Aggregate Report On The Greek Banks' Comprehensive Assessment 2015
Aggregate Report On The Greek Banks' Comprehensive Assessment 2015
Aggregate Report On The Greek Banks' Comprehensive Assessment 2015
GREEK COMPREHENSIVE
ASSESSMENT 2015
The document at hand constitutes an analysis of the disclosure data published on 31 October
2015 conducted by the ECB. In case of discrepancies, the disclosure data supersedes this report.
All rights reserved. Reproduction for educational and non-commercial purposes is permitted provided that the source
is acknowledged.
ISBN
EU catalogue number
Contents
1
Executive summary
1.2 Outcomes
2.1 Rationale
2.2 Scope
2.4 Execution
13
15
16
19
19
24
AQR outcomes
26
26
29
38
43
Stress test
45
45
5.2 Scenarios
56
5.3 Aggregate impact by risk drivers under the baseline and adverse scenarios
58
Appendices
61
61
62
6.3 Bibliography
64
65
List of figures
Figure 1
Figure 2
Figure 3
Figure 4
10
Figure 5
13
Figure 6
Project governance
14
Figure 7
18
Figure 8
19
Figure 9
20
Figure 10
21
Figure 11
22
Figure 12
23
Figure 13
25
Figure 14
27
Figure 15
28
Figure 16
30
Figure 17
31
Figure 18
Change in provisions by asset class for non-retail reclassified NPE (in the sample)32
Figure 19
Change in provisions by asset class for non-retail existing NPE (in the sample) 33
Figure 20
34
Figure 21
35
Figure 22
36
Figure 23
38
Figure 24
39
Figure 25
40
Figure 26
42
Figure 27
44
Figure 28
47
Figure 29
49
Figure 30
52
Figure 31
54
Figure 32
55
Figure 33
Cumulative stress test impact on the CET1 % by risk driver under the baseline
scenario
59
Figure 34
Cumulative stress test impact on CET1% by risk driver under the adverse scenario60
List of tables
Table 1
Table 2
24
Table 3
24
Table 4
27
Table 5
29
Table 6
32
Table 7
Table 8
49
Table 9
55
Table 10
56
Table 11
57
Table 12
61
Table 13
61
Table 14
62
EXECUTIVE SUMMARY
One of the main objectives of the Memorandum of Understanding (MoU) signed by the
European Stability Mechanism, the Hellenic Republic and the Bank of Greece on 19 August
2015 is to implement all necessary policy actions to preserve financial stability and strengthen
the viability of the banking system in Greece. In particular, solvency and liquidity of the
banking sector is to be preserved. Against this backdrop, a buffer of up to 25billion has been
envisaged under the Programme to address potential bank recapitalisation needs of viable
banks and resolution costs of non-viable banks, in full compliance with EU competition and
state aid rules. A forward-looking evaluation of each of the four core banks capital needs has
thus been requested of the ECB in its supervisory function. As a result, a Comprehensive
Assessment (CA) has been conducted by the ECB, based on end of June 2015 data and
comprising both an asset quality review (AQR) and a stress test with baseline and adverse
scenarios.
The exercise is based on updated macroeconomic data and scenarios that reflect the changed
market environment in Greece and has resulted in aggregate AQR-adjustments of 9.2 billion to
participating banks' asset carrying value. Overall, the assessment has identified capital needs
totalling, post AQR, 4.4 billion in the base scenario and 14.4 billion in the adverse scenario.
Covering the shortfalls by raising capital would then result in the creation of prudential buffers
in the four Greek banks, which will facilitate their capacity to address potential adverse
macroeconomic shocks in the short and medium term and their capacity to improve the
resilience of their balance sheet, keeping an adequate level of solvency.
Banks have to propose remedial actions (capital plans) in order to cover the entire shortfall
(14.4 billion), out of which a minimum of 4.4 billion (corresponding to the AQR plus
baseline shortfall) is expected to be covered by private means.
The recapitalisation process under the Programme will follow this exercise. The forthcoming
capital planning and recapitalisation process are not covered in this report.
1.1
Following the 19 August 2015 agreement between the European Stability Mechanism, the
Hellenic Republic and the Bank of Greece, the ECB was requested to provide a forward-looking
view of the capital needs of the four Greek systemic banks (Alpha Bank, Eurobank, National
Bank of Greece and Piraeus Bank). The key objective was to review the status of the banks
under a given set of macroeconomic scenarios. This report provides an overview of the
approach taken and presents the results of the exercise.
