Capping bank size

By Felix Salmon
December 7, 2009
Simon Johnson joins in the chorus advocating a hard cap on bank size:

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

Simon Johnson joins in the chorus advocating a hard cap on bank size:

There is a strong precedent for capping the size of an individual bank: The United States already has a long-standing rule that no bank can have more than 10 percent of total national retail deposits. This limitation is not for antitrust reasons, as 10 percent is too low to have pricing power. Rather, its origins lie in early worries about what is now called “macroprudential regulation” or, more bluntly, “don’t put too many eggs in one basket.”

This cap was set at an arbitrary level — as part of the deal that relaxed most of the rules on interstate banking — and it worked well (until Bank of America received a waiver).

Probably the best way forward is to set a hard cap on bank liabilities as a percent of gross domestic product; this is the appropriate scale for thinking about potential bank failures and the cost they can impose on the economy. Of course, there are technical details to work out — including how the new risk-adjustment rules will be enacted and the precise way that derivatives positions will be regarded in terms of affecting size. But such a hard cap would the benchmark around which all the specifics can be worked out.

What is the right number: 1 percent, 2 percent, or 5 percent of G.D.P.? No one can say for sure, but it needs to be a number so small that we all agree any politician who cares about our future would have no qualm letting it fail, and when doing so have confidence that our entire financial system is not at risk as it fails.

A hard cap at 4 percent of G.D.P. seems about right for a bank with the most conservative possible portfolio. This would mean no bank in our country would have no more than about $500 billion of liabilities, even with a relatively low risk portfolio. On a risk-adjusted basis, most investment banks would face a cap around 2 percent of GDP.

This is very much in line with the number of $300 billion which I pulled out of thin air in March: the key is to get a number, any number, and to make sure that any banks of that size are small enough to fail. Arbitrary rules are fine, just so long as they’re stuck with: we’re not trying to optimize the regulatory structure, we’re just trying to make it so that too-big-to-fail isn’t a systemically-devastating problem.

TED, however, still isn’t happy:

Mr. Johnson fails to mention what I consider to be an equally if not more important point: what financial leverage we will allow these institutions to maintain. It will do us no good whatsoever to have a bunch of globally interconnected $500 billion financial intermediaries if they are all levered 15-, 20-, or 30-to-1. At least in the subsector known as investment banking, one can make a strong argument that it was leverage which killed Bear Stearns and Lehman Brothers, not simply their sheer size.

This is true, but it’s a slightly different issue. The good thing about Simon’s use of liabilities is that it automatically includes a lot of the simplest leverage: the more you borrow, by definition, the more that your liabilities increase. The cap should I think be on contingent liabilities, and should try to recognize the use of embedded leverage as much as possible, with a principles-based approach which doesn’t allow banks to try to do an end-run around regulations by using off-balance-sheet vehicles and the like.
TED’s right, however, that we still need a leverage cap over and above the hard cap on liabilities: just because banks are small enough to fail, that doesn’t mean that we want many such banks all failing at the same time because they’re all small and over-leveraged. And capital ratios should be higher for big banks, with liabilities over $100 billion, than they are for smaller banks with lower leverage.
The problem, of course, is that all the distressed M&A over the course of the crisis means that US banks are bigger than ever — and that there are no plans at all for shrinking them. How would you even start taking behemoths like JP Morgan Chase or Bank of America Merrill Lynch and turning them into small-enough-to-fail institutions with a mere half-trillion in liabilities? It’s getting there from here which is the hard bit, and I haven’t seen anything which looks like a workable roadmap yet.


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

$300 billion is close to $500 billion? you are a funny man.

Posted by q_is_too_short | Report as abusive

What’s worse for the financial system, a $100B bank that is severely under-capitalized, or a $500B that is responsibly funded? Rather than focus on the size of the bank, we should be concerned with how much exposure the FDIC, the Fed, and other banks have to any one bank. If we limit the amount a bank can borrow from the Fed as a percentage of their capital (and that limit should increase with size), and also limit the percentage of capital that any bank can loan to any one bank, the risk of systemic failure is reduced. We should also take into account downstream lending, where Bank A lends to Bank B and Bank C, and bank B also lends to bank C.

I just don’t think we’re better off with a bunch of small under-capitalized banks than a few giant well-managed ones (of course, we’ll have to find those well-managed giant banks, but let’s not rule it out).

Posted by OnTheTimes | Report as abusive

Yes both a liabilities cap and a risk-adjusted leverage cap are necessary, but the devil is in the details as risk-adjusted leverage can be measured in many ways.

Some suggestions from my recent paper:

Regulators should be given the powers necessary to regulate the leverage cycle to account for the problems of systemic risk, moral hazard, informational asymmetries, and heterogeneity of assets within a risk-weighting that plague current regulations. Policies such as harsher penalties for firms that do not comply with the minimum capital-to-asset ratio, minimum margin requirements, diversification of cash flow, and increase the compensation for government regulation jobs would all help to mitigate the negative effects of the leverage cycle.

Posted by BZ-RealEcon | Report as abusive

Here are my thoughts on the matter — please ignore the annoyed thoughts at the bottom about electing Fed governors. egulate-the-banks-and-other-financials/

Posted by DavidMerkel | Report as abusive