Streetwise Professor

June 26, 2008

James Hamilton’s Analysis of Speculative Influence: A Counterexample?

Filed under: Commodities,Derivatives,Economics,Energy,Exchanges — The Professor @ 3:54 am

James Hamilton is a world-class econometrician who has done pathbreaking work on the relations between oil prices and the real economy. Recently, he has blogged extensively about the potential contribution of speculation to the price of oil. Although he argues that most of the runup is fundamentals-driven, he is open to the suggestion that the increase in price since January is not primarily the result of fundamentals, but is instead speculative in origin. He is sensitive to the objection that a speculative distortion in prices should also manifest itself in increases in stocks. He attempts to counter this criticism with an example:

A much more important way in which the spot price of crude would affect the refiner’s demand for the product is through an intertemporal calculation. Given my customers’ demand, I’m going to need to buy the product sooner or later. If you charge me a lower price today than you’re going to charge me next month, I’d choose to buy more today to put it into inventory. If you charge me a higher price today, I’d rather run down my inventory and buy the oil next month, and of course the futures market allows me an opportunity to lock in a price for doing just that. Thus by far the most important factor in refiner’s demand for July oil will be the August futures price. If my production plans left me willing to buy July oil for $124.25/barrel when August oil was selling for $124/barrel, I’ll probably want to buy July oil for $126.25/barrel now that I’m forced to pay $126/barrel for August oil. Thus to a first approximation, the spot price would move by exactly the same amount as the near-term futures price. A $1 increase in the August futures price would shift the demand curve for July spot oil up by $1. In this fashion, an ever-increasing volume of speculative purchases of the near-term futures contracts would drive the spot price up with them.

There are two fundamental problems with this example, one logical the other empirical.

With regards to the logical problem. The premise of the hypothetical reminds me of the old Steve Martin routine “How to make a million dollars without paying taxes! First, make a million dollars.” That is, James pulls the $1 increase in the August price out of a hat, like a magician producing a rabbit. But why would an increase in speculative futures purchases lead to a dollar increase in the price of the August contract? Setting aside that the speculators might be informed–in which case, the price response is likely rational and efficient–why would other market participants believe that an (uninformed) speculator’s purchase should lead to an increase in the price of oil in August? Perhaps they anticipate that the August contract will be offset, but the speculator will buy even more for September delivery. This just kicks the can down the road a bit. Due to the ultimate connection between the futures price and the price of physical oil (that results from the delivery mechanism) the only way an uninformed speculative purchase will lead to a price increase is if market participants anticipate that some portion of the speculator’s purchase will not be offset. That instead, the speculator will take delivery of oil, and USE THAT OIL INEFFICIENTLY.

Inefficient use can occur as a result of a corner or squeeze. The speculators could squeeze some future contract, causing distortions in the production and transportation of oil. (A squeezer demands excessive deliveries, and essentially forces the delivery market up the supply curve to outputs beyond the competitive equilibrium point.) The problem here is that many speculators are unlikely to execute a squeeze, although A Speculator might do so. Multiple speculators may collectively have the ability to squeeze the contract, but they also face a collective action problem. Each one reasons: if everybody else is going to take to many deliveries to drive up prices, why should I go along? Why don’t I just liquidate my entire position at the artificially high price, and let everybody else bury the corpse (i.e., dispose of the excessive deliveries at a depressed post-squeeze price.) Individual rationality of the speculators results in a non-cooperative equilibrium where no squeeze takes place. Thus, given that myriad hedge funds, pension funds, investment banks, etc., make up the speculative crowd, unless one or two of them possess market power it is highly unlikely that anticipation of a squeeze could cause the price to rise in response to an increase in speculative purchases.

Another story is that somewhere along the line, speculators will take excessive deliveries, and just hoard the oil, keeping it off the market to enhance the price. This scenario also faces the prisoners’ dilemma issue, and another problem as well. Even for an individual speculator, it is difficult–and arguably impossible–to make a credible commitment to carry out this inefficient action. This is analogous to the Coase durable goods monopoly problem. (My 1993 Journal of Business paper discusses this issue in detail in the section titled “Pure Monopoly Manipulation.”)

