July 2014

Sun Mon Tue Wed Thu Fri Sat
    1 2 3 4 5
6 7 8 9 10 11 12
13 14 15 16 17 18 19
20 21 22 23 24 25 26
27 28 29 30 31    
Blog powered by Typepad

« Still Looking for a Job -- Scramble Is On | Main | Short Film on Life and Career of James M. Buchanan »


TrackBack URL for this entry:

Listed below are links to weblogs that reference The "Too Big To Fail" Fallacy:


Feed You can follow this conversation by subscribing to the comment feed for this post.

Dave says:

"Cut down the mighty oak trees so that we could all enjoy living from the perspective of mosses and lichens."

And the trees were all kept equal by hatchet, axe, and saw? ;)


"Welcome to the USSA, where everyone is systemic, but some are more systemic than others."

Dave, I made this same argument to someone recently about the decline in 401K's and the decline in consumption which, according to "some" economists, would cause a recession. Clearly "systemic risk" is in the eye of the beholder. Unfortunately the Feds and Goldman Sachs get to do most of the beholding.

The best alternative to too-big-to-fail is let-em-fail. Absent that, however, is it really so bad to get out the hatchet, axe, and saw? Relatively simple restrictions such as the old restriction on branch bank might be sufficient. Sure, that's not so great, but mightn't it be better than too-big-to-fail, which creates the rather immediate danger that a genuinely fascist economic model will be employed?

I don't mean to use "fascist" as a general term of abuse, which has been the word's principle use for a long time. I mean that that too-big-to-fail points us toward corporatism, which is dangerous economically and, especially, politically.

Umm, probably a time to remind everybody the history behind the concept: May to August 1931, the largest financial collapse in world history that turned a bad recession into the Great Depression (ultimately giving us Adolf Hitler in charge of Germany, and... ). It began with the failure of the Creditanstalt in Vienna (you know those darned Austrians!), followed by a wave of bank failures in Germany, followed by a wave in France and Britain, then in the US, then in Japan, and, well, by then it was August and the UK went off the gold standard and unemployment was really soaring through the roof in most country as the money supply collapsed with banks failing and no deposit insurance entities...

We all remember the history Barkley. The question is whether "too big to fail" is the best way to prevent a repeat performance. Other actions (or other non-actions) at the time might have prevented those problems without the need to generate a "too big to fail" doctrine.

Roger -

The restrictions on branch banking were *one reason why so many banks failed in the first place.* How is "too small to be sufficiently diversified" any better than "too big to fail"? Is it really better to have 10,000 small banks fail than several big ones?


"Is it really better to have 10,000 small banks fail than several big ones?"

-- yes, as you have 10,000 experiments versus a handful; it is unlikely that all would fail, or fail in exactly the same way, and so different actors can try different things and as some fail conditions would evolve. Some can fail, and be sold off to others, some management styles will be better, some of the banks would not have jumped in over their head as much, etc.

It would also have been unlikely that bailout could have been implemented, as the number of actors would be too large for negotitions not to get bogged down.


I think "arare litus" provides a good answer. (A useful labor that, IMHO.) You have more cognitive diversity with 10,000 banks. Also, I think it's harder to transform 10,000 or 1,000 or 100 failed institutions into a giant state enterprise than to thus transform 10 giant banks. Shouldn't we be happy to trade off some efficiency in exchange for insurance against coporatism?

No, actually. Because a system with 10,000 small banks is MUCH more prone to failures than one with a smaller number of large banks. Again, compare the US with Canada in the 1920s and 30s. Large Canadian banks closed branches but had essentially no failures, while the US lost almost half its banks. The inconvenience of going to the next town over to do my banking is a price well worth paying if the alternative is my bank failing.

If you have the right rules of the game, a system with a small number of large banks is so much less prone to failure that it's well worth the risk that concerns you.


I'm not seeing the real force in your comments. It seems like you have one historical comparison to support the claim that a more decentralized system is "MUCH more prone to failures." That seems like a lot of conclusion stacked on a relatively small base of experience.

