Proceedings of The 10th International Conference on Modern Research in Management, Economics and Accounting
Idiosyncratic Volatility and the Intertemporal Capital Asset Pricing Model
When the true asset pricing model cannot be identified, the idiosyncratic volatility obtained from a misspecified model contains information of the hedge portfolio in Merton’s (1973) ICAPM. Empirically, I find that from 1815 to 2018, more than two centuries, neither equal-weighted idiosyncratic volatility (EWIV) nor value-weighted idiosyncratic volatility (VWIV) can forecast stock market returns. However, EWIV and VWIV when applied together are strong predictors of stock market returns over short- and long-term horizons. The explanatory power is economically significant with an out-of-sample forecasting 𝑅2 around 1% for one month and 12% for one year.
This finding suggests that EWIV and VWIV together are linked to state variables that capture time-varying investment opportunities. I argue that the combination of EWIV and VWIV is a proxy for the conditional covariance risk in the ICAPM. I revisit the debate between Goyal and Santa-Clara (2003) and Bali, Cakici, Yan, and Zhang (2005) and reconcile their mixed findings between aggregate idiosyncratic volatility and future stock market returns.
Keywords: idiosyncratic volatility, stock market variance, conditional covariance, time-series stock return predictability, expected stock returns, intertemporal capital asset pricing model,economic state variable, risk-return tradeoff.
University of Toronto