December 17, 2011
Where’s the Data?: The Speculative Debate over High-Frequency Trading Regulation
Over the last few years, the SEC and the EU have toyed with the idea of regulating flash trading. They’ve floated a few tentative thoughts and proposals for public digestion, see, e.g., http://www.sec.gov/rules/proposed/2009/34-60684.pdf, but because of vigorous debate over the unquantified benefits and harms of high-frequency trades (and the sheer volume of trades and the proprietary automated programs such reforms may affect), it has been difficult for regulators to be decisive.
Many of the reforms thus far have been informational---aimed at facilitating transparency and gathering data for after-the-fact analysis, rather than banning any particular activity. For example, in response to the May 6, 2010, flash crash, which many blamed on flash traders dropping out of the market and creating a liquidity vacuum, the SEC adopted a large trader rule earlier this year requiring broker-dealers to maintain records of transactions available for easy reconstruction of market activity and investigation into manipulative activity. http://www.sec.gov/news/press/2011/2011-154.htm. (It is currently trying to push for a consolidated audit trail for tracking orders across securities markets.)
Recently, with the EU’s October proposal for putting a financial transaction tax on stock and bond purchases, http://dealbook.nytimes.com/2011/09/27/europe-readies-plan-for-tax-on-financial-transactions/, the focus has turned back to substantive measures for discouraging high-frequency traders. Although anticipated to reduce the GDP, proponents seem to think it’s worth the cost to deter speculative transactions that allegedly do not contribute positive value to markets. The small tax percentage would be designed to curb the profitability of high-frequency traders, who would have to pay the transaction costs more often to pay for the volatility their activities may introduce into the market. Opponents express concern that without an all-or-nothing agreement between the major world markets, the tax would simply mean that the EU would scare away investors to other markets without such regulation. The regulation would have the potential effect of turning away trade volume without shielding their market from the volatility that may exist in other markets that still encourage high-frequency trading.
Despite the liquidity arguments in favor of high-frequency trading, both sides of the debate have to at least admit that whatever benefits that come from HFT are incidental---a product of historical coincidences rather than the intended effects of reasoned policy. According to the SEC, Regulation NMS’s Rule 602 exception, which permits this nonpublic flash-trading side market to exist, employs language that has remained unaltered since 1978. http://www.sec.gov/rules/proposed/2009/34-60684.pdf. Its original intent was to facilitate manual trading on exchange floors and to exclude “ephemeral” exchanges and cancellations because it would be impractical to include these deals in the consolidated quotation data. But these assumptions have changed with a highly automated trading environment, and there is a concern that the flash trading market is a significant enough one to elicit fears that we are creating a two-tiered market where the public does not have access to information on the best prices. And it doesn’t seem quite right to characterize one approach as pro-market or pro-innovation, as opposed to another, merely because of one’s unease with further regulation or upsetting the large amounts invested in proprietary trading algorithms based on the status quo. The advantages of the flash traders do not come from new insights into market value; it is a product of innovators taking advantage of carveouts in the current public price reporting regime, arbitraging the advantages of the unregulated at the expense of the regulated.
But despite calls for prompt action, perhaps the sluggish regulatory response isn’t such a bad thing for now, given the absence of consensus or data. There are a lot of proposals on the table, and part of the difficulty of creating a good response to all of the claimed ills of flash trading is the “black-box” nature of the proprietary algorithms and the difficulty of reverse engineering how any particular algorithm functions to create liquidity advantages or volatility crises. And aside from the actual logical effects of such trading, how should we measure causality when it comes to perceptions: loss of public confidence or the discouragement of small-time investors at what they believe to be a rigged system? I’ve been rummaging around the Internet for data to back up conclusions about causality, but such data seems quite elusive. And furthermore, even discussion of methodology---how we should begin to gather useful data separating out confounding variables and establishing useful comparison points---seems hard to come by. (The CFTC Commissioner suggested getting high-frequency traders to register and cooperate in disclosing information to the government to help regulators evaluate the effect of their activities http://www.cftc.gov/PressRoom/SpeechesTestimony/opachilton-53.)
In the absence of governmental regulation, various exchanges are experimenting with their own rules of minimizing the volatility effects of flash trading (e.g., circuit breaker rules, clerical error checks, and even outright bans of flash trading). Perhaps the lack of uniformity, for now is a good thing, allowing for regulatory experimentation that could lead to comparative data. (For example, if we took a sample basket of stocks and banned flash trading in those stocks, would we see a noticeable market preference in those stocks? Over time, would we see a difference in liquidity in some types of trades over others?) This problem, itself, was created by unanticipated effects of regulation; we might as well take a studied approach the second time around before unsettling a market that has adapted to the rule.