A key argument in Caplin and Leahy (1997) states that the correlation between monetary shocks and output is falling in the variance of the money supply. We demonstrate that this conclusion depends on solving for the correlation in the nonstationary state of the model. In the stationary state, that correlation is initially rising.