It is a tenet of urban planning that transportation networks help shape the spatial configuration of cities. In the case of heavy rail systems, a common belief is that building a subway system will promote employment and population density, thereby discouraging urban sprawl and its negative consequences. This dissertation examines the impact of the Washington Metro rail system in 1990 and 2000 on the distribution of employment and population in two counties in the Washington, DC metropolitan area--Montgomery County and Prince Georges County. It asks whether employment and residential construction increased more rapidly near Metro rail stations than in other parts of the metropolitan area. It also examines the impact of the Metro on the socio-demographic composition of population near Metro stations.

Evaluating the impact of the Metro system on employment and population density is complicated by the fact that stations along the Metro line may be located in areas of high population and/or employment density to begin with, or in areas with significant amounts of developable land available. To deal with this issue I use a propensity score matching estimator. The technique is an improvement over the traditional methods of evaluation as it acknowledges the endogeneity of the location of Metro stations. Furthermore, matching estimators relax the functional form assumptions of OLS estimators.

The research finds statistically significant impacts on employment and overall development density from proximity to a Metro station and does not find consistent impacts on population or dwelling unit densities. However, for Prince George's County a negative impact on the percentage of the population belonging to a minority is found. The results also suggest that impacts on development are greater closer to the station than farther away and that they are greater the longer the stations have been in operation.