Over on the New York Times' Economix blog, Harvard economics professor Edward Glaeser has a post urging a tax on systemically risky institutions to solve the too big to fail problem. He believes the tax will reduce the incentive to take excessive risk as well as provide a fund to bail out these institutions if history repeats itself. I think attempting to avoid regulation sets him on the right track, but he misses something key: we should structure an economy where bailouts never occur. I would propose a similar, but different, alternative.

Let's start with Glaeser's explanation of why a tax is preferable to regulation:

For example, some places have tried to regulate their way out of traffic congestion by limiting access to central city roads based on license plate numbers -- for example, saying that cars with plates that end in 5 or 6 cannot drive on Wednesday. These policies do nothing to separate drivers who really need to get downtown from unmotivated motorists. The rule does wonders, though, to inefficiently encourage the buying of extra cars. A congestion charge is better. It allocates roads to the people who value them most and generates extra revenue.

Right. And I agree that taxes are often the easy alternative. But that's in a simple scenario. Financial institutions are extremely complex. Glaeser addresses this problem:

In principle, the appropriate systemic-risk tax on a financial institution is the product of two figures: the probability that this entity will receive a bailout, times the expected cost of that bailout. Obviously, that number is impossible to know without better understanding the rules concerning future bailouts.

Calculating such a tax may seem incredibly hard, but well-designed regulations require even more knowledge. Consider the proposal to ban certain firms from trading in derivatives. To determine whether this ban makes sense, we would need to know both the impact of such trading on expected bailout costs and the private benefits that such trading yields for each firm. To micromanage a firm's trading strategy, a regulator needs to know private benefits and social costs. Designing a sensible tax just requires knowing the costs to the taxpayer.

Right, but how exactly is it easy to determine the cost to taxpayers? Would it have been easy to know how much money AIG would have needed? Given the fact that bailouts were provided on several separate occasions, it must not have been. Each firm would have to be taxed based on the types of assets or securities it holds or has exposure to, not its size. You might have two firms that are exactly the same size with entirely different risk profiles. Figuring out accurate loss scenarios for a tax to cover might be easier than regulation, but it would still be extremely difficult.

But there's an even more important problem with the tax. Even if you could accurately devise one, it allows firms to receive bailouts, no matter how reprehensible or stupid their actions. As a result I would assert that these taxes would still cause firms to take excessive risk, given the prospect of bailout. Do we really want an economy where it's okay for firms to be bailed out because they might have paid enough taxes into a fund to keep them solvent after acting irrationally?

From a market perspective, bailouts are completely undesirable. The entire capitalist system is based on the concept that, if you make intelligent business decisions, then you will succeed. If you make stupid business decisions, then you will fail. Failure must be an option for capitalism to function. As a result, I have trouble supporting any measure that helps ensure firms' right to a bailout. So what's the alternative?

I would point to the plan recently noted by Treasury Secretary Timothy Geithner that a new resolution authority should force financial firms to devise plans for their own failure -- with an addendum. Firms should also pay upfront fees for whatever cost those plans for their resolution would cause the financial system to incur. This is a better alternative than taxes.

First, this provides a fee specifically tailored to each firm's complexity. Second, with a real prospect for failure, risk would be better kept in check. Instead of the tax providing for its bailout, a firm would pay a fee to facilitate its failure. Like a tax, this alternative also has the nice side effect of not stifling innovation like regulation would.

This really isn't that different from what the Federal Deposit Insurance Company (FDIC) does right now with banks. They collect fees for depository insurance -- but that insurance doesn't provide them with a bailout: it provides for their resolution. As a result, the new government agency in charge of collecting firms' failure plans should collect similar fees for those firms' potential resolutions.