Data centers are the hubs of a variety of services that businesses and consumers rely on daily, from cloud-based applications and file storage to digital communications and more. They are also increasingly becoming the hubs of the financial world, powering what is known as high-frequency trading (HFT) in the stock market. It’s certainly unsurprising that computers now perform many tasks in the financial world—just as they do in many other areas of life—but the potential dangers, especially in a weak economy, are shocking.
Computers Running the Markets
High-frequency trading, sometimes called algorithmic trading, “is the biggest thing to hit Wall Street in years. On any given day, this lightning-quick, computer-driven form of trading accounts for upward of half of all of the business transacted on the nation’s stock markets,” according to The New York Times (“Searching for a Speed Limit in High-Frequency Trading”). These speedy trades generally don’t net huge profits, but when thousands or millions are conducted each second, small individual gains can quickly add up to large aggregate gains. Obviously, to conduct these kinds of trades, computers must analyze available data quickly, but this information doesn’t involve information like business plans, individuals in leadership positions of companies and so on—information that often drives the market decisions of savvy (human) investors.
With such a large volume of trades being carried out by computers according to a formula (which is necessarily bereft of critical business data that would normally inform a sound investment), individual investors are increasingly becoming an outside group for whom the stock market provides fewer and fewer opportunities to garner profits. Issues of the benefit of the stock market to individual investors aside, the question is whether this mix of computers and the stock market poses serious dangers not only to investors but to the economy at large.
High-Frequency Trading and Flash Crashes
With the advent of algorithmic trading, where millions of trades are conducted by computers in seconds or less, has come a new danger: the flash crash. A flash crash was reflected in the Dow Jones index on May 6, 2010. As Wired (“Nanosecond Trading Could Make Markets Go Haywire”) relates, “Starting at 2:42 p.m. EDT, the Dow Jones stock index fell 600 points in just 6 minutes. Its nadir represented the deepest single-day decline in that market’s 114-year history. By 3:07 p.m., the index had rebounded.” At one point that day, the Dow Jones index had lost more than 1,000 points. To be sure, the world didn’t stop turning; the market regained its footing, and the problems were attributed to computer glitches. But such tremendous swings in the market are nearly impossible when trades are conducted in a slower and, arguably, more rational manner.
A more narrow, but equally disturbing, example of the dangers of high-frequency trading manifested itself in early August of this year when brokerage firm Knight Capital Partners was nearly wiped out financially as a result of “a bug in one of its high-frequency trading algorithms [that] caused the firm to lose $440 million,” according to Time (“High Frequency Trading: Wall Street’s Doomsday Machine?”). Of course, companies engaging in high-frequency trading are virtually certain to add safeguards to their programs to prevent these kinds of catastrophic losses, but computers do exactly what they’re told, which isn’t always what the programmer wants or expects.
Perhaps the greatest danger of high-frequency trading isn’t the individual examples of a buggy program that loses a company some large chunk of money, but the aggregate behavior of all these programs together affecting and being affected by market conditions. A single program may behave in a fairly predictable manner, but when many of these programs interact, the results can be less predictable (this may fall into the category of “emergent behavior”). Flash crashes are a potential consequence of these interactions.
Add Sour Economy
In a strong economy, flash crashes are likely more of an annoyance than a real danger. Given the already precarious situation much of the western world currently resides in, however, flash crashes pose more of a danger. The current extremely low interest rates encourage borrowing—meaning greater debt—at all levels of the economy, from the consumer to businesses to governments. And at all levels (generally speaking), this debt is staggering. Furthermore, the fractional-reserve banking system means that if depositors decide en masse to withdraw their money, banks will quickly lose their reserves; not everyone will be able to get their money.
If a flash crash (say it lasted more than minutes—maybe a day or two) causes a panic, leading to a bank run, the situation could easily snowball into an financial calamity similar to, or even worse than, the events leading up to the Great Recession of 2008 and 2009. Thus, although high-frequency trading has its own inherent dangers as exemplified by flash crashes and losses like those suffered by Knight Capital Partners, the greater concern is the economic tinderbox that could go up in flames in response to this kind of stimulus.
So, What’s the Answer?
The involvement of computers and data centers in stock markets certainly isn’t wholly undesirable: it is a means of increasing the ability of individual investors and small companies to the market, and in doing so, it reduces the costs of trading by increasing competition. But anything can be taken to an extreme. High-frequency trading more closely resembles a game of poker played by master strategists than a means of profiting by providing a fiscal foundation for a company to do real, productive work. Does that mean regulators should tighten controls on the market? No, but it should be a warning to investors—particularly individuals. The stock market is largely rigged: individual investors are generally unable to make significant profits from the market, in part because of inflation (which, ironically, is a strategy of the Federal Reserve to prop up the market’s value). And inflation means a bubble: eventually, the credit card must reach its limit, at which time the bubble will burst.
Data centers aren’t the problem, per se. The problem is that high-frequency trading is more or less a sneaky way to make profits without even nominally creating any value. A business ideally turns investments in its operations into profits by producing something that is in demand. Investors then reap a portion of these profits. Algorithmic trading looks not to business productivity but market quirks to make money off the system, not off productivity. People should be allowed legally to play petty games with their money; investors, however, should be aware of such games. In this case, however, the scope of high-frequency trading poses a danger to an already fragile economy.
This may all seem like a matter of economics, but many companies are involved to some extent—perhaps through customers—in finance by way of their data centers. Data center space around cities like New York is at a premium as financial companies seek to minimize their distance from the market floor, thereby maximizing the potential speed of their trades. Data centers, in and of themselves, are not at fault for the situation with high-frequency trading, but they are at the center of it.
Photo courtesy of francisco.j.gonzalez