The Power

The whole of market power is contained in a single difference between two demand curves. In perfect competition, the firm is so small against the market that it faces a flat, horizontal demand: it may sell all it can produce at the going price and not a penny above, for the smallest premium would send every customer to an identical rival. Such a firm takes the price; it is handed to it by the market, and in equilibrium it equals the marginal cost of the last unit made — the equality, P equals MC, that is the signature of efficiency and the reason the competitive ideal is so cherished. A firm with market power faces, instead, a demand curve that slopes downward — in the limit of pure monopoly, the entire market's own demand — so that to sell more it must lower its price, and to charge more it loses some custom but not all. Such a firm does not take the price; it makes it. It chooses. That choice is power, and the downward slope of the demand it faces is power's entire origin.

From the downward slope the rest follows by an iron logic, drawn in the figure. Because selling one more unit forces the price down on every unit already sold, the extra revenue from that unit — its marginal revenue — is less than its price, and the marginal-revenue curve lies below the demand curve. The firm, seeking the greatest profit, pushes output only to the point where the marginal revenue of the last unit just covers its marginal cost; beyond that, each further unit would add more to cost than to revenue. That point lies at a quantity short of the competitive one — where price, not marginal revenue, equals marginal cost — and at a price above marginal cost. The monopolist, in a phrase, restricts output to raise the price. And the restriction has two consequences the figure separates: a rectangle of surplus transferred from buyers to the firm, the profit that power extracts; and a triangle of surplus simply destroyed — the deadweight loss, the gains from trade on the units never made, the exchanges that would have benefited both buyer and seller but for the price held high to protect the rest.

The signature of power, then, is the inequality the competitive ideal forbids: price greater than marginal cost. Where a market is competitive, the price a buyer pays equals the cost of making the last unit, so nothing is extracted and nothing is wasted; where there is power, the price exceeds that cost, and the gap measures at once the extraction and the inefficiency, the transfer and the loss. This single inequality, P greater than MC, is the deviation from the competitive benchmark that the whole of this volume sets out to measure, to source, to cost, and to remedy. And it is already tied to the question the volume travels under, for the monopolist's power to hold price above cost rests entirely on the buyer's having nowhere else to go — no substitute to flee to, no rival selling the same thing for less. The steeper and less yielding the demand the firm faces, the further above cost it may set its price; and how far, exactly, is the burden of the next plate.

If the signature of power is that price exceeds marginal cost, the natural measure of power is the size of the excess — the markup. Abba Lerner, in 1934, made the measure precise and gave it his name: the Lerner index, the difference between price and marginal cost expressed as a fraction of the price. It runs from zero, where price equals marginal cost and there is no power at all, toward one, as marginal cost shrinks to nothing beside the price. It is the single cleanest number for the degree of market power a firm possesses, and it is exactly the markup that The Share watched rising across the economy, the degree of monopoly that Kalecki made the determinant of the wage share. To measure power is, in the first place, to measure how far above cost the price is held.

And here the measure yields a result of startling economy, which ties the whole volume to its animating question. The profit-maximizing markup, Lerner showed, equals the inverse of the elasticity of the demand the firm faces. The reasoning is brief: a firm's marginal revenue is its price reduced in proportion to the elasticity of its demand, so setting marginal revenue equal to marginal cost and rearranging leaves the markup as one divided by that elasticity. Power is therefore, precisely and quantitatively, the inverse of the elasticity of demand — which is to say the scarcity of the buyer's alternatives, made into a number. A firm whose customers have many close substitutes faces highly elastic demand, for they will flee at the smallest premium, and so can mark up almost nothing: it has little power. A firm whose customers have few substitutes faces inelastic demand, for they have nowhere else to go, and so may mark up a great deal: it has much power. The figure is the whole of it. "Nowhere else to go," the phrase of the volume's question, is not a metaphor but the very quantity that sets the markup: the less elastic the demand, the fewer the alternatives, the greater the power, the wider the gap between price and cost.

This single relation unifies the threads the table has been spinning. The markup is the Lerner index is the inverse of the elasticity is the measure of how captive the buyer is — and the same logic, turned to the buying side, gives the monopsonist's markdown as the inverse of the elasticity of supply, the measure of how captive the seller is, which is the wage-below-product of The Wage's eighth plate and the subject of this volume's seventh. Power, on either side of any exchange, is the inverse of the other party's elasticity — the inverse of how readily they can walk away. What sets that elasticity, in turn, is the availability of substitutes and the freedom of entry, which the structure and barriers of the next plates determine. The markup measures power; the elasticity explains it; the barriers produce it.

A price held above cost is an advertisement. It announces profit, and profit summons entrants, who shave the price to win custom until it falls back to cost and the power is gone. It follows that market power, to last, requires something that keeps the advertisement from being answered — a barrier to entry, a reason the rivals do not come or the substitutes do not appear. Without such a barrier, any power is fleeting, competed away as fast as it arises; with one, the price can rest above cost indefinitely. The sources of durable power are therefore the sources of barriers, and they are worth cataloguing, for they are the organs by which the body of power is fed.

The deepest barrier is economies of scale, drawn in the figure as the natural monopoly. When the average cost of production falls across the whole reach of the market — because the fixed costs are vast and the cost of each further unit small, as in a railway, a power grid, a pipeline, or a piece of software written once and copied for nothing — then a single firm supplies the entire demand more cheaply than two could, for splitting the output would send each producer back up its falling cost curve. Here competition is not merely kept out but is positively wasteful, and the barrier is inseparable from a genuine efficiency, which is why the remedy for such a monopoly is not to break it up but to regulate it — a tension the tenth plate takes up. Beside scale stand the other classical barriers: the control of an essential input or facility that rivals must have and cannot obtain; the patent and the copyright, which are monopolies the state itself grants, by deliberate design, to reward invention; and the brand built by sunk advertising, which makes a product non-substitutable in the buyer's mind even where a chemical twin sells for less.

The modern age has raised one barrier above all the others: the network effect, where a product grows more valuable as more people use it — the telephone, the marketplace, the social network, the payment system. A large incumbent then offers something a small entrant simply cannot, an existing crowd, and users will not switch to a better-engineered rival that no one else has joined; the barrier is the incumbent's own scale, self-reinforcing and nearly insurmountable, and it is the subject of this volume's eleventh plate. To it one may add the switching cost, which holds a customer captive not by love but by the expense of leaving — the migrated data, the learned interface, the locked contract — and the strategic barriers a firm erects by its own conduct, the excess capacity held ready to flood a market upon entry, the price cut below cost to drive a rival out and warn the next away.

A crucial refinement, owed to William Baumol and his colleagues in the theory of contestable markets, corrects the instinct to count firms. What disciplines a price, they argued, is not the number of incumbents but the freedom of entry and exit. If entry were perfectly free and exit perfectly costless — if a firm could swoop in, undercut, profit, and depart before the incumbent could react — then even a single seller would be forced to price at cost, for any premium would invite such hit-and-run entry. The barrier that confers lasting power is therefore not concentration as such but the presence of sunk costs, the costs an entrant cannot recover if it leaves, which make hit-and-run impossible and entry a gamble. This sharpens the whole inquiry to its proper variable. A market with one firm and no sunk costs may be competitive; a market with several firms behind high sunk costs may not. Power lives not in the count of sellers but in the height of the wall around them.

Perfect competition and pure monopoly are the poles of a spectrum, and almost every real market lies somewhere along it. Two intermediate structures deserve names. The first, described independently by Edward Chamberlin and Joan Robinson in the same year, 1933, is monopolistic competition: many firms, each selling a slightly differentiated product — a restaurant, a brand of coffee, a neighborhood shop — so that each faces a gently downward-sloping demand and holds a sliver of power, able to charge a little above cost because its product is no one's perfect substitute. Yet entry into such a market is free, and entry competes away the profit that the sliver of power would otherwise yield.

The figure shows the long-run resting place. Entry continues until the typical firm's demand curve is merely tangent to its average-cost curve, touching at a single point and lying below it everywhere else — which means the firm earns exactly zero economic profit, its price just covering its average cost. But the tangency falls on the falling part of the cost curve, to the left of the lowest-cost scale, and this has two consequences that define the structure. The price still exceeds marginal cost, so a residue of allocative inefficiency remains even with profits competed away; and the firm produces less than the quantity that would minimize its average cost, leaving "excess capacity" — the unused scale that is the standing price of variety, the efficiency a society forgoes in exchange for the differentiation it enjoys. Whether that exchange is worth it — whether we are over-supplied with nearly identical brands at a small cost in efficiency, or properly supplied with the variety we value — is a question the model poses sharply and does not answer.

The second intermediate structure is oligopoly, the market of a few, where each firm is large enough that its choices move the market and it must therefore anticipate the reactions of its rivals — the strategic interdependence that is the subject of the next plate. Between monopolistic competition's crowd and oligopoly's handful, the mid-century economists of the Harvard school, Edward Mason and Joe Bain, drew a paradigm to organize the whole field: structure shapes conduct, and conduct determines performance. The concentration of a market and its barriers (structure) would govern how firms priced and whether they colluded (conduct), which would fix their profits and efficiency (performance). To make structure measurable they offered the concentration ratio, the combined share of the largest few firms, and later the more refined Herfindahl–Hirschman index, the sum of the squared market shares, which weights the giants heavily and survives today in the thresholds of merger control.

But the structure-conduct-performance paradigm carried a hidden assumption that the Chicago school exposed and that the previous plate already undermined: that concentration implies power. Harold Demsetz and others objected that a concentrated, profitable industry need not be the symptom of power at all — it may be the reward of efficiency, for the best firms grow large and earn high returns precisely because they are better, and the very correlation between concentration and profit that the Harvard school read as power is equally consistent with the dominance of superior firms. The objection broke the simple inference. One cannot read power off a concentration ratio, because the same ratio may describe a cosy cartel sheltered behind barriers or a contest won by merit in an open field. What distinguishes them is not the count of firms but the height of the wall — the barriers of the third plate — and the conduct of the firms within, which only the game theory of the next plate can model. Concentration is a clue to power, never a verdict.

Between the crowd and the single seller lies the market of a few, and it is the hardest of all to read, because each firm is now large enough that its own choices move the market and must therefore reckon with how the others will respond. This is strategy in the exact sense — the outcome for each depends on the choices of all — and its proper language is the theory of games, to which the table devotes a later element. The oligopolist cannot simply read its optimum off a demand curve, as the monopolist does, for the demand it faces depends on what its rivals do, which depends on what they expect it to do. The price that emerges from this mutual watching is the subject of the oldest and still the most instructive models in the field.

Augustin Cournot, writing in 1838 with a mathematical sophistication a full century ahead of his readers, gave the first answer. Suppose the firms choose quantities, each taking its rivals' output as given and producing its own best response; then each firm's optimal output is a declining function of the others', the reaction curves of the figure, and the market settles where these curves intersect — a point at which every firm is doing the best it can given what the others do, and none wishes to change. This is the Nash equilibrium, discovered by Cournot more than a century before John Nash gave it a general theory, and its location is the most important fact about the market of a few: it lies between monopoly and competition, more output and a lower price than a single seller would set, less output and a higher price than perfect competition would yield. And it moves: as the number of firms grows, the intersection slides toward the competitive corner, price approaching marginal cost; as it shrinks toward one, toward monopoly. Power, in the Cournot world, is a smooth function of the number of competitors.

Joseph Bertrand, reviewing Cournot in 1883, objected that firms set prices, not quantities, and that the change wrecks the result. If two firms sell an identical product at constant cost and compete in price, then each can seize the entire market by undercutting the other by a penny, and the undercutting cannot stop until price equals marginal cost — so that two firms alone suffice to reproduce the competitive outcome. This is the Bertrand paradox, and it is troubling precisely because the world does not behave so: real duopolies do not generally price at cost. The resolution is that the mode of competition matters as much as the number of competitors. Let the firms first choose capacity and then compete in price, and the outcome reverts to Cournot's; let their products differ even slightly, and undercutting no longer captures the whole market, so price floats back above cost. The lesson is not that one model is right but that the price the few will set depends on the fine grain of how they compete — quantities or prices, once or repeatedly, with capacities fixed or free.

The deepest feature of the market of a few is the standing temptation to stop competing altogether. The firms would jointly do best by colluding — restricting their combined output to the monopoly level and dividing the monopoly profit between them, the interior point in the figure. But that point is not an equilibrium of a single encounter, for each firm, taking the others' restraint as given, can profit by quietly expanding, and if all reason alike the restraint collapses and output swells back to the competitive level. Collusion is a prisoner's dilemma: collectively rational, individually unstable, betrayed by the very logic that recommends it. What rescues it, in the real economy, is repetition. When firms meet in the same market period after period, collusion can be sustained without a word of agreement, held together by the threat of future retaliation — the tacit understanding that to cheat today is to invite a price war tomorrow. The folk theorem of repeated games confirms that if firms value the future enough, a whole range of outcomes up to full monopoly can be quietly maintained. This is why cartels form and why they crack, why antitrust watches for the parallel pricing and price leadership that signal coordination without contract, and why the number of firms, the frequency of their dealings, and the visibility of their prices all bear on how much power the few can jointly wield.

Power does not stay where it is born. A firm secure in one market is forever tempted to use that security to win an adjacent one — to leverage the position it holds into a position it covets. The instruments are familiar. There is tying, the requirement that a buyer take a second product to obtain the first; bundling, the sale of two goods only together; exclusive dealing, the contract forbidding a distributor to carry a rival; and, most consequential in the modern economy, vertical foreclosure, in which a firm controlling an essential input, channel, or platform denies its rivals access, or degrades it, to disadvantage them in a market where it also competes. In each, the lever is the power already held, and the prize is power in the market next door. The historic instances are landmarks of the law: the tying of punch cards to tabulating machines, and, most famously, a software giant's binding of its web browser to the operating system that nearly every computer required, the better to foreclose the rival browser that might one day have grown into a platform of its own.

Against the instinct that such leverage is plainly anticompetitive, the Chicago school raised an argument of real power, the theorem of the single monopoly profit. A firm with one monopoly, it observed, can extract its monopoly profit only once; it cannot multiply that profit by reaching into a second, competitive market, because it already captures the available surplus through the price of the monopoly good itself. Tie a second product to the first, and the monopolist does not enlarge its take but merely relocates it — so tying and bundling, far from being engines of conquest, must be sought for their efficiencies: the cost savings of selling together, the assurance of quality, the quiet practice of price discrimination. On this reasoning the law should presume such conduct benign, and for a generation it largely did, softening the old per se prohibition of tying into a permissive rule of reason.

The theorem was an insight, and it disciplined a great deal of loose talk about leverage. But it holds only within walls the real economy does not respect. Its logic assumes that the second market is genuinely competitive and unprotected, that the goods are consumed in fixed proportions, and — crucially — that the first monopoly faces no threat from the second. Lift any of these and the conclusion reverses. When the adjacent market has scale economies or network effects of its own, foreclosing a rival there can defend or extend the original monopoly, for the rival, left to grow, might have become a platform that eroded it; the leverage is then not a relocation of profit but an insurance of power, and Michael Whinston showed formally that tying can be a profitable strategy precisely when it deters the entry that would otherwise threaten the monopolist's home market. This is the foreclosure concern that the Chicago argument had prematurely buried, and it has returned in its sharpest modern form as the question of platform self-preferencing — the marketplace that ranks its own goods above its rivals', the search engine that favors its own services — which the eleventh plate takes up as the central battleground of present competition policy. The doctrine has swung with the argument: from the per se illegality of the mid-century, through the Chicago-inspired permissiveness, to a renewed and unsettled scrutiny as the platforms revive the old fear.

Every exchange has two sides, and so does power. The volume has so far watched the seller — the firm that restricts what it offers to lift the price above cost. Its mirror is the buyer that restricts what it purchases to push the price below value: the monopsonist, the single buyer in the limit, of whom The Wage met one organ in the single employer of labor. That plate showed an employer paying its workers less than their marginal product; this plate shows that organ to belong to a whole body, for the same power operates wherever a dominant buyer faces sellers who have nowhere else to take their wares.

The mechanics are the monopoly diagram reflected. A buyer facing an upward-sloping supply curve knows that to purchase one more unit it must raise the price on every unit it already buys, so the true marginal cost of the input — the marginal factor cost — lies above the supply curve, just as the monopolist's marginal revenue lay below demand. The buyer purchases only up to the point where this marginal factor cost equals the marginal value of the input to it, a quantity short of the competitive one, and pays the price read off the supply curve at that quantity — a price below the input's marginal value. The gap between value and price paid is the markdown, the precise mirror of the monopolist's markup, and by the logic of the second plate it equals the inverse of the elasticity of supply, the measure of how few alternatives the seller has, as the markup equalled the inverse of the elasticity of demand. Monopoly restricts what it sells; monopsony restricts what it buys; both transfer surplus to the powerful side and destroy some at the margin; and both are governed by a single equation read from opposite ends.

The body of buyer power is as large as the body of seller power, and was long the more neglected. In the labor market — the case The Wage anatomized — it reaches far beyond the company town that first defined it: the concentration of employers in a local market, the occupational license that bars entry to a trade, the non-compete clause that chains a worker to a firm, the no-poach agreement by which employers quietly promise not to hire one another's people, each confers the power to hold wages below the worker's product. In the supply chain it is the dominant retailer squeezing its manufacturers, the processor dictating terms to the farmers who must sell to it, the platform setting the rate it will pay the drivers and merchants and makers who depend upon it. And it carries a complication seller power does not, which Galbraith named countervailing power: a mighty buyer that beats down a mighty seller may pass the savings to consumers as lower prices, so that buyer power can shield the public even as it squeezes the supplier. The antitrust authorities, having focused for decades on seller power and the price the consumer pays, have only lately turned to the buyer — prosecuting the no-poach agreement, weighing the effect of mergers on workers and suppliers, moving against the non-compete — and the recognition that the single-buyer model describes far more of the labor market than the company town it was built for has made buyer power one of the liveliest frontiers in the field.

