Your intrepid writer, when not toiling for free in the basement of The Sling, does a fair amount of testifying as an expert economic witness. Many of these cases involve alleged price-fixing (or wage-fixing) conspiracies. One would think there would be no need to define the relevant market in such cases, as the law condemns price-fixing under the per se standard. But because of certain legal niceties—such as whether the scheme involved an intermediary (or ringleader) that allegedly coached and coaxed the parties with price-setting power—we often spend reams of paper and hundreds of billable hours engaging in what amounts to navel inspection to determine the contours of the relevant market. The idea is that if the defendants do not collectively possess market power in a relevant antitrust market, then the challenged conduct cannot possibly generate anticompetitive effects.
A traditional method of defining the relevant market asks the following question: Could a hypothetical monopolist who controlled the supply of the good (or services) that allegedly comprise the relevant market profitably raise prices over competitive levels? The test has been shortened to the hypothetical monopolist test (HMT).
It bears noting that there are other ways to define relevant markets, including by assessing the Brown Shoe factors or practical indicia of the market boundaries. The Brown Shoe test can be used independently or in conjunction with the HMT. But this alternative is beyond the scope of this essay.
Published in the Harvard Law Review in 2010, Louis Kaplow’s essay was provocatively titled “Why (Ever) Define Markets”? It’s a great question, and having spent 25-odd years in the antitrust business, I can provide a smart-alecky and jaded answer: The market definition exercise is a way for defendants to deflect attention away from the harms inflicted on consumers (or workers) and towards an academic exercise, which is admittedly entertaining for antitrust nerds. Don’t look at the body on the ground, with goo spewing out of the victim’s forehead. Focus instead on this shiny object over here!
And it works. The HMT commands undue influence in antitrust cases, with some courts employing the market-definition exercise as a make-or-break evidentiary criterion for plaintiffs, before considering anticompetitive effects. Other classic examples of market definition serving as a distraction include American Express (2018), where the Supreme Court even acknowledged evidence a net price increase yet got hung up over market definition, or Sabre/Farelogic (2020), where the court acknowledged that the merging parties competed in practice but not per the theory of two-sided markets.
A better way forward
When it comes to retrospective monopolization cases (aka “conduct” cases), there is a more probative question to be answered. Rather than focusing on hypotheticals, courts should be asking whether a not-so-hypothetical monopolist—or collection of defendants that could mimic monopoly behavior—could profitably raise price above competitive levels by virtue of the scheme. Or in a monopsony case, did the not-so-hypothetical monopsonist—or collection of defendants assembled here—profitably reduce wages below competitive levels by virtue of the scheme? Let’s call this alternative the NSHMT, as we can’t compete against the HMT without our own clever acronym.
Consider this fact pattern. A ringleader, who gathered and then shared competitively sensitive information from horizontal rivals, has been accused of orchestrating a scheme to raise prices in a given industry. After years of engaging in the scheme, an antitrust authority began investigating, and the ring was disbanded. On behalf of plaintiffs, an economist builds an econometric model that links the prices paid to the customers at issue—typically a dummy variable equal to one when the defendant was part of the scheme and zero otherwise—plus a host of control variables that also explain movements in prices. After controlling for many relevant (i.e., motivated by record evidence or economic theory) and measurable confounding factors, eliminating any variables that might serve as mediators of the scheme itself, the econometric model shows that the scheme had an economically and statistically significantly effect of artificially raising prices.
Setting aside any quibbles that defendants’ economists might have with the model—it is their job to quibble over modeling choices while accepting that the challenged conduct occurred—the clear inference is that this collection of defendants was in fact able to raise prices while coordinating their affairs through the scheme. Importantly, they could not have achieved such an outcome of inflated prices unless they collectively possessed selling power. (Indeed, why would defendants engage in the scheme in the first place, risking antitrust liability, if higher profits could not be achieved?) So, if we are trying to assemble the smallest collection of products such that a (not-so) hypothetical seller of such products could exercise selling power, we have our answer! The NSHMT is satisfied, which should end the inquiry over market power.
(Note that fringe firms in the same industry might weakly impose some discipline on the collection of firms in the hypothetical. But the fringe firms were apparently not needed to exercise power. Hence, defining the market slightly more broadly to include the fringe is a conservative adjustment.)
At this point, the marginal utility of performing a formal HMT to define the relevant market based on what some hypothetical monopolist could pull off is dubious. I use the modifier “formal” to connote a quantitative test as to whether a hypothetical monopolist who controlled the purported relevant market could increase prices by (say) five percent above competitive levels.
The formal HMT has a few variants, but a standard formulation proceeds as follows. Step 1: Measure the actual elasticity of demand faced by defendants. Step 2: Estimate the critical elasticity of demand, which is the elasticity that would make the hypothetical monopolist just indifferent between raising and not raising prices. Step 3: Compare the actual to the critical elasticity; if the former is less than the latter, then the HMT is satisfied and you have yourself a relevant antitrust market! An analogous test compares the “predicted loss” to the “critical loss” of a hypothetical monopolist.
