On Tuesday Tim Geithner put the case for a tax on banks to the Senate Finance Committee. Congress Daily reports that he also seemed open to including the measure in the financial regulation bill, as urged by some Democrats. The tax he proposes is the so-called Financial Crisis Responsibility Fee, first pitched in January, which would raise $90 billion over ten years from financial institutions with assets of more than $50 billion. The principle is “simple and fair”, he said: “Banks, not the taxpayer, should pay for bank failures.”

Simple and fair. Let’s think about that. A new tax on banks–several new taxes on banks, in fact–might make good sense, but fairness and simplicity have little to do with it.

The costs of past or prospective bank failure are hard to measure. Direct losses under the TARP, if any, may end up small. Wider losses of the financial crisis (even if we confine ourselves to fiscal losses) are huge. What does the projected $90 billion have to do with either of those numbers? If the aim is to punish the banks, and this is a big part of the idea’s political appeal, fairness obliges you to worry about the incidence of the tax: who will actually pay it? Not a simple question. Will it be shareholders and employees of badly run institutions? Not those of Lehman or Bear Stearns, at any rate: too late for that. How fair is it to punish well-run banks–or to make credit more expensive for borrowers?

Ken Rogoff, discussing some recent IMF proposals, argues that taxing firms in proportion to their lending makes sense. I agree, but the point is not to punish, and getting the details right will not be simple. Read on–or read the IMF paper, which was leaked to the BBC, and especially Appendix 3.