04 August 2011

Do We Need Bank Stress Testing At All?

Recently we discussed in a LinkedIn Basel II | CEBS networking group the reliability of the 2011 EU-wide Bank Stress Testing Exercise (the EBA Stress Test as it is also called). “Do we need stress testing at all?” is precisely the question which was asked at the end. Considering that while commenting on their capitalisation, banks are increasingly referring to the stress tests, both those performed based on the methodologies and scenarios defined by the regulators as well as their own Pillar 2 / ICAAP stress tests, I think it is worth to spend some time on this topic.

Not surprisingly, in our discussion group we did not find the published EBA Stress Test results to be something to which investors, policymakers and other stakeholders could rely without questioning the conclusions and the accompanying communication by the banking regulators. Apart from the fact that stress test is still only a stress test for a defined scenario and thus not a quality assurance in itself, we identified at least four issues:
1) limited scope
2) methodical limitations of the used Core Tier 1 capital ratio as pass/fail criterion
3) banks applying the very same scenario and the very same methodical assumptions evidently differently
4) the results were outdated even before publishing
Furthermore, later it turned out that we should take even the words of banking regulators with a reservation. But let’s consider the above points one by one.

Limited scope of the stress test got quite a lot of attention even before the results were announced. Basically there are two main issues:
* first, liquidity stress testing was not included, and
* secondly, sovereign defaults (although considered highly likely by the market participants already  at the time when the scenario was published) were not included into the scenario.
Although EBA’s Chairperson Mr Andre Enria addressed both of these criticism in his presentation of the results (by saying that liquidity was stress tested separately with results not being published, and that sovereign stress conformed to the solutions being discussed for the European debt crisis), the fact is fact: at least for broader audience the scope of the stress test remained limited.

What concerns the suitability of Core Tier 1 capital ratio as pass/fail criterion, then I have explained it thoroughly in my earlier blog post “2011 EU-Wide Bank Stress Test: Scenario May Not Be Main Issue”. The point is that if the deductions from Core Tier 1 capital had been made properly, several more banks would have failed or put on watch based on the very same “not stressful enough” scenario.

An example would best illustrate the different implementation of the very same scenario and the very same methodical assumptions by the individual banks. We took the adverse scenario and compared the estimated losses from the defined sovereign shock for BNP Paribas (France) and RBS (UK). The following picture appeared.
* The total losses from sovereign shock for 2011-2012 were estimated to be EUR 615m for BNP Paribas and EUR 1,019.5 for RBS (estimated valuation losses for trading book exposures and additional loan loss provisions for banking book exposures are added up).
* Notably, at the same time BNP Paribas held much larger sovereign exposure (EUR 213,1bn) than RBS (EUR 113,5bn) as of 31 December 2010. Furthermore, BNP Paribas held not just much larger sovereign exposure in total, but also larger exposure towards the so-called PIIGS countries (EUR 37.0bn for BNP Paribas versus EUR 6.7bn for RBS). The same is true for the sovereign exposures towards Greece, Portugal and Ireland only (EUR 8.1bn versus EUR 1.7bn for BNP Paribas and RBS respectively).
The French bank’s estimates clearly look more optimistic; the differences between the assessments of these two banks based the very same stress test assumptions need a good explanation indeed.

 It’s ironical that after all the efforts that go into such a complex stress testing exercise, the results are outdated even before published. Furthermore, not only the results are outdated (which may happen if there are lots of uncertainties on financial markets); perhaps the most disturbing aspect by the 2011 EU-wide Bank Stress Test is that also the communication by the regulators at the time of presenting the results appears to have been out of date.

More specifically, I’m keeping in mind the EBA’s assurance on 15 July 2011 (when the stress test results were published) that the stress test included haircuts and provisions for sovereign stress in line with the discussed solutions to the EU sovereign debt crisis at the time. Note that well ahead of 30 June 2011, the second assistance plan for Greek was prepared (finalised on 21 July 2011). This plan also includes a contribution from private investors amounting to an estimated EUR 135 billion. In effect, using again the example of BNP Paribas, in the second quarter of 2011 the bank recognised a 21% loss for Greek bonds with maturities before or in 2020 which is more than the adverse scenario haircut for Greek sovereign bonds with the maturities up to 10 years (Sources: EBA’s stress test scenarios and methodology, and Q2 report of BNP Paribas). In light of this example, can we believe what the banking regulators are saying about the stress test? By 15th of July they should have had at least quite a strong feeling that Greek Assistance Programme could have larger impacts to the banks’ results than foreseen in the scenario.

Considering the above it very much looks like that the stress test results and conclusions are confusing and even misleading rather than helpful. If there are so many issues even in the regulatory stress tests subjected to rigorous reviews by various authorities, can we believe what bankers say on the analyst meetings and press conferences about the results of their internal stress tests? In other words: do we need bank stress testing at all?

I think that we still do, yet for a different reason than using the results as an argument in assuring everyone the robustness of a bank’s balance sheet (which is bound to lead to distorted conclusions).

The value of stress testing (and scenario analysis in broader sense) arises from the analysis of the collected data and information, and the qualitative discussions around it. The aim is to urge bankers themselves to learn from the past, understand better the present and to think about more than just one possible outcome for the future. The good hope is that people start to think more long term and act more responsibly than seen in the run-up of the crisis.

In order to provide motivation for thinking beyond making more profits here and now, increasing the transparency of such exercises is extremely important too. Furthermore, the more there is transparency, the less there is room for accounting tricks such as subjective valuations of the assets and liabilities (which is a problem when one really tries to estimate the value of a bank or compare the finances of several banks).

Such discussions and increased transparency are bound to lead to the consideration of the more fundamental issues in the current Western financial system, and to think about the real solutions – assuming that this system will not collapse before.

No comments:

Post a Comment