Commentary Type

"Bad Banks" for Beginners

Simon Johnson (PIIE) and James Kwak (Yale Law School)

Post on Baseline Scenario


What is a bad bank? . . . No, I do not mean that kind of bad bank, with which we are all much too familiar. I mean the kind of "bad bank" that is being discussed as a possible solution to the problems in our banking sector.

In this sense, a bad bank is a bank that holds bad, or "toxic," assets, allowing some other bank to get rid of these assets and thereby become a "good bank."

To understand this, you need to understand what a bank balance sheet looks like. I have covered this elsewhere, but for now a simple example should do. Let us say that the Bank of Middle-Earth has $105 in assets (mortgages, commercial loans, cash, etc.), $95 in liabilities (deposit accounts, bonds issued, other financing), and therefore $10 in capital. The assets are things that have value and theoretically could be sold to raise cash; the liabilities are promises to pay money to other people; and the capital, or the difference between the two, is therefore the net amount of value that is "owned" by the common shareholders. Next assume that the assets fall into two categories: There are $60 of "good" assets, such as loans that are still worth what they were when they were made (no defaults and no increased probability of default) and $45 of "bad" assets, such as loans that are delinquent, or mortgage-backed securities where the underlying loans are delinquent, etc. Say the bank takes a $5 write-down on these bad assets, so it now counts them as $40 of assets, but if it actually had to sell them right now they would only sell for $20 because no one wants to buy them. (When a bank has to take a write-down and for how much is a complicated subject; suffice it to say that in many cases banks have assets on their balance sheet at values that everyone knows could not be realized in the current market, and this is completely legal.)

Right now the bank balance sheet has $100 in assets, $95 in liabilities, and $5 in capital, so it is still solvent. However, everyone looking at the bank thinks that those $40 in bad assets are really only worth $20, and is afraid that the bank may need to take another $20 write-down in the future. So no one wants to buy the stock and, more importantly, no one wants to lend it money, because a $20 write-down would make the bank insolvent, it could go bankrupt, stockholders would get nothing, and creditors (lenders to the bank) would not get all their money back. Because no one wants to lend it money, the bank hoards cash and does not lend to people who need money.

Although not necessarily to scale, this is roughly what the banking systems of the United States and several other major economies look like right now.

How does a bad bank solve this problem? There are two basic models: one in which each sick bank splits into a good bank and a bad bank, the other in which the government creates one big bad bank and multiple sick banks unload their toxic assets onto it.

Bank Mitosis

In the first model, the Bank of Middle-Earth splits into two: a Bank of Gandalf and a Bank of Sauron. The Bank of Gandalf gets the $60 in good assets, and the Bank of Sauron gets the $40 in bad assets (that may only be worth $20). The Bank of Sauron will probably fail. But the Bank of Gandalf no longer has any bad assets, so people will invest in it and lend money to it, and it will start lending again.

This model has one tricky problem, though: How do you allocate the liabilities of the old bank between the two new banks? Luigi Zingales says the simplest solution is to do so on a proportional basis. Because the Bank of Gandalf gets 60 percent of the assets, it gets 60 percent of the liabilities. So if the Bank of Middle-Earth owed someone $1, now the Bank of Gandalf owes him 60 cents and the Bank of Sauron owes him 40 cents. Now the Bank of Gandalf has $60 in assets, $57 in liabilities (60 percent of $95), and $3 in capital; the Bank of Sauron has $40 in bad assets (that are really only worth $20) and $38 in liabilities. Instead of one sick bank with $100 in assets that isn't doing any lending, you have a healthy bank with $60 of assets that is lending, and what Zingales calls a "closed-end fund holding the toxic assets," whose creditors will probably get some but not all of their money back. The tricky part is that this is a good deal for shareholders in the Bank of Middle-Earth and a bad deal for creditors to the Bank of Middle-Earth, and so it is illegal for banks to divide up the liabilities like this. Zingales recommends legislation to make it possible, but I suspect that even were Congress to pass such a bill, there would still be a lot of lawsuits challenging its constitutionality.

I started with Zingales's version of bank mitosis because it illustrates the principle neatly, but the legal complication makes it difficult to implement in practice. Another way to divide one bank into two is to find separate funding for the Bank of Sauron. This is what UBS did in November, with the support of the Swiss government. UBS had $60 billion in bad assets that it unloaded onto the new bad bank. To pay for those bad assets, however, the bad bank needed $60 billion. How did it get it? First UBS raised $6 billion in new capital by selling shares to the Swiss government. Then it invested those $6 billion in the bad bank-that became the bad bank's capital. Then the Swiss central bank loaned the bad bank $54 billion. (There is little chance that any private-sector entity would lend a self-confessed bad bank money, but this was in the public interest.) Because shareholders get wiped out first, that effectively means that UBS was taking the first $6 billion in losses, and any losses after that would be borne by the Swiss government. This constitutes a subsidy by the Swiss government to UBS, but one that was justified by the need to stabilize the financial system. At the end of the transaction, UBS had diluted its shareholders by 9 percent (because of the new shares sold to the government) and had a $6 billion investment in the bad bank it was likely to lose, but it had cleaned its balance sheet of $60 billion in toxic assets.

