(Click Here for Paper
Since the conference version of this report in February 2011, bank stress tests have been almost continuously in the news. In the United States, the Dodd-Frank Act mandates annual stress tests for key institutions. In early 2011, the Federal Reserve conducted the fi rst test under the Act on major banks, and is currently conducting the second test, the results of which will be announced in early 2012. Europe completed a stress test in July 2011 that ignored many principles of this report. Like the prior European tests in 2010, the 2011 version is now deemed to have failed, so that yet another European stress test exercise is contemplated. We offer a framework for evaluating these exercises. The starting point is to contrast micro- and macro-prudential principles for stress tests. Microprudential stress tests emphasize the traditional role of bank capital as a buffer against loss, shielding the deposit insurance agency. The focus is on resolving insolvent banks and on “prompt corrective action” to protect taxpayers. The Basel capital ratio is key. The U.S. savings and loan crisis of the early 1980s is the motivating event. Macroprudential stress tests focus on whether the banking system as a whole has the balance sheet capacity to support the economy. A central goal is averting runs on systemic banks by wholesale creditors that lead to a contraction of credit and damage to the broader economy. To avoid aggregate deleveraging in periods of distress, remedies focus on raising new capital measured in total dollars (or euros), rather than on merely satisfying capital ratios. We argue for the macroprudential approach and propose five principles.
First, banks have to be suffi ciently solvent to avert runs. The “run point” of a systemic bank often entails significantly higher capital than the bare solvency point.
Second, even solvent banks may be required to refrain from depleting capital if the system as a whole does not meet the higher macroprudential criteria. For shareholders, one dollar inside the bank should be worth more than one dollar in dividends. But, in any case, supervisors should consider more than just private benefits and costs. Had U.S. supervisors suspended dividends in the summer of 2007, $80 billion of capital could have been retained in the 19 banks that were subject to the 2009 Supervisory Capital Assessment Program. That sum is roughly half of the public recapitalization funds that these banks received.
Third, the remedy to undercapitalization should be stated in dollar (or euro) amounts, not capital ratios. The objective is to maintain the balance sheet capacity of the banking system as a whole. That capacity depends on the level of equity in the system. After a shock, targeting the ratio of equity to assets invites banks to meet the goal by shedding assets and exacerbating the credit crunch.
Fourth, stress scenarios should consider both sides of the balance sheet, and explicitly consider fi re sales, runs by wholesale creditors, common exposures and credit crunch risks.
Fifth, liquidity rules, in addition to capital requirements, should be part of the overall framework of macroprudential oversight. By means of an illustrative example, this paper highlights the need for a macroprudential approach, and explains why the potential economic costs are much higher when banks are systemic. We then provide a diagnostic framework using bank equity and CDS prices (and correlations between them) that can help in the formulation of corrective measures. We also review a range of empirical evidence regarding large intermediaries in the light of our framework. Taken together, this evidence suggests that the fi nancial systems in both the United States and Europe remained stressed even in early 2011, well after the crisis peaked: First, based on the level of CDS prices, the default risk of U.S. intermediaries at the onset of the 2009 stress tests was perceived to be much higher than before the crisis. After the U.S. stress test, CDS prices fell substantially, but remained elevated as of early 2011 compared to highly rated non financial companies. European banks also faced elevated CDS prices at the time of the 2010 stress test, and only a few institutions have experienced signifi cant declines, indicating that perceived default risk was persistently high.
Second, the correlations between equity returns of intermediaries remained higher after the stress tests than they were in 2006, prior to the crisis. Even banks with less elevated CDS prices show equity returns
that are positively correlated with their competitors bearing higher CDS prices. Thus, bad news for one bank is bad news for all. This persistent pattern suggests that, even in early 2011, there were virtually no institutions seen as capable of absorbing weaker organizations in a fi re sale. Otherwise, bad news for weaker fi rms might be viewed as good news for potential saviors.
Third, the correlations between equity returns and banks' own-CDS prices remained negative for almost all of the institutions. A substantial negative correlation is to be expected if an institution is thinly capitalized because news that raises the value of its equity will lower the default risk on debt. For well capitalized fi rms, however, this correlation should be weak or non-existent.
Finally, the macroprudential perspective also provides important lessons for the 2011 Europe financial crisis. The bulk of the report analyzes the state of conditions in the runup to early 2011. This analysis anticipated the failure of the July 2011 European tests. We have added an epilogue that updates the empirical work to cover the developments in Europe through 2011. We continue to argue that the troubles in Europe will persist until the European banks are capitalized to the degree that they can withstand losses on their sovereign bond holdings without triggering a run or a widespread deleveraging that undermines the supply of credit.