An econometric study on the dynamics of urban spatial structure

Using data on 43 randomly selected census tracts for each of 39 U.S. cities in 1970, we present a comprehensive econometric analysis of a varying parameter density gradient (VCM) model in which the functional form, together with the variance components of an implied error structure, is freely estimated. We generalize a recent model of Johnson and Kau in which certain city- and tract-specific socioeconomic variables are introduced to depict the dynamic process of urban growth. Since these variables can be either controlled by policy actions or projected on the basis of time, we reexamine the usefulness of VCM framework for both forecasting and policy simulation purposes.