A MULTINOMIAL LOGIT MODEL OF RACE AND URBAN STRUCTURE

This paper examines the influence of race on urban spatial structure in a multinomial logit model of residential and workplace location choice. A unique data set that provides the residence and workplace census tract of workers in the Dallas-Fort Worth CMSA, and their demographic characteristics, makes it possible to explore the joint choice of residence and workplace for workers of different race and ethnicity. The results support established hypotheses in urban economics and urban ecology regarding the disutility of commuting; residential and workplace suburbanization; spatial job search; housing-commuting tradeoff; residential clustering by race, stage of life cycle, and income; and housing tenure choice. With separate models for whites, blacks, and Hispanics, we find evidence of both residential and workplace avoidance of blacks by whites, with less white avoidance of Hispanics. Both Hispanics and blacks are found to seek integrated locations as their incomes rise.