Booms and Busts in the UK Housing Market

This is an updated and non-technical summary of an earlier paper which appeared in The Economic Journal in November 1997.

In the past 40 years, there have been two major booms in the UK's owner-occupied housing market: in the early 1970s and in the late 1980s. There were also smaller booms in the 1960s and, more briefly, in the late 1970s, while the early 1990s saw a bust on an unprecedented scale, at least for the UK. In the paper, we examine the causes of these booms and busts with an econometric model for the prices of second-hand UK houses in the period 1957-1994.

It is now recognized that the increases in housing wealth which took place in the 1980s contributed significantly to the consumer boom of the 1980s. Indeed, that none of the major econometric models of the UK incorporated housing wealth in their consumption functions at that time, was a major reason for the failure to forecast consumer expenditure which led to costly errors in macroeconomic policy. The major econometric models of the UK economy do now incorporate housing wealth alongside financial wealth in their consumption functions. This makes it all the more important to have an econometric model which increases our understanding of the determinants of house prices and of the effects on house prices of various policies, fiscal, monetary or supply side.

Our paper derives an equation for real house prices as an inverted housing demand function. The theory of housing demand is examined in an intertemporal context taking into account expectations, credit constraints, lumpy transactions costs and uncertainty. The theory predicts several shifts in parameters as a result of the financial liberalization of the 1980's. One of these concerns shifts in wealth effects which we have already analysed in work on UK consumer spending at a national and regional level. The empirical evidence supports similar shifts in the housing demand function. Furthermore, the theory predicts an increased role for income growth expectations and real interest rates in the 1980s. This is strongly borne out by the empirical evidence.

The presence of lumpy transactions costs is shown to result in important non-linearities or threshold effects in the aggregate demand for housing. This arises from the extensive margin of housing demand: the greater is appreciation of house prices, actual and prospective, the more households are pulled over the transactions cost hurdle to engage in trade. At these times of heightened activity or `frenzy', sharply increased demand feeds back into higher prices and, as in 1971-3, 1978-9 and 1986-9, substantial increases in house prices then occur. These spikes in the data can be successfully modelled with a non-linearity in the predicted rate of return. Indeed Hendry's (1984) specification of a cubic (though applied by him to last quarter's capital appreciation) is shown to provide an excellent empirical approximation to the non-linearity. Without such a non-linearity or dummies for the spikes in the data, the equation standard error more than doubles. We also find evidence that sharp falls in the rate of return makes households more cautious about entering the housing market.

Our treatment of expectations takes care to make reasonable assumptions about the information agents are likely to have and permits both forward looking and extrapolative elements in behaviour. The strong evidence that both house prices and relative rates of return in housing are forecastable is consistent with the hypothesis that housing markets are far from efficient. Indeed, our evidence is for an important extrapolative element, as well as a rational element, in the formation of rate of return expectations. Similarly, for income expectations, our results support our findings in the consumption context, that forward looking expectations are important but that many households appear to feel constrained by current income.

Relative to previous work, there are a number of other innovations in the paper including the treatment of composition biases in the house price index and the incorporation of an index of demographic change. This rises with an increase in the shares of the population in the key house buying age groups. Another significant improvement is to allow the supply, both of owner-occupied and of rented property, to play a role.

According to our model, many factors conspired to produce the house price boom of the late 1980s. Initial debt levels were low as were real house prices, giving scope for rises in both. Income growth after the early 1980s recession was strong, as were income growth expectations and these became more important as a result of financial liberalization, though partly offset by bigger real interest rate effects. Wealth to income ratios grew and the spendability of illiquid assets was enhanced by financial liberalization. Financial liberalization also permitted higher gearing levels. Demographic trends were favourable with stronger population growth in the key house buying age group. The supply of houses grew more slowly, with construction of social housing falling to a small fraction of its level in the 1970s. Finally, in 1987-8 interest rates fell and the proposed abolition of property taxes in favour of the Poll Tax gave a further impetus to valuations.

The bust in the early 1990s was the result of the reversal of most of these factors. Interest rates rose from 1988-90. Income growth and growth expectations weakened. Demographic trends reversed. The revolt against the Poll Tax resulted in a new property tax, the Council Tax, being reintroduced. Debt levels and real house prices had reached very high levels, while wealth to income ratios then fell and recently experienced rates of return became negative and made households more cautious. Mortgage lenders tightened up their lending criteria, in a partial reversal of financial liberalization. Under these conditions, not even the major falls in nominal interest rates that took place in the early 1990s, while real interest rates remained high, were sufficient to revive UK house prices. Our results suggest an important lagged endogenous and indeed, non-linear, element in behaviour which implies a potential for volatility. Evidence for such volatility can be found in the years 1989 to 1995 when house price to income ratios have gone from the second highest peak in the post-war period to the lowest level, probably since before the War.

The model suggests that, fundamentally, the potential for volatility remains. It implies that three dampening forces are operating on the current upturn: less favourable demographic trends, relatively high levels of debt and relatively high real after tax interest rates. To this one can add the greater awareness by mortgage lenders of default risk and, by the authorities, of the UK housing market as a potential factor in macroeconomic instability.

On the other hand, there are signs of the former being forgotten in some quarters and the Monetary Policy Committee is currently constrained by the strength of Sterling and the outcry over plant closures in manufacturing. Furthermore, the model quantifies the feed-through from financial wealth to the housing market: effective financial wealth relative to income is currently close to the high point of the last 35 years and the effects of this are not yet fully reflected in the market. Moreover, the closest thing the UK has to a property tax - Council Tax - is poorly related to market values, and therefore dampens house prices less than did Domestic Rates, abolished in 1989. Finally, the overshooting on the downside of house prices relative to incomes in the early to mid 1990s has helped to give the rises since then a momentum which, in turn, increases the risk of overshooting at the next peak.

Download this paper
Back