Pricing Mistakes and the Non-Neutrality of Money
[Online Appendix]
Standard sticky-price models assume firms reset prices without error. Using UK and Argentine price microdata, I document that consecutive regular price changes are negatively correlated, more strongly when prices adjust more frequently, and with negligible correlation beyond one spell. A large class of standard models cannot jointly match these facts, but models with idiosyncratic pricing mistakes can. I ask how mistakes affect monetary policy’s real effects and their measurement using price microdata. First, even though firms perfectly observe monetary shocks, adding mistakes to a standard model can increase cumulative output effects by changing which firms adjust prices. Second, once mistakes are possible, these real effects are completely unidentified from standard micro moments (frequency and kurtosis of price changes) because mistakes distort them. I restore identification with a sufficient statistics result that incorporates regression coefficients measuring the documented serial correlation of price changes. Accounting for mistakes increases inferred real effects from the UK microdata by around 50%.