Is the US Phillips Curve Stable? Evidence from Bayesian VARs

Inflation did not fall as much as many economists expected as the Great Recession hit the US economy. One explanation suggested for this phenomenon is that the Phillips curve has become flatter. In this paper we investigate the stability of the US Phillips curve, employing Bayesian VARs to quarterly data from 1990Q1 to 2017Q3. We estimate bivariate models for PCE inflation and the unemployment rate under a number of different assumptions concerning the dynamics and covariance matrix. Specifically, we assess the importance of time-varying parameters and stochastic volatility. Using new tools for model selection, we find support for both time-varying parameters and stochastic volatility. Interpreting the Phillips curve as the inflation equation of our Bayesian VAR, we conclude that the US Phillips curve has been unstable. Our results also indicate that the Phillips curve may have been somewhat flatter between 2005 and 2013 than in the decade preceding that period. However, while the dynamic relations of the model appear to be subject to time variation, we note that the effect of a shock to the unemployment rate on inflation is not fundamentally different over time. Finally, a conditional forecasting exercise suggests that as far as the models are concerned, inflation may not have been unexpectedly high around the Great Recession.