Bulletin – September 1995 Trends in the Housing Sector

Introduction

Activity in the housing sector is particularly cyclical. The current downturn has many of the usual features: the number of new dwellings being constructed has declined substantially, sales of existing houses are down and the growth rate of housing credit has fallen. In addition to these cyclical features, the housing market continues to be affected by Australia's return to a low-inflation environment. This is seen most clearly in smaller rises in house prices. It can be seen also in an increase in the measures of housing affordability. By making possible lower nominal interest rates, low inflation has increased the access of many borrowers to housing loans. The result has been a significant increase in household indebtedness.

These developments have attracted considerable attention, not only by people concerned about the normal cyclical behaviour of the housing sector, but also by those concerned about the implications of higher indebtedness and lower house-price inflation. This article addresses some of these concerns and examines trends in the housing sector. It begins by presenting some broad features of the housing cycle and its determinants, and then reviews recent trends in housing construction and finance. It concludes with a discussion of the implications for the housing sector of low inflation.

Some Facts

Housing construction is characterised by regular phases of expansion and contraction, with a typical duration of about four years. Falls in new construction spending from peak to trough of 20 to 40 per cent, and increases in the upswing of 40 to 60 per cent, are typical. While dwelling construction accounts for only a small proportion of total spending and production in the economy (around 3 per cent of GDP), the magnitude of these swings means that it can contribute significantly to the overall business cycle (Graph 1). The direct impact of these swings can be the equivalent of around 1½ per cent of GDP. Housing activity usually leads the cycle in overall activity (Graph 2).

There is much more to the housing sector, however, than the construction of new dwellings. The dwelling stock in Australia numbers about seven million units, and the bulk of transactions in the housing market is related to turnover of this existing stock. On some estimates, about 5 per cent of the stock is turned over each year, whereas new construction of dwellings in any year typically adds only about 2 per cent to the stock. Improvements to the stock are also an important element of the market. In recent years, spending on alterations and additions has accounted for 35 to 40 per cent of total gross dwelling investment.

A mixture of demographic and socio-economic forces drives activity in the housing market. The demand for new dwellings is ultimately determined by population growth and rates of household formation. These factors also have a bearing on trading in and improvements to the existing stock. Other important determinants include the general upward trend in living standards, the relatively favourable taxation treatment of housing, and the traditional preference for real estate over other forms of investment. Around 70 per cent of households are owner-occupiers, on a par with the Anglo-Saxon countries but significantly above a number of European countries and Japan. Surveys regularly identify residential property or home extensions/renovations as the preferred use for discretionary savings. Around 10 to 15 per cent of owner-occupiers also own at least one rental property. Housing accounts for about one-third of the total capital stock, and represents about 55 per cent of private-sector wealth.

Expenditure on housing – either new or established – is sensitive to changes in interest rates, both actual and expected. Rises in interest rates are generally associated with reductions in both housing construction and purchases of existing houses. To the extent that this is followed by a slowing in economic activity more generally, the downward forces on housing spending can be intensified. The slowing in housing activity itself acts to slow spending on some consumer durable items. (Spending on household appliances and other household durables accounts for about 7 per cent of private consumption expenditure.)

Typically, the deferment of investment expenditure leads to an unsatisfied or ‘pent-up’ demand for housing emerging towards the low point of the cycle. The subsequent upswing in housing activity is intensified by the release of this backlog. Conversely, the existence of any overhang resulting from over-building during a boom tends to accentuate the downward phase of the cycle.

The Construction Cycle

During the early 1990s, housing construction proceeded at a very rapid pace. The upswing in the construction cycle was unusually long, and between mid 1991 and mid 1994, more new dwellings were constructed than in any three-year period since the early 1970s. Several factors were behind this strength. For a start, there were significant levels of pent-up demand left over from the late 1980s, while the pattern of internal migration added to the demand for new housing – particularly in Queensland and Western Australia. Affordability conditions were also very favourable, a result of stable house prices and the lowest nominal interest rates for a generation (Graph 3).

In an environment of increasing overall confidence, these conditions encouraged the bringing forward of home purchase decisions by prospective owner-occupiers, as well as strong investor interest. This was further accentuated by more aggressive targeting of household borrowers by banks and other lenders, and active marketing initiatives by developers (including, in some instances, ‘guaranteed’ rental returns). Another factor was the lower rates of return apparently on offer from some alternative investments, particularly bank deposits.

By the middle of 1994, however, these forces were beginning to turn against the housing construction cycle. High levels of construction activity, in excess of the underlying demand for new dwellings, rapidly eroded unsatisfied demand and an oversupply of dwellings began to emerge (Graph 4). Interstate migration slowed as the economic recovery became more broadly based, and there was some decline in investor interest because of high vacancy rates in some areas, poor prospects for capital gains, and better rates of return on some other assets.

Subsequently, the reduction in affordability that followed the rise in mortgage rates during the second half of 1994, together with the expectation of further rate rises and a tough Budget, added to these downward pressures during the first half of 1995. Lending for housing slowed, and construction activity began to fall away.

By the middle of 1995, the number of private dwelling commencements was about 26 per cent below its late-1994 peak. Sales activity, as proxied by real estate transfer expenses^[1] from the national accounts, has fallen by a similar amount. Construction activity has fallen in all States, although the downturn has not been uniform. NSW, to date, has proved to be more resilient than other States. Low vacancy rates and a pick-up in dwelling rents in NSW suggest that this is something more than just residual strength and could help to moderate the overall downturn. In Queensland, Western Australia and the ACT, the falls have been relatively large. These areas have been more prone to over-building.

Further falls in dwelling commencements are likely. The government's Indicative Planning Council (IPC) for the Housing Industry expects 130,000 new dwellings to be commenced in 1995/96. This is below the IPC's estimate for underlying demand of around 150,000 new dwellings per year. While the fall in dwelling commencements is substantial, it needs to be assessed against the background of the earlier strength.

The decline in the number of commencements will lead to a significant fall in dwelling investment, although this fall will be moderated by spending on alterations and additions – which accounts for over one-third of total residential investment expenditure. This spending is less cyclical than that on the construction of new houses, and could hold up at relatively high levels, particularly if overall income growth in the community is maintained, as seems likely. The 1995/96 Budget includes a forecast for a decline of 13 per cent in dwelling investment in 1995/96; this follows increases of 13.5 per cent and 4.6 per cent in 1993/94 and 1994/95, respectively.

It is worth noting that divergent trends are now evident between the residential and non-residential parts of the overall construction sector (Graph 5). In contrast to previous cycles, the value of non-residential building approvals is moving up, roughly offsetting the decline in residential approvals, so that the total value of approvals has shown little change over the past year. Business and other surveys show that planned spending on buildings and structures is rising. These factors may lessen the degree of contraction in the construction sector overall. Over recent years, high levels of housing investment have been accommodated, in part, by the slack in the non-residential sector, and employment levels have remained below previous peaks. The reverse is likely in the coming year or two. This ‘de-synchronisation’ is a considerable advantage in smoothing the overall cycle, and suggests that the construction industry, as a whole, may not suffer to the same extent as in some other cycles.

Finance for Housing

Lending and household debt

A good deal of attention has been given to the fall in loan approvals over the past year (Graph 6). They have fallen by about a third from their peak in mid 1994. Again, however, the extent of the previous upswing is important in assessing this fall: at the peak, loan approvals were double their level of 1991 when the upswing began. Their 1994 level was unsustainable, and certainly undesirable, associated as it was with growth in housing credit outstanding of 25 per cent per annum. Current levels of approvals (which include a rise over the past couple of months) are consistent with a pick-up in growth of housing credit, from recent rates of 10–12 per cent (already a solid rate of growth in an economy with inflation running at 2–3 per cent).

The increase in interest rates during 1994 (including for fixed-rate loans, which move with interest rates in the bond market) played a role in the slowing, as did the widespread expectation that further increases would follow. But these mechanisms were more powerful because the ‘fundamental’ preconditions for a cyclical peak in construction were in place: too many dwellings were being constructed relative to ongoing requirements.

