From Micro to Macro: A New Methodology to Discriminate Among Models

Cross-sectional variation in micro data can be used to empirically evaluate sufficient statistics for the response of aggregate variables to policy shocks of interest. We demonstrate an easy-to-use approach through a detailed example. We evaluate the sufficiency of micro pricing moments for the aggregate real effects of monetary policy shocks. Our analysis shows how a widely held notion about the kurtosis of price changes, as sufficient for summarizing the selection effect, turns out not to hold empirically. On theoretical grounds, we show how a small change in assumptions - removing random menu costs - can nonetheless reconcile the predictions of the existing theoretical literature with our empirical regularities.