An empirical investigation of high end-of-day transaction returns between 1978-1985

Abstract

Using a random sample of transactions data from the time period of September 1, 1978 through August 31, 1985, the high end-of-day transaction returns noted by Wood, Mclnish and Ord and by Harris were examined to determine their persistence over time and their relationship to a commonly used measure of daily security performance. Additionally, final transactions were classified by type of price change-reversal or continuation-in order to document whether the high end-of-day returns are the result of security price appreciation or the result of increases in transactions at the ask price. New information provided by this study can be summarized as follows:

1. The end-of-day anomaly persisted over the time period of the study and appeared to be strongest in the last three years.

2. A Friday effect was found in that the mean return to the final transaction on Friday was at least as great or greater than the mean final transaction returns on the other days of the week.

3. A relationship was found to exist between CRSP excess return level (good day/bad day) and the final transaction return, and there was evidence that the final transaction may have had a large impact on the CRSP excess return.

4. Reversals are more frequent than continuations on the final trade, particularly after 3:56pm, and the mean return to reversals is greater than the mean return to continuations.