2 - A Logit Model of Demand for Neighborhood

Abstract

This chapter examines a logit model of demand for neighborhoods. It highlights the research conducted in 1974–1975 to estimate housing demand functions for use in the National Bureau of Economic Research Urban Simulation Model and presents many of the assumptions of that model concerning the operation of housing markets. The model is empirically estimated using conditional logit analysis and is based on data concerning the residential choices of 5823 Pittsburgh households who changed residence between 1962 and 1967. The logit model provides a method of estimating models with discrete dependent variables, which avoids the problems of predicted values being outside the permissible range, lack of independence of the error terms, and loss of efficiency besetting linear least squares techniques. The results are generally consistent with prior expectations concerning the influence of the explanatory variables. In particular, measures of neighborhood quality exhibit fairly consistent and significant effects on residential choice. No study appears to have considered the quality of the neighborhood in which the housing unit is located to be a relevant attribute. Neighborhood quality has many dimensions. The average dwelling unit quality, the economic well-being of the residents, the quality and quantity of local public services, and the racial composition of the population are all aspects of the quality of a neighborhood as perceived by individuals choosing a place to live. Neighborhood quality would also be expected to vary with the size of households. Neighborhood quality is, thus, a superior good that enters the household's utility function. Even though it is difficult to measure with accuracy, its exclusion from housing demand estimations is likely to result in inaccurate predictions of residential location.