Three essays on trading volume

This thesis addresses the relation between trading volume and various information variables. The first essay presents a general method to analyze the dynamic portfolio choices of rational investors around public announcements of earnings. The analysis is based on a Dynamic Noisy Rational Expectations Equilibrium Model. A major result of our paper is that trading volume is positively related with the prior dispersion of beliefs. This result is independent of the error structures implied by whether the public signal is an independent signal or a sufficient signal. We find that when the public signal is an independent signal, it is the difference between the maximization problems at time 0 and time 1 that generates trading volume. On the other hand, when the public announcement is a sufficient signal the prior dispersion of beliefs completely disappears after the public announcement. It is this convergence of beliefs that generates trading volume in this case. The second essay empirically shows that abnormal trading volume around earnings announcements is positively related with: (1) the absolute value of unexpected earnings, (2) the prior dispersion of beliefs. There is a high correlation between two explanatory variables. But one makes an incremental contribution to the explanation of the abnormal trading volume made by the other. The third essay provides a theoretical model that captures a new aspect of trading volume which cannot be captured in traditional single risky asset models. A multiple risky asset model is proposed and studied to show that it is possible for trading volume of one asset to be induced by the acquisition of information about another asset. The analysis shows that the trading volumes of risky assets in such an economy depend on the magnitudes of the absolute value of differences in the variances of error terms