Estimating the Cross-sectional Distribution of Price Stickiness from Aggregate Data

We estimate sticky-price models for the U.S. economy in which the degree of price stickiness is allowed to vary across sectors. Perhaps surprisingly, we use only aggregate data on nominal and real output. In our models, identification of the cross-sectional distribution of price stickiness is made possible by the fact that different sectors are relatively more important in determining the response of aggregate variables to shocks at different frequencies. We find that the distribution of price stickiness inferred from aggregate data is strikingly similar to the distribution obtained from the recent empirical literature on microeconomic aspects of price setting in the U.S. economy. We also employ a Bayesian approach to combine time-series data on aggregate nominal and real output with such microeconomic information. Our results show that heterogeneity in price stickiness is of critical importance for understanding the joint dynamics of output and prices. Moreover, allowing for enough heterogeneity - in particular for prices in some sectors to last beyond one year - is crucial to avoid producing estimates that imply too little average nominal rigidity at the expense of too much real rigidity, relative to the specification of the model favored by the data.