by Calculated Risk on 3/27/2010 12:59:00 PM

From Fed Governor Daniel Tarullo: Lessons from the Crisis Stress Tests

The Supervisory Capital Assessment Program (SCAP) was fashioned in early 2009 as a key element of a crucial plan to stabilize the U.S. financial system. The stress tests, as they have been popularly called, required development on the fly, and under enormous pressure, of ideas that academics and supervisors had been considering for some time. After describing the concept, design, and implementation of last year's tests, I will explain how our experience has helped prompt major changes in Federal Reserve supervision of the nation's largest financial institutions. Then I will discuss how this experience has stimulated debate over the merits of publicly releasing supervisory information.

Tarullo reviews the stress tests, and then argues that the public release of data was helpful (I agree):

As you know, unlike other countries that conducted stress exercises, we took the highly unusual step of publicly reporting the findings of the SCAP, including the capital needs and loss estimates for each of the 19 banks. This departure from the standard practice of keeping examination information confidential was based on the belief that greater transparency of the process and findings would help restore confidence in U.S. banks at a time of great uncertainty. Supervisors released the methodology and assumptions underlying the stress test first and then, two weeks later, the results for individual institutions. ...

The merits of publicly releasing firm-specific SCAP results were much debated within the Federal Reserve. In particular, some feared that weaker banks might be significantly harmed by the disclosures. In the end, though, market participants vindicated our decision.

And then Tarullo argues for regular stress tests (I agree again):

To this end, the Federal Reserve is now implementing a more closely coordinated supervisory system in which a cross-firm, horizontal perspective is an organizing supervisory principle. We will concentrate on all activities within the holding companies that can create risk to the firm and the financial system, not just those that increase risk for insured depository institutions.

An essential component of this new system will be a quantitative surveillance mechanism for large, complex financial organizations that will combine a more macroprudential, multidisciplinary approach with the horizontal perspective. Quantitative surveillance will use supervisory information, firm-specific data analysis, and market-based indicators to identify developing strains and imbalances that may affect multiple institutions, as well as emerging risks to specific firms. Periodic forward-looking scenario analyses will enhance our understanding of the potential effects of adverse changes in the operating environment on individual firms and on the system as a whole.

In fact, I believe that the most useful steps toward creating a practical, macroprudential supervisory perspective will be those that connect the firm-specific information and insight gained from traditional microprudential supervision to analysis of systemwide developments and emerging stresses. Here, precisely, is where our SCAP experience has helped lead the way.

And Turallo argues the macro assumptions and the individual stress test results should be made public:

[T]he release of details about assumptions, methods, and conclusions would expose the supervisory approach to greater outside scrutiny and discussion. Sometimes those discussions will help us improve our assumptions or methodology. At other times disclosure might reassure investors about the quality of the tests. Either way, the public's reaction to our assumptions and methods would be useful.
...
[To increase transparency, the supervisors could] follow the SCAP precedent, with periodic release of detailed information about the assumptions, methods, and results of a cross-firm, horizontal, forward-looking exercise, including firm-specific outcomes. This approach would probably maximize both the potential benefits and potential risks. Note, however, that the possibility of a destabilizing market reaction may be lower if such information is released frequently, as major unpleasant surprises would be less likely with frequent, detailed disclosures.

I supported the stress tests of the largest financial institutions and I think this would be a helpful regular exercise (probably on an annual basis). I think another set of macro assumptions should be released (base case and severe), and the same level of detailed company specific information be released as for the SCAP.

The FSA is already doing regular stress tests in the U.K.:

We have now embedded our new approach to stress testing into our normal supervisory process. This includes supplementing firms’ own stress testing with supervisory stress testing of major firms. This involves regularly updating the stress test scenarios.

The initial stress tests were very helpful, I think it is time for the Treasury and Fed to release another set of macro assumptions and stress test the banks again - and also release the company specific results.