An empirical investigation of alternative approaches to estimating the equilibrium demand for housing

Abstract

Two approaches to estimating the desired or equilibrium housing demand parameters have been utilized in recent literature. One approach, commonly referred to as the conventional model, assumes that households continuously consume their temporal equilibrium housing services, where a temporal equilibrium is defined with respect to the household’s current economic and demographic characteristics. Based upon this assumption, conventional demand studies have employed random samples which include a majority of households that have not recently moved. (See Lee and Kong [6] or Carliner [2].) A problem with this approach is its failure to recognize moving costs households must pay to alter their housing service consumption. The existence of moving costs suggests that households may not respond rapidly to changes in their desired consumption of housing, but instead remain in their dwelling unit until the cost incurred from consuming the non-equilibrium level of services exceeds the cost of relocating. If so, demand samples employed in conventional studies may include a significant number of disequilibrium observations. This causes two problems. First, there exist measurement errors in independent variables that have changed since the household was last in equilibrium. Second, the true equilibrium effect of an independent variable may be incorrectly estimated due to its correlation with either the occurrence of disequilibrium or the length of the housing adjustment lag. The second approach to estimating the equilibrium parameters attempts to minimize these problems by restricting the sample to households that have recently moved. (See Hanushek and Quigley [3] or Aaron [l].)