One wire, one choice
The reader who has tried to leave an internet provider and found nowhere to go has already met this chapter’s subject personally. The service disappoints, the price climbs, and the market’s advertised remedy, take your business elsewhere, turns out to require an elsewhere. For one address that means one cable company; for another, one power utility, one hospital system within reach, two app stores, a single buyer for the region’s harvest. The question this raises is sharper than a complaint about any one bill. Competition is the load-bearing assumption of the market case that the last chapter’s argument otherwise strengthened: prices only carry honest information when someone can undercut a dishonest one. So the reader’s question, why do I get exactly one choice, is really a question about when the market’s central mechanism can be trusted to run. (The other half of the question as people usually ask it, why the middleman takes the biggest cut even where there are choices, is the next chapter’s.)
The promise the textbook makes
The promise deserves a fair statement before the exceptions, because much of the time it delivers. Wherever a seller earns more than costs justify, the excess is an advertisement: entrants come, capacity grows, prices fall, and the profit erodes toward the cost of staying in business. Nobody plans this and nobody needs to. It is why bread, haircuts, furniture, and takeout run on thin margins in every open economy, and the strongest version of the claim says the incumbent does not even need to see a rival’s face: where entry and exit are genuinely free, the mere threat of a raid disciplines prices as effectively as the raid itself (William Baumol’s contestable-markets argument, 1982). The textbook is not wrong about the machine. It is quiet about the machine’s operating condition, which is that entry must remain possible at a cost an entrant can recover. Every durable monopoly in the reader’s life is a place where that condition has failed, and the failures come in four repeatable shapes.
The four moats
Scale. Some businesses get cheaper per unit the bigger they run, without practical limit inside their market. The wire to the reader’s house is the pure case: the second cable laid down a street costs nearly what the first did and can win at most the customers the first one loses, so the arithmetic of recovering the dig never closes, and the street gets one wire. The same logic, softened, covers power grids, pipelines, rail, chip fabrication, and any trade where the fixed cost towers over the marginal one. Against a scale moat, entry is not merely hard; at natural-monopoly scale it is arithmetically self-defeating, which is why these industries end up regulated, municipal, or concentrated wherever they appear.
Networks. Some products are worth using because of who else uses them: the telephone in its first century, the marketplace where the buyers already are, the social platform holding everyone the reader knows. Where value rides on the user base, the network effect taxes every attempt to leave for something smaller (Katz & Shapiro, 1985), and an entrant is selling an empty room. How platforms turned this into the dominant business model of the reader’s pocket is treated with the attention economy in Volume II: History, in the chapter on how attention became a product.
Switching costs. Even without networks, staying can be engineered to beat leaving: the number that cannot move, the files in a proprietary format, the years of learning a system, the early-termination fee, the integrations that would all break at once. Each customer’s decision to stay is individually rational and collectively the incumbent’s fortress (Klemperer, 1987). Switching costs explain the otherwise puzzling generosity of introductory offers: the discount is not a gift, it is the purchase price of a customer who will be expensive to un-become.
Captured rules. The three moats above erode. Technology moves, fixed costs fall, networks tip. The moat that lasts is written into law: the license that caps how many competitors may exist, the franchise agreement for the right-of-way, the safety standard drafted so that only the incumbent’s process passes, the permitting queue an entrant cannot survive. George Stigler’s observation (1971) was that this is not an abuse of regulation so much as a market for it: regulation is, as a rule, “acquired by the industry” it governs and operated for its benefit. Why the industry reliably outbids the public for the pen, a logic of concentrated and diffuse stakes, is the machinery of Chapter 8, where capture gets its full political treatment; here it matters as the moat the other three moats buy.
The moats matter jointly, not severally, because a real incumbent runs them as a system. Scale profits pay for the lobbying; the network holds customers still long enough for switching costs to accumulate; the captured rule buys time for the network to tip. And a moated position converts time itself into an ally, since entry, where it is attempted, takes years of building, while an incumbent’s price response takes an afternoon. An entrant’s investors can do this arithmetic too, which is the quiet way most monopolies defend themselves: the fights never happen. Chapter 1 filed fortunes built on enclosure on the zero-sum side of its ledger; the four loops are the enclosure’s working parts.