The comprehensive assessment of the Greek banks was broad in scope: the participating banks
have total group assets of 296 billion which account for approximately 90% of the assets of
credit institutions in Greece. Substantially, the comprehensive assessment followed the
methodology of the 2014 exercise of the 130 banks. However, further specifications to the
methodology were required given the specificities of the Greek banking system and the current
macroeconomic situation. These specifications are explained in detail in Chapters 4 and 5.
The comprehensive assessment of the Greek banks consisted of two components.
1.
The asset quality review was a point-in-time assessment of the accuracy of the
carrying value of banks assets as of 30 June 2015 and provided a starting point
for the stress test. The AQR was undertaken centrally by the ECB, and was based on a
uniform methodology and harmonised definitions. Under the AQR, banks were required
to have a minimum Common Equity Tier 1 (CET1) ratio of 9.5%.
2.
The AQR respected current accounting and prudential regulation, including the Capital
Requirements Regulation (CRR) / Capital Requirements Directive IV (CRD IV) rules. The
Greek comprehensive assessment 2015 was in line with the AQR methodology applied in the
comprehensive assessment 2014 as outlined in the AQR Phase 2 manual1. In some areas, the
ECBs methodology involved additional prudential prescription to accounting concepts; the
results are thus of a prudential nature. Consequently, AQR-adjustments were made in cases
where banks were not breaching accounting rules. However, it is expected that many banks will
reflect many of these adjustments in their accounts in agreement with their statutory auditors.
Examples of areas in which additional prescription was provided include impairment triggers,
the calculation of individual specific provisions, and collateral valuations.
Given the constrained timeline of the exercise, prioritisation of portfolios based on their size and
materiality was required, while applying appropriate rigour to the wider process. Within the
https://www.bankingsupervision.europa.eu/ecb/pub/pdf/assetqualityreviewphase2manual201403en.pdf
AQR, a detailed asset-level review was performed for over 26 specific portfolios making up
92% of the banks' risk-weighted assets in Greece.
The stress test is not a forecast of future events, but a prudential exercise to assess banks ability
to withstand weaker economic conditions. In the Greek comprehensive assessment 2015, the
stress test was undertaken centrally based on data templates and loan tapes provided by the
banks. The projections were produced by the ECB following a centrally agreed methodology.
Throughout the exercise, there has been an appropriate level of interaction with the banks
technical teams in order for the ECB to understand the banks submissions.
In order to maintain consistency and equal treatment across both the AQR and stress test, Single
Supervisory Mechanism (SSM) teams independently performed quality assurance on the data
provided by the banks and work of external auditors. On the AQR side, the ECB was in close
contact with the external auditors, responding to over 150 methodology and process questions.
The ECB reviewed and challenged outcomes from an SSM-wide perspective using comparative
benchmarking, as well as engaging SSM on-site teams to investigate specific issues that arose.
Over 300 experts (including external auditors, appraisers, consultants, and SSM staff) were
involved in the AQR. On the stress test side, the ECB has followed a pre-approved quality
assurance process including benchmarking of the results and comparing all projections with
banks own projections, with the material differences having been understood in detail. For the
avoidance of doubt, the ECBs centrally calculated projections were always applied, however
the banks projections were used extensively in order to identify material differences and justify
the rationale for those.
1.2
OUTCOMES
Significant AQR findings have been found in this exercise, despite the already material AQR
findings from 2014 being captured in banks accounts. This has primarily been driven by the
deterioration in the macro-economic environment in Greece which has led to higher NPE
volumes as well as lower collateral values and cashflow valuations which has led to material
reductions in carrying values. Additionally, further standardization of the definition of key
metrics across the EU has led to further NPE and impairment recognition in the AQR. As an
example the full implementation of the EBA ITS on NPE has meant that forborne cases could
be better identified and tested for impairment. Finally, the fact that tax offsets were not allowed
from the AQR has amplified the findings of the AQR vis a vis 2014.
