A residence time model of housing markets

This paper estimates a semi-Markov model of housing markets in which families move from one home to another, spend a random amount of time in each home, and choose whether to own or to rent at each move. Limiting conditions of the semi-Markov process are used to characterize the steady state frequency of owner-occupied housing, and to simulate the impact of changes in housing tax policy. This approach differs markedly from previous studies of housing tax policy, which focus on the entire population (as opposed to the movers) and ignore the effect of household residence times. [See King (1980), Rosen (1979), White and White (1977), or Laidler (1969).]. Residence times further influence the analysis by affecting the relative cost of owning to renting in the tenure choice model. Homeowners pay legal and realtor fees at the time they move out of their homes which renters do not have to pay. The discounted value of these fees declines with length of stay and provides a structural explanation of why families with longer residence times have a greater propensity to own. Results from the empirical analysis suggest that the semi-Markov model closely predicts conditions in the United States, and provide important insights into the effects of housing tax policy. Residence times are also found to influence household tenure choice through their impact on the discounted legal and realtor fees paid by homeowners.