But leaving aside function-based restrictions, how would a regulator determine how big is too big -- literally, from the standpoint of creating systemic risk? Peter Boone and Simon Johnson, posting on The Baseline Scenario, provide a legislative update, an analogy, a guideline, and a recommendation:
The Kanjorski amendment recognizes that the systemic and societal danger posed by banks can be hard to recognize, and it proposes a number of potential objective criteria that could be used by the Financial Services Oversight Council (to be created by legislation in progress) to determine when banks need to be broken up, including the “scope, scale, exposure, leverage, interconnectedness of financial activities, as well as size of the financial company.”
The Kanjorski amendment does not impose a hard size cap on banks, but lawmakers in the House are discussing amendments that would do so.
There is, of course, 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.
So to us, 2 percent of G.D.P. seems about right. This would mean every bank in our country would have no more than about $300 billion of liabilities.
Though this sounds like a common sense proposal (leaving aside the right cap size), I wouldn't pretend to be qualified to judge. But the clarity of exposition here is admirable.