The AQR resulted in direct aggregate adjustments of 9.2 billion to participating banks'
asset carrying values as of 30 June 20152. Including indirect impacts on CET1%, post
AQR, the average CET1% for the system was 7.9%. The direct adjustments originated
primarily from accrual accounted assets, particularly adjustments to specific provisions on nonretail exposures and retail mortgages. Additionally, non-performing exposure (NPE) stocks
were increased by 7 billion across the in-scope institutions, as NPE definitions were moved
onto a harmonised and comparable basis, including the examination of forbearance as a trigger
of NPE status.
In the base scenario of the stress test, this capital impact leads to a decrease of the CET1 ratio
for the system of 0.3 percentage points from 7.9% (post AQR) to 7.6% in 2017. In the adverse
scenario, the impact is a decrease of the CET1 ratio for the system of 7.8 percentage points from
7.9% (post AQR) to 0.1%.
Overall, the comprehensive assessment identified a capital shortfall of 4.4 billion and
14.4 billion in the baseline and adverse scenario, respectively, across the four
participating banks after comparing these projected solvency ratios against the thresholds
defined for the exercise.
The results of the exercise, including the reduction in the CET1 ratio projected as of December
2017 for each bank in both the base and adverse scenarios are shown in Table 13.
The direct impact of AQR on provisions and CVA is equal to 9.2 billion. The indirect impact of AQR on CET1
(e.g. DTA deductions) and RWA is equal to 401 million and 8.8 billion, respectively.
3
The shortfall shown in the table reflects the lowest capital level over the 2.5 year period. In the baseline scenario
this is not necessarily December 2017, for which the projected CET 1 ratio is indicated.
Table 1
Bank Name
CET1 ratio
starting
point post
AQR
CET1 ratio
baseline
scenario
CET1 ratio
adverse
scenario
Capital
shortfall
baseline
scenario
( billion)
Capital
shortfall
adverse
scenario
( billion)
12.7%
9.6%
9.6%
2.1%
0.26
2.74
Eurobank
Ergasias, S.A.
13.7%
8.6%
8.7%
1.3%
0.34
2.12
National Bank of
Greece, S.A.
11.6%
8.1%
7.3%
-0.2%
1.58
4.60
Piraeus Bank,
S.A.
10.8%
5.5%
5.2%
-2.3%
2.21
4.93
System wide
12.1%
7.9%
7.6%
0.1%
4.39
14.40
Nature and methodology of the exercise: Objectives and guiding principles, a high-level
view of its approach and the execution of the project
Outcomes of the AQR and further analysis: review of the detailed drivers of the AQR
results and details on the methodology applied
Outcomes of the stress test and further analysis: review of the detailed drivers of the
stress test results and details on the methodology applied
This chapter explains the rationale for the Greek comprehensive assessment, its components,
provides a high-level overview of the methodology and describes how it was executed. Finally
this chapter introduces some key features of the exercise that the reader should be aware of
when interpreting the results.
2.1
RATIONALE
Following the agreement from 19 August 2015 between the European Stability Mechanism, the
Hellenic Republic and the Bank of Greece, the ECB was requested to provide a forward looking
view of the capital needs of the four Greek systemic banks (Alpha Bank, Eurobank, National
Bank of Greece and Piraeus Bank). The key objective was to review the status of the banks,
under given new macroeconomic scenarios and measure the capital needs for a subsequent
recapitalisation exercise.
2.2
SCOPE
The ECB has undertaken a comprehensive assessment of the four Greek significant institutions
with total group assets of 296 billion at the end of June 2015, accounting for approximately
90% of total banking assets in Greece. These banks were identified based on significance
criteria referred to in Article 6(4) of the SSM Regulation. The four banks in scope were the
following:
2.3
The exercise comprised two pillars, namely the AQR and the stress test.
The AQR aimed to review the carrying value of assets on the participating banks' balance sheets
as of 30 June 2015. The result was an indication of the need for additional provisions for losses
on exposures on banks' balance sheets, leading to prudently calculated AQR-adjusted capital
ratios, which allowed for the meaningful comparison of all participating banks on a like-for-like
basis.
Based on the AQR-adjusted balance sheet, the stress test examined the resilience of banks
against two separate scenarios a baseline and adverse scenario starting in H2 2015 and
running to the end of 2017. Under both scenarios, the solvency ratio of each bank was analysed
over that period to understand bank sensitivities given prescribed stressed economic conditions.
The baseline scenario was provided by the European Commission and reflected then-prevailing
official macroeconomic forecasts while the adverse scenario represented a severe economic
downturn triggered by a materialisation of the main economic risks as provided by the ECB.
