On the Relation between EGARCH Idiosyncratic Volatility and Expected Stock Returns

Abstract:     
Ang, Hodrick, Xing, and Zhang (2006, 2009) report that lagged realized idiosyncratic volatility is negatively correlated with returns in the cross-section. Fu (2009) reports EGARCH idiosyncratic volatility is positively related to returns in the cross-section. This paper shows both analytically and empirically that Fu’s conflicting evidence reflects a look-ahead bias accidentally introduced by standard methods of estimating month t EGARCH idiosyncratic volatility. Furthermore, we show that when month t EGARCH idiosyncratic volatility is forecasted using returns only up through month t-1, there is no significant cross-sectional relation between EGARCH idiosyncratic volatility and returns. Moreover, we show that even when controlling for out-of-sample EGARCH idiosyncratic volatility, lagged realized idiosyncratic volatility remains a significant predictor of one-month-ahead realized idiosyncratic volatility. Finally, we illustrate via Monte Carlo simulations that the look-ahead bias is problematic for empirically observed 1) degrees of stock return skewness and 2) monthly return series lengths.

Number of Pages in PDF File: 40

Keywords: EGARCH, idiosyncratic volatility, cross-section of stock returns

JEL Classification: G1

working papers series