Today I’m writing about how one of the key leaders of the antitrust bar is starting to attack the proposed Federal Trade Commission rule that bans non-compete agreements. Plus, there’s a round-up of monopoly news.

And now…

For years, we’ve been told that while drinking too much is bad for your health, having just a glass of wine with dinner, as the French do, can be good for your heart. And many studies do indeed show such an association. But what we’re discovering is that this correlation is an illusion. In fact people who drink a small amount of wine every night are healthier, but only because it means they aren’t drinking soda. Even small amounts of alcohol are bad for you, but as academics are coming to understand, if you drink a glass of wine every evening, you also tend to exercise, eat fruits and vegetables, and refrain from smoking. And it is those habits, not the wine, that improve health.

This logical fallacy is known as confusing correlation with causation, or assuming that because A happened and then B happened, that A caused B. The media loves these kinds of associations, because they seem correct, but they are often just coincidence, or a result of a variable that’s not being observed. Sometimes correlations are silly, like saying that “there are more sick people at hospitals, therefore hospitals cause sickness.” They can even be ridiculous; one famous study found that shark attacks are common on beaches with more ice cream sales, the theory being that sharks like to eat people full of ice cream.

Every aspiring law student encounters this fallacy; it’s fundamental in law to understand when something is causal vs coincidental or associative. It’s such a commonly taught error to avoid that it was the basis for one of the first episodes of the famous show The West Wing, an episode named “Post Hoc, Ergo Propter Hoc” after the latin phrase describing this fallacy. Lawyers understand the difference between correlation and causation. So do academics. An entire generation of would-be lawyers who watched the West Wing understand it. It’s not a mystery.

And this brings me to the antitrust bar, and the coming attack on FTC Chair Lina Khan’s attempt to ban non-compete agreements. Non-competes are agreements prohibiting workers from going to work for rivals. They are a soft form of indentured servitude. Such agreements would seem unlawful on their face - they are called ‘non-competes.’ Yet, many antitrust scholars really dislike that Khan is trying to eliminate them. But interestingly, as these scholars defend such agreements, some of them try to exploit the fallacy of confusing causation and correlation to do so.

And this choice goes all the way to the top. Take, for instance, one of the most important antitrust scholars of the last forty years, Herb Hovenkamp, nicknamed ‘the Dean of the Antitrust Bar.’ You may have heard of Hovenkamp, you may not have. But if you’re an antitrust lawyer, he’s a legend, the prime shaper of interpretation of the Sherman Antitrust Act, in some ways more important than Robert Bork.

A historian and antitrust law professor, he specializes in 19th century regulatory history, as well as being the guardian of a widely respected treatise of modern antitrust doctrine. He has been cited by the Supreme Court at least 38 times; Stephen Breyer once said most advocates would rather have “two paragraphs of [Hovenkamp’s] treatise on their side than three courts of appeals or four supreme court justices.” So how he argues matters, because how he has argued has set the template for how our economy functions.

Hovenkamp has published almost 200 SSRN papers on antitrust since 2003, and churns out legal content at an impressive, almost superhuman, rate. While Bork sat on the right, Hovenkamp is a centrist technocrat, beloved by big tech boosters at the Federalist Society, while signing amicus briefs on behalf of stronger enforcement, which he presumes is what the left favors.

Hovenkamp is also on Twitter, and engages in back-and-forth debates over antitrust policy, pulling citations out of his formidable memory. Those who praise Hovenkamp’s immense influence are correct. Since the 1980s, he has pushed to interpret the law primarily in terms of economic analysis, and sought to craft a legal framework in which questions of justice are irrelevant in favor of a narrow understanding in which the only value that matters in any antitrust question is output. The current status of antitrust law, with the focus on technocratic analysis that generally avoids bright lines and clear rules to constrain monopolies, is Hovenkamp’s legacy. It’s perhaps only a bit of an exaggeration to note that the anti-monopoly movement emerged in response to the monopolized economy Hovenkamp’s scholarship helped create.

We can see his approach when it comes to this non-compete debate. As a good progressive technocrat, Hovenkamp often expresses a desire for more enforcement. But in practice, when someone proposes a clear way of doing it, he objects. His preference is for highly complex and difficult to administer rules that do not interfere with private contracts. For instance, Hovenkamp, though he likes to imagine himself in favor of more enforcement, got frustrated when FTC Chair Lina Khan sought to strengthen merger enforcement, co-authoring a piece with Google consultant Carl Shapiro on how Khan simply didn’t understand economics. And so too with this rule against non-competes, which Hovenkamp found over-broad.

I’m going to walk you through the debate, in detail. It is going to seem a little petty, but understanding it will help illustrate why judges who listen to Hovenkamp might strike down the rule. First, there are some legal elements as to why he doesn’t like the rule, like he fears that the FTC might be engaged in a risky legal strategy. But the more substantive disagreement is the underlying policy choice. Hovenkamp thinks, as he told me on Twitter this weekend, that non-competes are good. “Many,” he wrote, “perhaps most [noncompetes], benefit workers and employers alike.”

