The Cyclical Behavior of Industrial Labor Markets:
A Comparison of the Prewar and Postwar Eras
WITH JAMES L. POWELL
This paper compares the cyclical behavior of a number of industrial labor markets of the prewar (1923–39) and postwar (1954–82) eras. The methodology follows that of the traditional Burns and Mitchell (1946) business cycle analysis in at least two ways. First, the data employed are relatively disaggregated (we use monthly data at the two- or three-digit industry level). Second, we have not formulated or tested a specific structural model of labor markets during the cycle but instead concentrate on measuring qualitative features of the data. As did Burns and Mitchell, we see descriptive analysis of the data as a useful prelude to theorizing about business cycles. Thus, although the research reported here permits no direct structural inferences, it should be useful in restricting the class of structural models or hypotheses that may subsequently be considered.
The principal questions we study are also two in number. First, what are the means by which labor input is varied over the business cycle? We consider the intensity of utilization (as measured by gross labor productivity), hours of work per week, and number of workers employed. Both the timing and the relative magnitudes of the changes in these quantities over the cycle are examined. Second, what are the relationships over the cycle of output and labor input to measures of labor compensation? We look at the cyclical behavior of product wages and real weekly earnings as well as of real wages.
As might be expected, many of our findings are not novel; rather, they tend to support and perhaps refine existing perceptions of cyclical labor market behavior. However, we do reveal some interesting differences between the prewar and postwar periods in the relative use of layoffs and short
Reprinted from Robert J. Gordon, ed., The American Business Cycle, (Chicago: University of
Chicago Press, 1986). Copyright © 1986 by the National Bureau of Economic Research. All
We thank Frank Brechling, Ken Rogoff, Larry Summers, and our discussants for useful