Asymmetric Adjustment Costs and The Dynamics of Housing Supply

This paper considers the potential effects of asymmetric adjustment costs on the dynamics of housing supply. The model in Section 2 provides explicit microfoundations for the divergence between long and short run supply elasticities and also predicts asymmetric adjustment whereby positive deviations from equilibrium are associated with faster adjustment as compared with corresponding negative deviations. The paper also tests for asymmetric adjustment costs by estimating a number of asymmetric and/or non-linear equilibrium correction models using data on the Irish housing market. A number of interesting insights into the dynamics of housing supply have been uncovered. Firstly, and most importantly, the empirical section estimated a unit elastic equilibrium housing supply curve which suggest Irish housing supply is significantly less elastic than housing supply in other economies such as the US. The finding of only a unit elastic long-run housing supply curve means that there would appear to be significant constraints on the supply side of the market even in the long-run. Secondly, of the six models considered, all are supportive of the proposition that the adjustment costs associated with an expansion in housing output are greater than the adjustment costs associated with a contraction. This gives rise to relatively slow upward adjustment of housing output in response to a surge in demand. Thirdly, a number of the estimated models support the belief that there are threshold points on the supply side of the housing market: large deviations from equilibrium appear to be associated with faster adjustment when compared with small deviations from equilibrium. Indeed, over a small interval about the estimated equilibrium, the adjustment of housing supply is not significantly different from zero. Such inertial supply behaviour is consistent with optimising behaviour under adjustment costs non-convexities. In conclusion, it appears that the above models with both asymmetries and non-linearities can capture important empirical features of the supply side of the housing market. One not insignificant shortcoming associated with these models, however, is that it is very difficult to distinguish them in-sample from corresponding linear symmetric specifications. Only in the case of the cubic polynomial adjustment mechanism was it possible to statistically distinguish the asymmetric non-linear adjustment from a nested model with symmetric linear adjustment. Future research should therefore examine the extent to which it is possible to distinguish between competing models in terms of out-of-sample forecasting.