Research

We investigate whether the tails of firm-level idiosyncratic return distributions are driven by common shocks. We use quantile factor analysis to extract such common idiosyncratic quantile factors with asymmetric pricing effects, and we find a significant premium for innovations to the lower-tail factor: high-exposure stocks outperform low-exposure stocks by approximately 7–8% per year. This premium remains significant even when controlling for standard factors, idiosyncratic volatility, and tail-risk measures. The downside factor strengthens when intermediary capital is weak and market liquidity is low, and the associated increase in risk predicts higher aggregate market excess returns.

This article investigates how two important sources of risk—market tail risk (TR) and extreme market volatility risk—are priced into the cross-section of asset returns across various investment horizons. To identify such risks, we propose a quantile spectral (QS) beta representation of risk based on the decomposition of covariance between indicator functions that capture fluctuations over various frequencies. We study the asymptotic behavior of the proposed estimators of such risk. Empirically, we find that TR is a short-term phenomenon, whereas extreme volatility risk is priced by investors in the long term when pricing a cross-section of individual stocks. In addition, we study popular industry, size and value, profit, investment, or book-to-market portfolios, as well as portfolios constructed from various asset classes, portfolios sorted on cash flow duration, and other strategies. These results reveal that tail-dependent and horizon-specific risks are priced heterogeneously across datasets and are important sources of risk for investors.

Do the many measures of asymmetric risk identify one downside factor or several priced dimensions? We study compensation for asymmetric risk in U.S. equities, separating robust premia, distinct return variation, and priced dimensions. Five measure families organize around three recurring priced dimensions rather than a single downside factor. Multiple dimensions receive compensation across periods and portfolio constructions, although their relative pricing strength varies over time. The resulting three-factor benchmark is not jointly spanned by leading factor models. The dimensions also exhibit distinct predictive relations with aggregate conditions. The results distinguish new measures from new priced risks.