That risk doesn't seem so bad anyway. In a fiat money regime, we can maintain MV even when banks fail. Anyway, part of the problem in the current crisis -- I mean in the current mess -- was that correlated risks were being treated as independent. That's an error more likely for institutions that are too big to fail (TBTF). As Hayek points out in LL&L vol. iii, when institutions are TBTF they are less risky investments and they gain a kind of artificial advantage. TBTF institutions gain political influence and are more likely to gain increasing special privileges and grow increasingly dangerous over time. The bigger they are the more sure they are of a bailout and the more risks they may freely take. TBTF is a such a big structural problem that we might at least examine rules to prevent financial firms form getting TBTF. I don't really see why that is such a non-starter for you.

"If you have the right rules of the game, a system with a small number of large banks is so much less prone to failure that it's well worth the risk that concerns you."

That is a big if - we live in a complex world, with complex regulations that can interact in unexpected ways. In particular, partial regulation can lead to serious problems - and as we are always discovering new ways of doing things we will always have regulation lag.

Canada did not have a central bank until 1935 - so one could argue that free banking is more stable, not that size is the determinant.

Alternatively, currently many suggest that the reason Canadian banks are not hurting as bad as US banks is due to them not being allowed to merge and grow in size. The opposite conclusion that many take away from the original depression - is there a window in optimal size?

We have little information to pin things down, it seems.

I agree with Roger Koppl regarding sparse experience base: a small sample size (Canada had 10 banks only in the depression) in one "experiment" is not sufficient information to make sharp claims, at least without a lot of additional evidence and theory to help. Some people pick stocks by making a small basket of companies that do very well, but even more do not. Only with iterated trials, or thought experiments, can one see problems with "small basket" investing.

Perhaps the sample size in Canada was small enough that sheer luck prevented failure. Perhaps the Fed messed up the US banks. Perhaps size is key. Perhaps the Scots dominated the banking industry in Canada, and they had a knack for banking. Perhaps Canadians are less likely to run on a bank.

"Because a system with 10,000 small banks is MUCH more prone to failures than one with a smaller number of large banks."

Contrast this to index funds - no one would make the claim in that case, yes you will have more failures in your fund, but that does not matter. If what you care about is systematic risk then diversity is good. Given the uncertainty in creating good rules (and that remain good), and given that it is hard to be sure what lessons to take home from the depression, it might be best to focus on systematic worse case risk and try to minimize that.

One thing that Steve's argument is based upon would be questioned by Schumpeter. So, to follow Roger's argument here, Schumpeter argued that a decentralized economy has the merits of a failure being localized rather than generalized. When the TBTFs collapse, the very fact of their centralized power and control over assets can lead to broad, systemic failure.


I think we're talking past each other. If by "TBTF" you include all the elements of politicization that you are talking about, then fine. My point was simply that, c.p., banking systems with a smaller number of large, more diversified banks are less prone to major problems than ones with a large number of small, unit, and therefore normally underdiversified, banks. The flip side of "too big to fail" is "too small not to fail."

There's no reason to use policy to make banks bigger, but there's equally no reason to use it to avoid bigness either.

That's a reassuring reply, Steve. I'm never entirely comfortable disagreeing with Horwitz on economics! I did intend to include all the politics. I did not mean to deny your basic point about diversification. I imagine you might make a point about the geographic component to diversification and I wouldn't deny it.

My question is whether we might accept all the problems of intervention and forcibly limit the growth of financial institutions. My sense is that we probably should take on those risks, which seem waaay lower than the risks of TBTF. I wish we could prevent Washington from creating monsters and then propping them up come what may. As far as I can tell, there is no realistic way to get that. Our next best solution, I suspect, is some sort of restrictive measure. Maybe that would be limits on branch banking. Maybe some other more clever measure should be employed. But it looks like some such measure should be employed as a makeshift against TBTF completely politicizing the economy. Is that a non-starter for you?

Now in defense of Steve: See pp. 150-155 of his Monetary Evolution, Free Banking, and Economic Order.

Second best problems aren't automatically non-starters for me Roger. I can see the argument for, in a politicized financial market, trying to do the sorts of things that you're talking about. My objections are at least two:

1. I think such limitations only *further* politicize the process such that the TBTF folks will still have the advantages in then responding with new interventions that favor them.

2. There are real costs to consumers of limiting branching and other benefits of large size. We've been there before and it was bad.

I'd much rather continue to work toward the first-best world in this case than to settle for what I think are really weak second-best solutions.

Steve: Would you like to summarize your argument from the pages in your book that I cited above? You'd be much better at it than I. I think it is quite good, and applies to this debate.


