Monday, August 23, 2010

High Frequency Trading and the "Flash Crash"

It requires only a glance at securities industry job sites to see that high frequency trading continues to grow rapidly, even though it is already thought to account for half of U.S. equity volume. The “Flash Crash” of May 6 and subsequent regulatory interest in these activities have attracted media attention to this corner of the markets. The technicalities of market microstructure and the use of computer algorithms, machine learning, etc. in an attempt to exploit it would seem unlikely to capture the public imagination. But the topic has a certain sci-fi glamour, which is enhanced by the perceptions that markets have escaped human understanding, that events such as the one pictured below could easily recur, and that participants in this activity are up to no good. Mystery, the threat of catastrophe and a suspicion of scandal – how could a journalist (or for that matter, a Congressman) resist?
There has been high frequency trading since the first stock exchanges opened in the seventeenth century – computerization has only accelerated its pace. As outlined in a recent IOSCO study, modern transaction speeds do raise concerns about pre-trade risk and compliance controls, but few if any of these are insurmountable. More trenchant criticisms note that high frequency trading, in the context of fragmented markets with multiple venues, each offering a somewhat different microstructure, opens the door to price manipulation and to meltdowns such as the “Flash Crash.” Further, the competition for liquidity among these venues has led to the practice of selling access to flash orders, which many commentators regard as tantamount to permission to front-run other market participants.

Dubious as the practice may be, offering firms access to flash orders is simply the latest in a long line of privileges that exchanges have offered to attract liquidity providers to their platforms. They argue that investors benefit from these, in terms of faster execution and price improvement: for example, the CBOE claims that flash orders saved its customers $3.6 million in a single month. However, if the practice is banned, as the SEC proposes, it is inevitable that exchanges will respond by devising some new trading privilege to attract liquidity providers. This is not just a cynical comment. Providing a venue for liquidity is exchanges’ business, but they are not in a position to provide it themselves, since that would transform them into broker-dealers. So they have no choice, in a competitive environment, but to find ways to attract third party liquidity to their transaction mechanisms. When exchanges argue that they perform a public service by offering privileges that enhance their liquidity, high frequency traders can hardly be blamed for making use of them.

Multiple, competing exchanges with different transaction mechanisms almost certainly do increase the opportunities for market failure, as has been apparent since the mechanics of the Crash of 1987 began to be understood. Whether computerized direct access to these execution venues exacerbates this problem is debatable. It is likely that modern market infrastructure would have prevented or at least greatly ameliorated the 1987 episode. On the other hand, while the causes of the “Flash Crash” remain obscure, it is certain (if only because of their share of transaction volume) that high frequency traders played an important role in it. While the information available to regulators may not, in the end, be sufficient firmly to establish the cause of the “Flash Crash,” this indicates a serious gap in their capabilities that must be filled – the sooner the better.

As the IOSCO paper notes, regulatory authorities are somewhat hampered with regard to the enforcement of rules against market manipulation by some of the ways that high frequency traders may obtain direct electronic access to markets. This is a situation that must be corrected. Since there is reason to suspect that tape-painting and possibly even more highly irregular activity by high frequency traders contributed significantly to the “Flash Crash,” addressing this issue must also be a high regulatory priority. Much of this can be accomplished by enforcing existing rules, provided that regulators are able to obtain the information needed in order to do so. Fortunately, this is essentially the same information they would require to determine the causes of such episodes.

No comments:

Post a Comment