Big Is Not the Same as Closed

In December 2024 a federal judge stopped the largest supermarket merger in American history. Nine months later, another federal judge declined to break up Google. One ruling kept two companies apart. The other let the biggest search engine in the world stay in one piece. Both turned on the same question, and it was not how big the companies were.

That is interesting, because almost every argument you hear about monopolies starts with size. Too big. Too few competitors. Too much power. The numbers get recited, the outrage follows, and somewhere in the middle the actual question disappears: what happens if the company raises its price, and someone tries to take its customers away?

The first judge, Adrienne Nelson, said something precise about this. Kroger had argued that it needed to buy Albertsons in order to compete with Walmart, Costco, and Amazon. Her answer was that this is not what the law is for.

The defense was not “we are small.” The defense was “we are smaller than somebody else.” A merger can be unlawful even when the company doing the buying is not the biggest firm in the room. Size was never the issue.

This is the part of the monopoly debate that gets skipped, and it is the part that decides almost every case. What follows is an attempt to put the question back where it belongs.

The Word Does Two Jobs

In ordinary speech, “monopoly” means “a company everyone resents.” In economics it means something narrower: one seller, no close substitutes, and something that keeps rivals out.

Each of those three clauses is a judgment, not an observation. “One seller” sounds simple until you ask what counts as the same thing. “No close substitutes” depends on what the buyer will accept instead. “Keeps rivals out” requires you to imagine a rival who does not currently exist, and say why he has not arrived.

That is why two reasonable people can look at the same company and disagree completely. They are not disagreeing about the facts. They are disagreeing about where the market ends. And the market is not a thing you find in nature. It is a definition someone chooses, and the choice often decides the case before any evidence is heard.

There is no cleaner illustration than a grocery merger that two federal courts answered in opposite directions.

Two Grocers and a Word

In 2007, Whole Foods agreed to buy Wild Oats. The two chains were not giants. Whole Foods had 194 stores and Wild Oats had 110. Together they were a rounding error next to Kroger or Walmart.

The Federal Trade Commission still tried to block the deal, on a specific theory. The two chains were the largest operators of what the commission called premium natural and organic supermarkets. Shoppers who go to Whole Foods for the prepared food, the produce, and the politics are not, the argument went, going to switch to a conventional supermarket because the price of salmon went up. They will pay. And in eighteen cities, the two chains were the only real options of that kind.

A district court rejected that. The relevant market, it said, was supermarkets. All of them. On that definition the merger was trivial, and the injunction was denied. The merger closed.

The D.C. Circuit reversed the following year, and its reasoning is the whole lesson:

Same merger. Same country. Same decade. Two federal courts, opposite answers, and nothing changed except the width of a definition. Either the commission drew the market too narrowly, or the district court drew it too broadly. Nothing in the facts settled which.

Now hold that thought, because the same problem runs in the other direction, and the other direction is where it gets interesting.

The Mirror Image

In 1956 the Supreme Court heard a case about cellophane. DuPont made most of the cellophane sold in the United States - roughly three quarters of it. On its face, that looks like the kind of share that ends an adverse ruling.

The Court found no monopoly. Its reason: buyers could switch to other flexible wrapping materials, so the market was not cellophane but flexible packaging, and in that market DuPont was one supplier among many.

Economists have been arguing about that decision ever since, under a name that tells you what they think of it: the cellophane fallacy. The objection is elegant. At the price DuPont was charging, other materials did look like reasonable substitutes. But that price was the price a company with power charges. Define the market using today’s prices, and a monopolist’s own markup becomes the evidence that he has no power.

Put the two cases side by side and the difficulty is plain. In one, the commission drew the market narrowly enough to make a 194-store chain look like a monopolist. In the other, the Court drew it broadly enough to make three quarters of a product look like one supplier among many. Neither court was bribed or stupid. The market definition did the work, and market definition is a judgment.

Which means the useful question is not “how big is this company.” It is: what would have to happen for a rival to take its customers, and is anything preventing that?

Three Sources of Advantage

When a firm holds customers and rivals stay away, the advantage usually comes from one of three places. They behave very differently, and only one of them tends to last.

Better and cheaper. This is the Walmart case, and it is one we have already made here. Sam Walton took enormous risks, built a logistics machine, and drove prices down for people who needed lower prices. He ended up with a share of American retail that would have made a 1950s antitrust lawyer reach for his pen. His customers were better off every year he did it. That advantage is real, and it is also temporary, because it is made of things a rival can copy. The penalty for getting lazy is losing the customers, and it arrives without a courtroom.

Network effects and switching costs. Here the advantage is not just being good. It is that every additional user makes the product more useful to the next one, and leaving costs the user something. A search engine with more queries gets better at answering queries. A marketplace with more sellers attracts more buyers, which attracts more sellers. A business that runs its ordering, inventory, and payroll on one company’s software does not switch on a Tuesday afternoon.

This is a genuine barrier, and it is where the “just wait for entry” answer gets thin. A rival can enter. Whether the rival can reach the scale needed to be worth using is a different question, and it can take a very long time to answer.

A law that limits who may compete. This is the oldest of the three. In 1600, Elizabeth I did not grant the East India Company a permission to trade. She granted it an exclusive right to trade. The state sold exclusivity to a favored few, and the consumer paid for it in dearer goods. The monopolies became so numerous - on salt, on starch, on playing cards - that Parliament voided most of them in the Statute of Monopolies in 1624. The pattern is older than the industrial revolution.

