Guarantees and Risk-Sharing
The fact that a consumer is frequently uncertain about the quality of a product that he purchases, and is therefore also unsure of the extent to which it will render him the services he might expect of it, is one that is gaining increasing recognition. In an earlier paper [5] I examined in a simple framework the effects of changes in the uncertainty about a product's quality on the consumer's demand, and also touched briefly on the effect of a guarantee. In this paper I want to examine in more detail some of the economic issues posed by the existence of guarantees. There are several useful distinctions that can be drawn as a preliminary to more detailed study. One is a distinction between situations where buyers and sellers have equal access to information, and a situation where sellers have superior access. The first situation is exemplified by a market where trade is in a product whose quality is genuinely random, with the distribution known to both buyers and sellers. Thus if a certain fraction p of the cars from a given factory are generally known to be faulty, then a transaction in which a retailer sells a car to a buyer comes into this category: each knows that the chance of the car being faulty is p, but, because the car is unused, neither knows whether it is actually faulty. Contrast this with a situation where the first owner of the car is reselling it: now there is an important asymmetry, in that the seller is much better informed about the quality of the product than the buyer. This is the case with which Akerlof's very interesting analysis [1] is largely concerned, and which I have also discussed in [4]. My main concern here, however, is with a situation of equal information. Thinking casually about such a situation, it is clear that one can distinguish between the incentive effects and the risk-sharing effects of a guarantee. Incentive effects arise because the existence of a guarantee provides the producer with an incentive to improve the quality of his product, at least to the extent of reducing the chances of its falling below the guaranteed level. If the compensation in the event of failure is less than complete, then the consumer also has an incentive to maintain the product. For example, a used car guarantee, under which the buyer and seller will each pay half of any repair bills, provides both parties with incentives to minimize these bills. Of course, if the guarantee is valid for only a limited period of time, then there is the further effect of providing the buyer with an incentive to ensure that if there is to be a failure, it occurs early in the product's life. This may act in opposition to the other effect, and reduce his eagerness to maintain the product. In addition to creating the incentive effects mentioned, a guarantee also acts as a way of sharing the risk associated with uncertainty about the quality of a product: to be efficient in this sense, it will apportion this risk in accordance with the risk-aversion of the participants. My main concern here is with the risk-sharing aspects of guarantees. This is partly because these seem to be the most tractable aspects of the problem, but also stems from a belief that these are the most important aspects. Casual empiricism suggests that except in rather unique cases, a product is usually designed and produced before any attention is given to the choice of guarantee terms: these are then chosen as part of a marketing package. In such situations, the reliability of the product will clearly be independent of the