We show that the negative relation between realized idiosyncratic volatility, measured over the prior month, and returns is robust in non-January months. Controlling for realized idiosyncratic volatility, we show that the relation between returns and expected idiosyncratic volatility is positive and robust. Realized and expected idiosyncratic volatility are separate and important effects describing the cross-section of returns. We find the negative return on a zero-investment portfolio that is long high realized idiosyncratic volatility stocks and short low realized idiosyncratic volatility stocks is dependent on aggregate investor sentiment. In cross-sectional tests, we find the negative relation is weaker for stocks with a large analyst following and stronger for stocks with high dispersion of analyst forecasts. The positive relation between expected idiosyncratic volatility and returns is not due to mispricing.

We propose a parsimonious, comprehensive proxy for innovations in limited arbitrage: innovations in ETFs’ premium. Consistent with a common component, we confirm limited arbitrage factors, LAFs, constructed from ETFs’ premium innovations spanning four asset classes are correlated. Further, we find that equity LAFs are negatively priced in the cross-section of stock returns. Our pricing tests also confirm that LAFs provide pricing information beyond well-known limits of arbitrage: illiquidity and idiosyncratic volatility. Overall, our findings suggest that limited arbitrage risk is priced and LAF is a relevant risk-factor.