Does Increasing the Riskiness of Choices Widen Gender Gaps?

We isolate the causal effect of changing the riskiness of choices on the gender gap in real world outcomes. We do so by exploiting a national reform to the regrade system of Norwegian universities that generated exogenous variation in the probabilities of the outcome of regrade requests. Using unique individual-level administrative data, we show that both the expected value of a regrade request, as well as the downside risk, increased substantially as a result of the policy. We then show how the ostensibly gender-neutral policy substantially increased gaps between men and women because they differed in their risk preferences and beliefs. Specifically, the exogenous change in the riskiness of requesting a regrade augmented the gender gap in regrade requests by 90%. We demonstrate that this has important implications for students through its impact on their grades. We disentangle the relative importance of mechanisms through auxiliary reduced-form analyses, structural estimation, and a supplemental laboratory experiment. Taken together, our mechanism analyses suggest that gender differences in risk preferences and loss aversion are not sufficient to explain the increased gender gap after the policy change, and that gender differences in confidence or beliefs are necessary to rationalize the data and more broadly consistent with the patterns we observe.