That power transfers surplus from the weak side to the strong is plain from the first plate; the harder question is how much it costs society as a whole — not who gains and who loses, but what is destroyed in the destroying. The textbook answer is the deadweight loss: the triangle, in the monopoly diagram, between the demand curve and the cost curve, spanning the units the monopolist declines to make. These are exchanges that would have benefited both buyer and seller, foregone because the firm holds output back to protect the price on the rest — gains from trade simply annihilated, belonging to no one. This allocative loss is the canonical cost of monopoly, and for a long while it was thought to be the whole of it.

Arnold Harberger put a number to the triangle in 1954, and the number was startlingly small. Measuring the deadweight losses across American manufacturing, he found them on the order of a tenth of one percent of national income — because the markups he observed, and the demand elasticities they implied, were modest, and the area of a triangle shrinks with the square of its small sides. The implication was reassuring and became a pillar of the case against vigorous antitrust: if monopoly costs a society so little, the apparatus of competition policy was a remedy in search of a disease. But the Harberger triangle measures only the allocative loss, and two later arguments found the cost of power to lie mostly elsewhere, in places his triangle could not see.

The first is rent-seeking, named by Gordon Tullock and pressed by Richard Posner. The rectangle of monopoly profit — the markup times the quantity, the surplus transferred from buyers to the firm — is conventionally not counted as a social loss at all, for it is a mere transfer, a movement of surplus from one pocket to another. But this overlooks the contest to possess it. If the monopoly rent is a prize, firms will spend real resources to win and to keep it: lobbying legislatures for protective regulation, litigating to exclude rivals, building excess capacity to threaten entrants, pouring money into advertising whose only function is to fortify position. To the extent the rectangle is competed for, it is dissipated — the resources poured into the fight produce nothing, and a transfer becomes a waste. Since the rectangle is typically far larger than the triangle, the social cost of monopoly, properly counted, may be an order of magnitude greater than Harberger found. The second argument is Harvey Leibenstein's X-inefficiency. A firm sheltered from competition lacks the discipline that forces costs to their minimum, and so runs slack — the padded payroll, the unpressed supplier, the comfortable routine, the quiet life that an unchallenged monopoly can afford. The cost of power is then not only that the firm makes too little, but that it makes it badly, at a cost above the necessary; and Leibenstein suspected this productive waste might dwarf the allocative triangle too.

The modern verdict has swung back toward the cost being large. As The Share's ninth plate recorded, markups across the economy appear to have risen substantially since 1980; and a bigger markup means a bigger triangle, a bigger rectangle to be dissipated, and more slack to hide behind. Economists now study the macroeconomic shadow of rising market power — output and investment held down, the labor share eroded, business dynamism and the founding of new firms in decline, inequality widened — and few now regard the cost of power as the trifle Harberger's triangle suggested. The full reckoning has three layers: the allocative loss of the triangle, modest and measurable; the rent-seeking dissipation of the rectangle, large and conjectural; and the productive slack of the unpressured firm, larger still and harder to see — to which the dynamic and distributional consequences must be added. Power is not cheap. It only looked cheap when only the smallest part of its cost was being measured.

Every cost the previous plate tallied is a static cost — measured at a frozen instant, with technology given and the only question how the fixed pie is priced and divided. The most powerful defense of market power denies that this is the question that matters. Joseph Schumpeter, in 1942, argued that the essential fact about capitalism is not the pricing of existing goods but the perennial gale of creative destruction — the ceaseless innovation that destroys old products, methods, and firms and throws up new ones — and that the competition which counts is not the textbook jostling at the margin of price but the competition of the new commodity, the new technology, the new form of organization, which strikes not at the profits of incumbents but at their foundations and their lives. And the engine of that gale, Schumpeter insisted, is the prospect of monopoly. The entrepreneur innovates in hope of the temporary monopoly that success confers; the rent is the prize, and without the prize the incentive to innovate withers. On this view the static deadweight triangle is a rounding error beside the dynamic gains from the innovation that the hope of monopoly calls forth — and the law itself ratifies the bargain in the patent, a monopoly the state grants by deliberate design, trading the static cost of monopoly pricing for the dynamic gain of invention rewarded.