One issue in this version is how to value the assets that are being sold to the bad bank. If they are sold at market value ($20 in the Middle-Earth example), then the parent bank has to take a write-down immediately, which arguably defeats the purpose of the whole transaction (because that could render the parent bank immediately insolvent). In that case, the parent bank would need to be recapitalized (presumably by the government) immediately, and the "bad bank" would actually be not that bad, since it is holding assets it bought on the cheap. If they are sold at the value at which they are carried on the parent bank's balance sheet, then the bad bank is essentially making a stupid purchase (overpaying for securities it expects to decline in value) for the public good. In the UBS case, forcing UBS to provide the $6 billion in capital was a way of forcing UBS to suffer at least some of the loss that the bad bank was expected to incur.

Big Bad Bank

The second model, which has been proposed by Sheila Bair, Ben Bernanke, and others, is the "aggregator" bank. Instead of splitting every sick back into a good bank and a bad bank, in this model the government creates one Big Bad Bank, which then takes bad assets off the balance sheets of many banks. (This doesn't necessarily have to be created by the government; the Master Liquidity Enhancement Conduit-bonus points for anyone who remembers what it was for-was supposed to be funded by private-sector banks. But in today's market conditions, the government is the only plausible solution.) In this plan, the capital for the Big Bad Bank is provided by the Treasury Department (perhaps out of TARP), and the loan comes from the Federal Reserve, which has virtually unlimited powers to lend money in a financial emergency. Once this Big Bad Bank is set up and funded, it will buy toxic assets from regular banks, which will hopefully remove the uncertainty that has hampered their operations.

Yes, the Big Bad Bank is similar in concept to the original TARP proposal, and it faces the same central question: What price will it pay for the assets (the issue discussed two paragraphs above)? If it pays market value, it could force the banks into immediate insolvency, so recapitalization would have to be part of the same transaction. If it pays current book value (the value on the banks' balance sheets), it will be giving a huge gift to the banks' shareholders. There has been talk of forcing participating banks to take equity in the Big Bad Bank (as in the UBS deal), presumably to make them shoulder some of the overpayment. In any case, the money the government puts in, up to the market value of the assets purchased, is a reasonable investment for the taxpayer; but there will need to be additional money, either to recapitalize the remaining banks (which, if done at market prices, would lead to government majority ownership), or to overpay for their assets. Pick one.

One last issue: Creating a bad bank works nicely if you can draw a clear line between the good assets and the bad assets. My theoretical Bank of Gandalf above only has good assets, so there are no doubts about its health. But what if you can not? This crisis started in subprime mortgage-backed securities, and it is pretty clear that things like second-order CDOs based on subprime debt are deeply troubled. But as the recession deepens, all sorts of asset-backed securities-such as those backed by credit card debt or auto loans-start losing value, and then even simple loan portfolios lose value as ordinary households and businesses that were creditworthy just a few years ago go into default. Put another way, if it were possible to neatly separate off the bad assets, then the second Citigroup bailout would have worked, since that provided a government guarantee for $300 billion in assets. Yet Citigroup's stock price, even after Wednesday's huge rally (up 31 percent), is still below the price on November 21, the last trading day before that bailout was announced. Clearly no one believes that Citigroup had only $300 billion in bad assets.

The goal of a bad bank is to restore confidence in the good bank, and it is not clear how much of the parent bank's assets have to be jettisoned before anyone will have confidence that only good assets are left. One potential problem with the Big Bad Bank is that banks could be tempted to underplay their problems-sell only some of their bad assets, hope the rest are all good, take the bump in their stock price, and then show up two quarters later with more bad assets. If investors suspect that is going on, and that the banks are still holding onto bad assets, then the scheme will fail. The solution to that problem is to overpay for the assets, which gives banks the incentive to dump all of them onto the Big Bad Bank, and then we are back where we started.

Update: Citigroup's division into a good bank (Citicorp) and a bad bank (Citi Holdings, which includes the $300 billion in assets guaranteed by you and me) is more symbolic than anything else at this point, because they are still just divisions of one company. So if Citi Holdings goes broke, the creditors can demand money from Citicorp, which defeats the purpose of a good/bad separation. The goal here was more to communicate the bank's long-term strategy (the hope is to either sell-off or run-off everything in Citi Holdings) in hopes of convincing shareholders that the management knows what they are doing.

Simon Johnson is a senior fellow at the Peterson Institute for International Economics and a professor at MIT.

More From