Essays on asset pricing

This thesis consists of three essays. The first chapter of the thesis analyzes the impact of trades on stock returns. Existing work on the impact of trades has been conducted at the transaction level and fails to identify the existence of the temporary price impact after controlling for bid ask bounce. Using daily money flow data for all NYSE and AMEX firms that is calculated from TAQ data set and longer return horizon, I reach a different conclusion: there exists temporary price impact of trades. If the impact of trades contains a temporary component, trading activity today will predict future returns. I find that high money flow predicts lower future returns at a relatively long horizon. I also investigate the cross section characteristics of temporary price impact and find evidence that is consistent with the prediction of Grossman and Miller (1988). In the second chapter, we develop a simple model where trading volume contains information about future stock return movements. Specifically, our model explains why high trading volume is observed when a firm announces its earnings news and how trading volume can be related to the initial underreaction in the stock price. Our model has a clear testable implication that high abnormal trading volume predicts stronger drifts. We test our model's implication and find strong evidence for the model in the case of good news. Weaker evidence is found in the case of bad news. Possible explanations of asymmetric informativeness of trading volume are discussed in the chapter. In the final chapter, we first test whether stock price reaction to earnings announcement news is path dependent. We find evidence that is consistent with the prediction of Barberis, Shleifer, Vishny's model (BSV model). We also test whether analyst forecast bias depends on the past history of forecast errors. Analyst forecast errors also show path dependency. However, the direction of bias is opposite to the BSV model's prediction.