Realized skewness and the short-term predictability for aggregate stock market volatility

Abstract

Forecasting stock volatility is of great interest to academics and practitioners because volatility has important implications for many areas such as risk management and portfolio allocation. Recent studies show that economic variables fail to predict stock volatility beyond lagged volatility. In this paper, we find that realized skewness shows significant predictive ability for future realized volatility. We use the daily price data of the S&P 500 index over a long sample period spanning 1928 to 2019 to construct skewness predictors, and reveal the negative relationship between realized skewness and volatility. The realized skewness significantly outperforms the benchmark of the autoregressive model in short horizons and contains different predictive information from macroeconomic indicators and volatility of volatility. The predictive ability of skewness is also found in most industry portfolios. The realized skewness predicts volatility mainly through risk transmission channel, and then through the business cycle channel.