In this paper we empirically investigate the time- and state-dependent behavior of aggregate price setting. We implement a testing procedure by means of a nonparametric representation of the structural form New Keynesian Phillips curve. By means of the so-called functional coefficient regression we allow for potential dependence of the Calvo (1983) parameter on inflation and inflation uncertainty. Thus, we can test for state-dependence of the Calvo parameter in a straightforward way. To address residual heteroscedasticity in the inference process regarding functional dependence, we make use of the factor-based bootstrap. We confirm that the Calvo scheme is a rather restrictive model of aggregate price setting. Moreover, it is documented that a number of shortcomings of empirical NKPC model representations in explaining inflation data may be addressed by means of a state-dependent pricing rule. In particular, problems of insignificant or even implausibly negative estimates of the relation between inflation and marginal costs are considerably reduced in the framework of our more general NKPC specification.
State- vs. Time-Dependence Phillips Curve Functional Coefficients