The Economist: Six Minute Madness
The author of this entry is responsible for this content, which is not edited by the Wilson County News or wilsoncountynews.com.
The recent, drastic temporary market drop in response to a tweet supposedly from the Associated Press regarding explosions in the White House and injuries to President Obama raised a number of questions about the market and how it reacts to information. When the false tweet went out, the Dow Jones Industrial Average fell by about 145 points in just over two minutes. When it became clear that AP’s Twitter account had been hacked and that there had been no such explosion, the market turned around and erased the losses from a few minutes previous. The drop (and associated loss in value of about $200 billion) and recovery took place in a span of about six minutes.
Volatility in stock price levels has existed for as long as the market itself, and prices fluctuate in response to a variety of factors. At the most basic level, at any point in time, share prices reflect the underlying potential of a firm to make money. Therefore, anything affecting a company’s ability to generate earnings (a global event, economic downturn, new product launch, or the death of a CEO) will have an impact on its stock price.
However, in the case of the AP tweet hoax, two things stick out in my mind: the speed of the fall and later rise and the fact that so much value was lost and regained over a single bit of information. The rapid reaction in the market was driven by computerized trading programs designed to analyze information in news stories (and apparently posts on social media) and initiate stock trades without human intervention. Twitter has been at the forefront of recent breaking news stories and AP is certainly viewed as a credible source. The announcement sparked a burst of activity, which then triggered other algorithms which were watching for a flurry of sales or a drop in prices. Human guidance is still the order of the day for many trading firms, and when the tweet regarding the explosion appeared, people exercised caution given the possibility of hacking.
High-frequency and other aspects of robotic trading have come under fire in recent years, particularly in the wake of 2010’s original Flash Crash. On that day in May, a large sell order (accidentally large, in fact) set off a free fall of almost 1,000 points in the Dow Jones Industrial Average in a matter of minutes. Markets were already on edge, concerned over European financial conditions and other issues, and the huge sell order sparked a frenzy of further computer-driven selling. Markets bounced back once the imbalance between sellers and buyers corrected itself, but the enormity of the market movement led to calls for investigation and regulation by the Securities and Exchange Commission (SEC).
High-frequency traders attempt to take advantage of tiny price differentials to make profits; automated strategies and sophisticated algorithms to analyze data are essential to success. The problem is that such systems are not (yet) perfect, and can exaggerate normal market movements. Given that industry experts estimate that half of stock trade volume stems from these robotic systems, it is easy to see how a mini-crash could occur.
The SEC has implemented some regulations which work to reduce the potential for flash crashing. One new rule which has proven its worth is a limit to the amount a single stock can fall before triggering a trading halt. Just this week, a large sell order for Symantec shares without a bottom limit set off a 10% decline in seconds and led to a five-minute trading halt, after which time the disconnect between buyers and sellers corrected itself and the stock came back.
A key question in all of this is how big the problem really is. Clearly, for individual investors, flash crashes are a source of concern and fear. For the unlucky ones who were burned by the market’s movements last week, there is obviously a financial toll. However, for the vast majority of people, the losses were purely blips on a computer screen with the situation at the end of the day actually better than when trading opened. One point which has been made since the AP tweet hoax is that companies should be restricted from using social media as a primary means to release corporate information to better guard against hacker-induced panics. If the people crafting algorithms know that valid corporate information will be coming from press releases and reliable media outlets through traditional means, they will likely place less weight on social media. Of course, we can assume that last week’s issue was a lesson learned, and that even now these sophisticated programs are being tweaked to guard against a repeat.
It is somewhat ironic that taking the human element out of trading has increased the chances for swift crashes. Back in the distant past, a panic mentality could wipe out market value in little time, and a more mechanical system with decision rules was a welcome restraint. Now, however, the lack of a human eye and discerning brain has generated a new risk for irrational swings. The ability of markets to efficiently process information has long been a basic tenet of the theory of stock behavior. Arbitrary decision rules, while beneficial in some context, clearly limit this efficiency when they are based on outcomes with no concern for the veracity of the initial stimulus. Moreover, an essential premise of thermodynamics is that maturing systems generate more instability and volatility than more primitive ones (when you had to dial up your local broker by phone, “six minute madness” was impossible). That is merely an unavoidable cost of progress, which is hopefully offset overall by a more responsive global trading mechanism.
Perryman is President and Chief Executive Officer of The Perryman Group (www.perrymangroup.com). He also serves as Institute Distinguished Professor of Economic Theory and Method at the International Institute for Advanced Studies.