The defense met a formal rebuttal from Kenneth Arrow, who in 1962 argued that Schumpeter had the incentive backwards. Consider who gains more from a given innovation: a monopolist, or a firm in a competitive market. When a monopolist innovates, it partly cannibalizes the profit it already earns — it replaces itself, so its net gain is only the increment of the new over the old. A competitive firm or a fresh entrant, having no existing rents to lose, captures the whole of the new market, and so has the larger incentive to innovate. By this replacement effect, monopoly should retard innovation, not drive it: the secure incumbent rests in Hicks's "quiet life," while the hungry challenger races ahead. Arrow's logic and Schumpeter's point in opposite directions, and for decades the evidence seemed to support now one, now the other.

The reconciliation, drawn in the figure, came from Philippe Aghion and his colleagues in 2005: the relationship between competition and innovation is an inverted-U. Where competition is faint and monopoly entrenched, innovation is low — Arrow's replacement effect and the quiet life, little pressure and much to cannibalize. Where competition is fierce and rents are instantly competed away, innovation is low again — Schumpeter's point, for there is no prize to reward the effort. Innovation is greatest at intermediate competition, where firms running neck-and-neck innovate hardest to escape their rivalry and seize a temporary lead — the escape-competition effect. Schumpeter and Arrow, it turns out, each described one arm of a single curve. And the lesson for the law is double-edged: a degree of market power, the hope of temporary rents, is indeed necessary to the innovation on which long-run welfare most depends — but too much power is as deadening as too little, and the entrenched, unchallengeable monopoly sits on the wrong, low slope of the hill. The innovation Schumpeter prized issues from the contest for power, not from its secure possession.

A society that judges market power a failure must decide what to do about it, and the instrument it has built is the law of antitrust — competition law, in the European phrase. Its founding text is the American Sherman Act of 1890, a child of the Gilded Age and its great trusts, the oil and railroad and sugar combinations whose power had grown to rival the state's; the Act forbade in its first section the contracts and combinations that restrain trade, the cartels and conspiracies of the few, and in its second the act of monopolizing itself. The Clayton Act of 1914 reached specific practices — the merger that lessens competition, the tie, the exclusive deal — and the same year's statute raised the Federal Trade Commission to police unfair methods of competition. Across the Atlantic, a parallel apparatus would grow around the prohibition of abuse of dominance and anticompetitive agreement. Two doctrinal tools order the whole edifice: some conduct, the naked fixing of prices or division of markets, is condemned per se, illegal without inquiry into its effects; the rest, the vertical restraint and the single firm's conduct and the merger, is weighed under the rule of reason, its harms and benefits balanced case by case. Much of antitrust history is the migration of practices between these two boxes.

The largest migration was the work of the Chicago school, and it amounted to a revolution in what antitrust was for. Through the 1960s and 1970s, Aaron Director, Richard Posner, and above all Robert Bork, in a book whose title named the perceived problem, The Antitrust Paradox, argued that the law had lost its way in a thicket of incompatible goals — protecting small competitors, resisting bigness as such, dispersing power for its own sake — and should be disciplined by a single lodestar: consumer welfare, which Bork largely identified with economic efficiency and measured by prices and output. Practices the courts had long condemned were, on the Chicago analysis, usually efficient or harmless; predatory pricing in particular was held to be irrational and rare, for the predator bleeds more than its prey and the prey may simply wait. Markets, the school taught, correct monopoly faster than courts can, so intervention should be sparing. The courts and the enforcement agencies embraced this view from around 1980, and the consequence was a generation of permissive antitrust: monopolization cases grew scarce, mergers passed easily, and the price-focused consumer-welfare standard reigned. To its admirers this was the maturation of a once-incoherent field into a rigorous one; to its later critics it was the law falling asleep while concentration and markups, the rising power of The Share's superstar firms, climbed unchecked.

