Multivariate Crash Risk

Abstract

This paper investigates whether multivariate crash risk (MCRASH), defined as exposure to extreme realizations of multiple systematic factors, is priced in the cross-section of expected stock returns. We derive an extended linear model with a positive premium for MCRASH and we empirically confirm that stocks with high MCRASH earn significantly higher future returns than stocks with low MCRASH. The premium is not explained by linear factor exposures, alternative downside risk measures or stock characteristics. Extending market-based definitions of crash risk to other well-established factors helps to determine the cross-section of expected stock returns without further expanding the factor zoo.

Keywords: Asset pricing, Non-linear dependence, Crash aversion, Downside risk, Tail risk, Lower tail dependence, Copulas

JEL Classification: C58, G01, G11, G12, G17