Evidence of Imitation Trading

Abstract

This article provides empirical evidence on how market making is affected by the existence of a crowd in a floor trading system. In particular, we test the extent to which traders imitate each others' actions. Our tests are based on data from the Toronto Stock Exchange which closed its trading floor in April 1997. We find that while effective bid-ask spreads, trading volume and average trade size are unchanged with the introduction of system-only trading; for floor-traded securities, there is much stronger evidence of a diagonal effect, i.e., a given type of event (trade or quote change) occurs with greater probability following an event of the same type than it does unconditionally. We then examine the three hypotheses put forward by Biais, Hillion and Spatt (1995) to explain this effect and find sufficient evidence to reject the hypotheses of strategic order splitting and, similar but successive reaction to the same events. However, we are unable to reject the hypothesis of traders engaging in imitating behavior which is more likely to arise when traders can better identify each other in the trading environment.