The relevant part, I think, is that the prohibitions on branch banking under the National Banking System (along with several other regulations) were a source of instability as they prevented banks from developing both diversified portfolios of their own and effective interbank clearing mechanisms. The lack of the latter certainly contributed to difficulties the system had in responding to exogenous changes in the absolute demand for money or relative demand for currency. One of the arguments for the Fed was that we needed to more centrally control bank reserves so as to be better able to respond to "panics" and the like. The fact that so many banks were "too small to succeed" was part of the problem.

One other comment: branch banking limitations, severe ones anyway, are to my knowledge a fairly unique US phenomenon. It's not just Canada that provides the evidence of the stability of systems with fewer, more diversified, large banks.

I can appreciate your reasoning, Steve, although I do come to a different conclusion. Where is the marginal return higher, investing in first best or in second best? It's not a complete tradeoff because you can't know what's second best if you don't know what's first best. We have an obligation to tell the truth and must therefore always be clear about what is first best. And so on. All in all though, it seems I see a higher marginal return of investing in second best. It's a point worth pursuing, but perhaps not here and now.

I don't think that ending branch banking would have had any impact on the recent crisis.

The "loan securitization" model is a way of protecting small, undiversified banks. They make loans, but rather than hold that portfoilio of assets, they sell them to Wall Street _investment banks_ that pool them, and then sell

Anyway, some of them go back to the small, undiversified banks. About 10% of the "aggregate" asset portfolio of the banking system is in these securities. Of course, it doesn't help when housing prices fall 30% and
they are mostly mortgage backed securities.

But, most of them were held elsewhere. Where? Well, investment banks, who don't have branch systems, sold commercial paper, including overnight commercial paper to fund portfolios of Collateralized Debt Obligations which were ultimately claims to pools of loans. These
were loans that were all made by a diverse
group of lenders.

All sorts of corporate treasurers held the commercial paper of investment banks, but also, money market mutual funds held them. These don't have branching systems either.

Citibank doesn't have a large branch system in the U.S. (they do have a lot over the world,) however,most of their liabilites aren't even deposits. Anyway, they got into trouble with the SIV's which was what the investment banks were doing. And as far as I know, these weren't being sold through local Citibank branches. They were all marketed on Wall Street.

Bank of America is a nationally branched commerical bank. And they got into big trouble by buying a failed investment bank, Merril Lynch. My understanding that it was a shotgun marriage.

It is true, however, that the three biggest banks are 45% of the banking system. Two of the
three, JP Morgan Chase and Citibank aren't really
nationally branched commerical banks. They
were huge money center banks when branching was
highly restricted.

There were giant banks on Wall Street when unit banking was the norm!

Bank of America (15% of the banking system) is a nationally branched bank. However, the next 20 or so are all heavily branched too. And while they, and all the little tiny banks hold securitized loans (and straght mortgages, 30%
of bank assets are in residential real estate one
way or another,) they were not at the center of
the crisis.

Anyway, most of the "too big to fail" institutions were investment banks. The branching restrictions didn't apply to them.
Oh, and branching restrictions didn't apply to AIG. It was an insurance company.

Koppl's "solution" is like thinking Glass Steagal is the answer.

I think Zingles has the right idea. Banking needs a system of expedited bankruptcy.

And, always remember, the problem is how all of
this impacts either the quantity of money or
the demand to hold money. I just reread Yeager's "Cash Balance Approach to Depression" from 1956. I guess I haven't learned too much
since. Make sure that monetary disequilibrium can be avoided, then, let them fail.

One final thing, all the commerical banks were under a regulatory regime where capital reguirementes were less for mortgages than other loans and even lower for CDOs with AAA ratings. That the banks are heavy in to real estate is no surprise.

That's a pretty strong answer, Bill. Thanks for that. I had not understood about differential capital requirements. Very interesting. Thanks for that bit, too.

I'm happy to give of branching restrictions per se. But mightn't there be some other contrivance that limits the growth of financial firms without introducing insane amounts of government discretion? I have grown downright phobic over TBTF and it's dangers. What is your take? Do you think it's hopeless and we'll never escape TBTF?

I wasn't suggesting that branch banking was somehow related to the current crisis. I was simply responding to Roger's argument by providing an example of why big can be better and that banks can be "too small to thrive."

The comments to this entry are closed.