Conditional Skewness in the Cross-Section of Stock Returns and its Macroeconomic Foundation

Recent literature places increasing emphasis on the asymmetry of stock returns, or skewness, and the role it plays in financial decision-making. On the one hand, several theoretical models have been developed that extend the standard portfolio-choice asset pricing paradigm in the presence of skewness, as in Mitton and Vorkink (2007) and Dahlquist, Farago, and Tédongap (2017). On the other hand, empirical studies have shown that the skewness of stock returns extends over long horizons and may contain useful information to forecast future stock returns, see Boyer, Mitton, and Vorkink (2010) and Conrad, Dittmar, and Ghysels (2013). Much of the extant studies, however, rely on the standard moment-based estimator of skewness, which is known to be particularly sensitive to outliers. This estimator is therefore not particularly suited to examine the skewness of individual stocks, which exceptionally noisy. A notable exception is the quantile-based estimator of conditional skewness of Ghysels, Plazzi, and Valkanov (2016), which use daily data to forecast the quantiles of long-horizon returns and construct an implied estimator of the third moment of aggregate stock returns. In this project, we want to deepen our understanding of the dynamics of conditional skewness in the cross-section of stock returns, and its link to macroeconomic fundamentals. In particular, we propose to extend the analysis in Ghysels, Plazzi, and Valkanov (2016) to explore how skewness varies in the cross-section of international individual stock returns. To be precise, we aim to address the following three questions. First, we would like to understand how do the up-side growth potentials and downside risk of individual stocks vary in the cross-section of stocks, and how strong are their commonalities at the country and industry level. This question has been explored so far only in the mean-variance context, see Heston and Rouwenhorst (1994). Using conditional skewness as conditioning variable in a portfolio allocation setting would allow us to measure in utility terms how the potentials for portfolio diversification at the micro/industry/country level change when taking cross-sectional variation in the conditional third return moment into account. The second question we aim to explore is how predictable patterns in skewness can be exploited to enhance the performance of strategies that are prone to crash risk. In particular, it is well known that the momentum strategy of buying stocks with recent positive top performance (winners) while shorting stocks with poor performance (losers) is subject to severe drawdowns, that is, it exhibits large negative skewness. If high-frequency information can be fruitfully used to capture time variation in individual stock’ skewness, this suggests that a strategy that sorts stocks based on the joint combination of predicted skewness and recent performance would have the potentials to avoid much of the downside risk. A final question we seek to examine is the nexus between skewness in equity returns and fundamentals. Starting with seminal work of Schwert (1989), several studies have investigated the relation between stock market volatility and macroeconomic activity. However, recent papers in macroeconomics have documented that business cycle fluctuations are unconditionally negatively skewed and procyclical. Using the above-mentioned estimation method would provide a novel framework to relate (at least, statistically) the downside risk of macro shocks (such as, the skewness of GDP growth or nonfarm payroll), industry-level and firm-level fundamentals (sales, cash flows, R&D expenditures) and stock returns.