Aggregate Production Functions and Productivity Measurement: A New Look

Many economists currently take a somewhat jaundiced view of the estimation of aggregate production functions. Three problems seem particularly troublesome; the unobservables problem, especially with regard to utilization, the aggregation problem and the simultaneous equation problems. This paper presents theoretical arguments and empirical evidence from British manufacturing for the view that the first of these is the most serious with important dimensions in the measurement of capital and output as well as that of utilization. New light is shed on two classic questions. One was first raised by Feldstein (1967) who observed in a cross-section context that the elasticity of output w. r. t. average observed hours of work significantly exceeded the elasticity w. r. t. employment. Craine (1973) observed a similar result for time-series data. The other question is one with which most researchers on productivity have struggled: how to correct productivity for cyclical variations in the utlization of inputs. A novel answer based on the use of overtime hours data is found to give excellent empirical results.