The reaction, gathering through the 2010s, took the name of Louis Brandeis, the old foe of bigness, and found its manifesto in a law student's article. Lina Khan's Amazon's Antitrust Paradox, in 2017, argued that the consumer-welfare standard is structurally blind to the power of platforms: a firm that keeps its prices low, even below cost, while integrating across markets and foreclosing rivals and amassing dominance, registers as no harm at all to a standard that watches only price — and yet its power grows. The New Brandeisians would return antitrust to a concern with structure, with the competitive process rather than the consumer price alone, with power over workers and suppliers and the conditions of innovation and even the health of democracy — the dispersal of concentrated economic might that the framers of the Sherman Act, suspicious of private power in every form, had chiefly sought. Khan came to chair the Federal Trade Commission; major cases issued against the great platforms; new and stricter merger guidelines lowered the thresholds of concern and widened the recognized theories of harm to reach labor markets and the acquisition of nascent rivals. The consumer-welfare standard, unquestioned for a generation, is now the central battleground of the field — and the contest beneath the doctrine is the oldest question of all: what should the law maximize? The consumer's price, precise and narrow; total efficiency, broader; or the competitive process and the dispersal of power, broadest and hardest to measure. The choice is a normative one wearing technical dress, and it decides what counts as a failure worth curing.

The digital platform has become the characteristic form of market power in our age, and it gathers nearly every theme of this volume into a single, novel shape. Its first peculiarity is that it is a two-sided market, in the sense Jean-Charles Rochet and Jean Tirole gave the term: it serves two or more groups whose participation each makes the other more valuable — buyers and sellers on a marketplace, riders and drivers on a ride service, users and advertisers on a search engine or social network, cardholders and merchants on a payment system. Because the value to each side rises with the participation of the other, the platform's pricing must balance the two, and it commonly charges one side heavily while subsidizing the other to nothing — free search paid for by advertisers, a free social network monetized by attention. This balancing breaks the intuitions the rest of the volume relied upon: a price below cost on one side may be no predation but the natural way to bring both sides aboard, and a high price on the other no simple mark of power. Antitrust built for one-sided markets has struggled even to define the relevant market here.

The platform's second peculiarity is the ferocity of its network effects, which turn the barrier of the third plate into something close to destiny. When a service grows more valuable as more people use it, and more valuable on each side as the other side fills, a market can tip decisively toward a single dominant platform — and once tipped, the dominant platform is nearly impossible to dislodge, for no user will cross to a rival the other side has not joined. The winner-take-most outcome of The Share's superstar firms is here a structural tendency, and it is reinforced by data: the feedback loop in which more users yield more data, more data a better product, and a better product more users, is self-propelling, and the incumbent's accumulated data becomes a barrier no entrant can vault — power and its own moat in one. The platform's third peculiarity defeats the chief instrument of the previous plate. When the service is free, there is no price for the consumer-welfare standard to examine, and yet power is plainly wielded — the true price being the user's attention and data, and the dimensions along which power tells being the quality of the service, the privacy surrendered, and the terms on which data is extracted, all of which a price-watching standard cannot see.

Upon these foundations the platform practices a power that is the digital reprise of this volume's earlier plates. It engages in self-preferencing — the marketplace ranking its own goods above its rivals', the search engine steering users to its own services — which is the leverage and foreclosure of the sixth plate in a new room. It defends itself by the kill zone, acquiring nascent competitors before they can grow into threats, the purchase of potential rivalry that the inverted-U of the ninth plate marks as innovation's enemy. It exercises buyer power, in the seventh plate's sense, over the merchants and drivers and creators who have nowhere else to reach their customers, dictating the rates and the terms. And it acts as a gatekeeper, the indispensable intermediary that can tax and exclude at will. The policy response is itself dividing along the fault line of the last plate: Europe's Digital Markets Act imposes structural rules on designated gatekeepers in advance, forbidding self-preferencing and mandating interoperability, a regulatory approach in the spirit of the natural-monopoly remedy; the United States proceeds case by case, with great suits against the dominant search, mobile, marketplace, and social platforms, and a live debate over whether to regulate them as utilities or to break them apart.

The volume opened with a definition and may close by collecting what the definition has yielded. A firm has market power when it can set its price away from the price competition would impose — above the cost of what it sells, below the value of what it buys — and the entire inquiry has been the anatomy of that single deviation. Power's signature is the gap between price and cost. Its measure is the markup, or on the buying side the markdown, which proves to be the inverse of the other party's elasticity, and so the scarcity of their alternatives made into a number. Its source is the barrier to entry, for power without a barrier is power competition dissolves. Its degrees run along a spectrum from the competitive crowd to the single seller, with concentration a clue to its presence but never a proof. Among the few it becomes strategic, a game of anticipation in which collusion beckons and betrayal tempts. It is extended by leverage from one market to the next, and mirrored exactly in the buyer who holds the seller below value. It costs a society the triangle of trade destroyed, the rectangle of rent dissipated in the seeking, and the slack of the firm no rival presses. It is defended, with real force, as the prize that pays for innovation — though the inverted-U shows that too much of it deadens what a little of it spurs. It is restrained, when a society chooses, by the contested instrument of antitrust. And in our age it gathers, above all, in the platform.

