefore, to have remained steady or to have fallen as a proportion of income for many households, especially those that remained continuously in work. However, the available data show that median rents rose strongly in real terms in 2007, implying that future HILDA waves are likely to show a decline in the position of renters.8 Table 3 shows the percentage of individuals in housing stress (conventionally defined) who were also in housing stress one or several years earlier. In each row, the cell entries are percentages of those originally experiencing housing stress in the year shown in the left-hand column that are also in housing stress in the year denoted by the column heading. Note, however, that such individuals may not necessarily have been in housing stress in each of the intervening years. Generally, the persistence of housing stress is lower than at least some may have anticipated. Of those found to be in housing stress in 2001, less than half were in housing stress in 2002, around 40 per cent in 2003 and about 35 per cent in 2006. Similar patterns were found for households initially identified in housing stress in 2002, 2003 and 2004 although there are fewer subsequent observations to assess persistence. Despite increases in house prices and interest rates, the persistence of housing stress (after one, two, three or four years) has not increased. This can be ascertained by comparing entries along each diagonal in Table 3. Persistence in housing stress is higher among renters than owners although the same pattern of declining persistence with time elapsed is evident for both groups. Among renters, the persistence level of housing stress three or four years after housing stress was first observed was more than 40 per cent compared with less than 30 per cent among owners. The standard indicator of housing stress used in many samples is essentially arbitrary. The cut-off for housing stress could be lower at say a quarter of household income or higher at 40 per cent. Furthermore, the ‘stress’ associated with paying 30 per cent of household income for housing would be much greater among low income than high income households. In addition, the standard indicator that is based on gross household income may be misleading because it does not take into account household size or government transfers such as rent assistance. In this section, three alternative indicators are used to assess whether or not there is an emerging housing crisis. Each year respondents to the HILDA Survey are asked if they ‘Could not pay rent or mortgage on time’. A high incidence or a substantial increase in the proportion of those who reported difficulty in paying their rent or mortgage would be indicative of a housing crisis. However, the incidence of difficulty in meeting housing costs has declined from 9 per cent in 2001 to 6 per cent in 2006 (Table 4). The proportion of renters behind in their payments is twice that of owners, at 13 per cent in 2006. However, for both groups there is a decline between 2001 and 2006 in the incidence of reported difficulties in meeting housing costs. The decline is especially pronounced among renters, declining from more than 20 per cent in 2001 to around 13 per cent in 2006. An increase in those falling behind in their payments would be indicative of increasing housing stress and a large increase would suggest an emerging housing crisis. The measure used also includes information on the proportions of mortgagees paying off their mortgages on time or at a faster or slower rate than minimally required by their lender. Households may assume higher levels of mortgage repayments by choice in order to pay off their loan ahead of time. Households that choose to pay faster than required are obviously less in need of policy support, even if they devote more than 30 per cent of their income to meeting these payments. Table 5 shows that the proportion behind in their payments is very low but increased from 3 to 5 per cent, indicating that the rise in interest rates has caused problems for a small proportion of householders. The proportion ahead in their repayments declined from 61 per cent in 2001 to 51 per cent in 2006. However, the decline in the proportion paying ahead of time was almost precisely matched by the increase in the proportion paying off their mortgage on time. This finding suggests that a substantial number of households have accommodated increases in interest rates over the period by no longer being ahead in their mortgage repayments. Even so, more than half of households in 2006 were still paying off their mortgages faster than lenders demanded. The final indicator examined is the proportion moving into and out of home ownership. If housing has reached crisis levels in the past few years, then the proportion moving from ownership would be expected to increase as owners were forced to sell up and the proportion moving into ownership would decline, because more households are forced to remain in rental or other accommodation as home purchase has become even more difficult. However, Table 6 indicates there has been little change in the proportions moving into and out of home ownership during the time period studied. Although a number of commentators argue that housing in Australia is in crisis and that the crisis can be expected to get worse in the next few years, this paper finds no evidence for a housing crisis in the years between 2001 and 2006. On the conventional measure, housing stress increased but not among the groups considered the most vulnerable and expected to suffer most—low-income owner households and renters. It was generally owners with higher household incomes that showed increased housing stress. It is more likely that this increase reflected choices about asset acquisition or consumption rather than difficulties they experienced in securing shelter. Furthermore, housing stress tends to be relatively short lived. More than half of those in housing stress in one year were not in housing stress the next year. Persistence in housing stress declines further with time and there is no evidence of higher levels of persistent housing stress in more recent years. Other indicators of housing reported here did not support the contention that there is an emerging housing crisis. The proportion of adults experiencing difficulties in paying their rent or mortgage declined rather than increased. Moreover there was no increase in the proportion of mortgagees behind in their payments and in the proportions moving into and out of home ownership. The ability of the owners and renters, overall, to weather the increases in house prices and interest rates that occurred in the five years to 2006 can be attributed to the strength of the Australian economy, with increases in labour force participation and real earnings and record low unemployment. The HILDA data is likely to tell a very different story should economic circumstances markedly deteriorate. The substantial increases in real rents in 2007 and further increases in interest rates throughout 2007 and increases in 2008 will have negative consequences on housing affordability in Australia. However, the analysis presented here shows that the position of renters, especially low income renters,9 is considerably more serious than that for owner occupiers, with in excess of 40 per cent of the lowest income quartile devoting 30 per cent or more of their equivalised disposable income to meeting housing costs. This suggests that the highest priority for policy development lies in the area of low income renters.