Antitrust Academy: What is the Philadelphia National Bank Structural Presumption?

The PNB Presumption is the FTC and DOJ's not-so-secret weapon in merger litigation. Litigation strategies and policy debates focus on it. But it is commonly misunderstood. Let's fix that.

Welcome back to Competition on the Merits! Today, a new kind of entry. I’m experimenting with mixing in some “Antitrust Academy” columns that provide a 101-level explainer for the law and economics underlying some important antitrust concepts, ideas, or debates. I’ve got the PNB structural presumption on my mind for reasons related to the HPE-Juniper merger challenge and a few others. Beyond that, I’ve always been fascinated with the feature of antitrust merger law that in the void of a substantive SCOTUS merger case (since 1974) a sort of parallel body of merger law consisting of lower courts and Merger Guidelines has become incredibly important. This is sometimes confusing for students. But it is fascinating. Especially when the two bodies are in tension. And even more so in light of recent Supreme Court decisions like Loper Bright undercutting arguments for deference to things like agency guidelines. And for those less interested in the intellectual foundations of antitrust law and more into the practical implications – this one is for you too! Nothing matters to modern merger litigation than the PNB Presumption.

So here we are with the first Antitrust Academy COTM entry.

The structural presumption in antitrust law is the Clayton Act’s secret weapon. A shortcut to understanding which party is likely to prevail in litigated merger challenged under Section 7 of the Clayton Act. The structural presumption is invoked in nearly all modern merger litigation. And it is invoked commonly in competition policy discussions. Just recently, for example, I explained here at COTM that the real issue in the HPE-Juniper merger and coming Tunney Act review is unlikely to be the heavy breathing around alleged lobbying at all, but rather the fact that the DOJ complaint did not even bother to allege shares that trigger the structural presumption.

One might be fooled into thinking the structural presumption’s primacy in antitrust merger litigation would mean it is well understood by practitioners, policy makers, and the antitrust commentariat. Dear COTM reader, if that were only so. In more than 20 years of teaching antitrust law to students, practitioners, and judges, I’d rank a failure to understand the structural presumption as one of the top 3 most commonly misunderstood concepts in antitrust. Ask me about the others some other time.

More importantly for contemporary practitioners, students, and observers of the field: understanding antitrust law’s structural presumption – despite the fact that it now more than 62 years old – remains critical to understanding outcomes and strategies in modern litigation. When the FTC or DOJ can successfully invoke the presumption that are exceedingly likely to win. When they cannot, the government is highly likely to fail. Many litigation strategies operate around this truth:

• Crafting market definitions that generate market shares that trigger (or avoid) the presumption;

• “Fix it first” strategies (the flavor of the day) in which the merging parties argue that the post-merger market share avoids the presumption if the court accounts for both the merger and the associated “fix,” or divestiture transaction;

  • A standalone COTM is coming on “fix it first” in merger litigation strategy soon. The agencies do not seem to be evolving at all to merging parties adopting this strategy and keep insisting that courts defer to the agencies’ expertise both on how to apply the Clayton Act and how to understand which divestitures will succeed. Any guesses how that has been turning out?

Unfortunately, the law and legal implications of the structural presumption are often misused and abused in antitrust discourse. Sometimes unintentionally conflated with a more general and fairly unobjectionable concept that “market structure matters” at all in antitrust. Sometimes the mistakes are less negligence and more sleight of hand, e.g. conflating the agencies’ own Merger Guidelines’ treatment of market structure with what I describe here as THE “structural presumption” under the law. If that last sentence is confusing to you, you are in the right place. We are going to get you straightened out.

So let’s do some learning today, Competition on the Merits readers.

But first, and as always, please do subscribe, upgrade to paid, and send to your friends. They need to learn some antitrust too.

What is the “PNB Structural Presumption” and Where Does it Come From?

Let’s start with what it is. The Supreme Court has endorsed a presumption of illegality of a transaction under Section 7 of the Clayton Act when the merger results in market shares greater than 30%. SCOTUS adopted that presumption US v. Philadelphia National Bank (1963), and thus the so-called “PNB Presumption” was born.

