"Hedging economic uncertainty from the cross section of stock returns," with Fengtian Yang and Hector Calvo-Pardo
Abstract: We examine the pricing of economic uncertainty in the cross-section of stock returns. Uncertainty is proxied by innovations to the macroeconomic and financial volatility measures in Jurado, Ludvigson and Ng (2015) and Ludvigson, Ma and Ng (2021). In contrast to existing literature, a negative uncertainty risk premium is found in calm periods, turning positive in turbulent ones. These findings are rationalized by means of a hedging portfolio that delivers positive returns in turbulent periods to compensate for the cost of insuring the portfolio in calm periods. We also provide statistical evidence of the uncertainty factors' added value and their relative predictive power across uncertainty regimes.
"Beyond the Short Run: The Term Structure of Implied Moment Risk Premia," with Meng Zhang
Abstract: This paper investigates the pricing of the term structure of risk-neutral moments. We construct slope factors by differencing six-month against one-month, and twelve-month against six-month implied moments to isolate incremental distributional risks. We document distinct pricing patterns across horizons: while variance commands a persistent negative premium, the pricing power for skewness and kurtosis is concentrated in the medium-term slope rather than the short-term level. Notably, the medium-term skewness slope carries a significant positive premium. Consistent with the ICAPM, these risk premia align with the factors' predictive power for future market returns. We also uncover a fundamental macro-dichotomy: short-term variance tracks real economic activity, whereas the term structure of higher-order moments is linked to monetary conditions and credit stress. These findings are robust to alternative test assets, the exclusion of small-cap stocks, and controls for liquidity and momentum.
"Realized partial probability measures," with Jesus Gil-Jaime
Abstract: This article introduces a novel decomposition of the realized partial variance originally proposed by Bollerslev et al. (2022) into two components: (i) an empirical probability measure capturing the frequency with which intraday returns fall into fixed regions of the distribution, and (ii) a realized variance term computed within those regions. With fixed thresholds, these probability components retain substantial time series variation rather than collapsing to constants; this variation is highly informative regarding the dynamics of realized variance. Leveraging these insights, we develop forecasting models that augment standard heterogeneous autoregressive realized volatility specifications with these empirical probability components. Our models deliver superior out-of-sample performance across both U.S. equity data and simulated stochastic volatility environments.