HOW LARGE IS THE OPTIMAL INFLATION TAX

I examine the title question in a competitive framework where the transaction demand for money is modeled using a cash-in-advance constraint and only distortionary taxes are possible. Empirical evidence and simulations are used to document the plausibility of the assumptions which produce the Friedman rule; these findings suggest that it is not a good benchmark. My results suggest that an optimal monetary policy calls for nominal interest rates that can exceed 6% per year.