For the lawyers, the “presumption,” is exactly that. When the PNB presumption is satisfied, the plaintiff’s prima facie burden under the Clayton Act is satisfied, thus shifting the burden of production to the merging parties. The presumption is rebuttable in practice. But as a practical matter, as we will discuss, merging parties rarely ever win under Section 7 once the PNB presumption has been invoked. Importantly, the PNB presumption – triggered once the plaintiff demonstrates post-merger market shares greater than 30% – is the law of the land as adopted by the Supreme Court interpreting Section 7 of the Clayton Act.

For the economists, I want you to be thinking about the economic translation of the PNB presumption. Or a better question – what economic statement would have to be true in order to substantiate the PNB presumption? We will come back to that. First, where did the PNB presumption come from?

Let’s start – as we always should – with the Clayton Act itself.

In a series of five merger cases starting with Brown Shoe (1962), running through Philadelphia National Bank a year later, Continental Can (1964), Von’s Grocery (1966), and ending with General Dynamics (1974), the Supreme Court turned the then-reigning paradigm in industrial organization economics into the law of the land.

In Brown Shoe, the Court interpreted the adopted the market shares of the would-be merged firm as the best indicator of “the effect of such acquisition” and whether it “may be substantially to lessen competition or tend to create monopoly.” The Court emphasized post-merger shares as an indicator of the effect on competition, but provided no real method for analyzing them in condemning the transaction that resulted in a 7.2% share of retail shoe outlets (and 5% of manufacturing).

Just one year later, in PNB, the Court considered a merger that would have created a bank with 30 percent of the relevant market—commercial banking in metropolitan Philadelphia—and would have raised the two-firm concentration ratio from 44 percent to 59 percent. In reversing the judgment of the district court, which had ruled in favor of the merger, the Court made the following fateful statement: “Without attempting to specify the smallest market share which would still be considered to threaten undue concentration, we are clear that 30% presents that threat.”

And that, dear readers, is the birth of the PNB presumption!

Of course, the history of the Clayton Act and the Supreme Court did not end there. (Though it did not go on much longer either). A few observations here are important:

First, by the standards then prevailing in the economics of industrial organization, the Court was undoubtedly on solid ground, as it made clear in a footnote appended to the sentence quoted above. Referring to the leading analysts of the day, the Court noted:

Kaysen and Turner . . . suggest that 20% should be the line of prima facie unlawfulness; Stigler suggests that any acquisition by a firm controlling 20% of the market after the merger is presumptively unlawful; Markham mentions 25%. Bok’s principal test is increase in market concentration, and he suggests a figure of 7% or 8%. . . . We intimate no view on the validity of such tests for we have no need to consider percentages smaller than those in the case at bar . . . .

The “Structure-Conduct-Performance” Paradigm within IO economics was at its zenith at the time. More on that later. But attempts to characterize the PNB presumption as devoid of economics are false. It’s just old and outdated economics. Even more pernicious are attempts to characterize SCOTUS’s creation of the PNB presumption as some sort of divine truth-telling exercise originating from the text of the Clayton Act. The Court explained exactly what it was doing and tethered its reading of the Clayton Act to economic thinking and empirical evidence available at the time.

Second, the Court decided a few more Section 7 cases. One year after deciding PNB, the Court held unlawful the acquisition by Continental Can, the second largest producer of metal containers with a 33 percent share of the market, of Hazel-Atlas Glass Co., the third largest producer of glass containers with a 9.6 percent share. The decision is best known for the Court’s grouping of metal and glass containers in the same relevant market and its focus upon “inter-industry” competition. Of more interest here, however, is the Court’s extension of the recently created PNB presumption. The merged company’s 25 percent of the combined glass and metal bottle market, the Court noted, “approaches that held presumptively bad in United States v. Philadelphia National Bank,” and as the Court had said in that case, “Where concentration is already great, the importance of preventing even slight increases in concentration and so preserving the possibility of eventual deconcentration is correspondingly great.”

Curiously, however, the Court held unlawful the merger in its next case, Von’s Grocery, without so much as mentioning either the 30 percent standard set in PNB or the possibility of prohibiting a lesser merger that nonetheless works a substantial increase in the concentration of the relevant market. The Von’s merger, as the dissent pointed out, produced a firm with 1.4% of the grocery stores and 7.5% of grocery sales in Los Angeles, and resulted in an increase of 1.1% in the market share enjoyed by the two largest firms in the market . . . . [These] figures are hardly the “undue percentage” of the market, nor . . . the “significant increase” in concentration, that would make this merger inherently suspect under the standard of [PNB].

