Determinants of the incidence of college graduates in United States cities

This paper examines both economic and amenity variables that may be used to explain the percentage of college educated persons that are maintained and attracted to U.S. metropolitan areas. A geographic information system (GIS) program identifies variables that visually coincide with population clusters of the college educated. OLS regression analysis of 304 metropolitan statistical areas in the U.S. for 1990 and 2000 is used to predict the influence of economic and amenity measures in a cross sectional model for each of the two years and, based on the two period panel data, a "fixed effects" model that predicts the change in college-educated population density among U.S. cities over the decade. Average wages and home values have a positive effect on college-educated population concentration. Unemployment rates and number of years in owner-occupied housing have a negative effect. Air quality and crime show negative coefficients in the panel estimation.