Two-side CVaRs and Cross-sectional Expected Stock Returns: Evidence from the Chinese Stock Market

Abstract

Recently, tail risks have attracted much attention in the literature for their role in predicting the cross-sectional expected returns of stocks. Using a modified conditional value at risk (CVaR), the extreme loss and gain of stocks can be measured using the left-tail CVaR- and the right-tail CVaR+, respectively. The left and right tail CVaRs are unified as two-side CVaRs that correspond to the two-side tails of returns. We empirically examine the relationship between two-side CVaRs and the cross-sectional expected returns of stocks, obtain the findings with significantly negative relations, and both economically and statistically. The empirical results are robust even after controlling for firm size, idiosyncratic volatility, liquidity risk, downside beta, and the maximum daily return in the previous month (MAX). The pricing powers of two-side CVaRs are strongly significant and cannot be explained by the Fama-French three- and five-factor models.

Keywords: Asset pricing, extreme loss and gain, two-side CVaRs, tail risk

JEL Classification: G11, G17, G12