International Evidence on the Efficacy of New-Keynesian Models of Inflation Persistence

Abstract

In this paper we take an agnostic view of the Phillips curve debate, and carry out an empirical investigation of the relative and absolute efficacy of Calvo sticky price (SP), sticky information (SI), and sticky price with indexation models (SPI), with emphasis on their ability to mimic inflationary dynamics. In particular, we look at evidence for a group of 13 OECD countries, and we consider three alternative measures of inflationary pressure, including the output gap, labor share, and unemployment. We find that the Calvo SP and the SI models essentially perform no better than a strawman constant inflation model, when used to explain inflation persistence. Indeed, virtually all inflationary dynamics end up being captured by the residuals of the estimated versions of these models. We find that SPI model is preferable because it captures the type of strong inflationary persistence that has in the past characterized the economies of the countries in our sample. However, two caveats to this conclusion are that improvement in performance is driven mostly by the time series part of the model (i.e. lagged inflation) and that the SPI model overemphasizes inflationary persistence. Thus, there appears to be room for improvement via either modified versions of the above models, or via development of new models, that better "track" inflation persistence.

Keywords: sticky price, sticky information, empirical distribution, model selection

JEL Classification: E12, E3, C32