This article investigates the relationship between inflation and output, in the data and in standard models. The article reports that empirical cross-country studies generally find a nonlinear, negative relationship between inflation and output, a relationship that standard models cannot come close to reproducing. The article demonstrates that the models' problem may be due to their standard narrow assumption that all money is held by the public for making transactions. When the models are adjusted to also assume that banks are required to hold money, the models do a much better job. The article concludes that researchers interested in studying the effects of monetary policy on growth should shift their attention away from printing money and toward the study of banking and financial regulations.