On Wednesday, the Joint Economic Committee held hearings on gasoline prices and whether they are on the up-and-up. Sen. Chuck Schumer (D-N.Y.), the committee chairman, made his position — and the position of many of his fellow senators — perfectly clear. The oil companies should be busted up, he said, and lower prices will naturally flow.
Really? The best witness he had on hand to back him up was Thomas McCool, director of applied research methods at the U.S. Government Accountability Office (GAO). McCool contended that mergers and acquisitions in the oil sector in the 1990s have increased wholesale prices by 1-7 cents per gallon. Now, it would appear on its face that tossing the economic equivalent of an atomic bomb into the oil sector to reduce wholesale prices by a few pennies a gallon might not be the best idea in the world. Nonetheless, a close read of McCool’s testimony suggests that it’s an awfully thin reed to hang public policy on.
The first thing we notice is that McCool’s testimony relied exclusively on past GAO reports. The fact that there is a mountain of peer-reviewed academic work on this subject was unacknowledged in his testimony. This, unfortunately, is par for the course at the GAO. The implicit attitude over there is “if we didn’t do the study, the study isn’t worth looking at.” As a consequence, most GAO analysts are horribly ignorant about many of the issues they discuss. Now, I don’t know if Thomas McCool is familiar with the economic literature on these questions or not, but given his job title, I would doubt it.
Luckily, not all federal agencies act as if they are the font of all conceivable wisdom. The Federal Trade Commission recently published a thorough study on oil markets with due attention paid to the external literature on the subject. In a paper commissioned for that study from University of Iowa economist John Geweke, we find that academic researchers have been unable to lay down any good evidence that mergers and acquisitions have, on balance, increased consumer prices,” a finding all the more telling given the higher quality of that work. As Geweke notes in passing regarding an earlier GAO study on mergers and acquisitions in the oil business, which used roughly the same methodology as the more recent study, “assessment of the technical work in the GAO report is hampered by the fact that the report’s documentation of data and estimation methods does not generally meet accepted academic standards.” Geweke’s criticism was echoed in the FTC’s analysis of GAO’s 2004 study, which was savaged [pdf] by the commission’s economists (see the appendix).
Second, it’s important to note the distinction between changes in posted rack prices (which is what GAO used to reflect wholesale prices) and retail prices. The two are not the same. As the staff of the Bureau of Economics of the FTC noted back in 2004,
Rack wholesale prices and retail prices do not always move together, in part because rack prices do not necessarily measure actual wholesale transaction price, which are also affected by discounts, and in part because significant quantities of gasoline reach the pump without going through jobbers.
Hence, GAO did not find that retail pump prices increased by 1-7 cents per gallon. I didn’t even find that wholesale prices increased by 1-7 cents per gallon. It purported to find that posted rack prices increased by 1-7 cents per gallon. That may - or may not - have increased retail pump prices. FTC economists, for instance, agree with GAO that the Marathon-Ashland merger increased posted rack prices, but found no evidence that retail pump prices increased as a result.
In sum, what GAO found is equivalent to finding that this or that led Ford to increase the suggested retail price of a car by x. Maybe it did, but that “suggested retail price” has little to do with actual prices paid by new car buyers on car lots. Did McCool make this distinction clear? Not on your life.
Third, McCool’s depiction of GAO’s 2004 findings is highly suspect even in the particulars. The 2004 GAO study that McCool relied upon for his claims actually were two separate studies packaged under one binding.
One analytic exercise provided a total of 10 estimates of the effects of mergers and acquisitions on posted rack prices. Those estimates cover three types of fuel (conventional, reformulated, and specially blended gasoline for the California market) and different geographic areas. Seven of the 10 estimates — all involving either conventional or reformulated gasoline — found that mergers and acquisitions increased wholesale fuel prices by 0.15 cents per gallon to 1.3 cents per gallon. Although mergers and acquisitions were found to increase wholesale California gasoline prices by 7-8 cents per gallon, that finding was not at a level of confidence normally thought of as statistically significant. And interestingly enough, the GAO study did not find a statistically significant increase in wholesale gasoline prices in the eastern part of the United States.
Another analytic exercise examined eight of the 2,600 mergers and acquisitions that occured between 1994-1999. GAO provided 28 estimates of the effects of those mergers on posted rack prices for branded and unbranded conventional, reformulated, and California-specific gasoline. In 16 cases, GAO found a positive and statistically significant impact on posted rack prices ranging from 0.4 cents per gallon to 6.9 cents per gallon. In seven cases, they found a negative and statistically significant effect, ranging from a price decline of 0.4 cents per gallon to 1.8 cents per gallon. In five other cases, they found no statistically significant effects at all.
Yet McCool glosses over these more careful observations in his oral presentation for the more arresting “1-7 cents per gallon” impact estimate. Media coverage might have been somewhat different had McCool said that GAO found no evidence that mergers and acquisitions have increased posted rack prices in the eastern half of the United States, but some evidence to suggest that mergers and acquisitions increased posted rack prices by somewhere between 0.15 and 3 cents per gallon in the western half of the United States (the findings of the more comprehensive study of the two undertaken by GAO) … but that posted rack prices and a quarter will buy you a cup of coffee for all the good they will do the analyst because posted rack prices and retail prices are two different things. But that wouldn’t have made the members of the committee very happy, and GAO is not in the business of going out of its way to offend the people funding their operations.
