Forecasting US inflation : Phillips curve, New Keynesian Phillips curve, or something else?

Abstract

I utilize and compare several common inflation forecasting models, including traditional Phillips curve models and the New Keynesian Phillips curve, as well as several other time series models. I evaluate these models using RMSE over several forecast horizons, using three different measures of inflation: CPI inflation, PPI inflation, and GDP deflator inflation. I find that the theoretical Phillips curve models outperform other time series models, however, the performance is sensitive to the inflation measure used in estimation.