That is, once he takes delivery and liquidates his futures position (at a high price), why should he just hold onto the commodity he received via delivery, reducing consumption below its efficient level and elevating price going forward? He would rationally sell the good, driving down its price to the efficient competitive level. But, forward looking counterparties, recognizing the speculator’s inability to precommit to the inefficient policy, would conjecture that he will not carry through on his threat, and hence are willing to sell the future at the competitive price today. The long would like to precommit to holding supplies off the market, but this is problematic, not to say impossible.

In a nutshell: distortion in the futures price today via the mechanism of speculative purchases requires an anticipation that the speculators will somehow take actions (a corner or hoarding) that reduce consumption below its efficient level in the future. There is no credible reason to believe that a large number of speculators can do this. Index funds that roll all their positions deterministically definitely cannot do this–they are effectively precommited NOT to take delivery. Ditto for those rading cash settled instruments, such as ICE WTI futures, or swaps (although the holders of such instruments could have an incentive to distort the physical market to enhance the value of their derivatives position.)

James’s example also poses other difficulties. The expectation that speculators will somehow take actions in the physical market that reduce supplies in the future (and such an expectation is a necessary condition for the price to move today, per the hypothetical) will induce immediate real responses. An anticipated increase in the demand for oil in the future will induce, at the margin, a decline in current consumption and an increase in current output, and an increase in stocks, in order to meet the anticipated future demand. That is, it is unrealistic to argue, as James’s example suggests, that an increase in future demand for the physical product will not lead to an increase in current inventories.

The empirical difficulty relates to the prediction inherent in the example: that there is a one-for-one relationship between the change in the spot and futures prices, especially when the market is in backwardation (as James deliberately specifies in the example, as he explains in his responses to comments). I have produced empirical evidence for industrial metals (J. of Bus., 1994) and oil (J. Futures Markets, 1996) that shows that when the market is in backwardation: (a) the correlation between spot and futures is well below one, and (b) the spot volatility is higher than the futures volatility. This implies that if one regresses the futures price change against the spot price change, the coefficient on the latter will be less than one–and often substantially so (e.g., on the order of .75 or .8). (Of course, if the regression is run the other way, the coefficient could be bigger or less than one, but in practice the coefficient is still less than one because the correlation effect dominates.)

Futures and spot move one-for-one when the market is in full carry (or nearly so)–but only because inventories (which are large when the market is in a carry) provides an intertemporal connection between spot and futures prices. This link is attenuated when the market is in backwardation. (I’ve also demonstrated this theoretically/numerically in solving dynamic storage models.) But when the market is in backwardation, the one-to-one relationship between spot and futures moves is unlikely to be observed in practice. Thus, if one treats the example seriously, as a source of empirical predictions about comovements of spot and futures prices, it is refuted by the data.

While musing about contango and backwardation, it is interesting to note that back in 2005-2006 those blaming speculation for distorting prices pointed to the strong contango as evidence that funds trading progressively more in the back end of the curve were distorting prices. These voices were notably silent when the market returned to backwardation last July. If anything, it appears that more activity is moving to the back end of the curve–so why did the contango go away? Moreover, although the price runup post-Katrina was accompanied by an increase in inventories–which is it at least superficially consistent with the price distortion story–the recent price runup was not accompanied by a jump in stocks. (And, as I said in my “Dance the Contango” post that is so loved by spam commentors, and in my recent working paper formalizing that intuition, in a world with stochastic fundamental volatility a positive co-movement in stocks and prices can occur in a rational expectations equilibrium, and hence is not necessarily a smoking gun of speculative distortion.)

In conclusion, I find James’s example unpersuasive. Although an increase in stocks that accompanies an increase in prices does not necessarily indicate speculative distortion, it is hard to imagine how the speculators could force prices above the competitive level without having to get their wingtips dirty with holdings of the gooey black stuff.

Print Friendly, PDF & Email

No Comments »

No comments yet.

RSS feed for comments on this post. TrackBack URI

Leave a comment

Powered by WordPress