In McCool’s defense, at the back of the written testimony he submitted to the committee, he breaks down the study’s findings by merger. According to GAO, the Exxon-Mobil and Marathon-Ashland mergers increased posted rack prices by 2 cents per gallon and reformulated gasoline (posted rack) prices by 1 cent per gallon. The Shell-Texaco merger, however, reduced reformulated gasoline (posted rack) prices by about a half cent per gallon. The Tosco-Unocal merger increased California gasoline (posted rack) prices by 7 cents per gallon.
A note about the Tosco-Unocal merger that provides the upper-bound estimate offered by Mr. McCool - the GAO finding pertains to the (posted rack) price of branded gasoline. The (posted rack) price of unbranded gasoline was actually found to decline. Economists at the FTC note that
Tosco had a branded presence in few of the cities affected by the merger and, where it did, Unocal typically did not have a significant branded presence. Under these circumstances, it is virtually impossible to imagine an anticompetitive theory that would be consistent with a large increase in branded prices but no increases in unbranded prices. Had the GAO researchers understood this problem, they would have recognized that their result must be flawed.
Fourth, McCool’s discussion of the mergers and acquisitions in the 1990s leaves much to be desired. For instance, 2,600 mergers and acquisitions are dutifully noted without the proper context. To wit, the mergers and acquisitions occurred because oil companies were hemorrhaging red ink due to historically low oil prices. Many of these companies simply could not survive on their own. Thus, the mergers and acquisitions. That is a vital aspect of the story that colors the mergers and acquisitions in a far different way than they are being colored by “the trust busters.”
McCool testified that increased consumer prices that followed from a merger can either be good or bad. Mergers will prove bad, he said, if they allow companies to exercise market power. Mergers will be good, however, if they allow for more efficient operations. Unfortunately, he does not tell us whether the mergers and acquisitions in the 1990s that he flagged as having driven up price were “good” or “bad.”
That aside, this sort of argument is a primitive construct. If a merger or acquisition improves efficiency, it will give that company greater pricing power by definition, so this isn’t an “either/or” game. Nonetheless, the observation that it might well be economically healthy if a merger increased fuel prices is quite important and well worth making in a more aggressive manner than it was in the testimony.
Fifth, McCool’s riffs about the oil market were so dodgy that one gains little confidence in GAO’s ability to sort any of these issues out. For instance, McCool contends that domestic refining capacity has not expanded enough to keep pace with demand. I don’t know what that is supposed to mean. Demand for gasoline is only manifested in response to price. If gasoline prices were zero, demand would be nearly infinite. If prices are around $3.00 per gallon, demand for gasoline would be less. So McCool can only be arguing that we don’t have enough domestic refining capacity to meet demand given current prices. Well, that’s flatly wrong. We do indeed have enough gasoline to go around given today’s price. If it were otherwise, service stations would be shutting down because they could not get enough gasoline from wholesalers to keep the pumps flowing. Obviously, they do.
McCool buttresses his contention that refining capacity is seriously constrained by noting that no refinery has been built in the United States since 1976 and then making a big deal of the fact that utilization rates have increased from 78% in the 1980s to 92% in the 1990s. But those observations prove nothing. Regarding the former, investors find it a lot cheaper to expand capacity in existing refineries than to build new refineries altogether - and that’s what they’ve been doing. Regarding the latter, high utilization rates = efficient operations. Excess refining capacity means capital is being wasted. It’s certainly true that if we had more slack refining capacity that we could respond to unexpected supply disruptions more quickly, but it costs money to maintain that reserve and McCool offers no analysis to suggest that this sort excess refining capacity “insurance policy” would be a good buy.
Another example: McCool observes that gasoline inventories are low and then spends some time discussing why the industry is generally inclined to minimize inventory levels as a cost-savings device. This is true enough, but is not particularly pertinent to the present situation. Inventory levels over the three month period of February - April 2007 fell by 15 percent, the steepest drop in history. EIA reports that this occurred because of labor strikes in Europe that disrupted fuel imports and an unusually large degree of refinery maintenance of late. In short, McCool told the wrong story.
McCool also indulged in the kinds of things that constantly grate on the nerves. For instance, he contends that “most of the increased U.S. gasoline consumption over the last two decades has been due to consumer preferences for larger, less-fuel efficiency vehicles ….” This is true in a sense but is a reflection of the underlying fact that real (inflation adjusted) gasoline prices in the 1990s were the lowest in U.S. history. Consumer preferences for gas guzzlers didn’t come out of the clear blue sky. Accordingly, it would be more accurate to say that “most of the increased U.S. gasoline consumption over the last two decades has been due to historically low gasoline prices in the 1990s.” But that would have been less pejorative.
GAO’s analysis is a lot less helpful to the mob than one might think given the number of times it has been offered up as a rationale for Hugo Chavez-style assaults on the U.S. oil sector.