In the last case, General Dynamics, the Court approved a merger even though the market shares “support[ed] a finding of ‘undue concentration’” under the approach of PNB because “other pertinent factors” affecting the relevant market indicated the merger would not work a “substantial lessening of competition.” In so doing, the Court provided the guidance missing from PNB itself about the administration of the 30 percent standard adopted in that case: Quoting Brown Shoe, the case that preceded PNB, the Court reiterated that market shares “controlled by the industry leaders and the parties to the merger” would remain the “primary index of market power; but only a further examination of the particular market—its structure, history and probable future—can provide the appropriate setting for judging the probable anticompetitive effect of the merger.”

And that, dear reader, is all she wrote. In the last 50 years the Court has not again passed upon the substantive aspect of merger analysis. The legacy of PNB endures: The plaintiff, ordinarily the Government, makes its prima facie case by showing 30 percent of the market is involved in the merger, which shifts to the merging parties the burden of showing competition in the market will not be diminished notwithstanding the increase in concentration.

Third, once invoked, the PNB presumption is rarely dispelled by merging parties. Back in 2012 when I co-authored this piece with Judge Ginsburg, the data showed that defendants had historically dispelled the presumption in only 17 of the 106 cases in which it was invoked. The success rate for plaintiffs conditional on the PNB presumption being invoked in cases since then is considerably higher.

Fourth, in this way, the PNB presumption looks like a bunch of other antitrust doctrines developed in the 1960s only to be revisited by the Supreme Court with updated economic thinking (e.g. the Supreme Court revisiting antitrust rules involving predatory pricing, tying, resale price maintenance, predatory overbidding, vertical restraints, etc.). Why the outlier (perhaps along with the per se tying rule in Jefferson Parish and the distorted reading of the injury to competition language of the Robinson Patman Act)?

The difference is that the Supreme Court has not taken a merger case since General Dynamics in 1974. When the FTC or the Antitrust Division obtains a preliminary injunction against consummation of a merger, it is now common practice for the parties to abandon the deal rather than pursue the litigation. (If the case is brought by the FTC, then the preliminary injunction will remain in effect at least until the completion of an administrative proceeding in which the Commission will inevitably rule against the merger, at which point the companies can return to district court.) Should the firms persist and prevail in district court, the agency will almost surely appeal and the court of appeals will likely stay the merger pending appeal lest it be impossible to effect a divestiture that restores the status quo ante. The appeal will take at least six and probably closer to 12 months. Prolonging the litigation by another year or 18 months to see the matter through the Supreme Court would mean a total time with the merger in suspension of at least four years. Only in the most peculiar situation would the parties to a merger persist for that long. (FWIW, even before the Expediting Act was amended in 1974 to eliminate direct review of antitrust cases by SCOTUS, it took about 2.5 years from the filing of the suit in district court to the SCOTUS decision.

The long and short of it is that the PNB presumption – for better or worse – is likely here to stay in the absence of an act of Congress or a very peculiar merger litigation. Because today is mostly about explaining the PNB presumption and why it matters – I’ll not spend as much time on why it is a bad thing that the PNB presumption persists. Though that is certainly my view. But persist it does.

The Merger Guidelines Matter, But They Are Not The Law

OK, we have talked about the PNB structural presumption. THAT structural presumption has one very special feature: the force of law.

Antitrust agencies promulgate merger guidelines that often discuss market shares and market concentration. The FTC and/or DOJ have issued such guidelines in 1968, 1982, 1992, 2010 and again in 2023. The most recent version, like the others, has a discussion about market concentration. Let’s look at that.

The 2023 Merger Guidelines first mention of concentration and a presumption go something like this:

So far so good. The second paragraph appeals to the PNB presumption and says the agency will use it. OK. Fair enough. But they go on to say that a merger that “creates of further consolidates a highly…”

The FTC and DOJ have declared that certain mergers that meet certain conditions (HHI > 1800 and Change in HHI > 100) are “presumed to substantially lessen competition or tend to create a monopoly,” and thus violate Section 7 of the Clayton Act. The FTC and DOJ are announcing a presumption of illegality. But THIS IS NOT THE PNB PRESUMPTION. And for those keeping score at home – mergers with post-merger shares below 30 percent can easily satisfy these conditions.

Thus the Merger Guidelines seek to broaden the PNB presumption of illegality.

