But the bigger problem may be systemic risk, as exemplified by the "flash crash" of May 6, 2010, and the $440 million loss incurred by Knight Capital in August as the result of a software glitch. There is an analogy to be made between the argument over HFT and that other tiresome acronym, TBTF. Too-big-to-fail banks defend themselves by arguing that they offer efficiencies smaller banks can't. Even Alan Greenspan has dismissed that claim. But even if the banks are right, there's the question—posed by Republican presidential candidate Jon Huntsman, among others—of whether the cost of the systemic risk they pose exceeds the benefits of the efficiencies. Likewise with HFT.
To be sure, some argue that they do not increase systemic risk. The most prominent defender of HFT practices, Tradeworx CEO Manoj Narang, argued in 2010 comments to the Securities and Exchange Commission that in a pinch, HFTs—unlike longer-term investors— can simply "turn off" their trading strategies and unwind their positions (something they effectively do at the close of each trading day anyway), "thereby eliminating any possibility of further losses" and immunizing them from the contagions that can afflict big quant firms. He also wrote that critics exaggerate the likelihood of "runaway algorithms" accelerating a meltdown, saying exchange rules and market regulations "eliminate most of the risk."
Narang concedes in emailed comments that "large orders executed algorithmically could pose a systemic risk if the technology is faulty, but this is not high-frequency trading. High-frequency traders hold small positions in their inventory and do not execute large trades." Flawed technology, he says, is what explains the flash crash, in which HFT "had at best a peripheral role." As for the Knight debacle, he says, "there was never any systemic risk to the stock market."
Maybe, but others argue that systemic risk is less a function of how HFTs actually operate and more about how markets perceive their activity, fairly or not.
"I don't think anyone believes that the default of an HFT would cause the international economy to collapse. That's not the point," says Wallace Turbeville, senior analyst at the progressive think tank Demos and author of a forthcoming study on HFTs. "HFT activity creates a perception of great liquidity in the market, but in fact HFT activity, because it's algorithmic, can switch from providing high levels of liquidity to consuming even higher levels of liquidity instantaneously. It's worse than no liquidity at all. If it can change by the flick of a switch, it's a very volatile situation."
In other words, HFTs enhance liquidity until they don't, and the uncertainty impedes investor confidence.
So, it could be that the market is in fact becoming an unpredictable casino rigged by the people who can buy the most computing power. Or it might be that investors are overreacting. As that eminent economist Stephen Colbert put it: "You destroy the global economy once, and everyone forgets all the times you didn't destroy it."