Over the period following the 2014 comprehensive assessment to 30 June 2015, Greece
experienced a deteriorating economic environment. In these 18 months from 2013 year-end to
H1 2015, there was a contraction of the economy with a cumulative GDP decline of -0.4%. In
addition, capital controls were announced on 28 June 2015. To reflect the recent
macroeconomic developments of the Greek economy over the period following the 2014 CA to
H1 2015, amendments and specifications to the AQR and stress test methodology were made.
These are explained in more detail in the following section as well as Chapter 4 and 5.
Figure 1
10%
YE 2013 H1 2015
5%
-0.4%
0%
-5%
-10%
-15%
2000
2002
2004
2006
2008
2010
2012
2014
Source: Eurostat
The stock of Deferred Tax Assets (DTAs) on the Greek bank balance sheets in the COREP
submission as at H1 2015 counted towards capital in line with the CRR rules over the course of
the stress test. The majority of the DTA stock (82%) was not dependent on future profitability
as a result of the tax regime in Greece. Of the remaining DTAs, for the purposes of the exercise,
the phase in of deductions was applied consistently across the Greek banks. Specifically, a 5
year phase in was applied for DTAs that rely on future profitability and do not arise from
temporary differences (12% of stock), whilst DTAs that rely on future profitability and arise
from temporary differences had either a 10 year or a 5 year phase in, depending on whether they
existed prior to year-end 2013 or not (4% of stock take a 10 year phase in, whilst 2% takes a 5
year phase in). No further DTA accumulation was allowed for the purposes of the exercise from
either the AQR or the stress test.
2.3.1
AQR METHODOLOGY
The Greek comprehensive assessment 2015 followed the same methodology applied in the
comprehensive assessment 2014 as outlined in the AQR Phase 2 manual. Given the constrained
timeline of the exercise, prioritisation of portfolios based on their size and materiality was
required, while applying appropriate rigour to the wider process.
For the purpose of this exercise, the portfolio selection from the 2014 AQR was taken as the
starting point, though the focus was put solely on exposures booked in Greek legal entities. In
total 92% of Greek legal entity credit Risk-Weighted Assets (cRWA) were reviewed as part of
the AQR.
Figure 2
100%
12%
90%
80%
70%
60%
50%
92%
95%
94%
94%
Eurobank
NBG
Piraeus Bank
85%
40%
30%
20%
10%
0%
Average
Alpha Bank
cRWA coverage %
Note: cRWA coverage by bank refers to total Greek legal entity cRWA covered within the AQR exercise per bank
For Alpha Bank, additional coverage with granular data (i.e. loan tapes) was achieved for adequate treatment in the
stress test.
The portfolio selection covered all major AQR asset segments, where the most significant in
terms of RWA are Residential Real Estate, Large Corporates and Large SME.
Figure 3
35,000
100%
96%
100%
100%
30,000
73%
25,000
80%
20,000
15,000
60%
29,018
25,394
40%
25,377
10,000
20%
5,000
9,057
7,689
7,083
Other
retail
Retail
SME
Shipping
96%
0%
0
Residential
real estate
Large
corporates
cRWA selected
Large
SME
Note 1: cRWA coverage by asset class refers to total Greek legal entity cRWA covered within the AQR exercise by each asset segment
Note 2: Alpha Other retail not included as covered under ST exercise
The AQR has nine interlinking workblocks with the final output of an AQR-adjusted CET1
ratio (workblock 9) to be compared to the threshold of 9.5%.
Figure 4
Phase
1
Phase 2
1
Projection of findings
of credit file review
Portfolio
selection
Processes,
policies and
accounting
review
(PP&A)
[Note: not
relevant for
this exercise
apart from
CVA
workblock]
Loan tape
creation
and data
integrity
validation
(DIV)
Credit
file review
(CFR)
Determination
of AQRadjusted CET1
for use in ECB
Collective provisioning
stress test and
definition of
remediation
Collateral and real estate valuation activities for
banks
following
the CA
Level 3 fair value exposures review
[Note: not relevant for this exercise]
8
CVA
challenger
model
adjustments
Sampling
i.
Level 3
revaluation
of non-derivative
assets
ii.
Core processes
review
iii.
Derivative
pricing
model review
2.
