The FTC's investigation of the Midwest Gasoline price spike found no evidence of any anti-trust violations. In my view the main culprit was the introduction of Phase II reformulated gasoline and the adjustment to that new requirement, particularly in the Chicago and Milwaukee area, where refiners use ethanol rather than MTBE in reformulated gasoline (RFG).
Wholesale prices for reformulated gasoline in the Chicago area were an average of 22 cents above normal, benchmarking to other markets, for a period of eight weeks. Conventional gasoline prices were also higher than elsewhere during May and June. While these price spikes were significant it is worth noting that prices never came close to the monopoly level.
Assuming that the price elasticity of demand was -.2 before the spike, that the wholesale price pre-spike was $1 per gallon, and the marginal cost of producing a barrel of gasoline is no less than the marginal cost of purchasing a barrel of oil, then standard economic calculations would imply a short-run monopoly price for gasoline of $3.30 to $5.60.(2) Therefore it is fair to say that even during the spike prices rose only by a fraction of the difference between normal and collusive prices(3). In this context, and given the obvious and legitimate explanations for the price spike, it would have been very surprising to find evidence of illegally coordinated supply reductions or other activity.
Adjusting to the new requirements was expensive. As happened in California with the introduction of California Air Resources Board- required gasoline, some refineries in the market concluded that they would be unable to recapture the investment required to fully upgrade their facilities to meet the new reformulated gasoline standards. One of these refineries closed down shortly thereafter, in January 2001. Without the full refinery upgrades these firms' only option was to switch part of their production from reformulated gasoline to conventional gasoline.
Even refiners who upgraded found their overall production possibilities reduced. Roughly, every gallon of reformulated gasoline a refinery produces reduces its potential output of conventional gasoline by more than a gallon. With Phase II RFG the "terms of trade" became worse, so that refineries that produced as much total gasoline as before would have to produce less RFG, and refineries that maintained their RFG production produced less total gasoline.
In addition to the capital expense involved in adjusting to the new regulations, there were also technical problems. While some refineries were able to adjust smoothly to the new standards others had a great deal of difficulty. These difficulties led to short term supply shortages. Had the refineries that were able to transition smoothly understood the degree to which their competitors were having problems then perhaps they could have done things to mitigate the price spike. But firms do not swap supply projections for both competitive and legal (antitrust) reasons, and the shortage was underestimated.
While the supply shortages might not have been noticed in other markets, they were highly noticeable in the gasoline market. The highly inelastic demand for gasoline meant that even a small shortage would lead to a large price increase. Ironically, a shortage of the same magnitude would not have led to nearly as large a price increase were the market monopolized. For example, a collusive industry would never raise prices by more than 5 percent in the face of a 5 percent production shortage, essentially because it would already be charging very high prices to begin with. If prices rise by 30 percent (at wholesale, 20 percent at retail) in the face of a 5 percent quantity reduction, the only implication is that prices are normally far below the collusive level, and the industry is generally highly competitive.
The price spike was made more noticeable by its timing. OPEC's coordinated production cuts had already led prices higher. Furthermore, gasoline prices generally rise in the summer because of the increased demand, which requires firms to build stocks in the winter, thus incurring inventorying expense. The higher demand also causes firms to employ higher cost production alternatives in order to maximize production. So, the additional 45-48 cent price increase in Chicago on the peak day of the price spike came at a time when prices were already fairly high. Of course, gasoline price increases are generally highly noticeable because such a large fraction of consumers will purchase gasoline within any given week. For these reasons the price spike received attention well beyond its true economic impact. Over the course of the spike, the average household in Chicago/Milwaukee that uses reformulated gasoline incurred $30 in extra expenses --- and midwestern households that use conventional gasoline paid an extra $20. While these sums must be added over a large number of households the total effect was still small relative to the exceptional publicity the price spike received.
Mitigation of the price spike was made more difficult by several factors. Chief among them again was the choice of ethanol as an oxygenate in Chicago-Milwaukee. This meant that unanticipated shortages in that region could not be met by shipments of RFG from the vast majority of the country that uses MTBE. Rather, refineries that were producing RFG with methanol would have to decide to switch to producing RFG with ethanol during their next production campaign and then ship the gasoline to Chicago --- creating a lag in supplying the market of several weeks. Again, there is an analogy with the California gasoline price spikes we have seen in the past. Refineries from outside the market must first switch over to producing the specialized gasoline, then produce it, and then ship it.
