An Empirical Analysis of Anonymous Trading in Equity Markets
Kar Mei Tang
This thesis presents three essays that examine the role of anonymous trading in equity markets, and its effects on liquidity, execution costs and market integrity.
The first essay presents evidence that liquidity increases, and order aggressiveness and execution costs decrease, following the removal of broker identifiers from the Australian Stock Exchange's (ASX) trading screens in 2005. This is consistent with the notion that limit order traders are more willing to expose their orders when they can do so anonymously. Anonymous markets also attract order flow from nonanonymous substitute markets, but this effect is only seen in large stocks.
The second essay examines how different market conditions and trading needs affect traders' decisions to trade anonymously. This analysis uses data from the Toronto Stock Exchange (TSX), an electronic exchange where identity disclosure is voluntary. The results show that most trading on the TSX remains non-anonymous, as various factors other than anonymity - such liquidity, information asymmetry, the time of day and expected execution costs - are also important considerations in traders' identity disclosure choices. The results suggest that informed traders are likely to use anonymous orders as part of their stealth trading strategies, in order to reduce their overall price impact costs. Anonymously-initiated trades are generally more informative about short-term price movements and, after controlling for selectivity bias, also have lower execution costs than trades initiated non-anonymously. The third essay reports the results of empirical tests that examine, using TSX data, whether anonymity facilitates abuses of client priority rules by allowing proprietarytraders to conceal frontrunning activity. Contrary to investor perceptions, there is no evidence that anonymous orders are commonly used to conceal frontrunning. The potential frontrunning identified by the frontrunning detection models in this essay is limited to only a small number of brokers, and generally more likely to be conducted non-anonymously. In the instances where systematic trading ahead of clients does occur, there is evidence of a positive association between the magnitude and frequency of frontrunning with broker size, order size and informed (profitable) client orders. There is also evidence of some asymmetry in the frequency and volume of frontrunning ahead of buy- and sell-side orders.
These results have implications for market participants seeking to better understand how anonymity can be used to reduce overall execution costs. They are also useful to market operators and regulators interested in assessing the effects of anonymous trading on liquidity and market integrity. Finally, compliance officers and market regulators may find the results informative in terms of assessing how anonymous trading affects traders? best execution and client priority obligations.
Associate Professor Carole Comerton-Forde and Professor David Johnstone