Systematic Asymmetry of Risk Premia in China

Abstract

In this study, we find that the new decomposition of the traditional market beta into four semi-betas, as proposed by Bollerslev et al. (2022), performs better than the upside and downside betas proposed by Ang et al. (2006) in the cross-section of Chinese stock returns. Based on the new decomposition components of beta, the semi-betas, we construct a new semi-beta, and find that it generates a statistically and economically significant risk premium in the Chinese stock market, which is larger than that of the semi-betas. To test the stability of this finding, we considered arbitrage risk, market liquidity, investor sentiment, and transaction costs, and found that the risk premium based on the semi-beta remains significant. Furthermore, using non-parametric estimation, we propose a new measure for capturing nonlinear asymmetric dependence, which we applied to the asymmetric dependence between stocks and the market. We found that smaller market capitalization stocks exhibit stronger asymmetric dependence, consistent with Jiang et al. (2018)’s findings. Additionally, we found that this nonlinear dependence measure outperforms the semi-beta in predicting long-term stock returns.

Keywords: semi-betas, asymmetric dependence, risk premia, cross-sectional analysis, Chinese stock market.