Purpose - In this paper, we conduct a out-of-sample test of Chordia and Subrahmanyam (2004) using the sample of 623 NYSE companies that made restatement on their previous financial results for the period of nine years (Jan, 1997 to Sept, 2005).
Design/Methodology - Using transaction data of the sample, we estimate two measures of order imbalance. To test the effects of lagged imbalances on return, we use time-series regression. Employing an event-study methodolgy, we also analyze the movement of trading activity, the imbalance measures, and return in the event period of restatement announcement.
Findings - The autocorrelation in order imbalance reduces substantially in our study period compared to their research period. In general, our results corroborate the predictions of CS's theory: a positive relation between the current return and the lagged imbalances and the reversal of the effect of the lagged imbalance in presence of the contemporaneous imbalance. But our results indicate that due to the weakened autocorrelation the effect of the lagged imbalances on the return diminish. Still the contemporaneous imbalance exerts significant influence on the contemporaneous return.
Implications - In the framework of an event study, we find excessive selling pressure with before the event day, which is considered informed trading. Examining the relation between changes in order imbalances and changes in return, we find that the effect of the lagged imbalances become negligible. Only the change in the contemporaneous imbalance is significantly and strongly associated with the change in the contemporaneous return.