Idiosyncratic Risk Matters! A Switching Regime Approach. Evidence from US and Japan.

The resent studies of Bali et al. (2005) and Wei and Zhang (2004) showed that there is no relation between idiosyncratic risk and value-weighted portfolio returns and therefore asset specific risk does not matter in asset pricing. We propose a different approach of examining the forecasting ability of idiosyncratic risk by using switching regime models and therefore to be able to separate the stock returns to either to high or to low volatility periods in order to examine if idiosyncratic risk is related to either or to both regimes. The empirical results suggest that for both markets (US and Japan) in the low variance regime there is a positive and statistically significant relation between future market returns and idiosyncratic risk. The proclaimed relation is not sample dependent as it holds before and after the 1999 year during in which it seems that a structural break has occurred