Loan tape creation and data integrity validation (DIV): The credit analysis (sample
selection and collective provisioning challenger model creation) was based on a "loan
tape" provided by the bank. This loan tape included basic account information such as
segment classification, missed payments status and identifiers of the loan / entity. The
data was required to be of sufficient quality to perform the required analysis, which
necessitated automated checks of the data set and a review of consistency across
internal IT systems.
3.
Sampling: Given the volume of analysis involved it was neither possible nor
appropriate to review all exposures in every portfolio within the scope of the credit file
review. Therefore, risk-based sampling was conducted in a manner that meant the
sample chosen was both large enough, and representative enough, to allow for robust
analysis and later projection back to cover the entire portfolio. The size of the sample
depended on: the homogeneity of the portfolio, the risk of the portfolio, the total
number of debtors and the level of debtor concentration. Portfolios were stratified based
on the riskiness and exposure size of debtors. The approach to sampling was consistent
with best practice as defined by adherence to ISA 530.
To achieve a timely completion of this exercise, the sample for the 2014 CA has been
taken as the starting point. In addition, fresh files were selected amounting to at least
10% for each portfolio.
4.
Credit file review (CFR): The credit file review involved external auditors working on
an exposure by exposure basis to verify that each credit exposure had been correctly
classified in the banks systems (e.g. correct regulatory segment, NPE status,
impairment status) and that, if a specific provision was required, it had been set at an
appropriate level. The CFR covered all loans, advances, financial leases and other off
balance sheet items in the selected portfolios.
To account for the current market environment, a number of additional elements were
included: in particular external auditors needed to make sure that information on
companies and collateral reflected the current market conditions. Quality assurance
findings from the 2014 CA were taken into account from the start. The credit file review
was also extended in order to capture any effects from the capital controls as an input to
the stress test. In addition minimum haircuts were applied across all major asset classes.
5.
Collateral and real estate valuation: A key input to determine appropriate carrying
amounts is the valuation of collateral. The results of these valuations were used as
inputs to credit file review and collective provisioning.
For the exercise to be feasible in the tight timeline available, the valuation of collateral
was limited to the areas deemed most important: All shipping collaterals were reviewed;
for residential mortgages, collateral values were updated from 2014 using an index with
a sample of 20% of properties being reappraised to verify the results of indexing; for
real estate related debtors, revaluation criteria as per the Manual were followed; for
other non-retail exposures reappraisals were conducted where deemed required by the
bank team.
6.
Projection of findings of credit file review: Findings of the credit file review were
then projected to the unsampled part of the portfolio. Specifically, projected metrics
were mainly provisions and NPE reclassifications. Projection of findings was applied to
homogeneous pools of exposure within each portfolio called "strata" (in line with audit
guidelines, see Sampling). In order to prevent overstating the projection of single credit
file review findings, a number of safeguards were implemented in the projection
methodology (e.g. flagging of anomalies, common risk stratum based projection using
results from the whole risk bucket rather than just from the stratum, and overrides in
rare cases where results from the sample were felt to be unrepresentative which had to
be approved centrally by the ECB).
7.
9.
Determination of AQR-adjusted CET1 ratio for use in ECB stress test: In order to
correctly account for all AQR adjustments, an AQR-adjusted CET1 ratio was
calculated for each bank. This AQR-adjusted CET1 ratio was calculated according to
the Single Rulebook, reflecting the implementation of the CRR / CRD IV rules (taking
into account transitional arrangements) as of year-end 2015 the main difference being
that no formation of new DTAs will be allowed, associated to the AQR adjustments
beyond those already booked by June 2015.
2.3.2
The stress test model projected CET1 ratios for each bank according to a baseline and adverse
scenario. The output projection of CET1 ratios was compared to the CET1 ratio thresholds
designed for the exercise and a shortfall calculated as the maximum CET1 shortfall in absolute
terms at June 2015 and December 2015, 2016 and 2017.
The high level modelling approach for the projection was as per the diagram below.
5
6
Provisions set aside for future expected losses on currently performing debtors.
International Accounting Standards.