From this anatomy the volume's thesis stands clear, and it inverts the order in which economics is usually taught. Perfect competition — every actor a price-taker, every price equal to marginal cost, power exactly zero — is not the normal state of markets from which monopoly is a departure. It is the rare and frictionless limit, the special case that obtains only when both sides of every exchange have somewhere else to go. Power, present in degree, is the general case, the ordinary condition; the question is never simply whether a market has power but how much, on which side, behind what barrier, and at what cost. The competitive ideal is the exception taught as the rule, and the price that bears power's mark — above cost where the seller commands it, below value where the buyer does — is not the verdict of competition but the outcome of a contest that the absence of alternatives has already settled.

This is why the volume answers, at last, the question it has carried: what is a fair price worth, when one side has nowhere else to go? The fair price of the competitive ideal, equal to cost, presupposes that both sides could walk away — that each disciplines the other by the credible refusal the bargaining theory of The Wage made decisive. Power is precisely the absence of that refusal, the missing alternative, the "nowhere else to go." A price set under power therefore measures not the worth of the thing but the want of an alternative, and to call it fair because the market yielded it is to mistake the verdict of power for the verdict of competition. Fairness in price presupposes alternatives; power is their absence; and so the "fair price" extracted from a party with nowhere else to go is a contradiction the market quietly conceals — the deepest instance of the whole table's lesson, that a price is a verdict on a contest and only as fair as the contest behind it.

And here the volume reveals its place beneath the three that concern distribution, for it is their foundation. They were each, in the end, accounts of how the product is divided, and each invoked power to explain a division that productivity alone could not. The Wage found the worker paid below his product by the single buyer of labor — and that buyer was monopsony, the buyer power of this volume's seventh plate. The Share found labor's portion of the whole falling, and its leading explanation the rise of markups and superstar firms, the loss flowing not to capital but to pure profit — and that markup is this volume's measure of power, that pure profit its rent. The Surplus found efficiency silent on who keeps the gains from trade, and the keeping decided by power, its cost the deadweight triangle and the rent-seeking rectangle — which are this volume's eighth plate exactly. The great result of The Share was that the division of the product has a degree of freedom technology leaves open, filled by power; this volume has named what that power is, structurally — the ability to set price away from cost, conferred by barriers, measured by the markup and the markdown. The distributional triangle stands on the anatomy of power; remove it, and the wage, the surplus, and the share have no account of the force that fills their open degree of freedom. The four volumes are one structure: three of distribution, and beneath them the power that distributes.

The foundation hands on what it cannot itself contain. It hands to its neighbors in the family of failures — to The Externality, The Asymmetry, and The Commons — the recognition that power is the first of the market's failures but not the last, and that they intertwine, the data that confers platform power being also the information asymmetry of its sister element. It hands to The Rent the profit that power yields, for the rent is power's revenue — the return to a protected position, the pure profit The Share watched rising — and that element will anatomize the rent itself, in land and patent and scarcity, where this one has only named its source. It hands to The Tax the state as the counterweight, the antitrust and the regulation and the levy by which a public answers private power. It hands to The Firm the truth that the firm is itself an island of power and planning within the market, its very boundary a question of where command displaces price. And it hands to The Rule the law that both confers power — in the patent it grants, the charter it issues, the property in essential things it protects — and restrains it, for power is made and unmade by rules, and is never the brute fact of nature it pretends to be.

So the series' conceit completes another turn. It began by denying that prices are facts of nature, insisting that each is a verdict on a hidden contest; and this volume has named the contest most directly of all. The price beneath every price is a measure of who had somewhere else to go. Where both parties did, the price rests at cost, and competition has spoken; where one did not, the price departs from cost by the measure of the other's power, and what has spoken is not the market's wisdom but the scarcity of an alternative. To read a price as fair because the market set it is to forget that the market's verdict is only as just as the contest behind it, and that the contest is decided by power — by the barrier, by the markup, by the want of a substitute. This volume has made the hidden contest visible, and it has found power not at the margins of economic life but at its center, the sovereign that the competitive story was written to forget. The argument remains open, as it must — for the measure of power is contested, its cost spans orders of magnitude, its part in innovation cuts both ways, and its remedy turns on a choice of ends that economics cannot make. But the price will no longer pass for nature. It is a verdict on a contest; and where one side had nowhere else to go, it is power's verdict, read at last for what it is.