Adjusted Long-Term Volatility and Stock Return Predictability

We design an adjusted long-term volatility (ADJ_LV) indicator by removing the interference information of short-term volatility from the simple long-term volatility indicator to investigate the level of predictive ability that ADJ_LV has for stock returns. In a sample spanning 2000 to 2019 and models considering 19 popular predictors, ADJ_LV positively predicts the next-month returns of the S&P 500 index, with the corresponding univariate model displaying the best forecasting performance with an adjusted in-sample R-squared of 3.825%, out-of-sample R-squared of 3.356%, return gains of 5.976%, certainty equivalent return (CER) gains of 4.708 and Sharpe ratio gains of 0.394. Adding ADJ_LV as an additional predictor to the other 19 univariate models generates significantly better forecasting performance in both the in-sample and the out-of-sample results. Furthermore, we find that ADJ_LV also has predictive power for long-term (1-12 month) stock returns and can forecast returns of industry portfolios and portfolios formed by size, book-to-market ratio, operating risk and investment risk. The predictive ability of ADJ_LV for equity returns is robust in forecasts of the return of the DJI index and to the use of an alternative estimated ADJ_LV