Thursday, May 05, 2011

Commodity Corrections

As we close in on the one year anniversary of the flash crash, there are some fireworks on display in the commodities markets:
Commodities plunged the most since 2009, led by oil and silver... The Standard & Poor’s GSCI index of 24 commodities sank 6.5 percent... and has lost 9.9 percent this week. Oil tumbled 8.6 percent, the most in two years, to $99.80 a barrel. Silver dropped 8 percent, extending the biggest four-day slump since 1983 to 25 percent...

Selling swept commodities markets as investors sold positions following gains of more than 23 percent in 2011 through April 29 by silver, oil, gasoline, coffee and cotton... Futures on Brent crude, crude oil, gas oil, heating oil, gasoline and natural gas plunged more than 6.9 percent today. Crude oil dropped below $100 a barrel for the first time since March 17. Copper futures slumped 3.3 percent, falling below $4 a pound for the first time in five months. Among agricultural commodities, cocoa, cotton, corn and weak retreated more than 2.3 percent in futures trading.
Adherents of the efficient markets hypothesis will look for fundamental explanations for the sell-off, and will doubtless come up with some plausible triggers. But as John Kemp observes in an excellent post, it's impossible to understand the plunge without first recognizing that prices in speculative asset markets can become disconnected from fundamental values from time to time:
It will be entertaining to read the thousands of gallons of ink spilled over the next couple of days as journalists and analysts try to rationalise the sudden turn around and identify that one or few factors that were the “tipping point.”

In reality, commodity prices and other assets rise because investors and hedgers anticipate further gains. The market needs a steady stream of net buying orders to keep rising. But at some point the risk of a setback outweighs the prospect of further gains. Long liquidation offsets fresh buying orders, and the process heads into reverse as the length cascades out of the market.

Given the powerful role of expectations and sentiment in building and sustaining coalitions of long (or on occasion short) investors and hedgers, there does not really have to be a rational cause for the market to turn on its tail, if by rational we are looking for a trigger that seems proportionate to the effect caused.

Even in retrospect, and after thousands of hours of econometric analysis, it has proved impossible to identify rational triggers for big market movements ranging from the stock market crashes of 1907, 1929 and 1987, to the flash crash of May 2010, the implosion of the technology bubble in 2000 or the sudden collapse of the subprime madness in 2007-2008.

None of the prior market movements was in any rational sense sustainable. But when it comes to identifying a specific trigger that caused the market to peak and then head into sudden reverse, it has proved impossible in every case to find the rational cause.
I have little to add to this, except to suggest a more disaggregated view of speculative behavior and an explicit recognition of belief heterogeneity. At any point in time there are a variety of price views within the population of speculators, and trading based on this distribution of beliefs causes prices to move. Prices rise if those expecting appreciation are more confident or better capitalized than those expecting depreciation. The rise then reinforces the price views of buyers and further increases their capitalization advantage relative to sellers. This propels further appreciation.

The main check on the process, as Kemp says, is the increasing perception among some investors that "the risk of a setback outweighs the prospect of further gains." When such fears become sufficiently widespread, further price appreciation is arrested. But the crash does not follow until selling is synchronized, an event whose precise timing is essentially impossible to predict. 

There is some evidence that bubbles can be identified in real time by examining the prices of securities that provide crash insurance. But regardless of whether or not this can be done, the presence of non-fundamental volatility in speculative asset prices is important to consider in the execution of monetary policy. Headline inflation has recently exceeded core inflation largely due to pressures from commodity prices. This has put Fed officials in a bind, uncertain of the relative weights to place on the two measures. If there's a lesson in today's events, it is that the speculative components of inflation measures should not have first order effects on monetary policy, at least until the economy is operating closer to its capacity.