Association for Public Policy Analysis & Management

Growing literatures in education economics and representative bureaucracy consider the effects of government workers’ racial and other demographic characteristics in the context of providing government services. For example, many studies consider whether clients experience better outcomes when served by a government employee who shares their own race, ethnicity, gender, or sexual orientation.

Largely absent from existing quantitative literature is a consideration of how these social identities can assume greater or lesser importance depending on one’s social context, although a few studies in the public administration literature have started to grapple with this issue (e.g., Keiser et al. 2002). When thinking about government service provision, one important determinant of identity salience may be the extent to which a particular identity is associated with unique service needs. For example, in neighborhoods or cities where being Latino is strongly associated with the need for foreign language accommodations, there may be particularly large benefits (average treatment effects) associated with being served by a Latino government employee.

This paper argues that the importance of a demographic characteristic (such as ethnicity) depends not only on the policy area (the main focus of prior public administration work looking at identity salience) but also on the degree to which the demographic characteristic is associated with substantial differences among individuals served by the bureaucracy. This argument is tested using a novel measure of the degree to which ethnic/racial identities are associated with differences in terms of socioeconomic status and English language proficiency. A dataset of California public schools is assembled at the school-by-grade-by-year level. The results of a panel data analysis with school-grade fixed effects indicate that for multiple ethnic/racial groups, the benefits associated with a student-teacher match in race/ethnicity are stronger when there are larger differences among students of different racial/ethnic identities. The finding suggests the importance of considering the local meaning of social identities when trying to understand bureaucratic representation.