In trades where the same people deal with one another year after year -- diamond dealers, fish markets, cotton merchants -- agreements of very substantial value are concluded orally and are kept. Such written contracts as exist at all are brief and are not consulted. Economists approaching such a trade from outside customarily describe it as operating on trust, which names the phenomenon and explains nothing.
What enforces these agreements is the identical thing that enforces any agreement, which is the prospect of a cost to breaking it. The cost is not a judgment but exclusion. A dealer who fails to deliver on a handshake is known to have done so within days: the trade is small and talks constantly. The consequence is that others will not deal with him, or will deal only on terms that assume he might not perform. In a trade where a person's whole livelihood consists of a stream of future transactions with the same counterparties, that consequence is far more expensive than a court could make it, and it arrives sooner.
The conditions under which this works are specific and can be enumerated. The population must be sufficiently small that information about conduct circulates effortlessly. Membership must be difficult to acquire and difficult to resume once lost, or exclusion costs nothing. Transactions must be repeated and expected to continue, since a dealer on his last deal has no future to lose. And the trade must have some way of distinguishing a default from an honest misfortune, or the mechanism punishes bad luck and everyone hedges.
Where the conditions hold, the arrangement outperforms formal contracting on every measure that admits of comparison. Agreements are concluded faster, disputes are settled in days by an arbitrator drawn from the trade, and the outcomes are accepted. Where any condition fails, it collapses, and it collapses suddenly rather than gradually, because each participant's willingness to rely on the mechanism depends on his belief that others are relying on it too.
The awkward finding concerns what happens when formal contracting is introduced alongside. It does not simply add a supplementary layer of security. A trade that commences writing detailed contracts also commences reading them, and reading them changes what a default means. A failure that would have been handled as a matter for the trade becomes a matter for a lawyer; the information that used to circulate becomes evidence somebody is advised not to discuss; and the reputational mechanism loses the thing it runs on. The written contract does not fail. It succeeds, and in succeeding it displaces something that was working better. Nor is the process reversible on any timescale that matters, for the conditions the informal mechanism needed took a generation to establish, and the beliefs that sustained them do not return on request.
