New paper

New paper (w/ Zhengyang Xu): Dynamics of Subjective Risk Premia. While objective risk premia extracted from standard in-sample predictive regressions are highly cyclical, we show that subjective risk premia extracted from individual and professional return forecasts in stock, bond, FX, and commodity futures markets are close to acylical. Out-of-sample forecasts of excess returns get close to matching the weak cyclicality of subjective risk premia.

New paper: Do Survey Expectations of Stock Returns Reflect Risk-Adjustments? with Klaus Adam and Dmitry Matveev. We reject the hypothesis that survey expectations of returns reflect risk-adjustments or marginal utility weighting (e.g., risk-neutral expectations). Theories of risk-adjusted survey expectations therefore do not help to reconcile survey expectations with expected returns implied by rational expectations asset-pricing models and return predictability regressions.

 

New paper: Judging

New paper: Judging Bank Risk by the Profits They Report, with Ben Meiselman and Amiyatosh Purnanandam. We show that high rates of bank profitability (ROA, ROE) in good times are predictors of systematic tail risk exposure in bad times—just like the yield of a risky bond portfolio in good times tends to be an indicator of systematic tail risk exposure.

New paper: Socioeconomic Status and Macroeconomic Expectations, with Sreyoshi Das and Camelia Kuhnen. We show that individuals of lower socioeconomic status (SES) have more pessimistic macroeconomic expectations. Helps explain, e.g., lower rates of stock market participation among low-SES individuals. The beliefs wedge between low- and high-SES individuals shrinks in recessions, consistent with a model in which low-SES individuals neglect good (macroeconomic) states of the world.

New paper: Shrinking the Cross-Section

New paper: Shrinking the Cross-Section, with Serhiy Kozak and Shrihari Santosh. We use SDFs to summarize the cross-section of expected stock returns in a high-dimensional setting with a huge number of characteristics and their interactions. Uses tools from machine learning combined with economically motivated objective functions.