A second reason the spike was not mitigated as quickly as it might have been was the pipeline disruptions that plagued the midwest. The Chicago and Milwaukee region is a net importer of gasoline. When pipeline problems reduce the amount of gasoline that can enter the market, unless the local refineries have spare capacity, there will have to be a price increase to clear the market. The pipeline disruptions created shortages which undoubtedly played a significant role in the price spike for conventional gasoline in the region, since that differential should have been much easier to mitigate than the RFG spike. Furthermore, the three refiners that did not upgrade sufficiently to maintain their RFG production were able to increase their conventional gasoline production (and maintain overall production) so, if anything, their actions would have mitigated the conventional gasoline price spike.
A third reason that mitigation may have been more difficult than in normal times was that inventories were lower than they might have been at the same time of year in the past. Again, some of this might have been a result of the changeover in gasoline that required firms to drain old gasoline from storage tanks before refilling with new gasoline. But, it was also caused by the general and systematic reductions in inventories that have taken place over the years as firms have attempted to adopt tighter controls. The high price of oil, or more specifically the high price of oil relative to futures prices, made the cost of holding inventories much more expensive than usual.
Fourth, the waivers the EPA granted to St. Louis, allowing conventional gasoline to substitute for RFG, contributed to the problem in two ways. First, they contributed to the conventional gasoline shortage in Chicago as conventional gasoline shipped up to St. Louis could not go on to Chicago. On the other hand, the RFG made with MTBE that was displaced by conventional gasoline in St. Louis could not be shipped to Chicago either, because Chicago could only use RFG made with ethanol. The second way in which the St. Louis waivers may have worsened the problem is that they created regulatory uncertainty. They added credibility to the possibility of waivers for Chicago and Milwaukee, reducing the incentive to produce RFG with ethanol for those markets.
Despite all this, the industry's response to the price spike was remarkably rapid. By the time Congress asked the FTC to investigate why there was a price spike firms had already shipped new supplies into the midwest and prices began dropping almost immediately thereafter. The total length of the spike was less than two months. This means that the industry was able to take refineries in the Gulf of Mexico and design new production campaigns, geared to producing a new kind of gasoline that those refineries had never produced before, ship that gasoline through a somewhat impaired pipeline system, get the gasoline distributed to retailers, and do all this in sufficient quantity to push the Chicago-area wholesale prices below national levels in about the same amount of time it took our agency to wordsmith its report.
There are two final considerations that contributed to the shortage. One is the enormous variety of gasolines currently produced for the market. The more a market is segmented the more inventory is required to achieve any particular probability that an individual consumer will face a stock-out. Furthermore, the more kinds of gasoline there are on the market the more likely it becomes that an individual supplier of a particular product will find itself with some ability to raise price. In the midwest gasoline investigation, one firm indicated that it might have sold off more inventory than it did had it not been concerned that the extra sales would reduce price and therefore the profitability of existing sales. In the Exxon-Mobil investigation the Commission required the divestiture of an Exxon refinery because the California gasoline market was viewed as separate from the national gasoline market, making imports difficult and making the post-merger market concentration unacceptably high. So multiple varieties of gasoline not only raise costs by, for example, increasing inventory and shipping expense, but also will inevitably lead to more highly concentrated markets.
The last consideration is that over the years the capacity of U.S. refiners has not kept pace with the growth in demand for refined products. Average capacity utilization in 2000 was 94 percent, and higher in the summer. So most refineries are running at close to full capacity most of the time. This also means that there is no slack capacity available to respond quickly to a price shock. This trend is likely to continue. Most areas in the U.S. are not particularly interested in having a new refinery built nearby, and the cost of meeting U.S. regulations is high. If demand continues to outstrip capacity then ultimately imported products will become the marginal source of supply for some oil products. Inevitably it will take importers longer to respond to supply imbalances than domestic refiners, so it is quite likely that we will see more price spikes in the future.
Endnotes:
1. This speech represents my views and those of some colleagues in the Bureau of Economics. It does not necessarily represent the views of the Federal Trade Commission, the Bureau of Competition, or any of our individual Commissioners.
2. If demand is assumed to be linear, which means that every nickel increase in price causes the same absolute reduction in demand, then the monopoly price is $3.30. If every nickel increase in price causes the same percentage reduction in demand then the monopoly price is $5.60. For other assumptions the monopoly price would be even higher.
3. That is, if the normal price is $1 and the monopoly price is $5 then an increase to $1.40 is 10 percent of the way from the normal price to the collusive price.