A method for analyzing backward recurrence time data on residential mobility.

Data on duration of current residence are used to estimate models of duration-dependent residential mobility rates. "We use the theory of backward recurrence times in renewal processes to infer the hazard rate of residential mobility from the density of the duration of current residence. We demonstrate that the inferences made using the asymptotic solution of the density of the backward recurrence times will be severely biased if the renewal process has not reached stability. We present an unbiased finite time solution and estimate univariate and multivariate models of residential mobility rates using 1980 U.S. census data on duration of current residence. These models include mover-stayer models represented by mixture densities. The multivariate analysis shows that the most important covariates of residential mobility are age, homeownership, race, and presence of school-aged children."