Herbert hovenkamp @Sherman1890

@AustinFrerick. If you received your entire antitrust education from Twitter over the last year you might have the impression that nearly all noncompetes are harmful. Many, perhaps most, benefit workers and employers alike. However, these are not low paid workers in fast food.

Cecile Kohrs Lindell @quiltsbycecile

@matthewstoller @Sherman1890 Delivering justice is what the rule of reason attempts to do. If you were a lawyer, you’d know in reality one size doesn’t fit all.

Hovenkamp believes that workers like non-competes because they get to bargain with employers for higher wages, offering to sign a non-compete in return. That assumption, and Hovenkamp’s statement, is astonishing, because it’s almost certainly untrue. How do we know? Well, you can look at comments submitted to the FTC so far. There are more than two thousand, and many of them are stories of people being trapped by non-competes. One would think that if non-competes were so great, you’d be hearing from workers who like them. But so far, I’ve read a few hundred comments, and I haven’t found any such workers. And since lawyers are exempt from non-competes, one would expect this highly paid class to be clamoring for the higher wages non-competes would bring. Unsurprisingly, they aren’t.

Still, I don’t want to do the correlation/causation error here myself. In terms of other evidence, the record supporting Hovenkamp is also thin. The FTC put out a 200-page accompanying document to its rule, and walked through the extensive economic literature on non-competes. On everything from consumer prices to wages to innovation to training, almost all of the recent evidence shows that non-competes have negative effects. Moreover, non-competes aren’t just bad for the workers bound by them. They are like secondhand smoke; non-competes also affect people who are adjacent to the smoker. In a labor market where a lot of jobs are held by people with non-competes, other workers are affected too, since wages across the labor market are held down.

We can see this dynamic already. Yesterday, the Wall Street Journal reported on how employers are planning to deal with the lack of noncompetes. And the answer is, they will pay workers more and treat them better, “using everything from compensation to career advancement to keep workers engaged and loyal to the company.” Some of the new tools involve “deferred compensation—such as retention bonuses or rolling stock options that vest after, say, three years —to give people incentives to stay.” That’s a pretty good argument against Hovenkamp’s view.

And yet, while I thought Hovenkamp’s statement was wrong, at first I thought it was just a difference of opinion. Since I figured I might have something to learn about non-competes, I asked him for evidence for his claim. As a scholar should, he cited a paper written by three respected academics, Evan Starr, J.J. Prescott, and Norman Bishara.

Herbert hovenkamp @Sherman1890

@matthewstoller @AustinFrerick A good empirical study (below) indicates wage premiums where noncompetes are properly used, and distinguishing proper from improper ones is not necessarily hard. Janitors cannot enforce the FTC Act, although the FTC might enforce it for them. journals.uchicago.edu/doi/epdf/10.10…

So far, so good. He believes something, I believe something else, he provided evidence. I eagerly read the paper, but was puzzled. It didn’t seem to say what he said it did. The paper made the point, as I noted, that “wages are relatively lower where noncompetes are easier to enforce.” I asked for clarification, and he followed up by saying that the paper is complex, but does also allege what he suggested.

Herbert hovenkamp @Sherman1890

@matthewstoller @AustinFrerick The piece is complex, but its baseline conclusion is "noncompete agreements are associated with positive differentials in wages and training. However, including our advanced controls ... causes the wages differential to fall from nearly 11 percent to 6.6 percent.

He’s an important scholar, so I presumed I was wrong. I sat down to read the paper again. What was I missing? As it turns out, I wasn’t missing anything. One of the authors of the paper, Evan Starr, helpfully chimed in to say the legendary Hovenkamp was the one who misread it, and made the correlation/causation error. Starr noted the evidence shows non-competes suppress everyone’s wages, but rich people sign non-competes more often than poor people, so it can look like non-competes are associated with higher wages. But it doesn’t show noncompetes help workers get higher wages, as Hovenkamp supposes.

Evan Starr @evanpstarr

@Sherman1890 @matthewstoller @AustinFrerick Thanks @Sherman1890. The early notice results are brought up regularly by pro-noncompete folks. We try to be clear that what we have uncovered is a correlation, not causal. A follow up paper suggests that the positive wage relationship is selection.

papers.ssrn.comBundling Employment Restrictions and Value Appropriation from EmployeesWe examine how firms bundle multiple postemployment restrictions and how such bundling is associated with value appropriation between firms and their employees.

That’s a really cool thing about Twitter, we can have this broad-based discussion, where the author of a paper can see people are arguing about his work, and then interject, a la Annie Hall.

Anyway, if I were Hovenkamp, I’d try to grapple with this error, and find some new evidence, or maybe even rethink my position. Is that what he did? Nope. In fact, the next day, he produced a piece for the Regulatory Review in which he wrote, “Workers who sign such agreements often receive higher wages than those who do not, because the noncompete agreement protects the employer’s investment.” And how did Hovenkamp justify this claim? He pointed to Starr’s paper. And he did so the day after Starr told him that the paper didn’t say what he thought it said.

At this point, I got frustrated, and asked Hovenkamp why he cited Starr’s paper after being corrected by Starr about it. But Hovenkamp didn’t think Starr had corrected him.

