Abstract: I construct firm-quarter measures of perceived risk and ambiguity from approximately 74 million management sentences in earnings calls from 2006 to 2025. Risk captures explicit uncertainty around a retained forecast, while ambiguity captures statements that uncertainty prevents management from forming or providing one. The measures are positively correlated but far from equivalent and share little variation with existing measures of uncertainty. Firms respond to them along different margins. Both predict lower investment, but the response to risk is stronger when assets are harder to redeploy, and only ambiguity predicts higher cash holdings. Investors also price them differently. Risk predicts a valuation discount and a higher required return, whereas ambiguity's association with required returns depends on the firm's liquidity, rising for firms with little net liquidity and falling for firms with ample net liquidity. Pooling the two conceals the larger investment response to ambiguity and its cash response. The findings suggest that the form of uncertainty matters for how firms respond and how investors price their exposure to it.

Abstract: Geopolitical risk reaches a firm through multiple channels, so its economic significance depends not only on how saliently it is disclosed but also on how it is interrelated with a firm’s other risks. Using over three million risk factor disclosures from SEC 10-K filings between 2010 and 2024, we develop network-based measures of geopolitical risk along three dimensions: whether the firm’s geopolitical risk profile resembles the cross-sectional average, how closely geopolitical risk is connected to other risk topics, and whether geopolitical risk is linked to the firm’s most systemically important risk topics. First, we find significant cross-industry heterogeneity in how geopolitical risk is interrelated with other risk topics. However, environmental and cybersecurity risks are the two topics most strongly and consistently related to geopolitical risk. Second, a firm’s risk network structure moderates the effect of geopolitical risk on future investment and valuation. The investment decline is sharpest for firms whose geopolitical risk is connected to their most broadly interconnected risks and the valuation decline is largest for firms whose geopolitical risk reaches broadly across their other disclosed risks. Third, using the Russian invasion of Ukraine as a plausibly exogenous shock, we show that a significant geopolitical event can reconfigure firms’ risk networks. Firms whose geopolitical risk was more interrelated with natural-resource and macroeconomic risks pre-war were most likely to report war-attributed asset impairments. Overall, this paper highlights that researchers as well as practitioners should adopt a network-based conceptualization of geopolitical risk to account for its full impact.

Abstract: Using the universe of account-level transactions in KOSPI 200 index options from 2010 to 2014—then one of the world’s largest option markets—we examine how retail demand shapes option prices and investor performance. Retail investors buy to open over 80% of their positions, calls and puts alike, whereas institutional opening trades are balanced between buys and sells. The buying tendency is a highly persistent retail investor trait: reforms that raised the cost of buying reshaped participation but barely moved this tilt, consistent with leverage demand rather than directional views or capital constraints. Options with greater retail net demand have higher implied volatility and larger variance risk premia. Following a 2011 reform that sharply reduced retail participation, implied volatility fell 5–7% where retail demand had concentrated. Buy-opened positions lose on average for retail and institutions alike, reflecting the side taken rather than timing or contract choice. Because retail investors overwhelmingly buy while institutions supply most opening sales, 76% of retail accounts lose money over their observed option-trading histories, whereas more than 60% of institutional accounts profit. The findings reveal a hidden cost of leverage: collective demand raises the price of the convex exposure retail investors seek.

Abstract: We propose a new method to separately quantify investor disagreement and subjective uncertainty at the firm level using equity analyst forecasts. Our approach exploits variation in how much forecast dispersion falls after the arrival of a signal that most agents use and interpret in the same way. Intuitively, for a given level of disagreement in point forecasts, a larger decrease in dispersion after the signal indicates greater subjective uncertainty in agents' prior beliefs. Using these measures, we document differences in the economic roles of disagreement and uncertainty. In the cross-section, stocks with higher disagreement earn lower subsequent returns and exhibit higher trading volume. Both effects are attenuated when uncertainty is high. In contrast, higher uncertainty is associated with higher expected returns and lower trading volume. Stock return volatility is strongly related to disagreement but only weakly related to uncertainty. We extend the extrapolative bubbles framework to show that the model's predictions closely match our empirical findings and provide intuition for the observed relationships. Finally, firm characteristics are more closely linked to disagreement than to uncertainty: smaller, less profitable, and more R&D-intensive firms exhibit systematically higher levels of disagreement.

Abstract: I show that dispersion among equity analysts' forecasts increases following earnings announcements. I provide evidence that this is due to information selection: analysts rely on different subsets of public information revealed in earnings announcements when revising their forecasts in a high-dimensional setting. Using text data from analysts' reports, I show that analysts cite different information when they revise their forecasts following earnings announcements, providing evidence of information selection. Results from elastic net regressions corroborate these findings but further imply the existence of a common benchmark to which analysts can compare their forecasts. The existence of a common benchmark explains the decrease in forecast dispersion in the periods between earnings announcements. To formalize these findings, I develop a simple model of forecast-making processes under a common benchmark to explain the dynamics of forecast dispersion. My results highlight the importance of information selection as a driver of disagreement, especially in high-dimensional environments.