The recent proliferation of algorithmic trading, new trading venues and innovative new trading products raise many issues about financial regulation and market design. One such trading product is flash orders introduced by Nasdaq on June 05, 2009. Flash orders are actionable indications of interest (IOI) that expose submitted marketable orders for a pre-defined period of time (500 milliseconds) to only Nasdaq participants, at or improving the national best bid or offer (NBBO) which is quoted at another trading venue.1 Thus, a “flashed” order may execute locally at the NBBO or better, while normally it would have been routed away to the other exchange offering the NBBO. Rule 602 of Regulation National Market System (NMS) allows for this occurrence.
The use of flash trading systems was first approved by the SEC under Chairman William Donaldson for the options market, Boston Options Exchange, in 2004.2 Flash trading was introduced into the equity market on January 27, 2006 by Direct Edge, followed by Nasdaq and BATS Global Markets in June 2009. Nasdaq and BATS voluntarily discontinued support for flash orders on September 1, 2009, due to political pressure and pending the review of flash orders by the SEC. According to Roseblatt Securities, during the period June - August 2009 executed flash orders constituted 3% of daily traded volume in the U.S. market, a market share as large as AMEX or Boston Stock Exchange at the time. NYSE is the only major market center that has not offered any flash-order functionality.3
Since August 2009, there has been wide media coverage and intense debates by regulators, industry analysts, and commentators over the impact of flash trading on financial markets and participants. Many important questions have been raised. Does flash trading undermine the integrity, fairness, and efficiency of the U.S. national market system? Does the practice of flash trading harm market liquidity and price discovery? Should the SEC remove the flash order exception? On September 18, 2009, the SEC proposed the elimination of the flash order exception from Rule 602 of Regulation NMS. No decision has been taken to date. Thus, answers to the above questions are informative for the SEC decision making and venues that continue to support actionable IOIs. In addition, addressing these questions has important implications for the information efficiency of prices, investors' trading strategies, market quality, market makers' behavior, and investors' welfare.
1 An actionable IOI expresses a trading interest with specified price, side, and number of shares and allows the buy-side trader to immediately trade on the indication directed to them, while submitters wait (or are willing to) for the counterparty to hit their IOI. An IOI functionality, frequently associated with “dark pool” liquidity, is mainly provided to facilitate trades among market participants with large orders and is an important trading outlet for long term retail and institutional investors.
2 Manual flash orders have long been practiced on floor-based exchanges, where brokers announce orders to the floor crowd for potential price improvements. Flash orders in electronic markets were introduced to replicate this auction market process.
3 NYSE has vehemently protested against the trading practices of their competitors, especially those related to flash and dark pool trading. NYSE's concerns and complaints induced New York Senator Charles Schumer to request the SEC to ban flash trading and to increase monitoring of dark pool trading. Any ban or restriction of the flash functionality and provision of dark pool liquidity may help NYSE to win back market share.
In a recent paper with Johannes Skjeltorp and Wing Wah Tham, we use the introduction and removal of the flash order facility by Nasdaq as a natural experiment to study the impact of flash orders on market quality. We consider the pre-routing feature of flash orders as a voluntary announcement of trading intent and essentially advertise liquidity needs in an attempt to trigger a response from other traders, similar in nature to sunshine trades. Admati and Pfleiderer (1991) theoretically show that trading costs can improve when liquidity demanders preannounce their liquidity needs, “sunshine trading”. They define sunshine trading as a strategy where a trader preannounces to other traders in the market that he or she will trade a specific number of shares before the order is actually submitted. Preannounced orders are unlikely to be based on private information because of the reputation cost for brokers, the potential delay cost of preannouncement, and the potential risk of information leakage for informed traders, implicit costs. Sunshine trading is beneficial because it allows for the coordination of liquidity supply and demand and the identification of informationless trades. Preannouncers inform potential counterparties of their demand for liquidity, facilitating the match between supply and demand. In addition, preannouncers indicate to the counterparty that they are uninformed by voluntarily disclosing their order, thus reducing adverse selection costs. Finally, sunshine trading reduces the risk-bearing costs for the market, as it reduces the uncertainty of the liquidity demand of uninformed traders and the amount of noise in the price.
We find that flash orders are mainly submitted by less informed market participants and that market participants regard these orders as less informative and are willing to trade against them quickly at favorable prices. We also show that flash orders are useful as an advertisement for liquidity demand, because they reduce transaction costs in Nasdaq. The saving in trading costs comes from reductions of the bid-ask spread in Nasdaq and price improvements offered by liquidity providers. Our results also show that flash order activity improves efficiency by narrowing the difference between the local Nasdaq quotes and the NBBO for individual stocks. Large deviations of Nasdaq quotes from the NBBO seem to trigger flash orders that help move the Nasdaq quotes quickly towards the NBBO.
Finally, we study the impact of flash orders on the overall market quality, which can be interpreted as the impact on all market participants. Comparing various liquidity and activity measures during the flash and non-flash period, overall market liquidity (measured by quoted spread, relative spread, and Amihud illiquidity ratio) improves (deteriorates) significantly when flash orders are introduced (removed). Finally, we find that market volatility and return autocorrelation improves (deteriorates) substantially when flash orders are introduced (removed).
The results seem to support the hypothesis that flash orders indicate to brokers that uninformed liquidity is available at a particular venue so that they can quickly route to it, if it represents the best available trading opportunity. Our findings indicate that advertising liquidity needs through flash orders successfully attracts liquidity providers and lowers price uncertainty and overall trading costs in the market. Hence, flashed orders appear to act as a coordinating mechanism for supply and demand and for identification of informationless trades, in line with what is predicted by the Admati and Pfleiderer (1991) model. The many concerns raised over flash orders do not appear to be vindicated by the data, on the contrary actionable IOIs and voluntary pre-trade transparency improves the market quality for all participants.
Admati, A. R. and P. Pfleiderer. Sunshine trading and financial market equilibrium. Review of Financial Studies, 4(3):443-481, 1991.
Skjeltorp, J., E. Sojli, and W. Tham. Sunshine trading: Flashes of trading intent at the NASDAQ, Working Paper