Environment and Planning A, 1986, volume 18, pages 763-788 Residential mobility in Dutch housing markets WAV Clark Department of Geography, University of California, Los Angeles, CA 90024, USA M C Deurloo University of Amsterdam, Spui 21, Amsterdam, The Netherlands F M Dieleman State University of Utrecht, Utrecht, Kromme Nieuwe Gracht 29, The Netherlands Received 6 May 1985; in revised form 11 November 1985 Abstract. A one percent stratified random sample of all Dutch households (comprising 62 000 households) is used to examine mobility in the Dutch housing market. Two techniques, proportional reduction in uncertainty (PRU), and logit analysis are used to evaluate the relative contribution of independent variables in explaining mobility across the tenure types and by housing market regions. The PRU technique is used to select the best variables and to simplify the categories of those variables before logit analyses. Given a reduced set of variables and combinations of categories, logit models are utilized to provide parameter estimates of the contributions of the independent variables. The PRU technique shows that it is possible to make considerable simplifications in the combinations of categories for variables, and the logit models (for owners, public renters, and private renters) indicate distinct differences in the combinations of variables which are most appropriate in each of these sectors. In particular, the logit analysis shows that the relationship between age and mobility is almost linear, but between income and mobility it is curvilinear, and that evaluating interaction effects adds significantly to our understanding of the reason to move or to stay. The models show that it is necessary to treat the three tenure sectors separately, even though age, just as we expected, dominates all sectors in terms of its explanatory power. Although there are not large regional effects, the models change for different regions of the country. 1 Introduction There is now general consensus that residential mobility generally, and the decision to move specifically, can be explained largely as a function of dissatisfaction with the present housing situation of the household. Although economists have stressed the importance of housing disequilibrium, and geographers and sociologists have focused on stress and stress-inertia trade-offs, there is in fact little difference in the conceptual formulation of these models. The differences come in the way in which stress, disequilibrium, or dissatisfaction are measured. In fact, the stress-threshold (Speare et al, 1975), stress-inertia (Clark et al, 1979), disequilibrium (Hanushek and Quigley, 1978a), and utility (Brummell, 1981) models can all be regarded as special applications of a general benefit-cost model. In such a model, residential mobility is assumed to occur if the benefits of moving outweigh the accompanying costs. The models that have been developed differ principally in the focus on types of benefits and the costs considered. So far, there has been somewhat greater empirical evaluation of housing disequilibrium models than of stress-inertia models. The greater emphasis on the former is in part related to the fact that they are more easily estimated, although Phipps and Carter (1984) have shown that the stress-inertia model can be estimated with utility functions derived from individual household preferences. Research on the housing disequilibrium approach, initially stimulated by the work of Quigley and Weinberg (1977) and Hanushek and Quigley (1978a; 1978b),