Tuesday, July 6, 2010

Bank Stress Tests in the GCC?

Bank Run 1933  USA National Archives  (In the Public Domain)

Tom Arnold over at The National reports that "Gulf governments have been urged to follow the EU and introduce stress tests of regional banks’ capacity to withstand financial shocks."

With some not unexpected remarks about the need for "transparency and disclosure"
“Stress tests would be welcomed but critical to their success would be having correct systemic inputs to make sure the assumptions going into stress-testing were sensible,” said Raj Madha, a senior banking analyst at Rasmala Investment Bank.

“We need fuller disclosure for stress testing to be independently verifiable.”
As usual, let's take a closer look.

Before my usual joyous descent into the details, an uncharacteristic exploration of the relevant meta topic. Framed as a question.  Just what do serious minds out there think these tests are designed to do?  If the national banking system is insolvent and unsafe, do they really expect the national regulator to start a bank panic or crater the economy by publicly announcing this?  Exercises like this are designed to reassure markets not disrupt them.  Purpose - as they say - informs practice.

Now to the details.  

There are three key areas in stress tests:
  1. The definition of stress
  2. The test methodology
  3. The interpretation of the results
First and foremost is the definition of the stress.  

As the Grey Lady reported regarding the US stress tests:
But analysts say the administration’s worst projections, which it describes as unlikely, are not much more dire than what many private forecasters already expect.
As the less than politically correct folks over at EurActiv put it about the EU tests:
"Some scenarios that should be included will not be included, at least in what will be publicly revealed, which will be sugar-coated to avoid market panic," according to an informed source close to policymakers.
Putting aside for the moment the goal of the exercise (reassurance), there is a very tricky "technical" problem here.  It is possible to define stress so that no bank passes or all pass.  With a very broad range of outcomes "in between".  

Just what is the right scenario - particularly when it will have policy implications (the need for banks to raise additional capital, for the authorities to "rescue" or shutter banks or some combination thereof) and market reactions (exuberance or panic).  There is no easy answer.  

No doubt regulators weigh carefully the potential implications of their choice of scenarios.  Use of a dire scenario may be taken by the market as the authorities' assessment that that outcome is likely.  And thus  become a self fulfilling prophecy.

Then there is the test process itself.

Given what might be described by some uncharitable souls (but definitely not AA) as "soft ball" downside scenarios, the banks are asked to grade themselves by coming up with projected loan and other losses.   Much better than an "open book" test.  

The banks could I suppose base their answers on the same internal models that served them so well prior to the subprime crisis.  Models whose worth has been tested and proven.  Well, if not exactly proven,  certainly tested!

A much surer method is to send in a team of flinty-eyed no nonsense examiners to each bank and do the work oneself.  An assurance of consistent methodology and standards as well as the opportunity to look closely at individual portfolios.  And strictly confidential.  This methodology is not applicable here as this is really not the purpose of the stress tests. 

Then there is the grading of results.

The "pioneer" of national stress tests after the Subprime or global Financial Crisis (Yes, it's still lower case "g") was of course the USA.   As the Washington Post reported:
"Some major banks managed to wrest concessions from the government in closed-door negotiations over their "stress tests" that helped them put the best face on their results, financial analysts, industry officials and sources said."
As the article notes, various mechanisms for remedying capital deficiencies were concurrent with the announcement.   Recognition of to-be-concluded sales, conversion of rescue "loans" to equity, and on and on.  All announced so that the impact of negative results would be muted.

Prodded by Spain, the EU will be releasing the results of its stress tests.  Unlike the US's tests, these will be "pass/fail"  (no quantification of capital shortfalls).  An innovative adaptation of one of the more remarkable advances in the theory of higher education. 

So, now that we're clear on the process, hopefully there's a bit more understanding of what stress tests for GCC banks would mean.  And more importantly what they would not.   

This takes us full circle back to the introductory calls for  شفافية  in any GCC stress tests.  An apparent sign of some concern about the integrity of the process in the GCC.   It's unclear if that is a relative or absolute judgment on the GCC as compared to the USA or the EU.   In any case, I'm fairly confident that the GCC will be able to match the "developed" world closely enough for it not to affect the purpose of the exercise.

Credibility will be another matter.  Short of declarations of financial Armageddon,  local regulators announcements are likely to be dismissed as less than  كلام  شريف .   Which in itself poses a rather fundamental problem.

And for those who haven't had their fill on this topic, a link to papers at a 2006 IMF Experts Forum on the topic of bank stress tests.

No comments: