Nominal and real wage cyclicality during the interwar period.

• Author: Silver, Stephen

1. Introduction

   Although there have been numerous attempts to study the cyclicality of real wages in the U.S., no consensus has formed as to the implications of the various findings for business cycle theory. This paper will demonstrate that separating the real wage into its nominal and price level components can shed new light on the interwar U.S. labor market. More specifically, we will show that an evolution from a relatively laissez-faire labor market during the 1920s, to a more highly regulated labor market during the 1930s, may have resulted in a more countercyclical real wage rate.

   If we could be certain that equilibrium business cycle models provided the correct specification of the labor market then interpretation of real wage cyclicality would be relatively straightforward; during periods dominated by labor demand shocks real wages would move procyclically and during periods dominated by labor supply shocks real wages would move countercyclically. In a disequilibrium model context, however, interpretation of real wage cyclicality is much more problematic. If nominal wages are sticky, then unanticipated price level changes can "induce" countercyclical real wage movements that mimic the real wage movements generated by shifts in the supply of labor. Perhaps the most famous (and controversial) example occurred during the first three years of the Great Depression when a massive decline in hours worked and production was associated with a sharp increase in the aggregate real wage rate.

   [ILLUSTRATION OMITTED]

   Even greater complexity results from governmental attempts to influence the aggregate nominal wage. Policies that attempt to artificially change the nominal wage rate can result in either procyclical or countercyclical real wages depending on whether the direct impact of changes in the price of labor on the quantity of labor demanded outweighs the impact of wage changes on aggregate demand, and output. Separating the aggregate real wage rate into its nominal and price level components will allow us to discriminate between these two hypotheses.

   If labor markets are correctly characterized by equilibrium models, then the price level and real wage are determined independently and separating the real wage into its nominal and price level components would provide no additional information. In an economy characterized by sticky nominal wages, however, bivariate studies of wage cyclicality essentially "throw away" useful information that can be captured in a trivariate analysis.

   After discussing the theoretical issues involved in evaluating nominal and real wage cyclicality, we present estimates of real and nominal wage cyclicality for the interwar period. Although our procedure does not allow us to discriminate among all of the various explanations of real wage cyclicality, the stylized facts generated in section III cast doubt on one recent interpretation of wage cyclicality during the Great Depression. Furthermore, our procedure for separating real wages into their nominal and price level components has important implications for comparing real wage cyclicality over time, and between countries.

2. Interpreting Real and Nominal Wage Cyclicality

   Previous research into real wage cyclicality has estimated bivariate relationships between real wages and either output or employment. Rather than look for evidence of causality between real wages and output, these studies have typically focused on whether or not observed real wage cyclicality is consistent with the predictions of existing business cycle theory. For instance, McCallum [19] argued that recent studies showing real wages to be procyclical favored sticky-price business cycle models over sticky-wage business cycle models. Procyclical real wages are also consistent with real business cycle models where productivity shocks generate shifts in the demand for labor.

   Hoehn [15] constructed a sticky-wage model in which aggregate demand shocks produce countercyclical real wage movements, while productivity shocks generate procyclical real wage movements. This model received empirical support from a study by Sumner and Silver [25] that found aggregate real wages move countercyclically during periods dominated by demand shocks and procyclically during periods dominated by supply shocks. Similarly, Leiderman [16] found evidence that monetary shocks generate countercyclical real wage movements.

   Recent papers by De Long and Summers [8] and Zarnowitz [29] asserted that wage and price flexibility contributed to the severity of the business cycle prior to World War II. De Long and Summers developed a model in which, because of the Mundell effect, falling wages and prices generate expectations of future deflation and therefore reduce output. They noted that the implementation of New Deal programs such as the National Industrial Recovery Act (NIRA) may have both reduced wage and price flexibility, and contributed to economic recovery.(1) Bernanke and Parkinson [4] speculated that higher real wages during the New Deal era may have raised output by increasing aggregate demand.

   In contrast to the preceding authors, Roos [23], and Weinstein [28], argued that the higher real wages generated by the NIRA retarded the recovery from the Great Depression. As noted in the introduction, if governmental attempts to raise aggregate wages did reduce output during the depression, then these programs would have been expected to make real wages move more countercyclically than otherwise.

   Because of the multiplicity of interpretations of real wage cyclicality it is difficult to test any one hypothesis against another. Evidence presented in section III confirms that real wages were, in fact, strongly countercyclical during the interwar period. This will simplify our analysis since it suggests that shifts in the demand for labor were a less important factor in explaining wage cyclicality than was the case during periods when real wages have moved procyclically.(2)

   There are at least three explanations for the strong countercylical tendency of real wages during the interwar period. First, aggregate demand shocks could generate (unanticipated) procyclical price level movements and induce countercyclical real wage movements if nominal wages were sticky. Second, shifts in the (equilibrium) supply of labor, perhaps due to worker preferences, would be expected to generate countercyclical real wages. Finally, governmental interventions into the labor market may have moved the actual wage rate (real and nominal) away from its equilibrium value and the resulting impact of changes in the price of labor on employment could produce the sort of countercyclical real wage movements hypothesized by Roos and Weinstein. This final explanation of real wage cyclicality is, of course, inconsistent with the hypothesis developed by De Long and Summers.

   If interwar real wage countercyclicality was caused by aggregate demand shocks, then one would expect a positive relationship between prices and output, but no necessary relationship between nominal wages and output. (As noted earlier there is no theoretical presumption that real wages should be either pro- or countercyclical.) In contrast to aggregate demand shocks, the second and third explanations discussed above (labor supply shifts and governmental interventions in the labor market) would be expected to generate countercyclical movements in the nominal wage rate.

   There are several implications of the preceding hypotheses for interwar real wage cyclicality. First, prices and output should have been positively correlated and that correlation should be stronger during periods dominated by aggregate demand shocks. Second, there would be no necessary correlation between nominal wages and output, but nominal wages would become more countercyclical during periods when government policies directly affected the aggregate wage rate or when there were autonomous shifts in labor supply. If we assume that there was no sudden shift in worker's preference for leisure between the 1920s and the 1930s, then a shift toward greater countercyclicality of nominal wages in the latter decade would provide evidence in favor of Roos's and Weinstein's view of the impact of governmental labor market interventions.

   In section III it will be shown that real wages were even more countercyclical during the 1930s than during the 1920s. With so many possible factors affecting real wage cyclicality, however, this finding is not, by itself, sufficient to reject the hypothesis that governmental interventions generated a procyclical bias to real wages. The trivariate approach to wage cyclicality employed in section III will allow us to examine whether New Deal policies made real wages even more countercyclical than otherwise, or whether they had a procyclical impact that partially offset the countercyclical pattern resulting from aggregate demand shocks.

3. Empirical Findings

   One of the most notable features of the Great Depression is the unevenness of the recovery. For instance, annual data show a steady increase in real GNP from 1933 to 1936, while monthly industrial production data suggest an extremely unstable economy. In this study we have relied solely on monthly data, since many of the wage and price shocks appear to have been of extremely short duration. The output series we used is industrial production as estimated by the Federal Reserve Board.(3) For our nominal wage series we used average hourly wages in manufacturing. (See the appendix for a description of data sources.) Although these series offer a somewhat...