Deputy Assistant Attorney General Michael Kades Delivers Remarks at GCR Live: Law Leaders Global 2024

Remarks as Prepared for Delivery

Thank you[1] , Josh, for the kind introduction. And thank you for not mentioning how much of a pain in the neck I was when we were both attorney advisors at the Federal Trade Commission.

I was recently in Milwaukee, Wisconsin. My daughter and I had tea at the top of the Pfister Hotel, where we looked out at the Federal Courthouse, where I clerked, including the window of my “office.”

I could not help but tell her various stories of my clerkship with Judge Reynolds: like the time a citizen upset with the judge’s busing decision ran into the court room brandishing a saber — that was before my time — or the time the judge threatened his children that if they didn’t stop bickering, I would get his tickets to the NFC championship game at Lambeau field — not surprisingly, they found common ground quickly.

My fondest memories, however, were sitting in his office, often late at night when the courthouse was empty, discussing the individual cases. Like wanderers in the countryside, we would meander from facts to policy to morality to research and back again. Then, at some point, Judge Reynolds would ask, “What does the statute (or case law) say?” At this point, I had long forgotten what the exact words were. I would respond, “I don’t remember exactly.” And, he would smile and say, “It never hurts to read the law.” Invariably, re-reading the law or the cases would reveal a key phrase I had overlooked or that I had added words that were not there.

You might think the point of my opening story is that policy and research do not matter. But the story has a second part. Finally, after one of these long sessions, I asked the Judge, why don’t we start with the text of the statute? Wouldn’t that be faster? With an indulgent smile and a twinkle in his eye, he would say, “Michael, we have to do both.”

In honor of Judge Reynolds, I want to focus on both: the text of Section 7 of the Clayton Act and the economic consensus supporting the structural presumption.

Why these two topics? Because they are two of the most contentious issues in merger litigation: the government’s burden and the value of the structural presumption.

First, the Clayton Act prohibits mergers that may substantially lessen competition. That text prohibits mergers as long as competitive harm is reasonably probable — even if harm is unlikely in the sense of less than 50%. That text requires skepticism of rebuttal evidence. And, that text reflects an error cost judgment in favor of enforcement.

Second, modern economic theory and empirical work support continued reliance on the structural presumption — the structural presumption is more than good law, its good policy.

Let’s start with the statute.[2] It says “may” — not that the effect “is to substantially lessen competition,”[3] not the effect “is possibly to substantially lessen competition,” and not that the effect “is certain to substantially lessen competition.” [4] Congress considered all these alternatives. But the word they chose was “may.” Specifically, Congress changed the language from “where the effect is” to “where the effect may be,” which Senator Chilton explained meant “where it is possible for the effect to be.”[5]

As the Supreme Court has repeatedly stated, this standard requires only a “reasonable probability” of substantially lessening competition. Reasonable probability does not mean any possibility. For example, the chance that the United States would defeat the USSR in Hockey at the 1980 Olympics was not a reasonable probability — there is reason it is called the Miracle on Ice.

In California v. American Stores, the Supreme Court’s last discussion of the standard, they not only quoted the “may be” in the Clayton Act, they italicized it, and they added that these words create “a relatively expansive definition of antitrust liability.”[6]

The Court has also described the standard as requiring plaintiffs to prove the acquisition creates a threat of competitive harm. As described in General Dynamics, successful rebuttal evidence “mandate[s] a conclusion that no substantial lessening of competition [is] threatened by the acquisition.”[7] The D.C. Circuit in Baker Hughes relied on a similar formulation.[8] Rebuttal evidence must ameliorate the threat of substantially lessening competition. The Court is talking about the risk of an anticompetitive effect.

The case law confirms the plain meaning of the statute: Section 7 does not require that the harm be more likely than not. As long as the harm is reasonably probable,[9] Section 7 bars some mergers that are more likely than not to have no — or even a positive — effect on competition.

And as Doctor Steven Salop has pointed out, when combined with the burden of proof, it means the government must prove by a preponderance of the evidence that the acquisition “may substantially lessen competition.”[10]

This insight has at least two important implications for merger review. First, those hostile to merger enforcement have long complained that the government wants it both ways. In its case, the government relies on the probability of harm but then demands greater certainty from defendants’ rebuttal evidence.[11]

The text of Section 7, not a policy judgment, mandates the difference. Evidence that entry or efficiencies may occur, or can occur or are even likely to occur can all be consistent with a transaction that may, or threatens to, substantially lessen competition.

There continues to be a debate about whether efficiencies are cognizable under Section 7. Even if they are, however, the legal question is whether evidence about efficiencies is strong enough that the acquisition does not threaten substantial harm.

Second, “may” reflects a policy judgment from Congress — that stopping harmful acquisitions is more important on the margin than allowing benign or even procompetitive ones. Ever since Judge Easterbrook’s “The Limits of Antitrust,”[12] introduced the concept of error-cost analysis, it has been de rigueur for scholars, attorneys and courts to intone gravely about “false positives,” “false negatives,” “type-one errors,” and “type-two errors” as intellectual badges of honor. Take away the technical jargon, Judge Easterbrook was simply saying overenforcement of the antitrust laws is more dangerous than their underenforcement.

