Discussion of “CEO compensation and corporate risk-taking: Evidence from a natural experiment”

•Gormley, Matsa, and Milbourn examine the design and causal effects of CEOs' equity portfolio incentives on firm risk.•The distinction between various treatment effects is important for causal inference.•Partial identification and sensitivity analysis are well-suited for providing causal inferences.

Gormley, Matsa, and Milbourn (in this issue) examine the design and causal effects of CEOs' equity portfolio incentives on firm risk in a novel research setting in which certain firms experience a large exogenous shock that increases their left-tail risk and reduces their investment opportunities. Gormley et al. find that boards and CEOs both make adjustments to CEOs' equity portfolios following the shock. They also find that CEOs with more convex equity portfolios (i.e., Vega) prior to the shock reduce risk less following the shock. Despite certain measurement and identification concerns, Gormley et al. is an innovative attempt to address an important and challenging research question. Partial identification and sensitivity analysis an important class of techniques that are well-suited for providing causal inferences about Gormley et al.'s and other important research questions that are impeded by endogeneity concerns.