Uncovering the Costs of High Inflation

Inflation gives rise to inefficient price dispersion in New Keynesian models, yet empirical analyses suggest that such costs are small (Nakamura et al., 2018). This paper argues they are significantly larger. We extend the canonical sticky-price framework with a “price-research” friction - the costly effort firms exert to measure their idiosyncratic marginal cost before resetting prices, in the spirit of Reis (2006); Caballero (1989). Under low inflation, firms invest heavily in price-research and reset prices accurately. Under high inflation, markups erode rapidly, shifting resources toward price-adjustment and away from research: prices are changed more frequently but with less precise information, generating a less efficient distribution of relative prices and larger welfare losses. We show that the second moment of inflation-adjusted price changes provides an upper bound on the volatility of idiosyncratic shocks, with the gap identifying the magnitude of the information friction (Proposition 1). We calibrate the model to a granular micro-price dataset from Turkey covering 2019–2024, which spans a well-identified transition from moderate to high inflation following the breakdown of the Turkish monetary policy framework in 2021. The model fits the elasticity of price-change frequency to inflation substantially better than a standard sticky-price model.

Joint work with Ken Miyahara and Alberto Cavallo