Fat-tailed stochastic volatility model and the stock market returns in China

Purpose - The purpose of this paper is to investigate how the selection of return distribution impacts estimated volatility in China’s stock market. Design/methodology/approach - The authors use a Bayesian analysis of fat-tailed stochastic volatility (SV) model with Student’st-distribution, and conduct an out-of-sample test with realized volatility. Findings - Empirical analysis results indicate that fat-tailed SV model performs better in capturing the dynamics of daily returns. The authors find that asymmetry, holiday and day of the week effects are detected in estimated volatility. However, the out-of-sample comparison shows that fat-tailed SV models fail to outperform SV models with normal distribution in fitting and predicting realized volatility. Originality/value - The contribution of this paper to existing literature is twofold. First, it proves that fat-tailed SV models with Student’st-distribution perform better than normally distributed SV models in fitting daily returns of China’s stock market. Second, this paper takes asymmetry, holiday and day of the week effects into consideration at the same time in the fat-tailed SV model.