RDP 2004-08: Housing Construction Cycles and Interest Rates 1. Introduction

Housing investment is one of the most cyclical components of output in many industrialised economies, and likewise one of the more interest-sensitive. It is likely that much of the cyclicality derives from the interest sensitivity, but construction lags and resultant sluggish supply might also lead to intrinsically cyclical responses of output to demand shocks. The particular interest sensitivity of housing demand seems to stem partly from frictions in capital markets. Since these can change over time, it is also likely that the interest sensitivity and cyclicality of housing construction can vary through time and across countries.

In this paper, we use econometric techniques to discern whether the observed cyclicality is intrinsic to the construction sector, or a consequence of its interest sensitivity, and to document differences in this interest sensitivity across four English-speaking countries. In doing so, we are effectively doing for quantities what Sutton (2002) did for prices. We then attempt to reconcile these differences based on institutional differences in the housing construction and mortgage finance markets. The goals of the paper are thus similar to those of Aoki, Proudman and Vlieghe (2002) and McCarthy and Peach (2002), who identified changes in the policy transmission mechanism in the United Kingdom and United States associated with financial sector deregulation.

Our focus on institutional factors places this paper within a substantial recent literature on the different macroeconomic effects of housing market developments, particularly in the context of European Monetary Union (Maclennan, Muellbauer and Stephens 1998; Tsatsaronis and Zhu 2004, for example). Our paper extends the analysis in that literature by proposing a new approach to structural modelling of the sector. This allows us to disentangle supply-side from demand-side factors, and in particular, establish the relative importance of intrinsic cyclicality caused by sluggish supply, and extrinsic cyclicality resulting from demand responses to the interest-rate cycle.

We find a dominant role for extrinsic interest-rate cyclicality in explaining the housing cycle. However, there is also some weak evidence of intrinsic cyclicality in some of the countries studied, driven by the interaction between sluggish supply and flexible demand. When a demand shock occurs, supply adjusts only gradually, thereby generating a hog-cycle type effect on both prices and quantities supplied. This is partly due to the fact that the change in the housing stock that households demand can be much larger than the feasible flow of new housing supplied in any one period, and partly due to time-to-build constraints on the construction of that new supply. The extent of the sluggishness in supply presumably depends on a range of factors, including the structure of the construction industry, land availability, regulatory policies and other country-specific factors. It is not feasible to reconcile all these supply-side differences with the quantitative crosscountry differences in our estimated supply functions. However, they clearly have considerable scope to affect the magnitude of the transmission of movements in interest rates to housing construction.

When comparing the results across countries, we find evidence of significant differences in the (extrinsic) cyclicality of housing investment, even after allowing for the different paths of interest rates experienced in different countries. That is, our structural modelling appears to identify cross-country differences in the direct response of housing demand to movements in interest rates. This is compounded by variations in the income sensitivity of housing demand across countries. Although we do not formally model the response of permanent income to interest rates, it appears that the direct interest-rate effect and the indirect effect via income represent two channels of the transmission mechanism from monetary policy to demand for new and improved housing.

Institutional arrangements in mortgage finance markets clearly matter in determining these different degrees of interest sensitivity and income sensitivity. Although financial deregulation did not appear to alter demand behaviour in Australia, our results confirm McCarthy and Peach's (2002) earlier findings of a structural change in the dynamic behaviour of housing demand following financial market deregulation in the United States. As well as explaining structural breaks in housing market behaviour within a country, institutional factors might help explain the cross-country differences identified in our empirical results. The prevalence of fixed-rate versus variable-rate mortgage finance appears particularly important, which in turn depends on a range of taxation, regulatory and other policies.