Speculation Redux
From Econbrowser, I learned that Paul Krugman (oy) made a point that I was going to blog about, but my procrastination gene kicked in. Specifically, Krugman opined that if speculation were really driving up prices above the equilibrium level, we should also observe an increase in inventories. In energy, and in some other commodity markets, inventories have actually been decreasing. So Krugman concludes–reasonably–that this is inconsistent with the ceaselessly repeated mantra that speculation is causing prices to diverge widely from the equilibrium fundamentals-driven value.
There are historical examples of this phenomenon. The International Tin Council attempted to prop up the price of this metal. The ITC had to buy increasing quantities of tin, as the high price discouraged consumption and encouraged production. Eventually, inventories became so large that the ITC could no longer afford its scheme; tin was dumped on the market; and the price plunged. The accumulation of stocks of grains by governments (including the US government) as the result of price support programs are also good examples.
James Hamilton at Econbrowser is skeptical, noting that in the short run supply and demand for most commodities is so inelastic that prices can diverge substantially from the equilibrium value without leading to a detectable increase in stocks.
If forced to choose, I would go with Krugman on this. The speculation-is-driving-the-market (nuts) mantra has been heard for nigh onto three years, which is more than enough time to see an accumulation of stocks in the hands of speculators (and it would be speculators that hold the stocks as they are the ones allegedly willing to pay the inflated price). Yet this has not happened. Hamilton argues that yes, the last three years have been fundamentals-driven, but the January-February spike may not be. In such a short time frame, he notes, given the inelasticities in supply and demand it may be very difficult to detect any speculative distortion in stocks given their natural volatility. That’s a fair point, but I am reluctant to conclude, based on this absence of evidence that speculators did not cause a distortion, that prices were irrationally high in two months after being fundamentally driven for years during which speculative activity was also intense.
I do have one quibble with Krugman (about this issue–about politics, quibble doesn’t come close to covering it.) Krugman’s analysis is a static, supply-demand one. Any analysis of inventories must incorporate dynamic effects.
Moreover, a dynamic analysis implies that it is possible that stocks and prices can rise simultaneously even if all speculation is rational, i.e., in a rational expectations equilibrium. I sketched the argument in the 2006 post “Care to Dance the Contango?” There I conjectured that increases in uncertainty could cause the simultaneous increase in inventories, prices, and contango in the oil market. I’ve recently completed a stochastic dynamic programming model that formalizes the intuition from that post. (A working paper is in progress, but its completion will have to await my return from Europe in June.)
Commodity storage dynamic programming models posit that equilibrium speculative storage decisions maximize welfare. Given current inventories, the state of the economy, and the stochastic processes governing the evolution of this state, speculative storage decisions maximize the expected present value of net surplus (consumer surplus plus consumer surplus). Equilibrium prices adjust to provide maximizing individual agents the incentive to implement the optimal plan.
One proceeds by conjecturing a function that relates forward prices to the current state of the economy and carry-in. One then solves for the amount of carry out such that the current spot price equals the present value of the forward price; if the spot price with zero carry-out (i.e., complete consumption of inventory) exceeds the forward price, it is optimal to store nothing, and to have a “stock out.” The possibility of stockouts (which will occur with positive probability in equilibrium) leads to backwardations.
Conventional storage models typically restrict attention to a single, mean reverting (i.e., stationary), homoskedastic net demand shock. Other models I have worked on posit two mean reverting homoskedastic shocks, with different speeds of mean reversion.
In this most current work, I have two state variables: a mean reverting net demand shock, and a mean reverting stochastic volatility shock. That is, I don’t assume that net demand is homoskedastic; the second state variable in the economy is the volatility (or variance) of the demand shock.
This framework formalizes the intuition in “Dance the Contango.” And the results of the model are exactly as conjectured in that post. When there is a positive volatility shock, agents respond by increasing storage, and thereby reducing consumption. This leads to a simultaneous increase in inventory and price–a phenomenon that has mystified many commentators on the energy markets in recent years.
I presented the results of the analysis at a conference in Italy in January. Here are the slides, complete with some graphs illustrating the results of the model.
In sum, although a simultaneous increase in stocks and prices could be a consequence of speculative distortion–but this is not necessarily the case. I would surmise that the absence of inventory accumulation is strong evidence against the excess speculation hypothesis, but that the existence of inventory accumulation when prices are rising does not imply that speculation is distorting prices.
I understand that this is not an especially satisfactory conclusion, but it illustrates that it is very difficult to identify implications that definitively distinguish the competing hypotheses. Methinks, however, that it is necessary to go beyond the crude correlation=causation argumentation that dominates the current debate about the impact of speculation: “There is a lot of speculation. Prices are high. Therefore the speculation caused the high prices.” The exact mechanism by which this works is seldom specified. Krugman is on the right track to point out that the excess speculation hypothesis should have implications for quantities as well as prices. His Micro 101 analysis is a good place to start, but when considering storable commodities, it is necessary to model the dynamics. When one does so, one finds that his test may generate false positives. That is, in a world where the amount of fundamental uncertainty is itself random, a simultaneous increase in stocks need not be definitive proof that speculation is distorting prices. It may instead show that speculation is in fact ensuring that resources are being allocated efficiently.
It is this very difficulty of coming up with empirical tests that cleanly distinguish between the rational speculation hypothesis and its irrational speculation alternative that make it so difficult to resolve this debate. Which is why it has raged for centuries, and will continue to rage into the foreseeable future.
