Maximum Score Estimates of the Determinants of Residential Mobility: Implications for the Value of Residential Attachment and Neighborhood Amenities

Preprint
Publisher Website
Google Scholar
    • preprint
      • Published in RePEc
Abstract
This paper examines the determinants of the decision of low-income renters to move out of their current dwelling. Maximum score estimation is shown t be superior to ordinary discrete choice estimation techniques (probit, logit) for this problem, ad for similar discrete choices that require revering a previously optimal decision. The estimation reveals psychological costs of moving for typical low income renters of at least 8 percent of their income; these costs are even higher for older, longer tenure, or minority households. Policies that displace low income renters will have large social costs. In addition, the estimation results are used to calculate implicit household willingness to pay (WTP) for neighborhood amenities. This WTP based on mobility behavior is much greater than WTP estimates derived using hedonic methods, and is argued to be more accurate. This paper uses a semiparametric empirical technique to estimate the determinants of the decision of low-income renters to move out of their dwelling. These estimates show that low-income residents highly value remaining in their dwelling. In addition, these estimates are used to illustrate an alternative method to measure willingness to pay for neighborhood amenities. Moving decisions are usually examined with standard discrete choice models such as probit or logit (e.g., Venti and Wise (1984), or Weinberg, Friedman, and Mayo (1981). But the moving decision presents econometric difficulties for standard discrete choice models. As will be explained in section 1, because the household decision about moving is conditional on having previously preferred the original location, the disturbance term in mobility models is unlikely to follow the simple distributional forms required for probit or logit estimation. Maximum score estimation is an alternative estimation technique for discrete choice models that is robust to unusual distributions of the disturbance term. Although the theoretical properties of maximum score e

This publication has 0 references indexed in Scilit: