¶ Are Phillips curves useful for forecasting inflation?

link to article at minneapolis fed

¶ BibTeX

@article{atkeson2001phillips,
  title={Are Phillips curves useful for forecasting inflation?},
  author={Atkeson, Andrew and Ohanian, Lee E and others},
  journal={Federal Reserve bank of Minneapolis quarterly review},
  volume={25},
  number={1},
  pages={2--11},
  year={2001},
  publisher={Citeseer}
}

¶ Abstract

This study evaluates the conventional wisdom that modern Phillips curve-based models are useful tools for forecasting inflation. These models are based on the non-accelerating inflation rate of unemployment (the NAIRU). The study compares the accuracy, over the last 15 years, of three sets of inflation forecasts from NAIRU models to the naive forecast that at any date inflation will be the same over the next year as it has been over the last year. The conventional wisdom is wrong; none of the NAIRU forecasts is more accurate than the naive forecast. The likelihood of accurately predicting a change in the inflation rate from these three forecasts is no better than the likelihood of accurately predicting a change based on a coin flip. The forecasts include those from a textbook NAIRU model, those from two models similar to Stock and Watson’s, and those produced by the Federal Reserve Board.

¶ Notes and Excerpts

How Terry described this to me: Imagine a monkey forecasting inflation. He just looks at current inflation and repeats what he sees. Is that monkey more useful than the Phillips curve? Turns out that maybe he is.

(Terry still discusses the Phillips curve in class because it’s one of those concepts where students should know what is, but he says the Minneapolis fed likes to trash it.)