The Trust

The farmer at the upper gate closes it when her turn is over. The water continues down the channel, and the farmer below receives what the village has agreed is his. An inspector could stand beside every gate; a bond could be demanded before every turn; a dispute could be carried to an authority after every suspected excess. Each arrangement would consume something the irrigation system was built to produce. Much of its ordinary working depends instead on an expectation: that the person beyond sight will keep the agreement. Trust enters precisely at that distance between an action required now and another person’s conduct, still unknown. The farmer who clears the channel before the others arrive advances labor against a promise. The rule specifies what is owed; the expectation that others will do their part helps make the first contribution possible.

Trust, reciprocity, and reputation perform different work within this arrangement. Trust is an expectation about another’s behavior. Reciprocity is a disposition or strategy for answering cooperation with cooperation, and a breach with some change of conduct. Reputation is the history, communicated or observed, through which others judge what to expect. In Elinor Ostrom’s account, these can reinforce one another: keeping a promise makes a reputation more useful, and confidence in reciprocal conduct encourages further cooperation.¹ The direction can also reverse. Once neighbors expect others to abandon the work, holding back becomes easier to defend, and each withdrawal supplies the next neighbor with a reason to withdraw. The channel can deteriorate while everyone waits for assurance that someone else will begin. Its physical repair then depends partly on repairing an expectation.

The value of that expectation can be isolated in a small bargain. Suppose a household advances ten units for a shared repair and receives a benefit worth fourteen if its counterpart performs; if the counterpart fails, the advance is lost. With risk-neutral valuation and a probability p of performance, the expected net return is 14p − 10. At a probability of nine-tenths it is 2.6; at three-fifths it is −1.6, and declining the bargain is preferable. Suppose a separately available assurance service guarantees performance for a real administrative cost of two. The assured return is then two. A reliable informal relationship can save that cost; an unreliable one makes the assurance valuable. The numbers expose the condition: confidence creates this saving when it is well founded. Raising someone’s belief while leaving actual performance unchanged can persuade the household to accept a losing bargain.²

Communication can help people create the performance on which such confidence rests. In experiments reported in 1992, Ostrom, James Walker, and Roy Gardner let groups of students discuss their use of a simulated common resource, while leaving individual investment decisions private. Repeated discussion raised and sustained joint returns in the settings studied. Participants worked out agreements and responded verbally when aggregate investment revealed departures; an external authority did not enforce their promises. A single opportunity to talk produced gains that subsequently weakened.³ The meetings did several things together: they helped participants find a joint strategy, say what each would contribute, and interpret what followed. Their effect therefore carries more content than the bare availability of speech. A conversation that never reaches an intelligible agreement leaves the parties uncertain about what counts as keeping it.

A reputation accumulates through occasions on which conduct can be judged, and those occasions differ in what they reveal. Ten promises kept when keeping them is easy may say less than one honored at a serious cost. A deliberate appropriation of a neighbor’s share may cast doubt on many future promises at once. This is one reason the making and breaking of trust can run at different speeds: the new event changes the meaning of the old record. Yet the inference depends on the event. A dry channel may reflect an upstream breach, a broken gate, or a drought. Punishing a neighbor for the weather converts uncertain evidence into a fresh injury. Repairing the relationship requires a way to distinguish these possibilities and to establish what future conduct would count as repair. Reputation needs usable information as well as memory.⁴

The institutions around the relationship can make that information cheaper and more credible. A record of turns makes a claim testable. A monitor accountable to the users gives the distant farmer a witness. A low-cost hearing lets an apparent breach be explained before it becomes a feud. Ostrom’s later synthesis places trust alongside rules fitted to the resource and arrangements for observing conduct.⁵ Such arrangements can extend cooperation beyond people who already know one another, because a newcomer can demonstrate compliance with an understood standard. They also make a community answerable for what it asks its members to trust. Loyal cooperation among insiders can preserve an unfair allocation or exclude an outsider with a legitimate claim. The capacity to keep an agreement and the merits of that agreement remain separate questions; a dependable arrangement may still need reform.

The social capital beneath the rules consists partly in these credible expectations, distributed among people who can act on them. Its return appears in work begun without a quarrel, water released without a guard, and disputes settled before they consume the harvest. Some of the means that support it can be paid for: records, meetings, inspection, and an accessible forum. The expectation itself has to remain warranted by what people do. A community that relies on reciprocal conduct must therefore decide how to respond when reciprocity fails. The next plate, The Sanction, takes up the cost of making that response credible, and the discipline required of those who would discipline their neighbors.