A promise asks someone to act before its truth can be known. The supplier builds to an order; the buyer sends an advance; the worker gives an afternoon that cannot be recalled when the wage falls due. At the moment for repayment, the other party may find a different choice attractive. The promise has named what he should do while leaving intact what he would gain by refusing. A pledge puts something of his own within reach of that refusal. It changes the later comparison on which the earlier confidence depends. The party who could freely change his mind acquires a way to be believed by making a particular change of mind more costly.
The value of commitment lies in the cooperation it can make possible before anyone is tempted to depart. The Collected Mast begins with one party performing first and another deciding afterward whether to reciprocate. If refusal is profitable and anticipated, the first act may never occur, though both parties would prefer an exchange faithfully completed. Schelling treats the capacity to assume an effective promise as an economic power. A party who can expose himself to a dependable consequence can obtain reliance unavailable to one whose assurance leaves every option open. The consequence may come through legal enforcement, an asset already in custody, or a valuable opportunity that nonperformance would destroy. What matters is what can actually happen when the promise falls due.
The bond makes the comparison especially plain because its consequence is counted in the same units as the temptation. Retain the invented bargain from The Hostage. Faithful performance gives each party ten, net of its ordinary costs. Once one has performed, the other can gain eighteen by refusing its contribution. It has an extra eight to take. Suppose it has already lodged a bond of twelve, returned in full on compliance and forfeited on a correctly established refusal. Refusal now leaves six, while performance still gives ten. The bond has made the promised action preferable. A bond of eight gives only indifference; a smaller bond leaves the specified temptation in place. Its usefulness follows from the choice it changes, rather than from the solemnity of the gesture.
The institution controlling forfeiture is part of the pledge. If refusal triggers the twelve-unit loss only half the time, a risk-neutral promisor expects twelve from refusing: eighteen less six. The original protection fails. At a three-quarter probability of forfeiture, expected refusal yields nine and performance again wins. A large promise to pay from an empty purse offers a different safeguard from money already under dependable control. So does a nominal penalty whose trigger cannot be established. Where the proposed safeguard merely requires the same unavailable proof as the original promise, the difficulty has been moved into another clause. Credibility requires a consequence that survives the moment when carrying it out becomes inconvenient.
A pledge can support investment by protecting the party who must commit resources first. Williamson’s hostage model concerns a supplier whose production can become cheaper through an advance investment that loses value outside the particular exchange. An unprotected right of cancellation can make that investment unattractive or force the supplier to recover its cost through a higher price on the orders that survive. A properly designed cancellation payment changes both the buyer’s later choice and the supplier’s initial willingness to invest. In the model, matching the safeguard to the specific investment can preserve the valuable technology and the appropriate decision about taking delivery. The point is productive: a credible commitment can enlarge what is worth creating before the parties divide the return.
The holder of the pledge may acquire a temptation of his own. If declaring a breach releases twelve into his hands, he may profit from an accusation, an obstruction or an insistence on a technicality that makes the other party fail. Williamson examines this danger of contrived cancellation, together with uncertainty about the investment supposedly being protected. More security can intensify the incentive to seize it. Evidence, independent custody, a trustworthy procedure for release and a proportionate response therefore belong to the safeguard’s design. The question is no longer merely whether the promisor has enough at stake. It is whether those entrusted with the stake have reason to administer the condition faithfully.
The asset’s value to its recipient depends on the function the pledge is meant to perform. A payment intended to cover an investment must have value to the party who incurred that investment. An asset exposed solely to distinguish a reliable type of counterparty can serve another purpose, and may be safer if the recipient gains little by seizing it. Williamson separates these cases. His investment model values a transfer the supplier can actually use; his distinct screening illustration examines how a pledge valuable to its owner but unattractive to its recipient can reduce an expropriation hazard. A single maxim about worthless hostages would erase the difference. One must identify the loss being covered and the decision being altered before deciding what should be placed at risk.
A pledge gives a promise weight only through the choices that can put the pledged thing in danger.
The ability to finance a pledge can determine who gets to make an otherwise valuable promise. In the constructed bargain, tying up the twelve may cost three, leaving the faithful participant seven of the original ten. If the cost is eleven, participation yields minus one and the exchange is declined, even though the bond would deter refusal once posted. The incentive condition has survived while the participation condition has failed. Lack of collateral can therefore reflect lack of wealth rather than a plan to cheat. A guarantee, smaller stages of performance or an advance tied to verifiable progress may reduce the exposure that needs to be secured. Each alternative has its own cost; none follows simply from calling the prospective counterparty trustworthy.
The freedom surrendered must also be distinguished from a valuable freedom to adapt. A promise can become inappropriate because needs change, costs rise or the parties discover a better arrangement. A safeguard that punishes every departure indiscriminately can trap them in performance neither now wants. The cancellation rule in Williamson’s model is useful precisely because it changes the price of withdrawal while preserving a meaningful choice about delivery. In a more complicated relationship, the parties may need to agree who can release a pledge, what evidence excuses performance and how a proposed adjustment will be considered. Binding oneself can support cooperation without making immobility a virtue. The promise needs enough weight to be relied upon and enough intelligible structure to be revised when revision serves its purpose.
The pledge makes reliance possible by giving a later decision an earlier history. Something has been committed before the temptation arrives, and the other party can judge what that commitment will do. Its force depends on valuation, resources, evidence and control; its success is measured by the exchange it supports after those burdens are counted. The same logic can operate without a separate bond. A relationship may hold future opportunities whose loss would make today’s refusal dear. Then the thing pledged is the prospect of dealing again, and the institution that administers the loss is the relationship itself. The next plate follows that prospect into the word given between parties who expect a future together.