A dynamic model of commutes

This paper studies the interaction between commuting, job mobility, and housing mobility. Many conventional models assume that the employment location has priority over the residential location and that the latter is adapted to the former. This implies that commutes which start with a job change will often be short lived because of a change in residential location that soon follows. It is also often supposed that the change in residential location is made with the intention to avoid long commutes. In this paper we test the empirical validity of these hypotheses. Our data are a sample of Dutch workers who report changes on the housing and labor market between 1990 and 1998. It appears from these data that both job mobility and housing mobility are often followed by repeat mobility on the same market, but also on the other market. Job mobility indeed triggers residential mobility, but the effect of residential mobility on job changes is of comparable magnitude. Moreover, both types of mobility lead to substantial repeat mobility. We specify duration models that focus on the time during which employment ¿ housing arrangements (hence, commutes) remain unchanged. Estimation results for these models confirm that commutes which start with housing mobility and those which start with job mobility have similar characteristics with respect to induced future mobility. We are unable to find evidence supporting the hypothesis that long commutes resulting from a job change induce additional residential mobility. Another result of the analysis is that workers belonging to dual-earner households are more mobile on both markets than other workers