Employment, Hours, and Earnings in the Depression:
An Analysis of Eight Manufacturing Industries
SEISMOLOGISTS LEARN MORE FROM one large earthquake than from a dozen small tremors. On the same principle, the Great Depression of the 1930’s would appear to present an important opportunity for the study of the effects of business cycles on the labor market. In no other period for which we have data do output, labor input, and labor compensation exhibit such severe short-run variations.
Despite this apparent opportunity, modern econometric analyses of labor markets have typically made little use of pre-World War II data. There are some significant exceptions. In the class of papers that assume continuous labor market equilibrium, the best known example is by Robert Lucas and Leonard Rapping (1969). This influential piece was followed by Michael Darby (1976), who basically supported the Lucas-Rapping approach, and by Joseph Altonji and Orley Ashenfelter (1980) and Altonji (1982), who were critical of it. Among papers that allow for market disequilibrium, work by Harvey Rosen and Richard Quandt (1978) and Ashenfelter (1980) should be noted.1 However, none of the papers cited, I think it is fair to say, is the definitive study of 1930’s labor markets. They have in common at least two deficiencies in this regard.
First, all of this work has employed annual and highly aggregated data. This reflects the fact that none of the papers is focused on the 1930’s per se but include prewar data only as part of a longer-period study. Since none of the papers uses data from before 1929, any conclusion drawn about the prewar period is based on at most a dozen or so observations.
Second, the papers are limited in their capacity to rationalize the movements of a number of key labor market time-series. For example, none of them addresses the radical fluctuations in the length of the workweek which occurred during the depression, a phenomenon which the present research
Reprinted with permission from American Economic Review, vol. 76 (March 1986), 82–109.
I thank participants in workshops at Stanford, MIT, Harvard, Chicago, Carnegie-Mellon,
Rochester, Princeton, and Pennsylvania for comments on the first draft of this paper. Numerous
colleagues were also helpful. The Center for Economic Policy Research and the Hoover Institu-
tion provided support.
1 Martin N. Baily (1983) gives an interesting discussion of labor markets in the 1930’s, but
does not estimate a structural econometric model.