Motivated by the implications from a stylized equilibrium pricing framework, we investigate empirically how individual equity prices respond to continuous, or "smooth," and jumpy, or "rough," market price moves, and how these different market price risks, or betas, are priced in the cross-section of expected returns. Based on a novel highfrequency dataset of almost one-thousand individual stocks over two decades, we find that the two rough betas associated with intraday discontinuous and overnight returns entail significant risk premiums, while the intraday continuous beta is not priced in the cross-section. An investment strategy that goes long stocks with high jump betas and short stocks with low jump betas produces significant average excess returns. These higher risk premiums for the discontinuous and overnight market betas remain significant after controlling for a long list of other firm characteristics and explanatory variables previously associated with the cross-section of expected stock returns.JEL classification: C13, C14, G11, G12 (2004), the habit persistence model of Campbell and Cochrane (1999), and the rare disaster model of Gabaix (2012), as special cases obtained by further restricting the functional form of the pricing kernel and the set of other priced risk factors.
1The statistical theory underlying our estimation of the separate betas builds on recent advances in financial econometrics related to the use of high-frequency intraday data and so-called realized volatilities. Bollerslev and Zhang (2003), Barndorff-Nielsen and Shephard (2004a), and Andersen et al. (2005, 2006, in particular, have previously explored the use of high-frequency data and the asymptotic notion of increasingly finer sampled returns over fixed time intervals for more accurately estimating realized betas. In contrast to these earlier studies, which do not differentiate among different types of market price moves, we rely on the theory originally developed by Todorov and Bollerslev (2010) for explicitly estimating separate continuous and discontinuous betas for the open-to-close active part of the trading day, together with overnight betas for the close-to-open returns. 5Our actual empirical investigations are based on a novel high-frequency dataset of all the 985 stocks included in the S&P 500 index over the 1993-2010 sample period. We begin by estimating the three separate betas as well as a standard CAPM regression-based beta for each of the individual stocks on a rolling one-year basis. Consistent with the basic tenets of the simple CAPM, we find that sorting the stocks in our sample on the basis of their betas, results in a positive return differential between the High-and Low-beta quantile portfolios for all of the four different beta estimates. 6 However, even though all of the return differentials are quite large numerically, the difference in the monthly returns between the High-and Low-beta portfolios constructed on the basis of the standard CAPM betas is not significantly different from zero at conventional lev...