Figure 5
P&L and
balance
sheet
projection
+ Balance
Sheet
projection
H1 2015
H2 2015
2016
2016
2017
2017
CET 1%
CET1
%
XX
XX
XX
XX
RWA
XX
XX
XX
XX
Balance Sheet
Assets
XX
XX
XX
XX
Liabilities
XX
XX
XX
XX
Equity
XX
XX
XX
XX
XX
XX
XX
XX
Expenses
XX
XX
XX
XX
Retained earnings
XX
XX
XX
XX
A projection of the P&L and balance sheet was produced based on the underlying performance
of each bank. The CET1 ratio projection then considered the roll forward of capital in a
mechanical manner, fully reflecting CRD IV phase in and other assumptions defined by the
relevant stakeholders. Detailed drill down analysis was performed on Net Interest Income
(NII), Other Operating Income and Expenses, RWA and provisions. The drill down analysis fed
the P&L and balance sheet projection. More details on the stress test methodology applied are
provided in Chapter 5.
2.4
EXECUTION
A number of parties were involved in the execution of the Greek comprehensive assessment:
The Steering Committee of the comprehensive assessment was the main decision and
steering body at the ECB / SSM level
At the ECB, the comprehensive assessment was conducted and coordinated under the
lead of a Central Programme Management Office
The ECB SSM Technical and Quality Assurance (QA) team performed central and onsite quality assurance on data provided by the banks and external auditors, prepared the
models and templates used and developed the methodology applied in the exercise
The Bank of Greece as the national supervisor has been closely involved in the process
(including the QA mentioned above), contributing with its local expertise about the
banks under review
External auditors, property appraisers and valuation advisers were involved in the
completion of the banks AQR, acted as a first line of defence in the quality assurance
and reported to the ECB. Note the auditor for each bank was not the statutory auditor
of the bank. Overall, more than 200 audit and valuation experts where involved in the
exercise, reviewing about 4,000 credit files and 12,000 collateral items
Participating banks were responsible for fulfilling their obligations to the ECB,
providing data for the AQR and the stress test
Figure 6
Project governance
Steering Committee
The AQR was executed by the external auditors, following a methodology designed and
published by the ECB. Central quality assurance was performed and requests were
made by the ECB and the external auditors to investigate certain results further details
of this process can be found in Chapter 2.6. Both the ECB and external auditors were
In this years comprehensive assessment, the stress test was a centrally led top-down
ECB exercise, following a methodology designed and published by the ECB. Banks
provided the ECB with data on their baseline projections, while the adverse scenario
projections were centrally performed by the ECB
2.5
questions (FAQ) and helpdesk process, as well as the thorough review of any outliers or
anomalies during the central quality assurance process.
The comprehensive assessment involved central oversight in both methodology definition
and quality assurance for consistency and transparency. This included the preparation of the
methodological manual and providing additional clarifying support.
The teams were aided in their work as both the external auditors and the Central Project
Management Office (CPMO) had experience of the QA process in 2014 and thus had the tools
and techniques to identify and achieve compliance with the manual.
The AQR was conducted using a general principle that an approach would be adopted
only where objective data was available to justify it. Whenever such data was not available a
conservative fall-back assumption was used and applied consistently across the Greek system.
An example of this is the use of loss emergence periods in the collective provisioning
workblock. Loss emergence periods have a direct impact in provisions required for performing
loans. A rebuttable assumption of 12 months was employed, which could only be lowered
where granular, objective data was analysed by the bank team and approved by the ECB to
show that a shorter period was appropriate.
The stress test is a forward-looking exercise that provides insight into the ability of a bank
to withstand pre-defined adverse economic conditions. It should be noted that the stress test
is not a forecast of future events, but rather a prudential gauge of participating banks resilience
under severe but plausible macroeconomic conditions. For example, a number of restrictive
rules were imposed by the stress test methodology that constrict the responses of the
participating banks to stress. These rules enhance the prudential nature of the exercise.
2.6
QUALITY ASSURANCE
This sub-section outlines the process of quality assurance conducted on the AQR and stress test,
including an overview of the types of checks conducted.
To ensure an accurate and timely delivery of the AQR and stress test results, a thorough QA
process has been put in place, involving ad hoc central and onsite ECB / SSM teams composed
of experts from different business areas of ECB and SSM. Those teams were regularly reporting
to the Operational Project Management Meeting and to the Steering Committee. In addition, the
banks themselves made significant efforts to supply data and other requested information to the
necessary standard on time.
The ECB Central QA team was mainly responsible for methodological oversight, crossbank benchmarking as well as performing automated QA checks on the data provided by
the external auditors (AQR) or the banks (stress test).
