A larger surplus is an achievement with a question still attached. There is more to gain from cooperation than there was before, more room for a bargain, perhaps more room to leave every person better off. Yet the number does not tell us which of those possibilities will be realized. It can rise while the benefits gather in hands already full; it can fall while an arrangement becomes one we have better reason to choose. The last task of this volume is to return the measure to its proper office: to show what is at stake, without letting the size of the stake pronounce upon its keeping.
That office is considerable. The surplus refuses the notion that every bargain merely passes an unchanged advantage from one party to another. It identifies a gain that price alone conceals, follows the division made by that price, and counts the opportunities lost when worthwhile trades are blocked. It can distinguish a payment received by someone from a cost that leaves no corresponding receipt. It can expose a seller who enriches himself while destroying more than he captures. These distinctions are indispensable to an argument about economic arrangements. To decline them because they do not answer every question would be to enter a dispute over the division of a gain without asking how much gain the proposed arrangement makes possible.
The trouble starts when a useful total is promoted into a complete social ordering. In the simple money-surplus setting, each dollar of gain enters the sum on the same terms, irrespective of the person receiving it. That is an intelligible accounting convention. If the sum is then used as the decisive rule for choosing between actual outcomes, it becomes an evaluative rule as well. It says that the destination of a dollar supplies no additional reason to prefer one result to another. People may defend that rule on grounds of practicality, liberty, the long-run effects of incentives, or an accompanying system of redistribution. But a defense must be made. The addition sign cannot make it on their behalf.
Consider two invented, feasible policies affecting two people. Relative to the same starting position, one yields complete money-equivalent gains of ninety dollars to a person with ample resources and ten to a person with little. The other yields fifty-five and thirty-five. Suppose the figures already deduct all real costs, valuations remain unchanged over the comparison, and no later redistribution is promised. The first has a surplus of a hundred; the second ninety. Both improve on doing nothing for both people, but neither improves on the other for both. A rule that maximizes the dollar total chooses the first. A rule that gives twice as much social weight to the poorer person's dollar chooses the second: its weighted score is a hundred and twenty-five against a hundred and ten. The arithmetic disagrees only because the questions disagree.
Neither rule has discovered its weights in the figures it adds. Equal dollar weights do not establish equal concern for people whose circumstances differ; unequal dollar weights require a reason for the particular inequality chosen. And a rule giving priority to the person worst off would need information about final circumstances, not merely the gains listed in this example. The purpose of making the choice explicit is not to prescribe the multiplier two, or any other ready-made coefficient. It is to prevent a preference about whose advantage counts from arriving disguised as the absence of any preference. A number becomes more useful when we can say exactly what changing it would mean.
The proposal to make the pie large first and divide it afterward is a serious answer to this difficulty. It asks production and exchange to find opportunities and a separate institution to direct the resulting means toward chosen ends. The second welfare theorem shows how an ideal separation can work under specified conditions, through the redistribution of purchasing power that supports a selected efficient allocation. But the theorem does not appoint the institution, determine its taxes, or ensure that the promised division occurs. Nor is the actual sequence always innocent. The distribution influences the demands to which production responds; the rules for sharing affect choices about effort and investment; and a transfer may have consequences beyond the sum that changes hands. The division is sometimes part of what determines the available gain.
It would be equally careless to turn that observation into a universal tax on fairness, as though every movement toward a better distribution must shrink the whole. A change can remove a wasteful privilege while broadening access; a transfer can make a previously unaffordable opportunity usable; an ill-designed restriction can sacrifice resources without helping the people whose name it invokes. Which description fits is a question about the particular institution and its alternatives. The example above assumed a real conflict between the unweighted total and the chosen distributional weights so that the judgment could be seen. It did not prove that such a conflict must exist. Before asking how much efficiency should be surrendered for equity, one should establish that the proposed alternatives require that surrender.
Even where a trade-off is real, the absence of a single perfect scale need not leave judgment helpless. Sen's account of social choice shows why additional information about people's circumstances can support comparisons that preference rankings alone cannot settle. We may have good reason to say that a particular deprivation is worse without claiming to measure every person's entire life on one exact ruler. We may agree on a comparison across several reasonable views while disagreeing about others. Knowledge can remain incomplete and still be decisive for the choice at hand. The demand for an exhaustive ranking of all possible societies would be an extravagant price to charge for recognizing an improvement in this one.
The same discipline changes how a surplus calculation should enter public reasoning. Show the benefits and costs, the people and periods to which they belong, and the consequences the number has not captured. Where the conclusion depends on a transfer, evaluate the transfer with the policy. Where it depends on a social weight, state the reason for the weight. Where the evidence supports only a range, ask whether the choice changes across that range. A decision that survives these disclosures has acquired a stronger basis, not a weaker appearance of certainty. A decision that does not survive them was leaning on something the total had concealed. This is how measurement can illuminate a judgment without pretending to replace it.
The earlier question of The Keeping returns here with its full force. A wage, a rent, a tax, a platform's charge: each settles some part of the contest over what cooperation makes possible. Economics can explain much about how those settlements arise, how alternatives would change them, and which opportunities an arrangement sacrifices. It can also reason carefully about distribution once the evaluative premises are supplied. What it cannot do is make those premises disappear by writing the result in dollars. The division is not rendered just because it is explicable, and a distributive argument does not become meaningless because a total cannot settle it.
The surplus can show us the room for a better settlement. The reasons for choosing that settlement must still be spoken.
Is its maximum the good? It is the maximum of a specified measure, an answer whose value depends on the question and whose limits belong beside it. The gain is real; avoidable waste is real; the claims of those who make and keep the gain are real. This volume has followed them far enough to show why none can be allowed to stand in for the others. The next volume, The Utility, turns toward the wanting beneath the willingness to pay — what it means to prefer, to be satisfied, and to compare. The account of the surplus ends where the judgment of a life begins. The argument remains open.