Information Content in Small and Large Trades


  • We gratefully acknowledge the comments we received from the seminar participants at the University of Connecticut and the Indian Institute of Management Calcutta. We thank NYSE for making the TORQ data available to us, B. Radhakrishna for his help with the TORQ data and Xin Tong for her assistance during the final revision of the paper. We sincerely thank the referee for countless suggestions for improving the paper. We are responsible for any remaining errors in the paper.

Malay K. Dey, NYIT and FINQ, 1091 Stillson Road, Fairfield, CT 06824. Tel No: 1-2032928072. Email:


We estimate the probabilities of informed trading (PIN) for small and large trades and then investigate their determinants. We model a competitive dealership market for equities with two order sizes using a Poisson process mixture model and use TORQ data to estimate the parameters for the model via the method of maximum likelihood. The PIN for small and large trades are functions of the resulting parameter estimates. In our empirical tests, we find that although for the majority of securities information contents in small and large trades are similar, the average PIN for small trades is significantly higher than that in large trades. We also find that trading volume and institutional trading are the primary determinants of information content in small and large trades, respectively, but not of both. A further investigation of the securities with the largest differences in terms of PINs for small and large trades reveals that trade size alone distinguishes those firms from the rest – all eight firms reside in the lowest quartile in terms of average trade size.