These trends in finance also need to be set in a slightly longer-term perspective. The stock of housing credit outstanding has doubled in five years. To some extent, this is part of a medium-term trend towards higher debt levels in the household sector, resulting from financial liberalisation which lifted earlier constraints on borrowing, and innovation which allows households to manage their finances in a more sophisticated way. Ten years ago, Australian households carried an unusually small amount of debt compared with their counterparts in other developed countries. Even now, debt levels are not high by international standards (Graph 7).

In addition, improved housing affordability has allowed households to carry more debt relative to their income. This includes allowing households to borrow who five years ago would not have been granted a loan at all. Indeed, the increase in housing debt has been attributable at least as much to a larger number of loans as to larger average size of loans (Graph 8).

Supply-side factors too played a role in the growth of lending for housing. With little demand for loans from business borrowers, and relatively high margins on standard variable-rate housing loans, banks have been keen to extend housing credit. The strongest competition for new loans has been in the ‘honeymoon products’, with the standard variable rate remaining relatively high.

Reflecting the rise in indebtedness, the share of overall household disposable income devoted to mortgage servicing has risen to about 6 per cent, which is higher than at the peak in 1990.^[2] For the typical recent home borrower, interest payments are estimated to be equivalent to 24 per cent of income. This is below the peak figure of 29 per cent reached at the end of the 1980s, but higher than the low levels in 1993.

House prices and household equity

To date, the trend towards higher levels of debt has co-existed with relative stability in house prices overall, although trends have differed by region (Graphs 9 and 10). Two years ago, it seemed unlikely that such a rapid increase in credit financing of the housing stock would not put upward pressure on prices, but those pressures have in fact been quite mild, and-mostly confined to a few areas.

Although compositional and sampling problems complicate the analysis of house-price movements by region, data obtained from the State-based Real Estate Institutes indicate that house prices in most regions in and around Sydney and Melbourne have changed little over the past five years. While falls have occurred in some regions – most noticeably in the ACT and South East Queensland – the falls have tended to be localised. These falls, together with the general lack of house price inflation, have led some observers to suggest that recent borrowers, with high levels of debt and the possibility that their house value might fall, are exposed to the risk of so-called ‘negative equity’ – a situation where the asset value falls below the amount of the debt obligation. The fact that the construction boom has left surplus stock in some areas, particularly in outlying suburbs in Sydney and Melbourne and South East Queensland, and that mortgage insurance companies report an increase in defaults and loans in arrears, has also attracted attention.

Such a situation, were it to be a widespread phenomenon, could carry serious implications; the experience of the United Kingdom in the late 1980s and early 1990s illustrates what can happen (Graph 11). A generalised fall in house prices could cause a much more serious slump in new house construction than is presently anticipated, and the adverse effects on household balance-sheets would also be harmful for growth prospects.

Our judgment at this time, however, is that a sequence of developments of this nature is not likely in Australia, for several reasons.

First, the pattern of prices seen in the UK in the 1980s – a very large rise, with many borrowers getting locked in at the top of the cycle, followed by a widespread slump in prices – is not in evidence. The last time there was a big increase in house prices was in 1988, and there has been little change since then. The period of greatest borrowing did not follow hard on the heels of that rise in prices, but occurred five or six years afterwards, and there is no evidence of a widespread fall in prices in the biggest urban areas.

Secondly, the vast bulk of borrowers has substantial equity. The ratio of housing credit to the value of the housing stock has risen from about 10 per cent in 1990 to 14 per cent at present – still a very low figure. Data for 1994 show that little more than one-quarter of all households were in the process of purchasing a dwelling.^[3] Those most ‘at risk’ within this group are recent home buyers, particularly first home buyers, who have had less time to build a significant amount of equity. Data for 1994 suggest that this group comprises perhaps only 5 per cent of all households, although it accounted for 37 per cent of home purchases.

Furthermore, many first-time buyers have a good deal of equity. In the June quarter this year, the average size of a loan taken out by a first home buyer was $93,000, while the median price paid by first home buyers was around $135,000. Looking across all home buyers, estimates indicate that only around 20 per cent of bank loans over the past year have been at loan-to-valuation ratios (LVRs) above 80 per cent, and the bulk of this proportion is thought to have been at LVRs below 90 per cent.

These proportions compare very favourably with the UK experience, where more than one-third of the original loans to first home buyers in 1989 exceeded the value of the dwelling (i.e. the LVR was over 100 per cent), and even the typical loan was equivalent to around 85 per cent of the value.

