Abstract
When estimating the welfare gains from moving from 2% inflation to price stability, Feldstein assumes that the compensated demand curve for retirement consumption is linear. Lucas has argued, in the shoe leather costs literature, that money demand functions are best specified as log-linear rather than linear. And the implications for the welfare costs of inflation are consequently very large. This note argues that linearity of the demand curve for retirement consumption is a good approximation for plausible parameter values and for the changes in inflation that Feldstein is considering.