New Evidence on Real Wage Cyclicality within Employer-employee Matches

In the most thorough study to date on wage cyclicality among job stayers, Devereux’s (2002) analysis of men in the Panel Study of Income Dynamics produced two puzzling findings: (1) the real wages of salaried workers are noncyclical, and (2) wage cyclicality among hourly workers differs between two alternative wage measures. We examine these puzzles with additional evidence from other sources. Devereux’s finding of noncyclical real wages among salaried job stayers is not replicated in the National Longitudinal Survey of Youth data. The NLSY data, however, do corroborate his finding of a discrepancy for hourly workers between the cyclicality of the two alternative wage measures. Evidence from the PSID Validation Study contradicts Devereux’s conjecture that the discrepancy might be due to a procyclical bias from measurement error in average hourly earnings. Evidence from the Bureau of Labor Statistics establishment survey supports his hypothesis that overtime work accounts for part (but not all) of the discrepancy. We conclude that job stayers’ real average hourly earnings are substantially procyclical and that an important portion of that procyclicality probably is due to compensation beyond base wages. New Evidence on Real Wage Cyclicality within Employer-Employee Matches Until a few years ago, macroeconomists firmly believed that real wages in the United States (and elsewhere) are nearly noncyclical, showing little systematic variation between recessions and expansions. This conventional wisdom was based on a long history of aggregate time series evidence. The stylized fact of weak wage cyclicality spawned numerous macroeconomic theories designed to explain how large cyclical swings in employment could be accompanied by relatively little wage variation. These theories included efficiency wage models, implicit contract models in which employers provide real wage insurance to workers, and insider-outsider models. More recently, a series of studies based on longitudinal microdata has demonstrated that real wages in the United States actually are quite procyclical, rising more rapidly in expansions than in recessions. As shown in Solon, Barsky, and Parker (1994), the true procyclicality of real wages is obscured in aggregate time series data because of a composition bias: the aggregate statistics are constructed in a way that gives more weight to low-skill workers during expansions than during recessions. In contrast, studies based on longitudinal microdata have been able to avoid composition bias by following the same workers over time. Some of these studies – such as Bils (1985) and Shin (1994) – have noted that the wage procyclicality revealed by longitudinal data is especially pronounced among workers who change employers. It is not surprising that workers who change employers in an 1 Blanchard and Fischer (1989:19), for example, declared, “The correlation between changes in real wages and changes in output or employment is usually slightly positive but often statistically insignificant,” and then they devoted much of their Chapters 7-9 to discussing theories designed to account for weak wage cyclicality. See Solon and Barsky (1989) for a detailed summary of the time series evidence and Solon, Barsky, and Parker (1994) for a discussion of the ways in which that evidence influenced macroeconomic theory. 2 Much of this literature is reviewed in Section IV of Solon, Barsky, and Parker (1992).