Trend Inflation and Evolving Inflation Dynamics: A Bayesian GMM Analysis of the Generalized New Keynesian Phillips Curve

Inflation dynamics in the U.S. and Japan are investigated by estimating a "generalized" version of the Galí and Gertler (1999) New Keynesian Phillips curve (NKPC) with Bayesian GMM. This generalized NKPC (GNKPC) differs from the original only in that, in line with the micro evidence, each period some prices remain unchanged even under non-zero trend inflation. Yet the GNKPC has features that are significantly distinct from those of the NKPC. Model selection using quasi-marginal likelihood shows that the GNKPC empirically outperforms the NKPC in both the U.S. and Japan. Moreover, it explains U.S. inflation dynamics better than a constant-trend-inflation variant of the Cogley and Sbordone (2008) GNKPC. According to our selected GNKPC, when trend inflation fell after the Great Inflation of the 1970s in the U.S., the probability of no price change rose. Consequently, the GNKPC’s slope flattened and its inflation- inertia coefficient decreased. As for Japan, when trend inflation turned slightly negative after the late 1990s (until the early 2010s), the fraction of backward-looking price setters increased; therefore, the GNKPC’s inflation-inertia coefficient increased and its slope flattened.