Real Wages over Short- and Long-Term Business Cycles: A Time-Series Evidence

This paper examines the correlation between aggregate real wages and real activity over various horizons of business cycles. Using a frequency band ï¬ ltering method, a business cycle indicator variable - the ï¬ rst diflerence of unemployment - is decomposed into high and low frequency components to represent short- and long-term business cycle movements, respectively. These components are used in real wage regressions to examine the characteristics of the correlation over various business cycle horizons. The main empirical ï¬ ndings of the study can be summarized as follows. 1) The real wage-real activity correlation has weakened signiï¬ cantly since 1980. 2) For the sample period 65:1 - 79:IV, real wages vary procyclically during short-term business cycles, but are countercyclical over the long-term if the cutofl period is longer than ï¬ ve years. ii) Short-term movements of the cycle indicator induce asymmetric real wage movements. Positive movements of short-term unemployment have much stronger negative correlations with real wages than negative movements before I980 but the converse holds aï¬ er I980. This suggests that real wages were more downwardly flexible and upwardly rigid before 1980 but became more upwardly flexible and downwardly rigid. 3) Supply driven cycle movements are more likely to induce countercyclical real wages than demand driven cycle movements before 1980 but the opposite is true aï¬ er I980.