Nonlinearities in the U.S. wage Phillips curve

We examine the relationship between wage inflation and the unemployment rate in the U.S. economy for the 1964-2014 period by means of a three-regime threshold regression model. The estimated threshold parameters suggest that this relationship changes when the unemployment rate transitions between regimes defined by 5.61% and 7.63%. During mild recessions and their subsequent recoveries, the time-varying estimates of the model indicate a negative relationship between both variables, consistent with the implications of a wage Phillips curve (WPC) derived from the standard New Keynesian model with staggered wage setting in Galí (2011). However, we find that this relationship breaks down during deep recessions and their recovery periods, which explains the difference between wage inflation predicted by standard New Keynesian models and the observed low wage growth in the aftermath of the 'Great Recession'. This finding and the fact that statistical tests strongly favor our three-regime model suggest that linear and two-regime models are insufficient to account for all the variability in the relationship between wage inflation and unemployment.