U.S. Leads in High-Frequency Trading, Trails in Rulesby
Given the missteps that have prolonged and deepened the European debt crisis, one wouldn’t necessarily expect the continent to be home to far-sighted financial reform. But that is exactly what seems to be happening in the realm of high-frequency, computer-driven trading.
High-speed trading comes with real benefits: lower trading expenses, better prices for investors and increased market liquidity. The costs, however, are fairly significant and can be seen in wild volatility and destabilizing trading snafus.
Consider two events in this year alone: In March, Bats Global Markets Inc., a computer-trading exchange, had to abandon its initial public offering, after its system crashed the day of its IPO and forced a halt in Apple Inc. shares. And in July, the Wall Street trading firm Knight Capital Group Inc. nearly collapsed after a flawed computer program ran out of control for 45 minutes and placed errant orders. Then there was the so-called flash crash of May 6, 2010, when U.S. markets plunged, losing about $1 trillion in value, before recovering within an hour.
All of this has contributed to the erosion of trust in U.S. markets. Today, only 15 percent of Americans express confidence in financial markets, the lowest since the Chicago Booth/Kellogg School Financial Trust Index was created in January 2009.
The Securities and Exchange Commission is holding a round table Oct. 2 to review what to do about high-frequency trading, but we aren’t optimistic. The gathering is little more than a fact-finding mission, and the agency hasn’t proposed significant fixes.
In contrast to this go-slow approach, Germany, Canada, Australia and the European Union are taking up some of the tools the U.S. should consider to keep computerized trading from running amok. To cite a few examples:
-- In Germany, legislation is pending to require high-frequency traders to register so regulators can better track their market moves.
-- Canada charges fees to firms that attempt to clog markets with buy and sell orders, as well as cancellations, a practice known as quote stuffing. High-frequency trading firms sometimes do this to overload the less-sophisticated trading systems of rivals and exploit minuscule and fleeting price discrepancies.
-- Australia will ask trading firms to conduct stress tests to gauge how they deal with market shocks.
-- The EU is reviewing a number of measures including one that would require a trading firm to honor a bid for half a second, a lifetime in a market where trades can be executed in microseconds.
It wouldn’t hurt if U.S. regulators took a look at these options and considered a few others, as well.
One is the creation of a tracking system to provide a consolidated audit trail of all trading in real time. The SEC has ordered markets to build such a system, but it will generate data only the next day. In a hyperspeed world, that’s too late.
High-frequency traders also should be required to make markets in securities, serving as buyers and sellers of last resort. One reason the flash crash was so deep was that traders pulled out as the market decline picked up speed. An obligation to buy when others sell can slow a fall. As part of this market-making duty, trading firms could be required to honor a buy or sell offer for a certain period of time.
Existing circuit breakers that temporarily halt trading and slow a market plunge are fine, but one idea worth considering is to have circuit breakers for individual high-frequency outfits. These would prevent a firm from sending out an inordinate number of orders to engage in stuffing.
Every firm also needs a “kill switch” to halt trading in an instant. If Knight Capital had one, it might have avoided calamity.
High-frequency trading is here to stay. It has the power to add real value to markets -- but only if it is tempered by sound rules to ensure that human beings, not machines, are in charge of the markets.
To contact the senior editor responsible for Bloomberg View’s editorials: David Shipley at firstname.lastname@example.org .