What is the legal authority for this presumption? Let’s start with what it is not. It is not the Supreme Court interpreting Section 7 of the Clayton Act. Here’s what the FTC cites in note 15:

The argument does not seem too persuasive from where I sit. For starters, it simply says that courts relied on previous iterations of the Merger Guidelines in some different cases. That’s fine. Courts rejected some parts of the Guidelines during that time period too. But there is no doubt courts relied heavily upon the 1992 and 1997 HMGs. But what about the argument that “the Agencies consider the original [1800] HHI to better reflect both the law and the risks of competitive harm…”

If that reads like a deference argument to you – it should. After Loper Bright it is not clear this kind of argument will get you too far with Article III courts. Is this a distinction without a difference? The expanded MG presumption is unlikely to matter practically since the agencies so infrequently bring cases with post-merger shares under 30 percent. But it does happen. HPE-Juniper is one example we have discussed. H&R Block was another. But it is rare. But the main point for now is that the PNB presumption enjoys SCOTUS endorsement and the Guidelines, to the extent they go beyond the PNB presumption, stand on “because the FTC says so” as the intellectual foundation. And that foundation simply does not fly with federal courts without substantial evidence to substantiate underlying claims. The FTC is well positioned and staffed to present that evidence in a new iteration of the Merger Guidelines should they be interested. But that line of thought is a distraction from today’s project: explaining what the PNB presumption is, and the difference between the PNB presumption (with the force of law) and the expanded presumption in the agency MGs (without the force of law).

Does the PNB Presumption Make Any Economic Sense?

Back to the quiz question. What is the economic translation of the PNB presumption? The presumption declares that a merger is more likely than not to violate the Clayton Act, that is more likely than not “substantially to lessen competition or tend to create a monopoly” if post-merger shares exceed 30 percent. I think the best available translation of the PNB presumption into an economic statement is the following:

Conditional upon post-merger market shares exceeding 30 percent, the probability that a proposed merger creates market power to the detriment of consumers is greater than 50 percent.

That is a very specific economic statement. The PNB presumption is a very specific presumption. It deems the plaintiff’s prima facie burden satisfied and shifts the burden of production to the defendant. The PNB presumption is a sufficient condition to dispel the plaintiff’s burden of production. Defenses of the PNB presumption and its economic relevance often rely upon some sleight of hand, e.g. “of course market structure matters!” Or, “there are some conditions under which market structure is a reliable predictor of market outcomes!”

But note that neither of those is the economic statement underlying the PNB Presumption. That PNB presumption embraces the economic idea that conditional only upon knowing the post-merger share is greater than 30 percent we have an empirical basis to believe a merger is more likely than not to harm competition.

To put it bluntly, there is no such economic evidence. None.

The Supreme Court itself points to the Structure-Conduct-Performance literature created by Joe Bain beginning in the 1940s and largely done by the early 1960s. The SCP approach posited a systematic relationship between market concentration and a variety of performance measures, including prices and margins. The PNB presumption thus, as others The SCP approach posited a systematic relationship between market concentration and a variety of performance measures, including prices and margins. The PNB presumption thus, as others have pointed out, was not devoid of economic thought but rather created “considerable consistency between judicial decisions and economic thinking during the 1940s, 1950s, and 1970s.” At its peak, the SCP paradigm—based upon the empirical work of Bain and his students—was the best economists had to offer for understanding whether markets would behave competitively. Even George Stigler, who would later recant upon learning of the analytical flaws in the SCP approach, adhered to the SCP view in supporting industry-wide deconcentration proposals in the 1950s.

Donald Turner, who was both a lawyer and a Ph.D. economist, incorporated the SCP paradigm’s heavy reliance upon market structure when, as Assistant Attorney General in charge of the Antitrust Division in 1968, he promulgated the Government’s first Horizontal Merger Guidelines. The SCP approach was also popularized and retailed to students in F.M. Scherer’s leading textbook, first published in 1970. The rise of the SCP paradigm represents a powerful example of the integration of economic learning into law. SCP’s reign over industrial organization economics, however, was not longlived. The SCP paradigm is now dead and has been for quite some time. Its intellectual influence on modern economics is nil. It is no longer taught in graduate economic courses in economics. To the contrary, a leading undergraduate text today teaches students that “the criticisms of [the SCP] approach are many, but perhaps the most significant criticism is that concentration itself is determined by the economic conditions of the industry and hence is not an industry characteristic that can be used to explain pricing or other conduct,” and notes that “[t]he barrage of criticism has caused most research in this area to cease.”

