Variability of local economic conditions underlie, in part, the default and prepayment risks of mortgages originated in a metropolitan area. In this study we examine the benefits of diversifying across metropolitan areas for reducing these risks. Employment data for the thirty largest metropolitan areas in the United States, divided into eight industry groups, are analyzed with the aid of factor and principal component analysis to determine if the variances of employment changes across the thirty areas are independent. Independence is investigated to assess the potential for diversification. Mean-variance portfolio analysis is then used to measure the gains from geographic diversification vis-a-vis a set of several alternative strategies for diversification. We conclude that geographic diversification is an important mortgage portfolio objective and that mean variance strategies outperform the alternative strategies tested here.