Idiosyncratic Volatility Measurement Frequency And Return Reversal

This paper examines whether the negative relation between idiosyncratic volatility and expected returns is due to stock return reversals as argued by Fu (2009) and Huang, Liu, Rhee and Zhang (2010). Controlling the return reversal effect, it shows that stocks with different past returns have different relations. The positive relation is mainly driven by stocks with low past returns, while the negative relation is result from stocks with high past returns. Additionally, the relation is very sensitive to the measurement frequency of idiosyncratic volatility, and the daily realized idiosyncratic volatility measure is a better proxy for the expected idiosyncratic volatility than the monthly measure. By employing an exponential generalized autoregressive conditional heteroskedascticity-in-mean (EGARCH-M) model, this paper finds a strong positive relation between time-varying risk premium and idiosyncratic volatility for portfolios containing stocks with low past returns and small portfolio, and a negative relation for growth portfolio.