Let’s set aside the broad range of scholars that have questioned Judge Easterbrook’s thesis as a general principle.[13] The problem with academics and enforcers deciding for themselves what error-cost judgment to make is that Congress has already made that judgment, and it is to err on the side of preventing anticompetitive mergers.[14]  

None of this means the government has no burden or that courts should ignore evidence on how an acquisition can improve competition in the relevant market. Rather, the points are:

1. If we read the statute, the word “may” exists and must be given meaning.
2. The legal question is not predicting the most likely future but whether an anticompetitive effect is reasonably probable.
3. Entry, efficiencies, and other rebuttal evidence may still be relevant but the government prevails if a threat to competition exists despite that evidence.

Reading the statute matters, but so does the research. For 60 years, courts have applied a structural presumption: mergers are presumptively illegal when they significantly increase concentration in highly concentrated markets.[15]

That is an undeniable statement of the law, whether one focuses on binding Supreme Court precedent or on recent district court opinions.[16]

As part of our guideline process, we also looked at the economic literature relating to the structural presumption. What we found — and I know this will be a surprise to at least some you in the audience — an economic consensus supports that presumption.

In a field still rife with debates and disagreements, a remarkable number of economists have written in recent years to defend reliance on structural presumptions. Let me give you a few examples:

1. “Economic theory provides support for the established legal presumption that a merger in a market is likely to have adverse competitive effects when it occurs in a concentrated market and makes it more concentrated (i.e., increases the HHI).”[17] 26 economists co-authored that paper in 2022, including leaders of economics at the antitrust agencies from both Republican and Democratic administrations.
2. “First and foremost, economic theory and a wide range of economic evidence support the conclusion that horizontal mergers that significantly increase market concentration are likely to lessen competition and harm consumers by raising prices, reducing output, or limiting product quality or innovation.”[18] That’s from a recent Yale Law Journal paper by Carl Shapiro and Herbert Hovenkamp.
3. “[O]ne important conclusion is straightforward . . . greater concentration can be expected to make coordination more likely, stronger, or more effective. Accordingly, our analysis supports a structural merger policy, by which mergers between rivals that increase concentration significantly in a concentrated market are presumed to harm competition.”[19] That’s from a recent contribution by Jon Baker and Joe Farrell.

Those three articles cover dozens of economists across the ideological divide, theorists and empiricists, academic economists and economists with public policy expertise.

A number of recent merger retrospectives link increased concentration to observed post-merger anticompetitive effects such as increased prices and decreased product availability. Those studies cover consumer package goods, grocery stores, hospitals, health care providers, health insurers and mergers in oligopoly markets.[20]

There are of course opponents to this view. For example, in a 2016 article by Joshua Wright and Douglas Ginsburg, the authors wrote “that defining the relevant market and calculating market shares are generally not useful tools for predicting the likely competitive effects of a proposed transaction.”[21] In some cases, these critics are loud and well-funded by companies with an interest in limiting merger enforcement.[22]

So, when we prepared the revisions to the merger guidelines, we looked closely across the bibliography of academic and economic work that all commenters, supportive or critical, submitted on this question. What we found is that the structural presumption is better supported in the empirical and economic literature today than it has ever been.

What is the disconnect? Here, we need a little history. In the 1950s, a number of papers found a correlation between structure, conduct and performance, which led to the Structure-Conduct-Performance paradigm in Industrial Organizations. Those papers suffered from serious flaws, however. In general, one cannot infer that concentration causes higher prices. This became well accepted in the 1970s and 1980s, as empirical work by Harold Demsetz (DEM sets) and others refuted the idea that the structure of a market determines its performance.[23] That was an important advance, recognizing that many factors — including the economics of the industry — can cause an industry to become concentrated.

What started as good and valuable economic work was eventually overread to argue for underenforcing the antitrust laws. Like some others, Ginsburg and Wright point to that economic literature and argue it refutes structural measures in merger review.[24] But of course, Demsetz’s empirical work refuted the correlation between all causes of concentration and performance, not the correlation between merger-induced concentration and the risk of that merger lessening competition.

The important finding of the Structure Conduct Performance work in the 1970s and 1980s was that competition on the merits can cause concentration. But merger enforcement depends on a different question: what happens when mergers cause concentration. On that point, the economic literature is clear. Both as a matter of theory and empirically, mergers that increase concentration in highly concentrated markets are likely to have adverse effects.[25] Not just that it may, but that is likely.  

Of course, any structural presumption will involve some amount of line drawing (why is a market with an HHI of 1800 highly concentrated but 1799 is not). It would be too much to ask empirical work to provide exact cut-offs. The empirical work available today, however, would arguably support even more stringent thresholds than we adopted in the 2023 Merger Guidelines.[26] We chose 1800 and 100 because, even if it is conservative, it is well-supported by this literature and by the many appellate decisions across multiple circuits that reference those thresholds.[27]

The very first merger-decision after we released the draft Guidelines, FTC v. IQVIA, explicitly rejected the defense bar’s criticisms of the 30% presumption in the guidelines.[28] The court relied on Philadelphia National Bank to find that the merging parties’ 30.6% market share triggered a structural presumption. Judge Ramos was correct on both the law and the economics.

Besides allowing me to wax nostalgic about clerking for Judge Reynolds in Milwaukee, returning to basic principles is valuable. It reminds us of what we know and allows us to re-evaluate them. Congress’s decision to use “may” matters. Those three letters define the government’s burden, frame the role of rebuttal evidence, and provide a policy judgment on the value of enforcement. Whether one likes the structural presumption or not, it is undeniably good law. It is also backed by an economic consensus, as strong as it has ever been.

Those two principles guide our work. Congress crafted Section 7 as a risk assessment and not a crystal ball. And the structural presumption remains an important tool in assessing mergers.

Thank you for your time.