A thorough autopsy of the SCP paradigm would reveal multiple causes of death, but the primary reasons include the rise of new economic theory, critical statistical work debunking the simple relationship identified by Bain and his followers, and the ascent and influence of the Chicago School in the 1970s and 1980s. Harold Demsetz and others questioned whether the observed correlation between market concentration and profitability were better explained by superior efficiency of the larger firms in the market. A number of scholars exposed the analytical flaws in the largely cross-sectional body of evidence that proponents of the SCP approach had cobbled together in support of the concentration-price hypothesis. As Professor Muris observed decades ago, “The SCP paradigm was overturned because its empirical support evaporated.” In reality – the stylized relationship between market concentration and price or other competitive outcomes looked much more like this:

Unsurprisingly, it turns out predicting market outcomes is much more complicated than counting the number of firms on one’s fingers. The key insight is that market concentration – the size and number of firms – is the OUTCOME of the competitive process. The dimensions and type of competition can determine the number of firms and their size; not just the other way around. This is not a new lesson in IO economics or to antitrust practitioners. As I said, the SCP paradigm has been absent from sophisticated (or even basic) economic thought for over 50 years.

And it turns out, modern empirical IO continues to suggest that the relationship between market concentration and markups (or price) does not show a decrease in competition. For example, here is Miller et al. showing that during the time period in which “industry” concentration increased significantly, markups increased but largely as a function of firms getting more efficient. In other words, markups rose because firms got more efficient, not less competitive.

This is not the place to discuss all of the relevant evidence – though Fauver & Wright do so here. For now, I just want to give a taste of the basic economic argument against the PNB presumption. The argument is not that market concentration or market shares never matter to predicting outcomes. Of course they do. The PNB Presumption is a much broader and more forceful proposition: are mergers to market share >30 percent more likely than not to harm competition by creating market power? THAT is the economic proposition underlying the PNB Presumption itself. I do not know many IO economists that believe that statement to be true. Maybe 1-2? This is not the column to present all of the evidence for and against. But simply to lay out the boundaries of the economic debate over the actual presumption. And if I’ve tipped my hand about where I come out on that debate — well, that is what you came here for I suppose. For a fuller treatment of the argument that the PNB presumption is bad law and bad economics, see Ginsburg and Wright here.

Ginsburg & Wright conclude thusly about the presumption:

Economic theory alone cannot provide a defense of the PNB presumption of illegality for a merger generating a post-merger market share of 30 percent or more; no general economic model is sufficient to support shifting the burden of justification to the defendant upon the demonstration that the post-merger firm’s market share will exceed any particular level. Others defend the much weaker proposition that economics provides some basis for allowing market structure to play some role in merger analysis.44 We agree that market concentration and market shares can be relevant to merger analysis. But we can find no serious defense of the proposition that a PNB-like presumption reflects the best of modern economic thinking about mergers or that presuming the illegality of transactions above any particular threshold is good economic policy for consumers.

Given the PNB Presumption’s primacy – indeed, its almost dispositive weight – in modern merger enforcement, it deserves a full evaluation of its costs and benefits in light of the best available evidence. At a time when nearly antitrust ideas are up for grabs – it seems the PNB presumption should be as well. Institutional features and market realities that discourage appeals from successful government challenges to proposed mergers render it a long shot that SCOTUS reconsiders the PNB presumption any time soon. Nor is it likely that the agencies defy their nature and refuse to rely on the presumption in merger litigation.

If the economic foundation of PNB’s structural presumption is no longer supported by sound economics—and it has not been for quite some time—then the presumption ought to go the way of the agencies’ policy decisions to drop enforcement of the Robinson-Patman Act and reliance upon the discredited antitrust theories approved by the courts in such cases as Brown Shoe, Von’s Grocery, and Utah Pie. But there is no mechanism to force SCOTUS review of a merger relying upon the PNB presumption and no more than a trivial likelihood that Congress will repeal the presumption through legislation. So the PNB Presumption lives on with outsized influence in modern merger litigation for the foreseeable future.

I plan on continuing a series of these “Antitrust Academy” entries on Competition on the Merits that present primers on a variety of legal and economic topics within antitrust. For subscribers I’ll soon include some video content as well. Let me know which topics you would like me to cover! See you soon.