A Note on John Kain's "Housing Segregation, Negro Employment, and Metropolitan Decentralization"

Abstract

The purpose of this note is (1) to criticize Kain's methodology with regard to testing hypotheses (2) and (3), and (2) to present alternative results that conflict with hypothesis (2). Kain's data are for geographic areas in Detroit (1957) and Chicago (1956), with ninety-eight observations for each city. With these data he runs regressions using the Negro percentage of total employment in the area as the dependent variable w. The independent variables are R, the Negro percentage of (employed) residents in the areaused as a proxy for employers' propensity to discriminate, and d, the airline distance from the area to either the nearest Negro residence area (more than 2 percent Negro) or to the nearest point in the Negro ghettowith this variable used as a proxy for transportation costs and effects of such costs on job information. The regression coefficients have the expected signs and are statistically significant. Over 75 percent of the total variance is explained for Chicago, and over 35 percent for Detroit. Kain estimates the effect of housing segregation on the level of black employment by assuming that with no housing segregation the proportion of Negro workers would be the same mR for each geographical area. Consequently, for each area R=mR and d=0. Substituting these values into the regression equation (1) (1) w==aR+bd+c,