6. Concluding
Remarks We provide a comprehensive reexamination of the lead-lag relationship between trading volume and stock returns. Our contribution to the literature rests importantly in the paper’s emphasis being on detailing out-of-sample evidence, thereby complementing in-sample findings in earlier empirical studies. In the U.S. market, higher trading volume, whether measured by aggregate time series of turnover or by the cross-sectionally constructed high volume return premium, is indeed followed by higher stock returns. However, such predictive power of trading volume should be interpreted with caution in that the associated economic gain is quantitatively small for the market as a whole. Furthermore, the predictive power of trading volume becomes insignificant even statistically in the more recent period featuring high-profile high-speed trading. Similarly, with only a few exceptions, the predictive power of trading volume for stock returns fail to pass the rigorous statistical and economic tests in out-of-sample regressions for most of the non-U.S. markets.
The lack of significant out-of-sample predictive power of trading volume for stock returns is in stark contrast to the existing in-sample analyses that have often found that a dynamic volume-return relationship exists. This empirical finding suggests that we may need rethink about the theoretical models which predict that trading volume is significantly related to future stock returns as reviewed earlier in the paper. Our finding is probably more important for practitioners who might otherwise consider exploiting the relation for timing the market and forming aggressive trading strategies.
