Quarterly Models of Wage Determination: Some New Efficient Estimates

The concept of stable Phillips curves and their economic implications have been the subject of a series of disputes in recent years. Since Phillips introduced his simple disequilibrium model,empirical research in the field of wage determination has indicated a wide spectrum of explanatory variables which expands and contracts almost randomly in published studies. Apart from this difficulty in specification, the intertemporal instability of estimated coefficients for Phillips curve variants is particularly disturbing. In addition, the theoretical bases for many of these variants are unclear, and several economists have participated in a neo-classical counter-attack on the existence and stability of Phillips curves from a theoretical viewpoint. The objectives of this paper are to examine the consequences of certain invalid statistical procedures which are employed in many of these studies and to provide some empirical evidence of their effect. In particular, we focus attention on the implications of the aggregate procedure which provides the basis for the use of a "four quarter overlapping change" representation of the dependent variable in quarterly studies of wage determination and, also, for the use of moving averages of explanatory variables.