Asymmetry Matters: A High-Frequency Risk-Reward Trade-Off

Abstract

Expected returns should not only include rewards for accepting the risk of a potential downside loss, but also discounts for potential upside gains. Since investors care differently about upside gains versus downside losses, they require a risk premium for bearing the relative downside risk. We validate this perception empirically as well as theoretically and show that conditional asymmetry forecasts equity market returns in the short run. The results hold not only for different return series and different data frequencies (daily and intra-daily) but also for various subsamples. Our short-term expected return predictor, the asymmetric realized volatility measure, captures more variation in equity returns than the variance risk premium, a forward-looking measure, or the price-earnings ratio, and can easily be extracted from realized return series. We formalize this intuition with a closed-form asset pricing model that incorporates disappointment aversion and time-varying macroeconomic uncertainty.

Keywords: realized variance, realized semivariance, stock market return predictability, asymmetric realized volatility, variance risk premium, high-frequency data, equilibrium asset pricing

JEL Classification: G1, G12, G11, C1, C5