High-frequency traders have a bad rap. One minute, they are being blamed for bizarre market behavior like the “flash crash” of 2010—which saw the Dow Jones Industrial Average lose and regain nearly 1,000 points over the course of 36 minutes. The next, they are being pilloried on best-seller lists for gaming the financial system with high-tech skulduggery, as in Michael Lewis’s 2014 nonfiction book Flash Boys. And Hillary Clinton’s latest tax plan aims to squelch the practice. Are high-frequency traders (HFTs) really the devils we make them out to be?
Check out more of our timely growth content here.
To get at the answer, Robert Korajczyk, a professor of finance at the Kellogg School of Management, asks a different, but related, question: “So how do they make their money?” Korajczyk, along with Dermot Murphy of the University of Illinois at Chicago, find that HFTs generally profit by providing liquidity to the market—suggesting that their unsavory reputation may be unwarranted.
HFTs as Back Runners
“There are all kinds of questions about the effect of HFTs on markets,” Korajczyk says, but they boil down to two basic suspicions.
The first is that HFTs profit by using their speed advantage to exploit large institutional trades, in which large “parent orders” are executed with many smaller “child orders” over the course of hours or days.
“If I wanted to buy 10,000 shares of Apple, I wouldn’t put out a single order—I’d put out smaller chunks over time,” Korajczyk explains. “HFTs are not privy to inside information, which would be illegal. But they do see the flow of these child orders. An HFT can infer that there’s a larger order happening, and then they can trade quickly in anticipation of the other orders that are coming.”
This behavior, known in academia as “back running,” can push up the prices of subsequent child orders within the large trade. (Back running also works the other way around: if an institution is trying to sell off a large number of holdings, HFTs can drive down prices by trading in anticipation of the child sell orders.)
HFTs perform a useful function in financial exchange markets by providing liquidity, since they never hold their positions for long. Ideally, this should work to the benefit of institutions seeking to make large or “stressful” trades—generally those involving more than half a million dollars. But back running invites suspicion that HFTs are actually incentivized to hinder stressful trades by buying or selling in anticipation of child orders, an activity which changes their pricing dynamics—presumably in HFTs’ favor.
While back running is not illegal, it is the kind of “dirty trick” that contributes to HFTs’ bad reputation. The trouble is, according to Korajczyk, the numbers behind that reputation don’t add up. After analyzing 150,000 large institutional trades on Canadian equities exchanges, Korajczyk and Murphy discovered that “HFTs are actually losing money on their trading activity,” he says.
Their analysis showed that in aggregate, the trading positions of HFTs resulted in a daily net loss of $100 per stock—precisely the opposite of what one would expect if back running were in HFTs’ economic interest. “Given all the allegations of how HFTs are hurting these large institutional trades, you’d expect them to be making profits on their own trading,” says Korajczyk. “But they’re not.”
“High frequency traders are actually losing money on their trading activity.”
Follow the Money
Instead, HFTs make their net profits by generating “maker’s fees”—per-transaction rebates issued by a financial exchange in order to incentivize liquidity provision in markets, leading to narrow spreads between bid and ask prices. Each maker’s fee “isn’t huge—it could be only 20 hundredths of a cent per share,” says Korajczyk. But those tiny amounts add up: so much so, in fact, that on average they cancel out the negative positions on the trades themselves and leave HFTs $300 in the black per stock each day. “Our evidence indicates that on average,” says Korajczyk, “the HFTs are making money by providing liquidity”—exactly how they are supposed to, in other words.
Which brings up the second suspicion about HFTs’ activity on financial exchanges: that they generate more “phantom liquidity” than the genuine article. The concept of phantom liquidity depends, again, on HFTs’ speed advantage in the context of large parent orders executed in smaller batches over time. If HFTs discern that a large trade in progress is stressful, then HFTs may start competing directly with that institution’s child orders.“They’ll provide liquidity for the trade at the beginning of the parent order, but later they’ll be trading in the same direction,” Korajczyk explains. “If Investor A wants to buy a lot of one stock, at the beginning the HFTs are selling—but later on, they’re buying more of that stock themselves than selling it to Company A.” Like a phantom, the expected liquidity in Investor A’s transaction suddenly disappears.
The research from Korajczyk and Murphy, who received his Ph.D. from Kellogg, does produce a smoking gun for this behavior: large stressful trades had an average bid–ask spread three to four times higher than non-stressful trades, due to HFTs “backing off” from providing liquidity to the stressful transactions. “Nevertheless,” he adds, “even for those stressful trades, on average the HFTs are still providing liquidity to the market because that’s where their profits are coming from.”
A Necessary Evil?
“Are HFTs evil, or just smart?” Korajczyk says. “It depends on your point of view. For the average big fund, they’re only happy if they can trade whatever they want with no price impact. But if you’re an HFT, why should you just take all their [child] orders when the information in the order flow is telling you something about the value of the underlying asset? There’s a lot of debate, but there isn’t a consensus about what would be an optimal market structure.”
High-frequency traders do have an incentive to maximize their speed advantage by any means necessary—even spending hundreds of millions of dollars to install their own fiber-optic links between exchanges in New York and Chicago, as Michael Lewis’s Flash Boys described. Korajczyk admits that “spending that much on a speed improvement that just transfers wealth from one group of investors to another is not necessarily socially optimal.” And yet, technology marches on. “Should we all still be using paper cards to trade stocks?” he says. “Obviously we need exchanges that can handle high volume and operate smoothly. But the debate is about whether we’ve gone too far in optimizing for speed. How much better is the world if prices can adjust in four milliseconds instead of a hundred? It’s an arms race.”
Just as killing all the mosquitos on earth in order to cure malaria could change the world’s ecosystems in unpredictable ways, removing or constraining HFTs—even for noble reasons—would have a ripple effect on the behavior of financial exchanges that could affect even individual investments. “If you’re a little guy like a day trader, having HFTs around is a good thing,” says Korajczyk, because their activity compresses the bid–ask spread on stocks down to mere pennies per share. “But if you’re another little guy who’s putting his money in a huge mutual fund, you’re going to be affected when that huge fund starts trading in and out of positions and the HFTs respond.”
Still, Korajczyk’s findings make it clear that policy makers who want to rein in high-frequency traders may have to find different crimes to pin on them than they first thought. HFTs may be good guys, bad guys, or something in between.