iOS app Android app More

Douglas J. Elliott

Douglas J. Elliott

Posted: May 6, 2009 05:47 PM

How to Interpret the Bank Stress Tests

Washington and Wall Street are waiting with bated breath for the results of the financial "stress tests" that the regulators have been running on the 19 largest banking groups in the U.S. The public may feel hopeless about the bailout, but it is worth paying attention to what is announced on Thursday afternoon.

The fundamental reason is that the banking system is the way we get money from savers to companies and people who want to put those funds to use, whether by growing a business or buying a home. This recession has driven home the basic truth that when the banking system doesn't work, neither does the economy. Economic studies confirm that recessions associated with severe banking crises are much uglier than normal recessions.

The financial system headed towards a meltdown last autumn because there was a loss of faith in the strength of the big financial institutions. Much has been done already to remedy this, but the stress tests are an important further step. The regulators have gone into the largest banks and put them through a comprehensive, consistent planning exercise. The object is to determine how much capital the banks need to protect themselves against a recession even worse than what it appears we will experience. (A note to the optimists: we are not nearly done with the pain of this recession. Even when we bottom out, unemployment will rise further and house prices fall more, in a delayed reaction that is completely normal in recessions.)

Why is capital so important? Capital consists of the funds held by a bank on which no one has a claim, except the shareholders. The larger the base of capital, the bigger a loss the bank can absorb without hurting depositors, creditors, customers, or the taxpayers. The big banks all have enough capital now to provide a significant layer of protection, but the stress test is intended to determine whether that layer is large enough given the pain still to come, in addition to covering any excessive optimism embedded in the banks' reported accounting numbers.

The stress test works like filling in a giant spreadsheet. The regulators specified a number of key variables, such as how badly the economy is assumed to perform in terms of declines in GDP, increases in unemployment, and declines in home prices. The banks are then required to estimate what their losses would be under those conditions on the loans they've made and the securities they own. In addition, they are projecting what their profits will be on new business and all the things they do that are not directly credit-related. The regulators have provided a great deal of guidance on the range of losses the banks should expect from different types of loans. In fact, the regulators are making the final calls on all estimates, so the stress test results will be more about how they see the future than about what the banks themselves think.

There is a wide range of outcomes that could theoretically result from these tests. Economists and stock market analysts differ among themselves by hundreds of billions of dollars on the key question of how much capital will need to be raised by the banks through the course of this recession. The banking system has well over $10 trillion of loans and investments, so a $500 billion disagreement represents less than a 5% valuation difference on those assets.

In practice, the regulators are likely to conclude that the 19 banks need to issue between $100 and $200 billion of common and preferred shares, most of which will actually be handled by converting existing preferred shares into common shares. (The regulators appear to be targeting a capital structure with a higher proportion of common shares, which are considered the strongest form of capital. Please see "Bank Capital and the Stress Tests," for more on the different kinds of capital.)

How do I arrive at this estimate? First, it would be a reasonable result. It is broadly in line with loss estimates from the International Monetary Fund and others. Consider this the "consensus" view -- one that appears to be shared by the Administration and regulators, as well as most private economists. There are much more pessimistic economists out there, such as Dr. Nouriel Roubini, but the regulators do not share their views. If they did, we would probably be seeing the nationalizations for which those analysts have been calling, because the banks would require more drastic restructuring.

Second, there are political constraints that provide a strong incentive not to require capital-raising much larger than my expected range. Bigger numbers would very likely require the taxpayer to supply much of the additional capital, as the Administration has promised to do if private capital is not readily available. The Troubled Asset Relief Program (TARP) only has $110 billion left to allocate, plus the possibility that some banks will repay TARP money which can then be redeployed. The Administration is jealously guarding these unallocated funds because they very much do not want to go back to a reluctant Congress for more. (Congress for its part would rather do almost anything than choose between voting for potential financial catastrophe by turning down more TARP funds or facing the wrath of the voters by authorizing that money.)

If the required capital-raising is much more than $200 billion, it will be a bad sign about the regulators' views on the stability of the banks. On the other hand, a figure below $100 billion would suggest considerably more optimism than the consensus view. Something in the expected range has less information content, because it could either reflect the regulators' true views or simply be as much as they were comfortable requiring given the political constraints (see "Interpreting the Bank Stress Tests" for more detail).

So look to see what the magic number is on Thursday because it will say a lot about the health of our nation's financial institutions, as well as the politics of very unpopular bailouts.