The executive compensation literature presumes that shareholders offer risk‐averse managers stock options to entice them to take on more risk, resulting in riskier investment decisions and thus a greater return on investment. However, recent empirical work challenges this assumption, and theoretical research even argues that high levels of option‐based compensation for generally under‐diversified managers may actually lead to greater risk aversion . We evaluate the incentive structure of employee stock options by examining the level of R&D investment and the return on that investment conditional on the portfolio “vega,” which captures the sensitivity of option value to stock price volatility. Our results suggest that both investment in R&D and the return on R&D, as measured by future earnings and patent awards, varies concavely with vega. That is, low to moderate levels of vega correspond to increasing investment in and returns on R&D, consistent with vega inducing more profitable investments, but marginal returns decline as vega increases. Collectively, these results, bolstered by several supplemental analyses, suggest that this surprising relation between vega and risky investment is driven by greater risk aversion at higher levels of vega. Overall, our results imply that employee stock options may not always align the incentives of managers and shareholders.

We examine commonly used indicators of aggressive non-GAAP exclusions and find that the majority perform poorly at identifying low-quality exclusions in terms of decision usefulness for investors. We propose a new firm-quarter-specific indicator that identifies instances in which GAAP earnings quality is high (i.e., when firms have less need to provide non-GAAP metrics) but managers disclose non-GAAP earnings anyway. Our new indicator is easy to calculate, requires minimal data, and performs far better at identifying low-quality exclusions than indicators used in prior research. Using our indicator, we find instances in which managers exclude earnings components that are decision useful, consistent with regulators’ concerns about the quality of some non-GAAP earnings disclosures. Our results are robust to a variety of specification checks. Data Availability: Data are derived from a combination of publicly available sources referenced in the article and third-party subscription data bases.