RDP 2004-08: Housing Construction Cycles and Interest Rates 6. Conclusion

In this paper, we have estimated structural models of housing supply and demand for several countries. The separate treatment of number and quality of dwellings has allowed us a reasonable identification of the two sides of the market, and pointed towards some possible reasons for differences between the behaviour of the housing markets in these countries. With only four countries in our data set, it is not possible to demonstrate conclusively that these cross-country differences are the results of the institutional and other differences discussed in the previous section, but they are likely to be important parts of the story.

In particular, the results for the UK seem to indicate that countries with supply constraints on land and generally slow population growth might have housing construction cycles with lower amplitude than those in countries where population growth is relatively higher and where there is more scope to expand residential development. If housing demand displayed similar interest sensitivity in all countries, we should expect that movements in rates would be translated more into price than volume movements in countries where these supply constraints are important. Notwithstanding recent developments in housing prices in Australia and elsewhere, we should therefore expect both structure and land prices to be less cyclical (and quantities more cyclical) on average in countries like Australia and the US than in continental Europe or Japan. It should also be expected that cycles in quantities would be more a feature of the Australian data, while the UK, with otherwise similar institutions in the mortgage market, would be more subject to cycles in housing prices.

Institutional details in the housing finance market also have a role in explaining the differing behaviour of construction and housing prices across countries and through time, and considerably complicate the story. The details of taxation arrangements and government intervention in the housing market might also make a difference, by determining where on the yield curve the mortgage market tends to operate.

We have estimated reasonably consistent, structurally motivated models for housing construction activity for a range of developed countries. However, these models do differ, and would be likely to have differed more if we had been trying to tune the individual country models closer to the data in each country. Quantitative findings about construction behaviour and policy transmission in one country are therefore likely to be inapplicable for other countries with different institutions. Although we have drawn out several likely factors explaining the differences in cyclicality and interest sensitivity in housing construction across countries, it is possible that these are not the only ones. Further research on these differences would therefore seem desirable in order to know whether lessons from one country are applicable to another.