Measuring segregation when hierarchy matters

This paper considers the problem of measuring segregation when groups form a hierarchy whereby some groups have greater economic status than others. While existing measures of segregation address the case where people are unequally distributed across groups with the same economic status, concern often focuses on groups with different status, e.g., occupational segregation where women have limited access to high wage occupations. This paper first defines a class of segregation indexes that encompasses both the same economic status and different economic status case. It then proposes two methods for incorporating economic status into empirical work. One is to rank groups from highest to lowest economic status and apply the dominance criteria in Theorem 2. The other is to invoke a cardinal measure of group economic status and then compute a numerical index. Finally, a numerical index of segregation is introduced, and both methods are used to analyze U.S. occupational segregation by gender and ethnicity.