Automated Trading Debate Intensifies
The U.S. stock market “flash crash” in May 2010 was the first episode to jolt investor and public consciousness over the risks associated with today’s highly automated, technology-driven marketplace. Two years later, investors are being shocked with alarming regularity. The list of technology-related mishaps so far this year includes the failed BATS IPO on its own market, the botched Facebook IPO on Nasdaq, the Knight Capital US$440 million rogue algorithm-induced trading loss, and in recent weeks a trading glitch hitting the stock of Kraft Foods and a “flash crash” in India. It’s little wonder that volumes remain depressed; confidence has been shaken, and many retail investors are, for now, steering clear of the stock market.
The regulatory response is stepping up with various initiatives globally to limit the propensity for errant technology to cause market instability. Such initiatives include tightening up controls over algorithms via more frequent and robust testing, regulatory authorisation and oversight, curbs on unfiltered electronic access to markets (such as by banning “naked” sponsored access), and more sophisticated circuit breakers to halt excessive trading volatility.
Despite the fact that the EU has escaped the worst of these mishaps, it is in this jurisdiction that policy proposals go furthest. The main piece of markets legislation in the EU, the Markets in Financial Instruments Directive (MiFID), is currently under political negotiation. MiFID encompasses everything from how markets are structured to how investment products are sold, yet it is the area of algorithmic and high-frequency trading (HFT) — absent from the original MiFID legislation implemented in 2007 — that is perhaps the most contentious.
At a conference in Brussels last week held by Finance Watch (the independent public interest association), a panel of senior practitioners discussed the issues of market structure, technology, and HFT, where the whole of gamut of views were expressed. The claims from the HFT camp are that it is not they and other professional market participants that have benefited from the technological revolution, but ordinary investors. Frictional trading costs have come down so dramatically, and markets have become so fast and efficient, that it’s hard for professionals to make money any more — the gains have all accrued to the retail investor. What’s more, if the HFT community is to be believed, the average pensioner will have 30% more in their pension fund at retirement as a result of cost savings induced by the activities of high-frequency traders keeping spreads wafer-thin. On the flip-side, the traditional buy-side investor claims that high-frequency traders are “front-running” (a term used loosely) their orders, resulting in high “slippage” (also referred to as implementation shortfall, or more simply, market impact). Consequently, large buy-side investors are increasingly seeking out the relative safety of dark pools, which can protect from information leakage and in some cases keep the high-frequency traders out.
On the issue of slippage, a recent paper by Oxera , under the U.K. government’s Foresight project on the future of computer trading in financial markets, suggests that high-frequency traders are not to blame. The Oxera paper finds that “the market impact being recorded now is similar to that recorded before the financial crisis, and this is similar for both U.S. large cap stocks and U.S. small cap stocks, while HFT is much less prevalent in the trading of small cap stocks.” In other words, high-frequency traders do not seem to affect institutions’ implementation shortfall, although Oxera also notes that “the absence of any large-scale trend on market impact (at least measured in this particular form), [and] the general reduction in spread… tend to reinforce the premise that, if there are negative consequences on market participants, these will vary both by the precise trading strategies undertaken by the high-frequency traders and the trading strategies used by the other market participants.” In other words, circumstances matter. In other studies, the academic literature mostly supports the argument that HFT has a benign or positive effect on markets, although many of these studies are focused somewhat narrowly on liquidity. The market abuse and financial stability aspects associated with HFT are less examined.
So what conclusions should be drawn from all this? While initiatives aimed at making the markets safer are needed, such as by strengthening the testing and controls around algorithms and improving network resiliency, there is a lack of clear evidence that HFT itself should be curbed. Yet that appears to be the aim of the latest MiFID proposals. For the time being, evidence-based policymaking in Brussels seems to have yielded to politics. On the table are proposals to place a limit on the ratio of orders to trades, a minimum 500 millisecond resting period for orders, possible market making requirements, the possibility of introducing a minimum tick size, and the other measures previously mentioned relating to the testing and review of algorithms. Not to mention the possibility of a financial transactions tax, an issue that regained momentum last week when 11 EU member states, including France and Germany, announced their intentions to press ahead with a financial transactions tax under the “enhanced cooperation” procedure.
Are these measures likely to be effective? Briefly, they are likely to induce unintended consequences. The financial transactions tax, for example, is most likely to impact end consumers the most, according to a CFA Institute member survey. The high-frequency traders have warned that such a tax will force them out of many markets (pensioners beware). As for the minimum resting time, Markus Ferber, the MEP steering the MiFID dossier through the European Parliament, commented at the Finance Watch conference that the current average resting time for orders is five milliseconds. Extending that by a factor of 100 is, therefore, far from trivial. Displayed limit orders resting in exchange order books effectively provide a free trading option; like any option, the longer the time that the option is valid, the greater the risk associated with it, and hence the higher its cost. In this case, it would likely lead to market makers widening bid-offer spreads to compensate for this order-longevity risk, or worse, encourage more “taking” of liquidity (using market orders to hit resting limit orders) and less “making” of liquidity (posting displayed limit orders). In other words, a minimum resting time could disincentivize market participants from using transparent order types. As for the limit on the ratio of orders to executions, ultimately this is a bandwidth issue that is likely best tackled by exchanges taxing (through higher fees) excessive message traffic, rather than an arbitrary cap set by regulators.
What form the final policy measures will take is not yet certain. But what is certain is that this debate has yet to hear the last word.