Accidental Incentives: Speculation and Reputational Enforcement

If a copper mine shuts down in Chile the price of copper goes up, giving other producers an incentive to produce more copper, consumers an incentive to consume less. The objective of the individual producer or consumer is to improve his own welfare, not the functioning of the economy, but he is led, as Adam Smith long ago pointed out, as by an invisible hand to achieve a desirable objective that is no part of his intent.

Seen from the individual's point of view, the effect on the market is an accident—indeed, an undesirable one. Producing more makes prices go down, which is not what the producer wants to happen. But seen in a broader sense, the market is a system of feedbacks, signals, that give the individual actors the right message, make it in their interest to produce more copper or consume less when and only when doing so improves the overall outcome. The gain to the individual actor is a measure of the social gain from his action.

There are some cases which look very similar but are actually quite different—where the link between what it pays an actor to do and the benefit his action produces is in some sense an accident.

One example is speculation. A successful speculator buys things when they are cheap, sells when they are expensive, and so both makes a profit for himself and smooths out price movements. The latter effect can be a very large benefit to other people. A speculator who sees a food shortage coming well in advance and takes the opportunity to buy up grain early gives other and less well informed people an incentive to use less grain, to plant more of other food crops, and thus to alleviate what might otherwise be a serious famine.

Unlike the usual case, however, the profit to the speculator is not a measure of the benefit produced. To see that, consider a case where the speculator learns of the shortage only a short time before everyone else would have learned—short enough so that the increase in price due to his activity does not have any significant effect on other people's behavior. He can still make a lot of money—not because he has produced valuable information but because goods belong to him instead of someone else when their price goes up. One implication, pointed out long ago in a classic article by Jack Hirshleifer, is the possibility of inefficient speculation. A rational speculator might spend a million dollars acquiring information about future price movements whose social value is zero—his whole profit is coming at the expense of whomever would have held the goods when their price went up if he hadn't bought them first.

The point is illustrated by a famous law case, Laidlaw v. Organ. A tobacco trader in New Orleans somehow got advance word of the treaty that ended the war of 1812 and took the opportunity to buy a large quantity of tobacco at the low price that had resulted from the British blockade. When the seller discovered that the war was over he attempted to reneg on the contract. The court held that the contract was binding.

The same pattern occurs in at least one other important context. I buy a jacket from a department store that guarantees to refund the purchase price if I am not satisfied with the produce, and when I return it they refuse to give me back my money. It is not worth suing them, but it is worth telling my friends—and anyone else who will listen—how badly I have been treated. The result is that other people stop buying from the store. That is a good reason why stores should live up to their promises, even if they are not at risk of being sued if they do not.

This sort of reputational enforcement surely plays a large role in encouraging commercial honesty. But, just as in the case of speculation, the incentive that makes it work is not linked to the actual usefulness of the behavior. The reason my friends—and even enemies—stop buying from the store is not to punish it for mistreating me but to protect themselves from similar mistreatment.

To see why this matters, imagine a case where it is not immediately obvious which party to a dispute is at fault. In order for interested third parties to punish the right person, they have to know who it is. But they have little incentive to investigate the claims of each, since they have the easier alternative of no longer doing business with either. That "punishes" one of us for cheating, the other for reporting the cheat. Anticipating that I realize that, having been cheated, I am better off saying nothing. At which point the mechanism for keeping firms honest stops working.

One conclusion is that reputational enforcement works only in contexts where it is cheap and easy for interested third parties to discover who was at fault; elsewhere I have argued that one function of arbitration is to lower the cost to third parties of doing so. My point here is a more general one—that it is worth distinguishing between those feedback mechanisms that work because the incentive to act measures the value of acting, and those that work, as it were, by accident.