Career Implications of Having a Female-Friendly Supervisor

Supervisor attributes, such as race or gender, are thought to play an important role in employee hiring, career advancement and turnover even in the absence of discrimination. For example, supervisors may be better able to infer the true ability of a worker if they have a common attribute (Cornell and Welch 1996) or it could be that mentoring is an important determinant of employees' career trajectories and is more effective if the workers and supervisors are of the same type (Athey, Avery and Zemsky 2000). Embedded in this idea are important policy implications for alleviating the gender wage gap and other adverse career outcomes for females. In our paper we introduce a type-based mentoring model of worker productivity and turnover in which human capital production is contingent on the worker's gender and the supervisor's type. We allow for a flexible definition of "type" based on the supervisor's attitude toward working with females. Using a longitudinal data set in which workers with a high level of managerial responsibility are observed at multiple establishments over time, we construct a measure of revealed supervisor female-friendliness above and beyond the institution-specific culture by comparing changes in the gender composition of workers at lower levels of the firm. We adjust the measure to account for its potential endogeneity with respect to the outcomes that we study and use it to test empirically the predictions of a type-based mentoring model against a model of pure taste-based discrimination. We add to the literature by proposing the idea that it may be necessary to be more flexible when defining "type" in the mentoring relationship. We argue that, in addition to demographics, supervisors can be characterized by a more complexly defined inherent attitude toward working with and mentoring females, which can vary within observable supervisor characteristics, and that the existing literature has left unexplored areas by limiting its focus on leaders' observable characteristics. Implementing our definition of female-friendliness empirically requires us to use a novel data set with information on both supervisors and lower-level workers, in which high-level managerial employees are followed across establishments. We link the theoretical idea of type-based mentoring to observed career outcomes and test whether females are more likely to benefit in terms of career progression when matched with a supervisor with a more favorable attitude toward mentoring women. While our study focuses on a very specific labor market setting, described in Section 3, our findings offer strong support for the importance of type-based mentoring in the labor market and in addition should motivate researchers to use broader definitions of "type" that include harder to infer attitudes in addition to directly observable supervisor and worker attributes such as gender.