| Abstract: | Facing the puzzling risk-return trade-off, this paper proposes a new model for risk premia to capture nonlinear and time-varying features under the influence of trading volume. Using high-frequency data for the US stock market in Wharton Research Data Services' Trade and Quote database, our empirical findings suggest a significant nonlinear and time-varying contemporary relationship between return and realized volatility, ranging from positive to negative with an up-down-up pattern, summarized as follows. First, the contemporary relationship is positive on inactive trading days when the trading volume is smaller than usual, in which case traders may face no new information or event uncertainty. Second, the relationship is significantly negative when the trading volume is large on active trading days, in which case traders may be overconfident and behave in a risk-seeking fashion. Third, the risk premium tends toward zero during extremely abnormal trading days. Finally, low and high levels of trading volume have asymmetrical influences on risk premia, with a larger absolute value of risk premia for high levels of trading volume. Furthermore, the nonlinear changing autocorrelation of returns is insignificant from zero on normal trading days and most likely different from zero on abnormal trading days. These results provide explanations for the conflicts between financial theoretic and empirical studies. |
