Illustration by Serifcan Özcan
Illustration by Serifcan Özcan

Last year, when the U.S. Securities and Exchange Commission came out with its final report on the flash crash, the stomach-churning event of May 6, 2010, that wiped $1 trillion of value from the markets in less than 30 minutes, it never managed to explain why the episode happened.

A large trade of stock futures by a Kansas firm had sparked it, the report said, and it detailed the ensuing chain of events, without offering much insight into why such a tumult was possible.

The report did say the activities of high-frequency traders, firms that use computers to buy and sell thousands of times per second, had helped the trouble spread. Investors today can’t be terribly comforted because hundreds of minor flash crashes involving only a few stocks at a time have struck since then. Earlier this year, for instance, the share price of an insurance company called Enstar Group Ltd. fell from roughly $100 to zero, then rose back to $100 in the space of just a few seconds. Since we don’t know what’s causing these sudden shifts, we can’t be sure that one as big as, or even bigger than, the May 2010 event won’t occur tomorrow, or next week.

The good news is, people are still studying the situation. One of the most illuminating recent analyses comes from Andrew Haldane, the Bank of England’s executive director for financial stability. High-frequency trading, he says, really is making the market less stable, and for understandable reasons. Unfortunately, identifying what is happening appears to be more straightforward than figuring out what to do about it.

High-frequency trading now accounts for almost 75 percent of all buying and selling of U.S. equities, and a race to trade even faster is under way. The technology to trade thousands of times per second has existed for several years, and it won’t be long before that rate is pushed to a million. The fastest exchanges are now executed in about 10 microseconds -- the time it takes a jetliner flying at full speed to travel one 10th of an inch.

This speed has clearly brought some benefits. Consider, for example, the so-called bid-ask spread: the small difference between the prices at which an investor can buy or sell a particular stock. Smaller spreads mean cheaper trading, and data suggest that spreads in the U.S. and the U.K. have, over the past decade, narrowed to about a 10th of their previous values. For slightly technical reasons, the bid-ask spread is also a convenient measure of market liquidity: how easy it is to sell a stock at any moment (without having to cut the price). Narrower bid-ask spreads mean higher liquidity and markets that run smoother.

Abrupt Price Changes

But a single measure never tells a complete story. Other evidence suggests that as markets become more liquid -- and hence, more stable most of the time -- they also become more prone to abrupt gyrations.

An informal study by the New York Times, reported a few weeks ago, found that single-day share-price changes as large as 3 percent to 4 percent are now more likely than at any other time in recent stock-market history. Academic studies show much the same thing: In the past six years, large and volatile movements in stock prices over periods of 15 minutes or so have become much more common.

This isn’t surprising. It just means that markets now behave the same way over short periods (minutes) as they have done over hours and days -- with a marked tendency toward sudden movements. It’s when you consider the consequences of this increasing volatility that Haldane’s analysis becomes most interesting: He argues that the precarious position in which some high-frequency traders find themselves, especially during stormy periods in the markets, may cause their actions to exacerbate the trouble.

Here’s why. The bid-ask spread is largely set by traders who act as so-called market makers. They don’t speculate on price movements, but simply stand ready to take either side of a trade, and aim to profit from the spread. Market makers can reduce spreads to attract trading, but they can’t go too far because they face risks -- for example, that a stock they buy might lose value before they can sell it. Ultimately, the bid-ask spread reflects how much market makers have to charge to take on such risks and still be profitable, while making it easier for everyone else to trade when they want.

Importantly, Haldane points out, the risks that market makers face grow with the likely size of price movements. In increasingly volatile periods, high-frequency market makers have to charge more because they get burned more often by unexpected movements. Hence, it’s quite natural to expect a market maker, during an episode of market chaos, to widen its bid-ask spreads, sometimes drastically.

Vanishing Liquidity

And this brings us to the main point -- that it’s the sudden evaporation of such liquidity that propels events such as the flash crash. All in all, then, it seems likely that the very high-frequency trading that makes the markets run so smoothly in quiet times does the opposite in stormy times, exaggerating the chaos.

Some critics say the liquidity these traders provide when markets are calm is elusive: It vanishes in rough periods. This seems to have helped make the 2010 flash crash so extreme, and is, in Haldane’s analysis, exactly what we should expect.

In principle, the very computing technology that has made high-frequency trading possible might also be put to work in controlling the instabilities it creates, possibly through circuit breakers that would kick in automatically on early-warning signs of liquidity disruption. A mechanism for doing that effectively may be a long way off, though, and simple speed limits on trading -- making every transaction take at least a time T, which might be one second, or something else -- would probably be much better in the short run.

But that, of course, would mean accepting a trade-off. In reducing the likelihood of flash crash events, we would have to accept wider bid-ask spreads during the market’s normal peaceful periods -- and that would mean more expensive trading. Very possibly, as Haldane has helpfully pointed out, we just can’t have it both ways. (For those interested in some of the technical details of Haldane’s argument, see my blog).

(Mark Buchanan is a theoretical physicist and the author of “The Social Atom: Why the Rich Get Richer, Cheaters Get Caught and Your Neighbor Usually Looks Like You,” is a Bloomberg View columnist. The opinions expressed are his own.)

To contact the writer of this article: Mark Buchanan at

To contact the editor responsible for this article: Mary Duenwald at