What Makes a Bank Too Big to Fail?

David, one of my readers, writes in with a question:

I'm curious about what actually makes a bank too big to fail. My guess is that it is because they are counter-parties to many deals, but would something like a CDS exchange help facilitate "canceling out" these deals? If Citi, for instance, was allowed to go bankrupt and quickly liquidated, and their operations sold to solvent banks, what calamity would befall us? Would this be a reasonable alternative to nationalization in which the taxpayers don't have to shoulder the burden of bad assets?

The first thing worth saying here is that too big to fail (TBTF) has nothing to do with counterparty risk or credit default swaps or other such arcana of this particular financial crisis. Instead, it's all about the sheer size of a bank's balance sheet -- its assets and liabilities. And when a balance sheet gets up into the hundreds of billions of dollars, or even into the trillions of dollars, you get big problems on both sides when that bank runs into trouble.

On the asset side, you can't quickly liquidate hundreds of billions of dollars of assets in some kind of fire sale; the Lehman bankruptcy alone is going to take years to work through, since there simply isn't the global capacity to pick up those assets for cash. If another bank were to join Lehman in that process, it could clog up the financial system for the foreseeable future.

What's more, when you do try to liquidate a bank's assets, you set a market price for them. So if the bank has a large leveraged loan, say, which ends up getting sold for 40 cents on the dollar, then there will be a lot of pressure on all the other banks who own that loan to mark their asset down to 40 cents on the dollar. Which could create a whole new round of write-downs and insolvencies.

The liability side is even worse. Lehman had relatively little in the way of unsecured liabilities -- a hundred billion dollars or so -- but when those went to zero, that contributed directly to the Reserve Fund breaking the buck and October's panicked flight to quality. When a stock goes to zero, financial markets can cope, although there are always plenty of unhappy shareholders. If hundreds of billions of dollars of bondholders are wiped out, however, there are nearly always systemic consequences.

And that's not the worst of it. At commercial banks, most of the unsecured liabilities are likely to take the form of deposits rather than bonds. Many of those deposits are guaranteed by the government, which means that a bank failure means a massive direct loss to the taxpayer. And then there's the deposits which aren't guaranteed by the government -- everything from Granny's life savings to the accounts out of which your employer makes payroll every month.

TBTF, then, means that both on the asset and the liability side of the balance sheet, any kind of bankruptcy or liquidation would have devastating systemic effects, and therefore it can't be allowed to happen. There's also a hint in the term that no other bank is big enough to be able to take on the failed bank's balance sheet.

As for credit default swaps, they're derivatives, and therefore zero-sum games. Investors can use them to shunt bank-failure risk elsewhere in the system, but they can't eradicate it. CDS might not be part of the problem, when it comes to TBTF banks, but neither are they part of the solution.