For those thinking the New Brandeisians dispensed with such formalism in the newly issued 2023 Merger Guidelines, I refer you to Section 4.3.C, which spells out the formal HMT in “Evidence and Tools for Carrying Out the Hypothetical Monopoly Test.” To their credit, however, the drafters of the new guidelines relegated the formal HMT to the fourth of four types of tools that can be used to assess market power. See Preamble to 4.3 at pages 40 to 41, placing the formal HMT beneath (1) direct evidence of competition between the merging parties, (2) direct evidence of the exercise of market power, and (3) the Brown Shoe factors. It bears noting that the Merger Guidelines were designed with assessing the competitive effects of a merger, which is necessarily a prospective endeavor. In these matters, the formal HMT arguably can play a bigger role.
Aside from generating lots of billable hours for economic consultants, the formal HMT in retrospective conduct cases bears little fruit because the test is often hard to implement and because the test is contaminated by the scheme itself. Regarding implementation, estimating demand elasticities—typically via a regression on units sold—is challenging because the key independent variable (price) in the regression is endogenous, which when not correctly may lead to biased estimates, and therefore requires the economist to identify instrumental variables that can stand in the shoes of prices. Fighting over the proper instruments in a potentially irrelevant thought experiment is the opposite of efficiency! Regarding the contamination of the formal test, we are all familiar with the Cellophane fallacy, which teaches that at elevated prices (owing to the anticompetitive scheme), distant substitutes will appear closer to the services in question, leading to inflated estimates of the actual elasticity of demand. Moreover, the formal HMT is a mechanical exercise that may not apply to all industries, particularly those that do not hold short-term profit maximization as their objective function.
The really interesting question is, What happens if the NSHMT finds an anticompetitive effect owing to the scheme—and hence an inference of market power—but the formal HMT finds a broader market is needed? Clearly the formal HMT would be wrong in that instance for any (or all) of the myriad reasons provided above, and it should be given zero weight by the factfinder.
A special form of direct proof
An astute reader might recognize the NSHMT as a type of direct proof of market power, which has been recognized as superior to indirect proof of market power—that is, showing high shares and entry barriers in a relevant market. As explained by Carl Shapiro, former Deputy Assistant Attorney General for Economics at DOJ: “IO economists know that the actual economic effects of a practice do not turn on where one draws market boundaries. I have been involved in many antitrust cases where a great deal of time was spent debating arcane details of market definition, distracting from the real economic issues in the case. I shudder to think about how much brain damage among antitrust lawyers and economists has been caused by arguing over market definition.” Aaron S. Edlin and Daniel L. Rubinfeld offered this endorsement of direct proof: “Market definition is only a traditional means to the end of determining whether power over price exists. Power over price is what matters . . . if power can be shown directly, there is no need for market definition: the value of market definition is in cases where power cannot be shown directly and must be inferred from sufficiently high market share in a relevant market.” More recently, John Newman, former Deputy Director of the Bureau of Competition at the FTC, remarked on Twitter: “Could a company that doesn’t exist impose a price increase that doesn’t exist of some undetermined amount—probably an arbitrarily selected percentage—above a price level that probably doesn’t exist and may have never existed? In my more cynical moments, I occasionally wonder if this question is the right one to be asking in conduct cases.”
I certainly agree with these antitrust titans that direct proof of power is superior to indirect proof. Let me humbly suggest that the NSHMT is distinct from and superior to common forms of direct proof. Common forms of direct proof include evidence that the defendant commands a pricing premium over its peers (or imposes a large markup), as determined by some competitive benchmark (or measure of incremental costs), or engages in price discrimination, which is only possible if it faces a downward-sloping demand curve. The NSHMT is distinct from these common forms of direct evidence because it is tethered to the challenged conduct. It is superior to these other forms because it addresses the profitability of an actual price increase owing to the scheme as opposed to levels of arguably inflated prices. Put differently, it is one thing to observe that a defendant is gouging customers or exploiting its workers. It is quite another to connect this exploitation to the scheme itself.
Regarding policy implications, when the NSMHT is satisfied, there should be no need to show market power indirectly via the market-definition exercise. To the extent that market definition is still required, when there is a clear case of the scheme causing inflated prices or lower output or exclusion in monopolization cases, plaintiffs should get a presumption that defendants possess market power in a relevant market.
In summary, for merger cases, where the analysis focuses on a prospective exercise of power, the HMT might play a more useful role. In merger cases, we are trying to predict the profitability of some future price increase. Even in merger cases, the economist might be able to exploit price increases (or wage suppression) owing to prior acquisitions, which would be a form of direct proof. For conduct cases, however, the NSHMT is superior to the HMT, which offers little marginal utility for the factfinder. The NSHMT just so happens to inform the profitability of an actual price hike by a collection of actual firms that wield monopoly power, as opposed to some hypothetical monopolist. And it also helpfully focuses attention on the anticompetitive harm, where it rightly belongs. Look at the body on the ground and not at the shiny object.