A MULTIPLE BREAK PANEL APPROACH TO ESTIMATING UNITED STATES PHILLIPS CURVES

Phillips curves have often been estimated without due attention to the underlying time series properties of the data. In particular, the consequences of inflation having discrete breaks in mean, for example caused by supply shocks and the corresponding responses of policymakers, have not been studied adequately. We show by means of simulations and a detailed empirical example based on United States data that not taking account of breaks may lead to spuriously upwardly biased estimates of the dynamic inflation terms of the Phillips curve. We suggest a method to account for the breaks in mean and obtain meaningful and unbiased estimates of the short- and long-run Phillips curves in the United States and contrast our results with those derived from more traditional approaches, most recently undertaken by Cogley and Sbordone (2008).