Emergent misbehavior

by on May 13th, 2010

How would you like a beer? How about a beer company along with it?

On Thursday, the 6th of May, for a few minutes, you could have bought a delicious Sam Adams plus a substantial interest in its maker, the Boston Beer Company, all for the price of a pint. Boston Beer stock, along with dozens of others on the major U.S. stock exchanges, plummeted to zero, while the Dow Jones Industrial Average nosedived 700 points in a matter of minutes.

To the great relief of most traders and to anyone whose financial well being is linked even indirectly to the stock market–and that’s pretty much all of us–the market rebounded almost as quickly. Still, the wild ride left even seasoned traders in shock.

It’s hard to overstate how much value was at risk during this ten-minute event. As just one example, Exelon, a utility worth about $30 billion at 2:49 p.m. was worth nothing three minutes later. It’s estimated that one trillion dollars of value evaporated during the “flash crash.” That’s three times what the U.S. spends on public education per year, $300 billion more the U.S. government bailout of the banking system in 2008, and about equal to the current European package to rescue Greece.

Some of the most extreme trades were eventually erased. The tech-heavy NASDAQ decided to annul trades that took place during those critical minutes and at more than 60 percent above or below a stock’s pre-anomaly value.

The grab-your-airsick-bag crash and rebound was an anomaly, but that’s not the same as saying that it was an error, in the sense that it was caused by some specific mistake or malfunction.

Economist and market analyst John Hussman points out that U.S. stock markets have hit similar “air pockets”– in 1955, 1987 and 1999. Like the Thursday event, those episodes resulted in roughly ten percent losses. The big difference is that they played out over weeks rather than minutes.

Since the Thursday debacle there’s been no shortage of fingerpointing.

Early speculation centered on a so-called “fat-fingered trade” as the trigger for the selloff. Instead of offering to sell a few million shares of Procter and Gamble, rumor had it that a trader mistakenly put up a few billion shares. Lacking buyers, the stock tumbled, starting a panic that took the rest of the market down with it.

The theory got a lot of attention, but like the infamous weapons of mass destruction in Iraq, there’s no evidence for it.

The most recent theory is that as the market started to fall a particular hedge fund placed a $7.5 million bet that the drop-off would continue, and the rest of the hedge funds followed suit. Do lemmings come to mind?

One suspect that most market gurus agree on is high frequency trading. Multiple firms now trade using high speed computers linked directly to the stock exchanges. These constantly analyse massive amounts of data and exploit fleeting opportunities by buying and sell huge quantities of stocks and futures in milliseconds. Experts estimate that these automated agents now make from sixty to seventy percent of all trades.

The existence of these computerized agents goes a long way towards explaining what happened, and the absence of an identifiable trigger.

If there’s one thing we’ve learned about complex systems since chaos theory pioneer Edward Lorenz popularized the idea of the “butterfly effect” in the 1960s, it’s that they are capable of amplifying the tiniest perturbation to virtually any scale. It takes just one last snowflake to unleash an avalanche.

The stock market is a classic example of a highly dynamic system driven by many independent but interacting agents. One state that it’s capable of occupyin–what system theorsists refer to as an attractor– is when the tug of war between buyers and sellers arrives efficiently at a stock’s current value. That’s the state that economists tell us represents the stock market’s raison d’etre.

It would be great if that were the only way the system functions. Unfortunately, history shows that the stock market can also wander into at least two other states or attractors. It’s prone to huge bubbles, in which contagious enthusiasm drives the prices of most stocks well above their “true” value, and, as we’ve just seen, “air pockets” in which contagious panic does the opposite.

That was bad enough when human traders were the ones calling the shots. Presumably they had some sense that a company valued at $30 billion one minute couldn’t really be worth zero a few minutes later. Their interaction led to dramatic booms and busts, but at least these had believable tops and bottoms and unfolded on a human time scale.

Over the years the markets have instituted various fixes to try to keep the market from manifesting its most unattrractive attractors. After the global “Black Friday” market crash of 1987, The New York Stock Exchange, for example, put in place “circuit breakers”–trading curbs that snap into place when the market falls too quickly and that are supposed to slow panic selling and so prevent a full-scale crash.

Some market analysts are blaming the circuit breakers themselves for the Thursday meltdown. They think that when the NYSE circuit breakers clicked in, the effect was to shunt the flood of sell orders to other markets that were even less able to find buyers for them .

(Just as a star needs to maintain a continuous flux of nuclear fusion to keep from collapsing under the force of gravity, stock markets need to continuously match sellers and buyers. If there are no buyers, stock prices start to fall. We now know that computerized trading can drive a sagging stock to zero in minutes, and can threaten to implode the entire market).

The circuit-breaker problem has gained traction. Six major exchanges have now agreed to strengthen and coordinate their circuit breakers. New rules are currently being negotiated and should be in place within a few weeks.

Those fixes may be good ideas, but they almost certainly are nothing but temporary patches. The system remains as complex, dynamic, and unpredictable as ever. It’s still shuttling hundreds of billions of dollars form buyers to sellers at inhuman speeds every day, impelled not just by humans vacillating between greed and fear, but increasingly by computerized agents impelled by abstruse algorithms. There’s no “beta testing” for these patches, leaving all of us as guinea pigs in a very high-stakes experiment.

Regulators and investors would like to believe that the proposed fixes will result in an efficient, reasonably stable market. I think it’s more accurate to view the market as something like a manic-depressive chef on speed–brilliant at what it does but capable of cooking up a disaster at any time.

Thursday’s collapse and rebound, and the current fix-it-on-the-fly patches, ought to make normal investors think hard about their nesteggs. Harry Truman’s aphorism about politics seems even more appropriate for investors in today’s market. “If you can’t stand the heat, get out of the kitchen.”

Robert Adler