Optimal Wage Bargains

Abstract

Recent attempts to provide a theoretical framework for macroeconomics have focused on the nature of wage-employment agreements between firms and workers. Workers are in general unable to purchase insurance against future income fluctuations on grounds of, for example, moral hazard, and consequently firms may find it profitable to offer workers a contract that provides them with partial insurance as well as employment compensation. The seminal papers are those of Azariadis (1975) and Baily (1974) (henceforth A-B) and a useful survey is provided by Azariadis (1981). In their model, hiring decisions must be made before uncertainty about product market conditions is resolved. If firms are risk-neutral and severance payments to the unemployed are disallowed, then the optimal contract entails a state-invariant real wage and the possibility of (ex post) involuntary unemployment. While this model provides important insights, it has been criticized on a number of grounds. First, Akerlof and Miyazaki (1980) and Pissarides (1981) have pointed out that unemployment can occur in the simple A-B model only when the marginal revenue product of labour falls below the reservation wage so that unemployment would also occur in a Walrasian model; and that, in general, the level of unemployment will be less than would be experienced in spot markets (although it does provide a clue to the involuntary nature of that unemployment). Second, the ad hoc exclusion of payments by the firm to the unemployed can be criticized. If these are allowed, then the optimal contract involves a topping-up of the reservation wage to provide complete income (rather than wage) stabilization and the ex post efficient level of employment will be sustained. These criticisms have prompted a number of authors to examine the implications of assuming asymmetric information about the state of nature. An early example of this literature is provided by Calvo and Phelps (1977), who note that, under an A-B contract, because there is a tendency to overemployment in bad states, there will be an incentive for firms to misrepresent the state of nature and lay off more workers than previously agreed, but that this incentive may be attenuated by making the wage schedule contingent on the level of employment that is assumed to be observed by workers. The properties of these contracts usually depend on the precise assumptions of the model. However, in some versions unemployment may be greater than would occur in the Walrasian model. Hart (1983), for instance, considers a model in which severance payments are allowed but firms as well as workers are risk-averse. Then the optimal contract will involve lower wages in bad states of nature in order to spread risk, but the only way this information can be conveyed to workers is by lowering the level of employment. The papers by Grossman and Hart (1981) and by Azariadis, Chari, Green and Kahn, and Grossman and Hart in the symposium in the Quarterly Journal of Economics