The informational content of trading activity

This dissertation investigates the idea that trading activity contains information regarding the evolution of stock prices. Specifically, we examine whether trading activity has directional prediction power over future prices. This research question is not trivial. The standard assumption in the finance literature is that stock prices reflect instantaneously all publicly available information. Thus, trading activity should not have any predictive power over and above an appropriate measure of risk.;The first chapter concentrates on intradaily trading activity. We find that 10-minute trading periods of unusually large buyer-initiated net order flow or unusually high volume tend to be followed by periods of price appreciation up to 24 hours after the initial price impact. A similar but opposite effect is detected for periods of unusually low trading volume and under certain conditions for periods of unusually large seller-initiated net order flow. We show that these results are consistent the existence of frictions in the market that cause markets to be slow to recognize some of the information revealed by traders' transactions and to incorporate this information into prices. On the other hand, the inventory control consideration of a specialist, the special structural features of the NYSE, time deformation and price discreetness fall short from explaining the results.;The second chapter extends the investigation of the information contained in trading volume to the daily and weekly frequency. We find that stocks experiencing unusually high (low) trading volume over a day or a week tend to appreciate (depreciate) over the course of the following month. This effect is consistent across firm sizes, portfolio formation strategies, volume measures, and is more prominent in stocks that underperformed in the prior months. Thus, also daily and weekly trading volume contains information regarding the evolution of future price. We show that these results cannot be attributed to firm announcements, market risk, liquidity risk or the return autocorrelation patterns that have been documented in the past. However, the results are consistent with the effects that shifts in a stock's visibility should have on returns.