The modern version is quieter. Economists Morris Kleiner and Alan Krueger, using a 2006 Gallup survey, found that 29 percent of American workers needed a government license to do their job. Estimates built from state-level data come in lower, around 25 percent, which is worth noting: the compilers disagree, and both figures are far above the less than 5 percent of the early 1950s. Licensed workers earn about 15 percent more than similar unlicensed workers. Some of that is skill. Some of it is the license doing what a tariff does.

We do not usually call that a monopoly, because the firm doing the excluding is a state licensing board rather than a company. It is the same mechanism. The cost is real and it is paid by the people who cannot afford the entry fee: the hair braider who needs 1,500 hours of training in some states, the mover who needs a permit that a competitor helped design.

And here is a detail that rarely makes the news. Under the state action doctrine, which goes back to a 1943 Supreme Court decision, restraints that a state actively supervises are largely immune from federal antitrust law. The one kind of monopoly that never gets competed away is the kind the government has agreed to protect.

Where the Usual Answer Runs Out

The answer we usually give, and the one we have argued for years, is that the antidote to monopoly is not breaking up successful firms. It is making sure newcomers can get in. That argument is right about Walmart, and it is right about most of what people call monopoly.

It is too slow for a hospital.

Hospitals are not like search engines or retailers. You do not drive past three of them on the way to work. A town of 80,000 people may have one hospital system, or two that have already merged. Building a new one takes years, requires a certificate of need in many states, and costs more than any entrant will risk for a market that may already be served.

So the evidence on consolidation should not surprise anyone. Zack Cooper, Stuart Craig, Martin Gaynor, and John Van Reenen, in a 2019 paper in the Quarterly Journal of Economics, looked at claims covering 28 percent of Americans with employer-sponsored insurance. Hospitals with no rival nearby charged about 12.5 percent more than hospitals in markets with several competitors. When two hospitals within five miles of each other merged, prices rose by more than 6 percent. The authors are careful to call their work descriptive, and that caution is worth repeating: they show a strong association, not a proven cause.

But the pattern is not subtle, and it is not confined to one study. This is the case where blocking a merger is cheaper than waiting for the market to correct itself, because the correction might take twenty years and a bond issue. A competition authority that does nothing here is not defending competition. It is defending the incumbent’s quiet decade.

That concession has a limit, and the limit matters. The barrier in American health care is not only private. The same states that review mergers also decide who may build a hospital, add a bed, or buy a scanner. If entry is the answer, then the rules that prevent entry deserve at least as much attention as the merger that consolidation makes easier.

What the Google Case Actually Decided

Which brings us back to the search engine.

Google handles about 85 percent of American searches, by StatCounter’s count. That number is not why it lost.

In August 2024, Judge Amit Mehta found that Google had violated Section 2 of the Sherman Act. The finding was not that Google was large, or that its product was too good. It was that Google had paid to lock up the access points where a search engine gets chosen, and that the payments had frozen the market. In the remedy ruling a year later, he described the mechanism:

The numbers behind that finding are large. Google’s payments to phone makers, carriers, and browser developers for default placement reached $26.3 billion in 2021, according to figures presented at trial and reported by the Associated Press. That is not a company winning on quality. That is a company buying the shelf.

So what did the court do about it? Not what the headlines expected. It declined to break up Google, and it said so in terms that could have been written by a skeptic of antitrust:

What it ordered instead was narrower, and more interesting. No more exclusive distribution contracts for Search, Chrome, Assistant, or Gemini. Payments are still allowed, but they may not be conditioned on exclusivity, and the deals may not run longer than a year. Google must make some of its search index and user-interaction data available to qualified competitors, and offer search results and search text ads for syndication while those rivals build their own systems. Material changes to its ad auctions must be disclosed. The judgment runs for six years, with a technical committee watching.

For all the talk of a new era of trust-busting, the biggest antitrust case in a generation landed roughly where the old economics would have put it: stop the exclusionary contract, open the data, and let the next firm try.

The Amazon case, now scheduled for trial in March 2027, will test whether that approach travels. The FTC alleges that Amazon’s rules for third-party sellers create a price floor across online retail. That is an allegation, not a finding, and it will be argued for months. But the shape of the question is the same one Judge Nelson asked in Portland: does this practice protect customers, or does it protect the company from customers leaving?

The Three Questions

So what does this all mean?

The next time someone tells you a company is a monopoly, ask three questions.

Can I leave? Not “is it convenient to leave.” Can you actually take your business elsewhere, at a cost you would accept. If the answer is yes, you are looking at a company with a good product, and the resentment is about the product.

Can a rival enter? Not “is there a rival today.” Could one appear, with money, time, and a better idea. If the answer is yes, the advantage is on a clock, even if the clock is slow.

Who wrote the rule that stops them? This is the one that separates the cases. A company that got big by serving people can be displaced by the next company that serves them better. A company that got big because a statute says you may not compete cannot be displaced by anything except the legislature that wrote the statute. That monopoly does not have customers. It has a license.

The first two questions are economics. The third is politics, and it is where most of the durable monopolies in your life actually live - in the licensing board, the certificate of need, the exclusive franchise, the tariff schedule, and the statute that says the state’s own restraints are beyond review.

Next time you hear that everything is a monopoly now, do not ask how big the companies are. Ask what would have to change for someone to take their customers, and then ask who has the power to prevent that change. The answer will usually be more specific than you expect, and it will usually be written down somewhere.

Related: Competition and Co-operation · Public Choice Theory: Why Government Fails · The Problems of Capitalism · The Regulation Tax