| Abstract |
Stocks with high trading volume outperform otherwise stocks for one week, but subsequently underperform at the longer horizon. We show that such time-varying predictability of trading volume is attributed to abnormal trading activity, which is not explained by past volume. Specifically, we find that the return forecasting power of abnormal trading activity is strongly positive up to five weeks ahead. In contrast, the predictive power of the expected trading activity is negative, and lasts for longer horizons. We further argue that behavioral biases and investors’ attention induces abnormal trading activity, but its price impact is primarily related to behavioral biases. Overall evidence emphasizes the role of behavioral biases and investors’ attention to explain trading volume |