Housing Location and the Supply of Local Public Services.

An econometric model of residential choice among communities. Using a sample of 682 white households whose head worked in San Francisco and resided in nine Peninsula cities of 10-50,000 population in 1965, the study employs multinomial logit maximum likelihood estimation methods. The sample was divided into six subsamples along three dimensions: household's income, size, and age of head. The estimated parameters are interpreted as marginal utilities. Results refute earlier studies by indicating that residential choice depends primarily on the quantity of housing available, not community characteristics--except that low-income households appear to locate where the most public services are provided. As expected, all subsamples dislike commuting, prefer lower crime rates, and desire parks and recreation. School expenditures matter to younger, lower income, and larger families. The parameter of the percentage of black residents was significantly negative only for low-income whites. Unexpectedly, the parameter of effective tax rate was significantly positive. An appendix applies discriminant analysis to classify households into communities; it performs better than chance, but not importantly so. (University of California, Berkeley, doctoral dissertation in economics.) 107 pp. Bibliog.