### BibTeX

@MISC{_bypierre-olivier,

author = {},

title = {By Pierre-Olivier Gourinchas [UC Berkeley]},

year = {}

}

### OpenURL

### Abstract

Cecchetti and Debelle have written a stimulating paper. It makes a deceptively simple point: inflation persistence is lower than previously believed. The central reason is the significant evidence of at least one break in the mean of the inflation process for many industrialized countries over the last twenty to thirty years. This break in means biases upwards estimates of the persistence of inflation rates. The authors find that the median inflation persistence –as measured by the sum of AR coefficients in an AR(12) specification for monthly inflation- falls from 0.86 to 0.57 once we allow for a break in means. This is an important observation, for policymakers and academics alike. For policymakers, it indicates that the sacrifice ratio –the output cost of reducing inflation – may be lower than expected if inflation tends to revert naturally back to its long run mean. Assuming that mean inflation is low, this implies that rooting out bursts in inflation may be less costly or lengthy than previously estimated. For academics, this raises the question of the origins of this decline in mean inflation. For both, this should also lead to a reassessment of the goodness of fit of the monetary models commonly used by central banks to form projections of future inflation rates in response to various shocks and under a variety of scenarios for current and future monetary policy. As the authors are well aware, others have previously documented breaks in the mean of the inflation process for many countries (see the references to the literature in the paper). Their contribution goes beyond previous work in looking at various sub-components of the CPI and in documenting breaks in means both for the various subcomponents series and inflation expectations. I organize my comments around two issues. First, I will discuss the methodology adopted in the paper. Second, I will make some observations on how their results should be interpreted in the context of modern monetary policy theory. 1. The Methodology. The approach followed in the paper consists in estimating the autocorrelation of the inflation process in a univariate setting. The approach is most easily understood if we suppose that the inflation rate π t follows an AR(1) with an autocorrelation coefficient 0< ρ <1, a mean π and an innovation u t ( π

### Keyphrases

pierre-olivier gourinchas uc berkeley inflation process inflation rate mean inflation output cost significant evidence various shock many industrialized country monetary model central reason various sub-components modern monetary policy theory last twenty ar coefficient previous work future inflation rate mean bias inflation persistence many country simple point autocorrelation coefficient various subcomponents series inflation expectation monthly inflation median inflation persistence stimulating paper univariate setting important observation future monetary policy long run mean central bank