11 April 2014

Notes on ECB’s Comprehensive Bank Assessment (Part 3)

The ECB’s Comprehensive Bank Assessment, as I have mentioned earlier, is the main thing to watch in European banking this year. (Recall Part 1 and Part 2 of this series.) A number of months have passed since the kick-start announcement by the ECB on 23 October 2013. Nations’ largest banks must now be busy with extracting the requested data, performing calculations and filling the templates, carefully prepared by the ECB and the National Competent Authorities (NCAs). Let’s take a closer look at what’s going on.

For a quick reminder, the whole assessment exercise consists of the three closely interlinked components:
1.    A supervisory risk assessment
2.    An asset quality review (AQR)
3.    A stress test
Just to note: While there have been many talks about the AQR and the stress test, I haven’t noticed anything specific about the supervisory risk assessment other than what was said in the initial communication on 23 October 2013 – and this wasn’t much. Are supervisors now hesitant when it comes to judging about banks’ liquidity, leverage and funding – or even providing specific assessment criteria? It’s my impression that the answer is “yes” – even if that’s what they are supposed to do all the time in the course of the Supervisory Review Process (SREP).

General communications

Back in November 2013, it was expected that by the end of January 2014, the ECB and the EBA will have announced key parameters of the stress test exercise.

Accordingly, on 31 January, the EBA published a paper of six pages entitled “Main features of the 2014 EU-wide stress test”. Well, that one paper wasn’t particularly informative from content point of view; it mainly informed that:
  • The exercise will be carried out on the basis of the year-end 2013 figures and the scenarios (baseline and adverse) will cover the period of 2014-2016;
  • Banks are required to stress test the following common set of risks: credit risk, market risk, sovereign risk, securitization, and cost of funding;
  • Capital hurdle rates for the baseline and adverse scenario are set at 8% (base) and 5.5% (adverse) CRR/CRD4 Common Equity Tier 1 capital (CET1%) respectively;
  • The stress test will be conducted on the assumptions of a static balance sheet and banks not changing their existing business mix and -model; and
  • Scenarios, key parameters and stress testing methodology will be disclosed in a later stage (presumably in April).
The provided technical details were fairly limited. In my view, the general tone of the paper was that the largest banks and their regulators in the euro area countries would not need to worry too much: everything will be discussed and agreed with them before making further stress test details public. That’s rather telling about what one might expect from the stress test conclusions: they will be in one or another way engineered.

Anyways, just to be clear about this one thing: the EBA’s 2014 EU-wide stress test is not the same as the stress test in the frame of the ECB’s Comprehensive Assessment; the two stress tests are simply performed in co-ordination.

Next Monday following the EBA’s press release (i.e. on 3 February), the ECB reported its progress with the AQR (“Collection of first set of data completed, portfolio selection closes in mid-February.”), confirmed the use of the discussed EBA’s stress test parameters (while highlighting that in addition the ECB’s stress test will incorporate the results of the on-going AQR, and that the definition of CET1% capital used for the stress test will be that applicable at the end of the horizon on 31 December 2016), gave some hints on how the capital shortfalls are going to be addressed, provided guidance about the next steps, and stressed its commitment to ensuring the quality and credibility of the entire process. Further, one key concern of the media and outside observers was explicitly addressed: treatment of sovereign exposures.

Specifically, information was provided that sovereign exposures will be treated in the same way as other exposures in the respective portfolios. That sounds fair enough. The trick will be in the allowance of gradual phasing-out of the available-for-sale (AFS) prudential filters which has been left to the NCAs to decide. It means that NCAs may allow their banks not to take in the capital adequacy calculations into account unrealized gains or losses on exposures to central governments classified as AFS. Effects will be disclosed, so let’s not forget to pay attention once the results are published. (See CRR Article 467 and read more about the “filter furor” from Risk.net.)

