Interpreting the Bank Stress Tests

Douglas J. Elliott
Douglas J. Elliott Former Brookings Expert, Partner - Oliver Wyman

May 4, 2009

The stress test results for individual banks have largely been leaked prior to the announcement late afternoon today of the official results. The total capital that will need to be raised seems to be under $75 billion, which is below my “expected” range of $100-200 billion. As explained in “interpreting the Bank Stress Test Results,” this is good news, but we have to be careful not to overinterpret.

It’s definitely good news that the regulators are this optimistic after reviewing each of the 19 banks in some detail. However, the stress test is essentially a prediction about an uncertain future and the regulators could easily be wrong. They used a great deal of subjective judgment in guessing future loan losses and the underlying profitability of the banks, so this is their estimate, not a pure technical result. These 19 banks have around $10 trillion in assets; a 3 point higher loss ratio would mean $300 billion more capital is needed than shown by the tests.

We do not yet know how tough the test was. The regulators should give us more detail about assumed loss rates this afternoon, in which case we can compare their estimates with those of the IMF, Nouriel Roubini, and others. My sense is that the regulators’ assumptions are more optimistic than those from these other analyses.

The real stress test will be making it through 2009 and 2010 without losing too much money. We will only truly know the capital needs after the fact.


We will hear on Thursday the results of the financial “stress tests” that the regulators have been running on the 19 largest banking groups in the U.S., those with assets of over $100 billion. We will directly learn which bank holding companies will be required to raise additional capital, or to convert one form of capital to another, and which are fine as is. The likely result is that a handful of the 19 banks will need to issue common or preferred stock totaling $100-200 billion in the aggregate, with some of that handled by converting preferred shares into common shares. Taxpayers would be on the hook as a “back stop” provider of capital if one or more of these firms cannot raise the capital privately or chooses not to.

Equally importantly, we may learn something about the government’s views of the financial state, and prospects, of the 19 banks, and of the likely depth of the banking crisis. These firms dominate the U.S. banking system, holding over two-thirds of the system’s assets among them. Good or bad news about them has profound implications for the overall economy, because the credit crunch created by the banking crisis is a major contributor to the severity and length of this recession. The “white paper” released by the Federal Reserve on April 24th makes clear that there is a great deal of regulatory judgment being applied in these tests, whether in judging loan quality or determining the capital requirements. Therefore, the results will give us a real sense of the thinking of the regulators about the financial state of the banks, based on a detailed, comprehensive analysis. Unfortunately, this view may be muddied a bit by political constraints.

The key figure to note is the size of the aggregate capital actions required for the 19 banks, measured as the amount of preferred stock converted to common stock plus the amount of incremental capital raised beyond the starting level. The range I expect, of $100-200 billion, would be broadly in line with consensus expectations for the depth of the recession and the resulting credit losses. A smaller requirement would likely indicate the regulators and the Administration are more optimistic about the banking crisis. The $100-200 billion range is feasible within political constraints and therefore would not be avoided solely for political reasons. On the other hand, a larger figure might indicate that things are worse than the consensus believes. Imposing a larger total capital requirement would hit significant political barriers, particularly since it would probably force an eventual return to an unwilling Congress for more money. Doing this despite those political constraints would seem to indicate serious concerns.

The trickiest situation to analyze would be if the aggregate figure is in the expected range. This could either mean that this is the level regulators genuinely see as the right additional buffer or could reflect the political reality that Congress would be extremely reluctant to authorize more funds and would almost certainly add strings that the Administration would find onerous. It could be more appealing to operate within the present authorization and wait to see if that proves to be enough.

This paper answers the following questions:

  • How does the stress test work?

  • How can we interpret the overall results?

  • Can we tell in advance which banks will need the most funds?