This research project asks fundamental questions about whether and, if so how, the returns to financial intermediation should be taxed. This has proved to be a difficult area of economic theory, in which economists have so far come to different conclusions. Part of the confusion appears to stem from there being two distinct types of financial intermediation services: savings intermediation, and payment services. Our approach distinguishes these two in a theoretical dynamic general equilibrium setting. The results indicate that, in general, there should be a tax on payment services. But when firms differ in the cost of intermediation services, then the tax on savings intermediation should be zero. Both of these results can be understood as applications of the well-known Diamond-Mirrlees production efficiency theorem.