State-Dependent or Time-Dependent Pricing: Does It Matter For Recent U.S. Inflation?

Inflation equals the product of two terms: the fraction of items with price changes (whose volatility figures prominently in state-dependent pricing models), and the average size of those price changes (the only source of fluctuations in time-dependent pricing models). The variance of inflation over time can be decomposed into contributions from the variance of each term and from their covariance. We use micro data collected by the U.S. Bureau of Labor Statistics to calculate this decomposition for consumer price inflation from February 1988 through April 2003. We find that 90% of the variance of monthly inflation stems solely from fluctuations in the average size of price changes. When we calibrate a prominent statedependent pricing model to match the empirical variance decomposition, we find that the model"s shock responses are very close to those in a standard time-dependent pricing model. We conclude that, at least for recent U.S. inflation, a realistic state-dependent pricing model has aggregate implications quite similar to time-dependent pricing models.