Working Papers

Leadership and Gender Discrimination: Experimental Evidence from Ethiopia

with Shibiru Ayalew and Ketki Sheth. [Draft coming soon]

Globally, women are underrepresented in top management. Using a novel "lab in the field" experiment in Ethiopia, we test whether leader gender influences the way subjects respond to leadership, and whether providing information about the leader's underlying ability changes this gender gap. We find evidence for discrimination against female leaders when subjects are given no information about leader's ability: subjects are less likely to follow female leaders. In contrast, when the leader is presented as highly trained and competent, the gender gap is reversed: subjects are \textit{more} likely to follow women than men. The findings are consistent with a dynamic model of discrimination in which women are, on average, less likely to obtain signals of high ability. An implication of the model is that we should not observe gender discrimination among the highly educated in Ethiopia, where there are large gender gaps in educational attainment. Consistent with this, we demonstrate a lack of discrimination in a resume evaluation experiment for a management position. And, using a large sample of university administrative employees, we show that there is no gender wage gap among the highly educated.

In markets where suppliers possess important, unobservable information, theory predicts that credible certification can improve welfare. I study sex work “legalization and regulation”, a common program that provides government certification of female sex workers’ health. Adapting a standard information disclosure model, I show that when sex workers’ health status is unobservable, certified providers should earn higher prices, and as long as certification costs are low, all suppliers should obtain certification. I test these predictions using a novel randomized experiment among uncertified female sex workers in Dakar, Senegal. In a randomly selected treatment group, I offered a one-time certification incentive that reduced monetary certification costs to zero. Contrary to the theoretical prediction, take-up of this incentive was very low: only 7 percent of the treatment group obtained certification, relative to 2 percent in control. Moreover, I find no evidence for a large price premium to certified sex workers. To explain this finding, I present new evidence that sex workers’ health status is partially observable and already priced into the sex market: sex workers earn 19 percent lower prices when they have visible STI symptoms. As a result, the value of certification is limited for both clients and sex workers. These results, combined with evidence for high non-monetary costs related to stigma, suggest that complementary services for uncertified sex workers are needed to achieve sexually transmitted infection control.

Standard expected utility models exhibit risk vulnerability: greater status quo uncertainty induces more risk averse decision-making. Empirical research has documented risk vulnerability with monetary lotteries, mostly in hypothetical or laboratory settings. Using data from a violent conflict in Ciudad Juarez, Mexico, I investigate how dramatic changes in mortality risk affect real-world behavior. Consistent with risk vulnerability, more intense conflict substantially reduces sexual risk-taking among female sex workers who inject drugs and among their clients. Evidence suggests client preferences drive the result. I show that changes in mortality risk have real implications for risky behavior and health.

Research in Progress

Learning about Commitment: Evidence from Contingency Management Programs, with Erin Giffin

The Impact of Stigma on Risk Preferences


Padian N, McCoy S, Manian S et al. (2011). Evaluation of Complex, Combination HIV Prevention Programs: Essential Issues. Journal of Acquired Immune Deficiency Syndromes.