Volatility forecasting: The role of lunch-break returns, overnight returns, trading volume and leverage effects

Abstract

This article extends the HAR-RV model to enable it to forecast volatility by including lunch-break returns, overnight returns, trading volume and leverage effects in the Chinese stock market. The findings show the significant role of additional leverage effects, captured by negative lunch-break returns and negative overnight returns, in volatility forecasting, in addition to the trading volume’s impact. Moreover, there is a strong significance of the usual leverage effects, which turn out to be persistent even for SHCI. Surprisingly, squared lunch-break returns, measured as additional volatilities during the lunch-break period, have a large long-run impact on the volatility for SHCI but not for SZCI. This new empirical evidence is robust to alternative realized measurements and unconditional variance, and, in particular, confirms the impact of intermittent trading, captured by the returns and volatilities outside the trading hours. Overall, our model performs much better than the benchmark HAR-RV model when various forecasting horizons are considered, and our findings have important implications for investors and policy makers.