I find that the implied dividend for a stock, calculated on its cum-dividend date, contains information on its subsequent one-day return. I suggest that this is at least partly due to option traders having an informed estimate of the ratio of stock price drop to discounted dividend for those stocks going ex-dividend. This ratio is found to be significantly less than 1 on average confirming the persistence of the so-called ex-dividend day anomaly, but it varies greatly across stocks.
Specifically, I show that the difference between the implied dividend and the actual dividend for a stock contains significant information on the cross-sectional variation in the stock's ex-dividend day unhedged and hedged returns. I find using portfolio sorts and regression tests that stocks with low (high) values of implied dividend relative to actual dividend are more (less) likely to experience high (low) abnormal unhedged and hedged returns over the ex-dividend day.
Furthermore, the representative options trader has higher quality information on ex-dividend day returns for stocks with lower idiosyncratic risk, higher liquidity and higher dividends which is consistent with the dividend clientele theory.