The ECB also reassured banks, NCAs and whoever else might be concerned that the exercise will fully respect the provisions laid down in the applicable accounting frameworks – be they the International Financial Reporting Standards (IFRSs) or generally accepted national accounting practices (national GAAPs) – and those set out in the CRR/CRD IV, even if for the stress test, Pillar 2 decisions may be used to go beyond regulatory minima. This means that:
  • Known issues in the existing accounting standards, even if in the process of being amended, are still going to be there.
  • Conceptual differences between the two accounting regimes (IFRS and GAAPs) are accepted and thus, banks in Germany, Luxembourg and Austria (non-IFRS countries) may report different figures on the same data fields from the banks in other euro area countries (IFRS countries). Not too transparent, but probably also not feasible otherwise.
On 6 February, final list of the 128 credit institutions (including 124 banking groups) subjected to the Comprehensive Assessment was published. Coverage seems to be fairly good: reportedly, 85% of the bank assets in the 18 euro area countries.

A shorter list of 29 banks included in the AQR trading book review was already published as an annex to the 3 February Note. These are the banks with material exposure of hard-to-value assets (the so-called level-3 assets) as well as banks with material trading book exposures.

Communications of details

On 11 March 2014, the detailed manual for the AQR, a 287-page PDF document, was published on the ECB’s website. In the accompanying press release, the ECB announced that this manual provides guidance for the National Competent Authorities (NCAs) and their third-party support on carrying out “Phase 2” (on-site inspection and the actual review of the banks’ assets) of the AQR. It’s the key part of the whole exercise. According to the current plan, this “Phase 2” will run until August 2014 – i.e. it will take two months or so longer than initially expected.

(AQR consists of three phases. “Phase 1” of the AQR was Portfolio Selection, i.e. selection of the portfolios with the highest risk for the in-depth review. “Phase 3” will be collation, i.e. final quality assurance, benchmarking and review at the central level.)

Broadly speaking, the “Phase 2” of the AQR consists of the three work blocks:
1.    The review of processes, policies and accounting (Chapter 1 in the manual)
2.    Credit file review and collateral value appraisal (Chapters 2-7)
3.    Review of level 3 fair value assets (Chapter 8)
This work will be followed by the determination of the AQR-adjusted CET1% and defining remediation activities for banks (Chapter 9).

From the perspective of the people who have to do the real work, were they in banks, in NCAs or in the ECB, that one paper doesn’t look like lots of fun ahead. A glance at the instructions in the manual, at the long list of AQR supporting tools that need to be developed and filled (not yet available, but according to the manual includes: 9 illustrative models and parameter sheets, 13 templates to be submitted at regular intervals to the Central Project Management Office (CPMO) at the ECB to provide an update on progress, 9 additional output reports in PowerPoint, PMO templates, FAQ templates, issue logs…), as well as at the time schedule (see Figure 1 below) and
at overall set-up of the project organization (illustrated in Figure 2) doesn’t make one jealous, especially in summer time…

Yeah, that’s a high degree of bureaucracy and centralization indeed:
  • NCA bank team consisting of the NCA staff and the third party experts is responsible for accurate and timely execution of the AQR in line with guidance issued by the CPMO. It is directly communicating with the relevant / dedicated staff in banks.
  • The NCA central team has been made responsible for closely monitoring progress and escalating issues to the CPMO as required.
  • CPMO has given the powers and responsibility to control the process and progress on the SSM level. 
The message to the banks is clear: no bullshit with doing exactly what you are being asked to do. As an example, there is even a funny but revealing paragraph in the AQR manual about banks selecting random data samples:
“Experience suggests that some parties can struggle to select samples randomly.”
A polite way of saying: often banks, when being asked to select random data samples for supervisory purposes etc., are very careful with picking and choosing “random” observations. This time responsible parties are forced to sign declarations that appropriate measures have been taken to ensure the sample is random. Really, that can be a lot of stress for some staff members in the banks and in the NCAs (assuming that even bankers have morals and personal values).

