The benefits of too big to fail

By Felix Salmon
October 5, 2009
Ryan Chittum applauds Gretchen Morgenson and Dean Baker for trying to calculate what they all refer to as the "costs" of having a too-big-to-fail "policy". But I don't like this lens at all.


" data-share-img="" data-share="twitter,facebook,linkedin,reddit,google" data-share-count="true">

Ryan Chittum applauds Gretchen Morgenson and Dean Baker for trying to calculate what they all refer to as the “costs” of having a too-big-to-fail “policy”. But I don’t like this lens at all.

For one thing, what Baker is totting up here aren’t costs at all, they’re benefits to the banks concerned. At no point is the key logical step spelled out: if a big bank benefits by being too big to fail to the tune of say $1 billion a year, then there must be an equal and opposite $1 billion a year cost to taxpayers. What’s more, Morgenson implies, that cost can be compared directly to other government spending:

If Mr. Baker is correct about the estimated size of the subsidy, the costs of too-big-to-fail are substantial when compared with other government programs. At $34.1 billion a year, the subsidy is more than twice the grant given under Temporary Assistance to Needy Families, a $16.5 billion program that helps recipients move from welfare to work. A $6.3 billion subsidy would be roughly what the government spent in 2008 on the Global Health and Child Survival program, an initiative aimed at preventing malaria, AIDS and tuberculosis.

What Morgenson’s talking about here aren’t contingent liabilities, they’re cash outlays. The government isn’t giving that level of cash to the banks; there’s not even an explicit government guarantee on the banks’ unsecured debt. Instead, there’s an implicit government guarantee, which, being implicit, can’t really be charged for. Instead, the government is seeking to increase large banks’ minimum capital ratios so as to reduce their profits by at least as much money as the extra profits that Baker is calculating.

Conceptually, then, I don’t like the idea of painting this benefit to the banks as a cost to the government, because it really isn’t. After all, if the government were forced to bail out any of these banks’ bondholders, the sums involved would be vastly greater than the numbers that Baker is throwing about. In that sense, the cost to the government is either zero, or it’s in the hundreds of billions of dollars. It’s not somewhere in between.

I also don’t like the idea of painting too-big-to-fail as a “policy”, as though it was a decision implemented by some technocrat somewhere down the line. It’s not. Pretty much every major economy in the history of capitalism has had too-big-to-fail banks. It’s a natural state, which governments can try to manage, but doing so carries its own costs. Morgenson quotes Baker:

“There is a subsidy here, and we either have to say we are going to break up the banks and get rid of the subsidy, or if we don’t do that, then we have to be confident that we have put in enough regulation to offset the subsidy.”

Breaking up too-big-to-fail banks is, in the best-case scenario, an extremely costly and disruptive thing to do. That doesn’t mean it shouldn’t be done, of course. But there is absolutely a cost-benefit tradeoff here. What’s more, it might not even be possible: if you break a trillion-dollar bank into two $300 billion parts and one $400 billion part, all three parts are still too big to fail, especially in a time of crisis. How big is small-enough-to-fail? And is it possible to grind America’s largest banks into chunks that small? These are non-trivial questions, especially when you’re dealing with broker-dealers which need large balance sheets just to be able to make liquid markets.

Baker’s alternative involves offsetting subsidy with regulation — how is that supposed to work? Can you calculate regulatory costs and debit them from too-big-to-fail benefits? Goldman Sachs currently has a tier-1 capital ratio around 16%; if the government increases its minimum capital ratio from 8% to 12%, what is the associated cost to Goldman? Yes, it loses a certain amount of option value — the ability, if times turn tough, to reduce its capital ratio sharply while still remaining in regulatory compliance. But how would you even begin to calculate that option value?

I think that Baker’s exercise is a useful one, especially when he comes to the conclusion that in the case of a bank like Capital One, all of its profits and then some might be a function of the implicit government guarantee. (Although, is Capital One really too big to fail? When WaMu and Lehman were not?) But let’s not pretend that we could, at a stroke, take the benefits that accrue to banks when they get too big to fail, and spend that money on something entirely different. We can’t.

More From Felix Salmon
Post Felix
The Piketty pessimist
The most expensive lottery ticket in the world
The problems of HFT, Joe Stiglitz edition
Private equity math, Nuveen edition
Five explanations for Greece’s bond yield
Comments
4 comments so far

“At no point is the key logical step spelled out: if a big bank benefits by being too big to fail to the tune of say $1 billion a year, then there must be an equal and opposite $1 billion a year cost to taxpayers.”

This need not be so. The benefits of FDIC insurance to the banks is huge and over the long run may cost the government little. Similarly, if the benefits to the firm of the subsidy cause the miss-allocation of capital, the dead-weight loss must be added. A billion dollars of benefits to the firm may cost the economy far more than a billion when the distortions are included.

Come on Felix,

just because we can’t get the number exactly right is that a reason not to try. Suppose the TBTF subsidy to Goldman is $1 billion and we raise their capital requirement by an amount that imposes an effective penalty of $1.2 billion. Is that worse than doing nothing and letting them effectively pocket the benefit, no questions asked?

I would agree that we don’t know exactly how big is too big to fail and that this would differ over time. Of course, that implies that the implicit subsidy would be lower also. While everyone may assume that BoA is too big too fail, we don’t know if that’s the case with the smallest of the large banks. That would suggest some sliding scale on capital requirements rather than a discrete break.

For some reason, the Obama administration has been pushing the discrete break view. That seems wrongheaded to me, unless the idea is that we will literally have an official policy that even a Capital One or other small big bank is TBTF.

Posted by Dean Baker | Report as abusive

Another policy response is possible. Make it explicit which bank debts need to be guaranteed for financial system soundness and require TBTF banks pay the government an insurance premium to have those debts guaranteed.

Re OneEyedMan: The subsidy is free insurance and the right way to measure the cost to the government is the ex ante correct premium not the ex post losses that actually occur.

Re Dean Baker: Someone should ask why debt is cheaper than equity for banks. For TBTF banks some of it is the cost of this insurance. For banks that are clearly not TBTF how much of the cost advantage of debt over equity is due to the belief that if the bank does fail, the FDIC will arrange something that makes even the unguaranteed lenders/depositors whole? Unless there is some other important reason for debt being cheaper than equity for banks, it is not feasible to impose an appropriate cost on banks by raising capital requirements.

Imposing these insurance premiums is not going to solve all the prudential regulatory problems. The level of the premium has to be set somehow. I do not see how a market mechanism could be used. Moreover, there does not appear to be enough publicly available information to intelligently price default risk for a bank. This means that the size of premium will have to be set by some government employees (possibly including congress). Thus the premium will probably be set too low. Moreover, it is less bad for the premium to be too low than for it to be too high.

This comment is too long already, so what else to do will have to wait for another time.

Posted by Thomas Spencer | Report as abusive

“The subsidy is free insurance and the right way to measure the cost to the government is the ex ante correct premium not the ex post losses that actually occur.”
FDIC insurance is not free. Banks pay premiums. It covers depositors against two eventualities. A bank run is costless to ensure against because the assets are worth more than liabilities if properly unwound. The government can just print the desired cash and then unwind the position. A bank failure from asset loss is also insured, but this insurance is costly to provide.

Post Your Comment

We welcome comments that advance the story through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can flag it to our editors by using the report abuse links. Views expressed in the comments do not represent those of Reuters. For more information on our comment policy, see http://blogs.reuters.com/fulldisclosure/2010/09/27/toward-a-more-thoughtful-conversation-on-stories/