A model forecasts skewness in US and international equity markets and enumerates the economic benefits to various investor classes. The authors demonstrate how forecasting skewness can improve risk-adjusted returns of the popular low-volatility strategy as well as improve portfolio diversification in down markets.
The authors expand on the dearth of articles that forecast skewness by focusing on a much longer period for the US stock market and then extend their dataset to seven other US and international equity classes. They investigate and demonstrate the importance of forecasting left-tail risk for investment strategies. Their study provides supportive empirical evidence and explains how investors can avoid large negative returns—without lowering Sharpe ratios—by applying skewness forecasts to common diversification strategies.
How Is This Research Useful to Practitioners?
Traditional portfolio strategies follow the notion that market returns have a normal distribution. But the concept of tail risk suggests that the distribution of returns is negatively skewed, with fatter tails, which is apparent in all the equity classes studied.
The authors forecast skewness for the ensuing six months and discover that it is negatively correlated with past returns and acceleration of returns (i.e., a return growth measure), which means that good past performance results in a negative skewness forecast and increases the probability of large negative returns. The study confirms that acceleration of returns and past 12-month average returns are good predictors of future skewness for each of the common equity classes but that past skewness is not a reliable predictor of future skewness. The latter observation should be of particular concern to industry practitioners owing to their propensity, according to the authors, to rely on past skewness when forecasting future skewness.
The study also uncovers a trade-off between volatility and skewness: High historical volatility forecasts lower future left-tail risk, whereas low historical volatility forecasts higher left-tail risk. This finding should be particularly troubling for the popular low-volatility strategies, which may bear more risk of complete failure (i.e., negative skewness).
The authors conclude that mean–variance investors should include skewness forecasts in their research. Diversification strategies, which rely on the assumption of normally distributed returns, should incorporate protection against skewness loss (i.e., non-normal returns) and underweight particular markets if the forecasted skewness is negative. Likewise, opportunistic traders can consider entering the market when skewness is very positive or after the market has suffered a huge loss.
How Did the Authors Conduct This Research?
The study is based on daily return data for the US market over the July 1926–December 2014 period, obtained from the CRSP Value-Weighted Stock Index Total Return series. For their asset class analysis, the authors use daily returns for the Russell, MSCI, and FTSE NAREIT indexes covering the early 1990s to 2014, obtained from Morningstar Direct.
The authors focus on 6-month skewness forecasts for the US equity market and run regression models to measure its sensitivity to independent variables, including historical 6-month skewness, past 12-month average returns, acceleration of returns, and standard deviation. Factors with negative sensitivity coefficients are interpreted as factors that increase the probability of incurring large negative returns. For each component, the statistical significance of the correlation is confirmed using t-tests.
In a time-series analysis, the authors run multiple regressions and look for structural changes in the t-statistics of the regression coefficients. The authors measure the predictive power of independent variables in each of seven popular equity classes (e.g., small and large cap, value/growth stocks, emerging markets, and REITs) and confirm relationship significance using t-tests.
The notion that asset returns have a normal distribution is becoming an obsolete theory. Financial market turbulence is difficult to capture using standard variation analysis, but its impact on portfolios is too substantial to ignore. The authors provide a structured approach to predicting and avoiding loss in an overheated market by forecasting skewness. The study would be more relevant if it included an assessment of strategy effectiveness in predicting a particular event—for example, the 2008–09 financial crisis or, more recently, the 2015–16 turbulence in China’s markets. The study is comprehensive and may be useful for investors who want to better diversify their portfolio or take advantage of the economic recovery.