My further thoughts and highlights from reading the manual are following:
  • Findings of the AQR can be used for much more than a health check of the euro area banks. For one thing, AQR provides the ECB insights about the quality of the job that auditors have done in auditing banks’ accounts. Will auditors have to take some responsibility for the identified issues in the banks’ reports? I also keep wondering about likely conflicts of interest in the Comprehensive Assessment process. Secondly, and more importantly when taking into account the over indebtedness and overcapacities of the European non-financial corporations, the thorough credit file review of sampled exposures provides the ECB with detailed information about the individual companies. I wonder how the ECB is going to use this information: is it going to put a pressure on banks for engineering bankruptcies of some zombie companies, or will it rather encourage banks to sell certain loans in the form of asset-backed securities (note that the ECB has started to promote “simplest form” of securitization) to the so-called shadow banking entities? (Hopefully the ECB is not keeping in mind creating Chinese sort of “Wealth Management Products” for the unsuspecting retail investors.) Or maybe the ECB is doing pre-screening of the assets that it might purchase into its own balance sheet? In light of the recent communications, all three alternatives seem to be under consideration. 
  • Instructions on collateral and real estate valuation and on loan loss provisioning are even this detailed that these could be used as training material for credit analysts. Indeed, also smaller banks that are not in the current scope of the Comprehensive Assessment would greatly benefit from taking a closer look and gaining access to the illustrative models, parameter sheets, various benchmark indexes and -yields etc.   
  • On the other hand, notably less specific instructions are being provided when it comes to level 3 fair value exposures review work block. Ok, it’s relevant for a smaller number of banks, but still… As for the first element of this block, level 3 revaluation for non-derivative assets, revaluations of the most material exposures are ought to be performed by the NCA bank team; general principles are being provided, but the exact methodologies are left for them to decide. The second element, trading book core processes review, is highly qualitative: how a bank validates and monitors pricing models, how it calculates credit valuation adjustments and other fair value adjustments, what’s the quality of the new product approval process etc. (A job so far poorly done by the NCAs?) The third element, level 3 derivative pricing models review, is once again first of all qualitative in nature: a Q&A about the model use, model assumptions, input data and calibration; as for the calculation of quantitative adjustments for issues identified, it is being recognized that “there is no single consistent methodology available to the NCA bank team that can be used for all issues identified”. Apparently the task of evaluating complex assets (or “assets” in quotation marks) is beyond the capabilities of the supervisory staff right now. (No wonder.) In that sense, a fair degree of obscurity remains also after the AQR. 
  • That’s about the technicalities. More intriguing questions are perhaps: “How and when the banks will have to adjust their financial accounts based on the AQR findings?” and “What will be the AQR-adjusted CET1% and how capital ratios are going to be re-calculated for the stress testing purposes?” Well, while the ECB is at least trying to get a reasonably good understanding of what’s there in the banks’ balance sheets and is working hard in that sense, the communicated answers to these questions seem odd / inconsistent enough to take a closer look.
AQR “Phase 2” output no. 1: Supervisory communication to each relevant bank (e.g. in the form of a letter) outlining any areas where the bank is found to be outside of accounting principles and the required remediation actions the bank would be expected to take

(The communication follows the completion of the Comprehensive Assessment, i.e. most probably takes place end of October / beginning of November 2014.)

More than once it is stated in bold that:
“No change in the 2013 certified accounts of banks will be required following the AQR.”
Any such adjustment is intended to be very exceptional. At the same time, a number of findings from the AQR should be expected to be reflected in bank’s accounts in the relevant accounting period in 2014 following the AQR, incl.:
  • Corrections to specific provisions for individually impaired credit facilities that were sampled in the file review;
  • Corrections to specific provisions for collectively impaired credit facilities, where the bank’s collective provisioning model is viewed by the NCA Bank team as missing crucial aspects required in accounting rules;
  • Creation of a credit valuation adjustment (CVA) for derivatives.
Wait a second: wouldn’t these findings be pointing to rather significant errors in the 2013 accounts and thus shouldn’t they be reflected already there? While it’s understandable that it may not be feasible for the banks to redo all their 2013 accounting, at minimum, I’d expect restated 2013 balance sheets as an output of the AQR. They in the ECB are doing all the calculations anyway; for the sake of transparency, why not publish the results? It would be very helpful indeed for the analysts and investors. It’s not yet quite clear, what will be published at the end; so let’s hope that behind the rhetoric and professional communication strategy, there is a real intention to be a bit more transparent.

