Too Big to Fail

There's an excellent article in this morning's New York Times which begins:

"The financial crisis is forcing regulators to encourage the creation of bigger, more interconnected institutions. In the short term, this may serve a useful purpose by allowing healthier, well-capitalized banks like Wells Fargo, Bank of America and JPMorgan Chase to shore up weaker ones.

"But it also presents a serious threat to the financial system by fostering financial behemoths that are, to use Federal Reserve Chairman Ben S. Bernanke's euphemism, 'systemically critical.' Policy makers need to start thinking about how to downsize institutions that are becoming 'too big to fail' before the situation comes to that."

"Systemically critical" is a good phrase even though it sounds like jargon. Those of us who have built complex interactive systems know that nodes which are significantly large compared to the network as a whole pose an outsized risk. The Internet is a triumph of decentralized relatively small nodes, none of which is "too big to fail". There are always nodes failing on the Internet; this is the normal state. A brilliantly simple architecture allows the network as a whole to remain functioning despite constant failure of nodes. Because the nodes themselves don't have to be made failure-proof (which is impossible anyway), the nodes are so cheap that redundant nodes are easily affordable.

The current crisis began with the failure of mammoth nodes – blaming it on improvident home-buyers is simply absurd. These mammoth nodes – especially Fannie Mae and Freddie Mac - accumulated risk in a way that would have been impossible in a more decentralized time. Some community banks would certainly have been imprudent even if they'd had to keep the mortgages they wrote on their own books (remember the S&L crisis); but the system could much more easily have tolerated the failure of a few of these small nodes AND their failure would have encouraged others to be more prudent.

Until the Riegle-Neal Interstate Banking and Branching Efficiency Act of 1994 branch banking across state lines was forbidden in the US; that was a significant barrier to horizontal concentration. Bank holding companies were not allowed to own non-financial institutions until the passage of the Gramm-Leach-Bliley Act in 1999, a barrier to vertical concentration. Both acts were passed not only because of huge lobbying efforts by financial institutions which wanted to grow but also because there was a perceived (and quite possibly real) need to allow US banks to grow large enough to compete with less-constrained foreign entities.

With the limits gone, the monster banks grew albeit with some regulation, especially around their FDIC-insured deposits. However, this regulation did not include institutions like Merrill Lynch, Morgan Stanley or Lehman Brothers which were non-bank holding companies and could neither own banks nor offer FDIC insurance to "depositors". These institutions had the advantage of light regulation but the disadvantage that competitors like Bank of America and Citicorp could offer a full range of bank and non-bank services and had a base of insured deposits which wouldn't flee in a crisis. In recent months all of these either failed (Lehman), merged into bank holding companies (Merrill), or converted and became "banks" subject to regulation (Morgan Stanley). But we got even more concentration.

There's no question anymore that institutions which are too big to fail are also too big to leave unregulated. There is significant question whether any degree of regulation will be sufficient to prevent failure. Nodes fail for unexpected reasons – usually not the ones you're watching for. Fannie Mae and Freddie Mac were regulated; that didn't help much. The problem of adequate regulation grows even worse when government owns a stake in what it regulates and, of course, that's exactly what's happening with the banks.

Perhaps, immediately after the immediate crisis we need a form of antitrust which limits the size of financial institutions. Part of this might be accomplished by reducing the amount of federal deposit insurance on each account (this was just raised) to force investors in search of the safety of such insurance to spread their wealth among banks. Maybe the total amount of insured deposits any institution can offer should be limited; this step would eliminate the risk of having institutions big enough to bring down the FDIC itself. In a global economy, such limits will be very hard without similar steps being taken worldwide. But it's a better problem for world financial leaders to work on than weekly rounds of "coordinated" bailouts.

Nothing should be too big to fail because nothing can be made failure-safe.

Comments

[ps]

Simon,
As I understand network theory the degree of interconnectedness makes the network more robust as there are more pathways available to route around the failed node. LB and BS not only had a high degreee of interconnectedness but also had a huge number of transactions. If we view number of connections as the measure of interconnectedness and dollar volume of transactions as the measure of size of the node, then the issue was indeed size not interconnectedness. And I recognize that your point was only a quibble.

I agree that the highly optimized system we had was a necessary condition to the "meltdown" (and I also notice that on my screen Taleb's Black Swan is immediately to the right and probably offers one of the best reads to understand what is going on in this market).

I disagree with your solution of restricting/banning 2nd order+ derivatives on the basis that even granting the government the power to regulate trade between large and reputedly expert parties is too much of a step down Hayek's "Road to Serfdom". Clearly this trade was ill advised and the parties were not nearly as smart as so many thought they were. Unfortunately the government is no smarter and granting them the power to decide who can do what is an even greater mistake. Let such large parties trade and negotiate the degree of transparency regarding those assets/derviatives they trade. Then, if they have made huge errors let them fail. I know its too late now since some have already been bailed out, but lets not compound that error by granting more power to the government. We need to grow up and act like big boys and girls. If not somebody else will.

[Patrizia Broghammer]

The point is not too big to fail, the point is too big for a democratic country.
Democratic means that the power IS NOT in one or a few, the power belongs to all.
But the economy needs big groups, because big means more power, more opportunity.
The BIG of a really democratic country is FEDERALISM.
Many small that make a big.
If one of them fails the system won't fail.
And it is a system easy to regulate, because it regulates itself.
One looks at the others.
Do you remember: Divide et Impera.
That is the best way to rule a huge Empire.
But the same rules apply also to a modern society.
Where nobody is too big, and everybody can be big.
Balance is the secret of peace and of a healthy economy.

[Bill Seitz]

Shades of early-depression FDR policies that led to the RooseveltRecession of 1937-1939.

[andy abraham]

We have been speaking on myinvestorsplace.com... in regard to Too big to Fail... Is the US Fed too big to fail?

The Fed transferred the debt of banks on to its shoulders... what do you think??/

Please let us hear your opinion..

Thanks

Andy
www.myinvestorsplace.com

[Simon Cast]

Slight quibble. I wouldn't worry about size so much as interconnectedness within the system and the nodes of the system. In the case of Lehman Brothers and Bear Sterns it wasn't so much being to big as being to interconnected which amplified the risk in the system.

Basically, we ended up with a highly optimised system that was highly sensitive input. To much input and the response quickly becomes non-linear (or chaotic). Which is what we are seeing.

Our aim should be for creating a robust system that can be driven by shocks (i.e. large credit loss) without shifting into non-linear response. I don't think regulation that has been proposed (salary risk effecting capital ratios etc.) will make much of a difference for that. Rather we need to look at measures such as restricting/banning 2nd order and higher derivatives.