What the toll collects
Persistence would matter less if a moated position were merely profitable. What it is, structurally, is a different relationship to price. A competitive seller takes the price the market sets and lives on volume; a moated seller sets the price, and sets it by a calculation with no line item for the customers it prices out: raise the toll until the buyers lost would cost more than the margin gained. The bill the reader keeps paying is the visible half of the cost, the transfer that Chapter 1 classified with the enclosure fortunes. The invisible half is the exchange that never happens: every buyer priced out of a good that costs less to make than they would gladly pay is surplus that simply fails to exist, which is why economists count monopoly as a loss to the whole ledger and not only a grievance of the payer.
The subtler charge is slack. A seller no customer can punish is a seller no signal disciplines, and the discipline does not only govern price. John Hicks’s summary (1935) has outlived every model built since: “the best of all monopoly profits is a quiet life.” The quiet life is recognizable from the customer’s side of the counter: the four-hour service window, the interface unchanged since the franchise was won, the queue that is the product’s real price. And where a moated firm does innovate, the incentive tilts toward innovations that deepen the moat, tighter integration, higher walls around the data, rather than toward whatever the customer would have chosen with somewhere else to go. None of this requires a villain in the corner office, which is the chapter’s recurring lesson: the moat prices, the moat relaxes, the moat defends itself, whoever sits behind it.
What the ledger shows
Mechanisms deserve measurement, and market power leaves a measurable trail: the share of an industry’s sales held by its few largest firms. In the United States, whose economic census makes the accounting unusually complete, that share has been moving one way for decades. The sector series assembled by Autor, Dorn, Katz, Patterson & Van Reenen (2017) follows six large sectors, manufacturing, retail, wholesale, services, finance, and a combined utilities and transportation sector, together covering about four-fifths of private-sector employment. Across four-digit industries averaged within each sector, the slice of sales taken by the top 4 firms rose between 1982 and 2012 in all six: in retail trade from 15% to 30%, in manufacturing from 38% to 43%, in finance from 24% to 35%. The pattern is the point: not one runaway industry but a broad tilt of the floor.
Two honesty notes belong to this ledger. Concentration is a symptom of market power, not a measurement of it: an industry can concentrate because its worst firms lost, which is competition working, and the researchers behind the sector series read part of the rise that way, as markets tipping toward genuinely more productive superstar firms. And the national numbers are not the reader’s street. Measured locally, concentration has in many industries fallen over the same decades, because the national giants’ arrival in a town adds a competitor to the ones already there (Rossi-Hansberg, Sarte & Trachter, 2021). The moats framework absorbs both notes without strain: it predicts power where the loops run, not wherever a share number is large. The wire, the platform, the locked-in system, and the licensed franchise are where the reader should expect the bill to behave like a toll; the concentrated-but-contested aisle of the supermarket is not.
Why entry alone does not clean up
The chapter’s question can now be answered in one sentence: monopolies persist where one of four loops keeps rebuilding the wall faster than entrants can climb it, and the textbook’s erosion story holds only where no loop has formed. History adds a blunter witness. The United States, the economy most rhetorically committed to free markets, found competition so far from self-sustaining that it built a permanent legal machine to manufacture it: the Sherman Act (1890), the breakup of Standard Oil (1911), the divestiture of AT&T (1984), each an admission that the moats do not drain themselves. The measured turn after 1997 has been read in the same light: the researchers who documented it point to slackened antitrust enforcement and rising technological barriers to entry as the likely movers (Grullon, Larkin & Michaely, 2019). On that reading, the quarter century of rising concentration is not a market failing by itself; it is maintenance deferred.
That framing settles the reader’s standing toward the modem. The single wire is not an aberration in an otherwise competitive world, and not proof that competition is a myth. It is what a market looks like at the places where competition has to be maintained and currently is not: where scale arithmetic, network gravity, engineered stickiness, or a friendly rulebook has suspended the erosion the textbook takes for granted. The diagnosis also prices the remedies, since each moat names its own treatment, and each treatment requires someone with authority and a motive to apply it, which is where the trouble moves from economics to politics.
Where the argument goes next
Politics is the right word for what this chapter leaves open. Captured rules are the durable moat, and capture is not an accident of bad character but a predictable equilibrium of who organizes and who cannot; that machinery, with the subsidies that demonstrate it in its purest form, is Chapter 8. The network moat’s modern giants, platforms that sell the reader’s attention rather than the reader’s subscription, have their own economics in Volume II: History, in the chapter on how attention became a product. And the question the reader actually opened this book with, why the middleman standing between maker and buyer so often captures the largest slice, gets the next chapter, where the answer turns out to have less to do with villainy than with who bears risk, owns the customer, and solves the matching problem.