Many business cycle models use a flat short-run Phillips curve, due to time-dependent pricing and strategic complementarities, to explain fluctuations in real output. But, in doing so, these models predict unrealistically high persistence and stability of US inflation in recent decades. We calculate “reset price inflation”—based on new prices chosen by the subsample of price changers—to dissect this discrepancy. We find that the models generate too much persistence and stability both in reset price inflation and in the way reset price inflation is converted into actual inflation. Our findings present a challenge to existing explanations for business cycles. (JEL E31, E52)