There is an ongoing debate between regulators and investors about the usefulness or lack of transparency of trading in dark pools (or dark markets). The authors argue that, although marketable orders executed in the dark have lower information content, smaller fill rates, and longer execution times, they offer more favorable prices (i.e., a lower bid–ask spread) than lit markets. Traders in dark markets also tend to be more sophisticated.
How Is This Research Useful to Practitioners?
As more and more trading volume (37% in 2013) is being shifted to dark markets, the authors aim to provide insights into why traders choose dark markets. One major insight is that about 80% of the orders transacted in dark markets are executed at a better price/lower spread than those in in lit markets, resulting in a total saving for traders in this study of around $6.3 million (or an average of slightly over a half cent per share).
The decision regarding where to execute orders appears to be driven largely by the trade-offs with respect to dimensions of order execution quality in both dark and lit markets. The effective bid–ask spreads are lower (higher), the order fill rates are lower (higher), and the execution times are slower (faster) in dark (lit) markets. One exception to these findings is that large block trades are executed more quickly in dark markets. Informed traders are less likely to go to dark markets because they tend to trade around the same time and in the same direction, making it difficult to find a quick match in dark markets, which are thus rendered less informative about future prices.
Other favorable factors for trading in dark markets arise from the higher volatility and uncertainty in lit markets. Seasoned traders with more skill, trade diversity, and patience are more likely to prefer dark markets.
How Did the Authors Conduct This Research?
The authors use proprietary data (market hours) from a US broker/dealer for NASDAQ-listed stocks over eight calendar years (October 1999–May 2006). The data consist of 3,014 US equity traders with 6.2 million orders (337,000 dark orders), 9.3 million trades, and 12.1 billion shares (with a value of about $104 billion). Matching data for lit markets are from Thomson Reuters and CRSP.
Using a two-stage selection model, the authors check the trading venue choice and execution quality. The first-stage regression uses a probit model to predict whether an order execution will occur in a dark market or a lit market. The authors emphasize that order characteristics, stock characteristics, market conditions, and volatility are important determinants of dark execution (i.e., when the bid–ask spread is wide, traders are likely to choose dark markets over lit markets).
The results are then incorporated with an ordinary least-squares Tobit second-stage regression that focuses on a dark dummy variable that distinguishes differences in effective spread, price impact, execution time, and fill rate between dark markets and lit markets. The authors find that both effective spread and price impact are lower when either marketable or nonmarketable orders are executed in the dark, despite average order size being more than double for the dark market sample.
The study does not take into account counterparty dark order executions or whether wholesalers are using their own capital to execute dark marketable orders. The study also fails to consider transaction costs.
Abstractor’s Viewpoint
Although the authors talk about cost savings as one of the major benefits of trading in dark markets, if overall trading costs are considered, traders might not realize savings as high as envisaged. The rise of high-frequency trading and algorithm strategies has made traders more sophisticated in terms of finding the best price or removing the arbitrage (if any) quickly. Moreover, because the study covers the period preceding the global financial crisis, the same results under either extreme volatility or a lack of total liquidity in the markets are not available.
The study covers only NASDAQ-listed stocks. It would be useful to see whether the dark-trading argument holds true for larger exchanges across borders, especially in Asian markets.