Surplus that leaves a buyer's hands need not disappear. The seller may receive it, or a government may collect it, and the payment can be followed into someone else's account. There is a further possibility: a surplus that could have existed is prevented from existing. A buyer values a thing at fifty pounds; providing it costs thirty. There are twenty pounds of gains available for the parties to divide. Now suppose a charge of twenty-five must be paid on the exchange, with no other circumstance changed. No price can leave the seller the thirty required to provide the thing while keeping the buyer's full payment within fifty. The trade does not occur. No twenty-five is collected, and no one receives the twenty of gains that the trade would have made possible. That unrealized benefit is the deadweight loss. Its name describes a comparison with a feasible alternative: value that could have been created, after meeting the real costs of creation, but is enjoyed by nobody.
The familiar triangle records the same event across many possible trades. Take a hypothetical market whose demand price is one hundred minus quantity, with a constant resource cost of twenty for each unit. Assume that willingness to pay provides the relevant money measure of benefit, that buyers receive the highest-valued available units, and that there are no costs or benefits outside the exchange. At a competitive price of twenty, eighty units are traded and the surplus is thirty-two hundred. A tax of twenty per unit raises the buyer's price to forty, leaves sellers twenty, and reduces quantity to sixty. Consumers now keep eighteen hundred of surplus, while twelve hundred goes to the treasury. The combined three thousand falls two hundred short of the original thirty-two hundred. The shortfall is the gain on the twenty units no longer traded: their valuations range from forty down to twenty, against a cost of twenty each. The revenue is a rectangle twenty high and sixty wide. The missing benefit is the triangle beside it, twenty high and twenty wide.1
The loss depends on how much conduct the wedge changes. Double the tax in this example to forty: quantity falls to forty, the revenue rises to sixteen hundred, and the lost surplus rises to eight hundred. The tax has doubled and the triangle has quadrupled because both its height and its width doubled. That square relation is exact for these linear schedules over the stated range; it is not a law that can be carried unchanged through every demand curve and every existing tax. If quantity responds less to the same wedge, fewer beneficial trades disappear. If the quantity truly remains fixed and the allocation is unchanged, a payment can change hands without this particular quantity loss. The relevant response may also take time: households can change habits, and firms can replace equipment or alter location. A diagram based on the first week's adjustment need not measure the eventual burden. The inquiry is therefore about the attainable alternatives and the response at the margin, rather than the size of the cash payment alone.2
A monopolist's restriction can leave the same triangle with a different recipient of the rectangle. In our market, a single-price seller maximizes profit by supplying forty units at sixty. The sixteen hundred above production cost goes to the seller; the eight hundred of gains on the missing units goes to nobody. Harberger's 1954 study brought such reasoning to estimates for American manufacturing in 1924–1928. He used differences in accounting returns, assumed long-run constant costs and specified demand responses, and arrived at a monopoly-allocation loss below a tenth of one per cent of national income. That was a model-based historical estimate, with explicit qualifications about the data and omitted selling costs. It was not a theorem that monopoly losses must be small. Applying the construction elsewhere requires evidence about costs, demand, and the relevant alternative allocation. Profit alone does not identify the triangle, just as the triangle alone does not measure the whole significance of a firm's power. The model makes one consequence measurable by specifying the particular consequence being measured.3
A trade that benefits its buyer and seller can still impose a larger cost on someone outside their bargain. The efficient benchmark must include that cost before a restriction can be called a loss. Suppose each unit in our example also causes twenty pounds of damage to others, absent from the seller's bill. The social cost of another unit is then forty. Only sixty units are worth supplying on the specified money measure, rather than eighty. A tax of twenty now brings the private price into line with that social cost and removes the last twenty units. Those units carried two hundred of gains for the trading parties, but caused four hundred of damage; suppressing them improves the total by two hundred. The same physical reduction that appeared wasteful when no harm was present is beneficial when the harm is counted. Pigou's distinction between private and social product gives the reason. Before drawing a deadweight triangle, one must establish which benefits and sacrifices its curves include. A wedge that corrects an omitted cost is doing different work from a wedge that merely prevents an otherwise worthwhile exchange.4
The accounts of lost output, resource use, and distribution must be kept distinct even when all matter to the decision. Calling tax revenue a transfer does not show that the public activity it finances is valuable, or that the people who pay can comfortably bear the charge. It means that the payment itself is received by another party, whereas the foregone gains have no recipient. The bridge earlier in this volume required a further comparison of the service financed and the resources committed. Collection and enforcement can also use real resources, and a contest to acquire a protected profit can consume effort beyond the triangle; the later Scramble examines that possibility. Nor does adding the buyer's pounds to the seller's pounds establish equal effects on their well-being. A transfer from a poor household to a wealthy owner can matter greatly even if the measured total is unchanged. Deadweight loss isolates an allocative question within this larger reckoning. It cannot settle the larger question by declaring everything outside its own boundary unimportant.5
The lost surplus is an absence with a definite comparison behind it. The good was worth more than the full sacrifice required to supply it, yet a rule or incentive prevented the exchange; the difference is what failed to arise. To measure that absence responsibly is to specify the demand, the costs, the adjustment, and the alternative against which the actual outcome is judged. The resulting triangle may be large or small. It may coexist with a transfer that matters more to the people involved, or sit beside other costs that its construction does not contain. Its enduring contribution is the separation itself: the gain kept by a customer, the gain received by a seller or treasury, and the gain that no one enjoys are different things. The next plate, The Control, puts the seller's power at the centre of the question. It asks how that power can be measured and constrained, once capture and destruction have been distinguished.