Herbert hovenkamp @Sherman1890

@matthewstoller @evanpstarr he didn't "correct" me at all. Most pieces, including this one, written over the last two decades find situations in which wages are higher noncompetes, and others where they are not. That's why this is not a case for a per se rule.

Over the course of the conversation, a few other antitrust scholars pushed back on Hovenkamp, but he remained strident that Starr’s paper, and other unspecified studies, illustrated that non-competes help workers get higher wages.

Now, you’d think that such an important scholar would see the value of admitting error, and to even appreciate those who point it out. The goal here, one would think, is to understand how the economy works so we can make better policy. But that is not, I suspect, what Hovenkamp seeks. His goal is to attack the FTC’s rule, because protecting a theory of how the economy works is more important than actually understanding how the economy works.

This is a bold claim, of bad faith. But I found evidence for it. Today, as I was writing this piece, I looked over Hovenkamp’s article in the Regulatory Review. It turns out he had quietly changed it. Instead of saying that Starr’s paper shows causality, it reads, “Workers who sign such agreements often receive higher wages than those who do not. Noncompete agreements can protect the employer’s investment.” Breaking the sentence into two is a subtle shift, but an important one. Now he recognizes the evidence just shows correlation. So he quietly admitted he was wrong, but with no note of error, or admission he updated the piece, and no change to his underlying theory. He hinted that non-competes raise wages, but no longer says it outright, recognizing he must play an intellectual trick because he now has no evidence for his argument, just theory. But he hoped no one will notice, certainly not the law clerks looking for a rationale for their bosses to strike down a rule. That’s bad faith, intentionally conflating correlation and causation.

Now, this whole piece may seem like a baroque spat among a group of weird insider nerds who focus on monopoly power. And for sure, it is that. But it also reveals intellectually dodgy behavior on the part of the most important antitrust scholar of his generation, a man with dozens of citations by the Supreme Court and over a thousand citations from lower courts. A man feted by the Department of Justice Antitrust Division and academia. A man considered the canonical source on how 19th century Americans thought about railroad and corporate regulation, and a man who holds editorial control over the most important antitrust treatise used by legal practitioners in America.

And this brings me to the social ecosystem of the antitrust bar. Hovenkamp is beloved by law professors, powerful lawyers, and Wall Street analysts. (In fact, one of his defenders on twitter talking up his impressive scholarship works for a Wall Street merger arbitrage hedge fund.) And the reason isn’t just these bad habits. He is in many ways an impressive man, clearly explaining the law and working extremely hard to produce timely articles.

Now, I don’t want to build Hovenkamp into some sort of superhero. He just represents a style common in antitrust, of elevating theory over substance so much that highly trained thinkers make basic errors. It’s now common to have antitrust scholars conflate causation and correlation, so they can attack the FTC’s rule on non-competes. It’s an error that pre-law students, or even casual viewers of the West Wing understand. It’s an error Hovenkamp made, knew he made, and quietly tried to avoid letting the world know he made. So the question is, why does the antitrust bar often make such a basic error, confusing correlation and causation, even though such an error helps to facilitate consolidation, higher prices, and less liberty for all of us?

And that is a question that answers itself.

Monopoly News Round-Up

There’s a very weird moment in crypto, with bitcoin and crypto prices rising about 25% over the last two months, with a sort of ‘turn those machines on kind of vibe. At the same time, the courts are about to rule in a significant case over Ripple Labs which could clarify the SEC’s authority to regulate crypto as a security.
The FTC is starting to go after big sellers and retailers of packaged goods and foods for discriminating against smaller stores. They are starting by targeting Coca Cola and Pepsi and their sale of beverages on favorable terms to larger outfits. Both Politico and Bloomberg have stories. The National Association of Convenience Stores have been lobbying to revive the law.
Calls for re-regulating the airlines entirely are getting louder. The New York Times had a piece from one of my colleagues, Bill McGee, on the structural problems in the industry. Meanwhile, Time Magazine showed that airline deregulation significantly harmed smaller cities and contributed to economic inequality.
The EU is readying an ‘antitrust warning’ to Microsoft over the Activision merger. I’m not exactly sure what that means, but given that the EU’s competition enforcers are motivated by two things. One, EU-wide enforcers want to let big firms do whatever they want. And two, EU-wide enforcers want to get press for being tough on big firms.
This is an excellent article about how the lobby for big restaurant chains has been able to get workers to pay a fee ostensibly to learn about food safety, but in reality to support lobbying against raising the minimum wage.
Florida Governor Ron DeSantis is targeting pharmacy benefit managers, a key set of consolidated middlemen in our drug supply chain. DeSantis isn’t the only or first Governor to do so, but he’s prominent and sees opportunity. “Among other things,” wrote WFSU, “the proposal would bar prescription benefit managers from forcing consumers to use mail programs for prescription drugs.” One of the people who spoke at the announcement was an official from the supermarket chain Publix, whose owners are big donors to the Florida governor.

BIG by Matt Stoller

Non-Compete Agreements and the Cult of the…