Recently the Foresight Programme, a research arm of the U.K. Government, published a working paper on “The Future of Computer Trading in the Financial Markets.” The paper focuses on the effects and risks of computer-based trading, including algorithmic and high-frequency trading — HFT — on financial market stability and efficiency.
The paper supports what many market participants suspected: since the rise of computer-based trading, markets may be less tidy, but they also appear to be more efficient. But the paper points out novel – and alarming – risks raised by computer-based trading. These include the risk of causing sharp market shocks that are difficult, if not impossible, to anticipate or address ex ante.
The topics addressed in the paper are closely related to those being assessed by the U.S. Securities and Exchange Commission and Commodity Futures Trading Commission in connection with the “Flash Crash” of May 6, 2010. The paper is consistent in some respects with findings of the SEC and CFTC staff in the Flash Crash Report in that it emphasizes the role computer-based trading may have in causing market instability. Overall, the paper may strengthen the resolve of regulators and legislators to take action to forestall endogenous shocks from computer-based trading.
Overview of the Working Paper
The working paper comprises three articles that draw conclusions from a series of academic papers commissioned by Foresight. The articles focus on separate, but closely related, aspects of computer-based trading. The first article analyzes the effects of computer-based trading on market volatility. It concludes that, to date, no evidence exists that ties computer-based trading to higher market volatility. Indeed, it observes that the increased prevalence of computer-based trading has coincided with decreased market volatility under normal market conditions.
The second article reviews the effects of computer-based trading on liquidity, price efficiency, and transaction costs. Consistent with the conclusions of the first article, the second asserts that computer-based trading has coincided, in normal market conditions, with improved market quality, including increased liquidity, lower transaction costs, and greater efficiency in market prices. But it emphasizes the likelihood of non-linear price movements shocking the system and undercutting investor confidence.
The third article focuses on the effects on financial markets of technological advances in trading systems, including cloud computing, customized CPUs, and autonomous adaptive trading algorithms, each of which reduce the need for human traders and increase the speed of trading systems. The article concludes that the rapid pace of technological change, and increasing use of computer-based trading systems, will continue for the foreseeable future. Importantly it notes that these technological advances mean that computer-based trading systems can relocate to markets anywhere in the world.
Each article, while pointing out likely benefits of computer-based trading, also provides sobering explanations of its potential risks, which include risks of unpredictable, system-wide disruptions with far-reaching effects. These disruptions could arise from various types of destructive feedback loops that have been observed in computer-based trading systems and which may be non-linear in their effects on the market. Increasing complexity of computer-based trading systems — and the associated increase in uncertainty inherent in these systems — also is a source of market instability. And interactions among computer-based trading patterns, and between computer-based trading systems and human traders, are not well understood. Each of these features of computer-based trading leads to risks that could result in significant, negative effects on financial markets that are difficult to predict.
The paper further notes that computer-based trading systems may exacerbate rapid decreases in liquidity in stressed markets. Although liquidity has always decreased during times of market stress — regardless of whether market makers were humans or computers — computer-based trading may speed, or cause more extreme, reductions in liquidity during periods of market stress.
U.S. Implications
The working paper is far from a jeremiad against computer-based trading. However, overall the report can be read as a call for stronger action to deal with the poorly understood risks of this trading. In particular, the paper could be read to support additional action by the SEC, CFTC, and other U.S. financial market regulators in the future.
Perhaps the foremost regulatory ellipsis highlighted by the paper is their need to monitor, and potentially restrain, the trading behavior of algorithms due to their interactions with each other, with human traders, and across trading venues. U.S. and non-U.S. regulators have considered, and even implemented, rules that partially address the issue. The SEC’s circuit breaker rules would, to some extent, mitigate the effects of algorithms gone wild. And the SEC’s market access rule would require a broker-dealer to have in place controls designed to detect malfunctions in an algorithm used by it or a client and to prevent delivery of erroneous orders.
Rules proposed by the Canadian Securities Administrators require more: a broker would need to ensure that it has the necessary knowledge and understanding with respect to the computer-based trading systems used by it or any client. A proposal by the European Securities and Markets Authority would require a market participant to establish controls over and monitor trades of a trading algorithm it employs. But none of these proposals would provide for system-wide monitoring or ex ante regulation to address risks posed by the interactions of algorithms.
In this respect, the working paper would seem to support a stronger duty for brokers to understand the algorithms they deploy and the need for the regulators themselves to understand the effects of algorithmic trading. This is consistent with the establishment of a consolidated audit trail system, as has been proposed by the SEC and FINRA, which would allow regulators to monitor transactions across market participants and trading venues.
The paper may also be relied upon by proponents of a transaction tax. Although the paper finds that the increased trading volume generated by HFT is correlated with lower spreads and faster execution speeds, which benefit would be diminished by a transaction tax, the significant and unpredictable risks associated with HFT could be argued to justify the chilling of HFT that would be result from a tax on the activity.
Finally, the paper calls for more coordinated regulation of trading, including better coordination between futures and equity markets regulation, the importance of which the SEC and CFTC articulated in the Flash Crash Report. And the paper warns that, as computer-based trading systems are not limited geographically, regulators must recognize that computer-based trading can readily relocate to another jurisdiction.
Conclusion
The Foresight working paper provides a potentially influential perspective on the potential effects of computer-based trading on financial markets, which could support regulatory and legislative initiatives to rein in the risks of unbridled computer-based trading.
Annette Nazareth, partner, Robert Colby, partner, and Jai Massari, associate, are in the financial institutions group at the law firm of Davis Polk & Wardwell. Nazareth served as a Commissioner at the SEC and headed the Division of Market Regulation and Colby was Deputy Director of the Commission’s Division of Trading & Markets.
