The missing disinflation puzzle: An investigation for the euro area

Many observers were surprised by the behavior of inflation since the financial crisis. According to commonly used empirical Phillips curve models, the increase in the unemployment rate should have implied much lower and even negative inflation rates (Ball and Mazumder, 2011; Murphy, 2014). This mismatch between the model predictions and the actual inflation outcome is referred to as the “missing disinflation” or “missing deflation” puzzle.

This paper focuses on the Phillips curve relationship between the unemployment rate and core inflation, i.e. excluding food and energy components, in the euro area. It addresses two questions: i) is there evidence of a missing disinflation puzzle, and ii) has the effect of the unemployment rate on inflation (the Phillips curve slope) weakened over time?

I replicate the missing disinflation puzzle for the euro area by using the single-equation Phillips curve model of Ball and Mazumder (2011) and Murphy (2014). The model is estimated on a 1999Q1-2007Q4 sample, coinciding with the enactment of the European Central Bank, and inflation is dynamically simulated over the post-2008 period. The simulations indicate that we should have seen strong deflation, while in reality inflation remained positive.

     Next, I show that the puzzle disappears when a vector autoregression (VAR) is used. Inflation forecasts from a simple VAR with inflation and the unemployment rate, conditional on the 1999-2007 sample and the actual post-2008 path of unemployment, are quite accurate and indicate that inflation was not abnormally high. Several factors behind the different results are discussed.

The third part concerns the time-variation in the Phillips curve slope. I construct a historical series of euro area core inflation which starts in 1976, and estimate the model of Chan et al. (2015) in order to identify different sources of time-variation. According to the results, the Phillips curve slope did not change much. However, both the persistence of the inflation gap and the volatility of the inflation equation error term did decline a lot over time.

Overall, these results imply that the relationship between euro area core inflation and the unemployment rate is weak, but has remained active during the crisis period. Moreover, the evidence that the Phillips curve slope flattened is not robust, but hinges importantly on the model assumptions.