Finally, while the interest rate increases in 1994 have increased the repayment burdens of those borrowers with variable rate loans, the number of loans with overdue repayments remains relatively low. The increase in the number of such loans that has been recorded is from the very low base established over recent years. Bank housing loans with repayments more than 90 days overdue (so-called ‘arrears’) have increased from $550 million in September 1994 to $750 million in June 1995. As a proportion of loans outstanding, this increase is quite small; currently less than 0.5 per cent of loans have been in arrears for more than 90 days.

Some households are no doubt suffering some financial difficulty from higher interest rates, and some others are seeing a fall in the value of their house, but in our judgment there is no large-scale problem of financial distress or ‘negative equity’ at present or in prospect.

Adjustment to Low Inflation

Low rates of inflation are having on-going effects on the housing sector. With lower inflation has come smaller rises in nominal house prices, increased access to debt, and repayment burdens which, in real terms, do not decline quickly through time. These changes are leading to adjustments in the mind-set of the community towards housing.

Under high rates of inflation, the nominal price of housing was rising continually, irrespective of what was happening to real prices (Graph 12). While there has been a general upward trend in the real price of housing through time, there have been extended periods during which real prices have been constant, or falling. Recent movements in real house prices are not out of line with this previous experience. What is different is that low inflation has meant that nominal house price increases have been much smaller than in the past. In the medium term, real house prices are likely to continue to rise but, in a low inflation environment, this will be associated with lower nominal increases. This change is only gradually being realised by the community.

A second effect of high inflation was that repayment burdens eased relatively quickly through time. Even if the interest payment burden early in the life of a loan was quite high, payments thereafter remained constant (changes in interest rates aside) while nominal incomes grew rapidly. It therefore became quite widely accepted that it was sensible to take on as much debt as possible and simply wait for inflation to help ease the burden. Lenders sought compensation in high nominal interest rates, of course, so that there was effectively a front-end loading of repayments, to the detriment of some people who struggled to pass the initial income tests applied by lenders.

With low inflation, there is less front-end loading of repayments, which allows larger loans to be taken on and more people to obtain loans. The counterpart of this is that real repayment burdens do not diminish as quickly through time. This change also has yet to be fully appreciated by the community, so that there is a risk that the aspirations of some borrowers may be disappointed. The same change might also lead, over time, to some reassessment by households of how much debt they wish to carry.

The lower rates of inflation and the associated slower rates of decline in the servicing burden mean that borrowers with high LVRs are at risk of encountering financial difficulties for a longer period of time. This risk was anticipated a year ago when the Reserve Bank removed the concessional risk weighting from housing loans with LVRs above 80 per cent. The provision of this clear signal to the market aroused controversy at the time, but the measure was introduced in such a way (on loans written after 5 September 1994) that there was no dramatic effect on lending patterns. Those regional banks (former building societies) which have maintained a rapid rate of high-LVR lending are now finding that the capital costs are beginning to bite.

Developments on all these fronts will need to be monitored carefully, but we do not expect serious deflationary pressures to arise. In fact, a strong appetite for new debt remains and there are some signs that the slowing in the take up of new loans might be tapering off. An early bottoming out in lending approvals would be consistent with the duration of previous cyclical downswings, which typically lasted a little over a year. While the demand for housing finance is not about to race away, credit is still rising noticeably faster than income. An ongoing trend of this nature would also warrant close monitoring.

Footnotes

Real estate transfer expenses cover the various fees incurred by buyers and sellers of real estate (e.g. legal fees, sales commission and stamp duty). [1]

Broader measures of debt servicing remain below their previous peaks. Allowing for personal debt, which has grown at a significantly slower rate, the rise in household debt relative to income has been less pronounced. The broadest measure of household debt, which includes the borrowings of unincorporated enterprises, has risen from 74 per cent to 86 per cent of household disposable income over the past five years. See the Reserve Bank of Australia's 1995 Annual Report for further details on trends in indebtedness. [2]

A further 42 per cent of households owned their dwelling outright; 28 per cent were renting, and the remainder were living rent free. Source: ABS Housing Survey, 1994. [3]