Academia: indices DIDN'T break the wheat market

A few weeks ago the Senate released a report blaming speculation and specifically index funds for high wheat prices and non-convergence of the futures and spot markets. It was interesting reading but received mixed reviews. At the time I stated:

It's very complicated to prove anything in this arena - my hunch is that the indices did play a large part in non-convergence of wheat contracts at the least, and maybe higher volatility as well, but I haven't seen anyone prove it to my own satisfaction.

Well, turns out I was wrong, according to Scott Irwin, a professor of agriculture at the University of Illinois. In a recent post on Econbrowser, he assembles the academic work on the topic and points the finger at two other culprits:

Our research at the University of Illinois pinpoints two major factors contributing to the lack of convergence between cash and futures price in wheat. The first factor is the tendency for spreads in the futures market to reflect a relatively high percent of full carry (contango) since 2006. The second factor is long-term structural deficiencies in the delivery system for CBOT wheat.

I won't do more than summarize (you should read his whole post if you're interested), but on the latter, the basic idea is that because of non-negligible delivery costs and an oligopoly controlling delivery locations, the market isn't entirely efficient (for reasons not yet understood).

It is not clear why these firms have apparently left so much money on the table by not arbitraging the very large differences between cash and futures prices. Understanding the motivation and strategies of these firms would be a far more informative line of inquiry than the current obsession with index funds.

As for the former, the following chart is worth a thousand words:


Days 5-9 are when the so-called "Goldman roll" takes place, i.e. where we would expect to see the impact of the influx of money into long-only commodity indices. And we see a bump... but the effect is not persistent (it subsides immediately in Days 10-13) and it dates back at least a decade. In light of this evidence, it's hard to believe the Senate report's assertion that index rolling was responsible for the recent and persistent widening of spreads.