Other findings from the AQR will not be included in 2014 accounts at all; for instance banks may reject third party or NCA valuations of level 3 securities. This, of course, underlines the point that financial supervisors are not yet in the position of judging about the valuation of such securities. Maybe they don’t even have to be – but in that case such securities shouldn’t be in the balance sheets of the institutions covered with the deposit insurance and/or benefiting from the explicit or implicit government guarantees either.

AQR “Phase 2” output no. 2: Key inputs to the stress test, incl. any adjustments to data segmentation, an AQR-adjusted CET1% parameter (to allow the impact of the AQR to be applied to stress test projections of the CET1%), Probability of Impairment (PI) and Loss Given Impairment (LGI) parameters for use in the stress test

Supervisors in the ECB are saying that in order to correctly account for all incurred losses, an “AQR-adjusted CET1%” will be calculated for each bank. Fair enough. Yet again, it’s there in bold and underlined:
“The bank would not be required to restate accounts or apply the AQR assumptions on an on-going basis, i.e. the AQR-adjusted CET1% is not a de facto alternative accounting standard.”
In other words: for the calculation of the banks’ capital need, an adjustment to the banks’ capital adequacy figures is needed, but not much will be done to correct banks’ misleading capital adequacy reports.

There is a nice figure about the AQR-adjusted CET1% in the AQR manual to be recalled when interpreting the results of the ECB’s Comprehensive Bank Assessment at the end:

 This illustrative example should read as follows:
  • There is a bank that has reported a CET1% of 11% as of at the end of December 2013.
  • Based on the AQR findings, the bank has made too little provisions for the individually impaired credit facilities and/or its collective provisioning model is missing crucial aspects required in accounting rules and/or it doesn’t have proper CVA adjustments for derivatives, i.e. the bank is underestimating the incurred losses. After adjusting for these issues, the so-called Pillar 1 CET1% for the bank is 9%. Pillar 1 capital impact: 11%-9%=2%.
  • Based on the AQR findings too, an additional Pillar 2 capital impact of 9%-6%=3% is calculated. This derives from the extrapolation of the sample credit file review results to the wider portfolio, revaluations of the level 3 securities and from other likely issues in the bank’s official financial statements that will not be included in 2014 accounts because of being estimates, not known for sure.
  • For this example bank, the AQR-adjusted CET1% ratio is 6% instead of the originally reported 11%. The banks’ capital shortfall is 2% (the AQR-adjusted 6% minus the defined CET1% threshold of 8%), [follows quote from the AQR manual] “stemming from AQR adjustments not incorporated in accounts or regulatory capital that will be applied in the first year of the stress test. Such a requirement will be applied in the ST and would have to be met by the bank in a form and timeframe to be determined by the ECB.”
  • An additional capital requirement may arise from the stress test; this requirement would have to be met over a more gradual timeframe. [Another quote from the AQR manual] “For example, if the stress test resulted in a reduction in CET1% of 3%, then in the example above, with a 5.5% threshold for the stress test, the bank would need to raise a further 0.5% of current RWA in capital or via other means over a period of time to be determined.”
Complexity is created by differentiating between the Pillar 1 and Pillar 2 capital impact – this leaves certain flexibility for the ECB to interpret the results (Pillar 2 capital shortfall sounds a bit softer than Pillar 1 shortfall for the starters). Differently from what the ECB appears to be suggesting, I’d look at 6% as the example bank’s true AQR-adjusted capital position at the end of 2013, and 2% as its additional capital need at that point.

A further limitation of the CET1% calculation in Comprehensive Assessment is reliance on the banks’ internal models when it comes to deriving risk-weighted assets; review of these models is not in the scope of the Comprehensive Assessment, even if issues in the models have been highlighted by the EBA and BCBS earlier.

To conclude: There is no doubt that the ECB is going to gain fairly good insights from the exercise. However, it very much looks like that “outsiders” are going to be misled with the interpretations unless they bother to look into the very details.

With excitement, we are waiting for more information about the stress test which should be published soon.

No comments:

Post a Comment