Cartel Behaviour and Adverse Selection

CARTELS ARE central to any general theory of market structure. For, given the prevailing notion of rationality based upon self-interest, the rational behaviour of a group of firms should be to form together into a cartel and maximize joint profits or, more generally, act in a manner efficient to the group. If other forms of conduct, or market structures compatible with such conduct, are to be rationalizable then an exploration of why firms fail to act in this way is central. Some constraints on firms can be imposed by a society's legal structure, broadly defined. If, as is common, collusive action by firms is deemed to be illegal, collusion may still occur but be hampered; for instance, side-payments made between firms to encourage members of a cartel to act in the group interest may be hard to disguise and in consequence may be ruled out. As well as these, what may be termed, external constraints, there are internal constraints which are part and parcel of a society's economic system. Here, one naturally thinks of the costs of communication which are often used as an argument against collusion. However, a more basic constraint is the informational structure of an economic system-in loose terms, what information about the economic system is known or can be found out by agents within the system. Assume that transfers of goods or income cannot take place between firms; then in a world of full information-all tastes, technologies and behaviour known to all-there is no need for direct communication between firms. Each firm can calculate what would be the result under costless communication and act accordingly; there is an incentive to do so because, as behaviour can be observed, firms can respond to any erring firm in exactly the same way as if there was direct communication. It is the lack of full information which makes communication between firms necessary or useful. A lack of full information can take different forms and have different implications. For present purposes, it is desirable to distinguish between moral hazard and adverse selection. Moral hazard exists when the behaviour of other agents is less than perfectly observable whereas adverse selection arises when it is the preferences of other agents which are unobservable. Following the seminal work by Stigler [I964], the extensive literature on the problems of implementing rules for cartels has tended to equate lack of full information with moral hazard.' The classic example of moral hazard occurs when the demand for a firm's product is stochastic and other firms can observe only the level of demand