U.S. Wage Growth and Nonlinearities: The Roles of Inflation and Unemployment
50 Pages Posted: 11 Sep 2016 Last revised: 18 Jun 2017
Date Written: June 17, 2017
Despite a low unemployment rate, wage growth in the U.S. was negligible during the 2013-2015 period. Conventional linear models of the relationship between wages and unemployment, the so- called wage Phillips curve (WPC), and previous models of the WPC that rely on regime-switching driven only by changes in unemployment, provide a poor fit in the aftermath of the Great Recession. Meanwhile, standard linear theoretical general equilibrium models are based on an assumption that economic agents take into account nominal wages relative to prices when making labor decisions, suggesting that there is a role for inflation in determining the empirical dynamics of the WPC. We employ a nonlinear empirical model to study how the relationship between U.S. wage growth and unemployment changes over the business cycle. In particular, we estimate a threshold vector autoregression with multiple threshold variables and multiple threshold parameters for each threshold variable for the 1965-2015 period. We find that the WPC changes according to the dynamics of both unemployment and inflation. Specifically, it changes as the unemployment rate transitions above or below the two estimated thresholds, defined by 5.03% and 7.77%. Simultaneously, it also evolves depending on whether inflation is above or below 0.38% relative to trend. The results show a strong negative relationship between wage growth and unemployment during periods of expansion when inflation is above its long-run trend. The relationship weakens, although remains negative, during periods of expansions with low inflation and during mild recessions. Our results indicate that the negligible wage growth observed during 2013-2015 was driven not only by labor market slack, as suggested by previous studies, but also by the low inflation environment.
Keywords: Wage Phillips Curve, Inflation, Unemployment, Threshold Vector Autoregression
JEL Classification: C32, E24, E52
Suggested Citation: Suggested Citation