A Test of the "Expectations Hypothesis" Using Directly Observed Wage and Price Expectations

I ECENT research directed at. improving the theoretical underpinning of the Phillips curve relationship, has given rise to several new models of wage determination. One theory receiving considerable attention is the "expectations hypothesis," which suggests that the rate of change of money wages depends upon expectations of the future rate of change of prices and/or money wages, as well as the unemployment rate.' The absence of money illusion implies that these expectational variables should en'ter the wage equation with a coefficient equal to unity. This in turn would imply the absence of a long run money wage-unernployment trade-off, or equivalently, the existence of a long run, vertical Phillips curve that passes through the "natural" rate of unemployment. Unfortunately, direct observations on expectations are not widely available in which case a proxy variable needs to be substittuted for the expectations variable. In the literature, the typical solution to this problem is to assume that expectations of price and/or money wage changes are generated by a distributed lag on past values of these variables. This procedure, however, has important well-known limitations. In this paper, direct data on price and wage expectations are introduced. This is done with a two-fold objective. First, the expectations data are used directly as explanatory variables in wage equations to test the expectations hypothesis. In view of the difficulties of proxy expectational variables, the use of such direct expectational data are important in an empirical investigation of the theory. Secondly, an attempt is made to explain the wage expectations series themselves.2