The competitive ideal returns at the end of a volume in which almost everyone has found some way to depart from it. The single seller withholds, the few watch one another, the differentiated seller acquires a small domain, the incumbent guards the gate and the platform controls the passage. Against them stands the firm that accepts a price it cannot move and chooses its output on those terms. That firm is useful to imagine because it makes the absence of power precise. It is harder to find in the world with every supporting condition intact. The question is whether the distance between the model and the market describes a temporary disturbance on the way to competition, or a continuing task of making alternatives available. Nothing in the elegance of the ideal can answer that question for us. We have to ask both what the ideal establishes and what would bring its conditions into being.
The decisive feature of the competitive firm is the price it treats as given. It chooses what to produce at that price rather than selecting a quantity in order to move the market's price. The consumer likewise chooses an affordable bundle at terms the individual decision does not alter. An equilibrium requires these independently chosen plans to fit: demand must meet supply, with the appropriate treatment of goods that remain abundant at a zero price. Many small firms, comparable goods and ready access to alternatives can help make price-taking a plausible approximation. A head count alone cannot establish it. A small firm may face captive customers; a large supplier may face a credible entrant. The assumption concerns the choices and constraints facing participants, not a magic number of establishments after which power disappears.
The familiar marginal-cost result is one consequence of that premise, with its own conditions attached. For a firm able to vary output smoothly at an interior optimum, an additional unit should be made while its price exceeds the additional cost, and should not be made when that cost exceeds the price. At the interior stopping point the two coincide. A capacity limit, indivisible output or a decision about whether to operate at all requires a different comparison. Fixed costs matter to the latter decision even though they do not change the cost of the next unit once the installation is in place. Thus price equal to marginal cost is a sharp description of a particular margin. It is not a complete set of instructions for financing every service, nor a finding that any observed departure from equality must have the same cause.
An invented market makes the gain from the competitive margin easy to see. Suppose willingness to pay for the qth unit is 10 − q, production costs two per unit, and there are no fixed costs, capacity limits or effects on outsiders. Units through the eighth are worth at least what they cost, so a price-taking equilibrium supplies eight at a price of two. The area between willingness to pay and cost is thirty-two. A single seller facing that demand maximizes (8 − q)q by supplying four at six. Its profit is sixteen; buyers retain eight in surplus; the combined gain is twenty-four. The missing eight is the gain from the four additional units that buyers value above cost but the seller leaves unmade. The calculation gives the ideal a definite use: it identifies a loss created by that particular restriction under those stated conditions.
The broader welfare result concerns feasible improvements, rather than a sum of money. Under the first welfare theorem's conditions, a competitive equilibrium is Pareto efficient: there is no feasible rearrangement that makes someone better off without making someone else worse off. The reasoning depends on people optimizing at common prices, firms optimizing over their production possibilities, markets clearing and preferences satisfying the relevant nonsatiation condition. It also depends on the goods and claims being represented well enough that one person's unpriced effect on another does not sit outside the exchange. Convexity is not required for this direction of the result. The conclusion is conditional but substantial. If everyone has already chosen the best affordable opportunity and firms cannot produce a more valuable feasible bundle at those prices, a proposed improvement for all runs into a resource or value constraint. The theorem shows what the assumptions accomplish; it does not certify their presence wherever a price is posted.
Efficiency leaves the distribution of claims to be judged. A person whose income buys little may be making the best choice available and still have very little to choose. An allocation can be Pareto efficient precisely because helping that person requires taking something from another who would object. The theorem has no instruction to treat the objection as a conclusive judgment of justice. Nor does a large willingness to pay establish a greater human need independent of the resources with which the offer is made. The second welfare theorem makes the separation especially clear: under additional conditions, including the relevant convexity, an efficient allocation can be supported by prices after a suitable distribution of purchasing power. It does not show that the existing distribution deserves endorsement, or that the transfers needed to support a chosen alternative can be designed without difficulty.
Even the efficient competitive arrangement can be difficult to sustain where the technology will not divide neatly. A service with a large fixed installation and low marginal costs may fail to cover its full cost at a single price equal to marginal cost. The earlier Barrier's one hundred customers cost three hundred to serve through one installation but would pay only two hundred at the marginal cost of two each. The missing hundred must come from some other term of the arrangement if the service is to continue. A fixed charge, differentiated contributions or a public payment changes the way users share that requirement; each brings further questions about participation, information and fairness. The example does not overturn the first welfare theorem. It warns that the proposed marginal-cost outcome may not be a competitive equilibrium available to this technology in the first place. Existence and efficiency are different assertions.
A proof that plans can fit together does not explain how strangers discover the terms at which they fit. This is where the competitive ideal meets the problem of knowledge. No participant ordinarily possesses the whole account of resources, demands and opportunities on which a theorist writes an equilibrium. Hayek's argument for prices begins from that dispersion. A rise in the price of a material can induce people to economize on it without requiring each to know whether a new use has appeared or a source of supply has failed. Local adjustments communicate through the terms others face. The argument offers a reason to value a coordinating process that nobody directs in full. It does not require every buyer to know everything, and it does not establish that every price communicates only scarcity innocent of power. A deliberately restricted supply can also make a price rise.
Competition consequently has a temporal difficulty that the picture of a resting state can conceal. The possibility of a return invites an experiment; a successful experiment can give its maker an advantage; entry and imitation can erode the advantage; a barrier can preserve it. None of those transitions follows merely from the presence of the preceding one. Rivals need access to resources, customers and usable knowledge, and must be able to bear the risks of arrival. Consumers and workers need alternatives they can actually take. The institution of competition consists partly in maintaining these possibilities while allowing discoveries to be made and costs to be recovered. Its success cannot be measured by freezing a market at the instant before anyone succeeds, or by treating every return after success as a permanent entitlement.
The practical benchmark is therefore a comparison between arrangements that can be made to work. A proposal to admit entry should say how entry becomes possible. A proposal to separate a firm should say what remains connected and how the connection is governed. A proposal to regulate a price should say how costs are covered and what incentive remains to improve the service. These demands do not relieve power of scrutiny; they make the scrutiny consequential. The alternative to an incumbent's terms must be more than a diagram with the difficult institutions left outside its frame. Conversely, a defender of the incumbent must identify which benefits actually require the protected position, rather than taking the existence of useful output as proof that every protection is necessary. The ideal helps ask the question because it makes the cost of a constrained alternative visible.
Perfect competition cannot finally serve as a promise that a market will repair every departure from itself. It can serve as a disciplined account of what follows when a specified set of choices remains open and no participant can improve its terms by restricting the choices of others. Whether a particular market approaches that condition is an inquiry into its technology, information, rights and actual alternatives. The inquiry can find vigorous rivalry where a simple count suggested power, and dependence where many firms seemed to promise freedom. It can also find that competition has done its own work while leaving another problem untouched. A price-taking firm may still impose an unpriced cost on a neighbor. Removing a markup does not by itself bring that neighbor into the bargain. That is why the next volume must begin where this one ends.
Competition is a condition to explain and sustain before it can become a result to celebrate.
The market's resting state is thus not something the ideal can announce in advance. The ideal describes a relation among choices; history and institutions determine whether people can inhabit it. Power can grow from an obstacle, a discovery, a necessary coordination or a permission whose purpose has outlived its terms. Competition can interrupt it when real alternatives become available, yet the interruption itself requires an account of how those alternatives are made and kept useful. The volume has followed a price from the solitary seller to the platform's crowded gate and found the same question in different forms: who can refuse the terms, and what can they do instead? The next question belongs to those who never agreed to the exchange at all, yet must live with its consequences. The argument passes to the externality.