Using a unique dataset of Korean listed companies for which trade initiators are correctly identifiable, we estimate bias-free PIN (probability of informed trading) that is no longer subject to the trade misspecification problem and test whether it is related to expected returns. Unlike prior studies, we find that bias-free AdjPIN, the adjusted PIN purged of a liquidity component, is positively related to implied cost of equity. Our findings suggest that the errors in PIN variables hamper a proper identification of PIN pricing in prior studies.