Structural Estimation and Evaluation of Calvo-Style Inflation Models

The authors structurally estimate and evaluate, for the U.S., the E.U., and Canada, various classes of recently-proposed Calvo-type models, using identification-robust methods. The models differ in their assumptions regarding price indexation (when firms cannot re-optimize their pices), in the way capital is used (homogenous or firm-specific), and in the elasicity of intermediate goods demand facing firms. Our approach is to obtain point and confidence-set structural parameter estimates, based on inverting identification-robust test statistics. Importantly, we maintain the focus on the structural aspect of the model, and formally impose the restrictions that map the theoretical model into the econometric one. In addition, we propose a test statistic that is invariant to the considered delay between the time firms re-optimize their prices and the time they implement these new prices. Results are as follows. For the U.S., we find no statistical support for the standard Calvo model. Instead, there is evidence in favour of a dynamic indexation model with firm-specific capital and an increasing elasicity of intermediate goods demand facing firms. For Canada, there is some support for a dynamic indexation Calvo model regardless of whether capital is assumed to be firm-specific or not, but only if no price implementation delays are present. For the E.U., the results are mixed. Overall, we also find that, in all cases, when firm-specific capital is assumed, results are almost identical whether adjustment costs are assumed to be zero or not. Second, outcomes are very different depending on the considered implementation delay. Third, except when allowing for uncertainty in the considered implementation delay, the uncertainty about the average frequency of price adjustment in the economy is dramatically large.