The SSM on-site QA team (especially devoted to the AQR) contributed heavily to the QA
by performing a file by file review in parallel to external auditors to challenge their work
by performing spot checks and liaising with the auditors on-site.
JSTs contributed with their supervisory experience and engaged with the central and onsite QA teams in the investigation of specific issues
with
the
aim
of
reaching
consensus
between
the
ECB
external auditors.
and
Figure 7
1
2
3
Identification of
QA issue
Desired action
communicated
to bank team
Further bank team
analysis and ECB
consideration
In case of disagreement on the issue from the bank team, the ECB is
willing to consider further evidence and the process would loop back
to step 2
The combined work of the off-site and on-site team gave the ECB a reasonably complete picture
of the work of the external auditors, and the ECB was thus able to be effective in QA despite the
short time-frame.
There were 4 critical elements to the Quality Assurance of the stress test analysis: (1) Basic data
quality checks on all of the input data (2) Comparison of ECB centrally led projections with
bank projections (3) Sense checks on projections based on benchmarking analysis and (4) ECB
expert review. The four elements are discussed more fully below:
1.
Basic data quality checks were performed on all the input data templates. This involved
basic tests e.g. check for segment volumes that dont add up to total; totals on different
templates that should match, do match; no missing data etc. Issues that were identified
were resolved or other prudent adjustments made in ECB projections.
2.
Once the input data set from the banks was verified, the ECB produced its own stress test
projections. To ensure that there were no issues around interpretation of input data in the
calculations the projections were compared to the banks projections. Where there were
divergences between bank and ECB projections that had a material impact on the CET1%,
investigations were carried out with the relevant bank.
3.
A peer benchmarking was carried of the four banks across a wide range of factors.
Unintuitive findings from a relative or absolute perspective were verified. If issues were
confirmed, inputs were revised or other prudent adjustments were made in ECB
projections.
4.
Finally, the stress test projection models were reviewed by a range of different experts
from the ECB not directly involved in the calculations. These ECB staff identified any
issues or concerns and highlighted them to the team managing the models. Any issues were
discussed and errors were fixed. Any alternative propositions on assumptions were
forwarded to the steering committee for decision.
This chapter first shows the aggregate change in available and required capital that was
projected by the comprehensive assessment under the base and adverse scenario, covering both
the AQR and stress test impact. Subsequently, it details the capital shortfall that arises after
comparing the capital impacts against the relevant thresholds.
3.1
The comprehensive assessment capital impact across the four Greek significant institutions is
10.6 billion in the base case and 25.6 billion in the adverse scenario. 9.6 billion represents
that AQR impact (9.2 billion relate to adjustments in the carrying value of assets and 0.4
billion to the additional indirect impact on capital deductions), while the capital impact from the
stress test was 1.0 billion in the base case and 16.0 billion in the adverse, as can be seen in
Figure 8 below.
Figure 8
Baseline scenario
Adverse scenario
14%
14%
12%
12%
10%
-0.3%
8%
6%
4%
10%
-4.2%
8%
6%
12.1%
7.9%
7.6%
-4.2%
7.6%
2%
4%
12.1%
7.9%
-7.8%
2%
0%
0.1%
0%
Starting point
Capital
CET1
depletion from
AQR
Capital
depletion in
Stress Test
Final CET1
Starting point
Capital
CET1
depletion from
AQR
0.1%
Capital
depletion in
Stress Test
Final CET1
Significant AQR findings have been found in this exercise, despite the already material AQR
findings from 2014 being captured in banks accounts. This has primarily been driven by the
deterioration in the macro-economic environment in Greece which has led to higher NPE
volumes as well as lower collateral values and cashflow valuations which has led to material
reductions in carrying values. Additionally, further standardization of the definition of key
Aggregate report on the Greek Comprehensive Assessment, October 2015
19
metrics across the EU has led to further NPE and impairment recognition in the AQR. As an
example the full implementation of the EBA ITS on NPE has meant that forborne cases could
be better identified and tested for impairment. Finally, the fact that tax offsets were not allowed
from the AQR has amplified the findings of the AQR vis a vis 2014.
The AQR part of the comprehensive assessment required changes to asset carrying values of
9.2 billion. This led to an adjustment of the aggregate June 2015 CET1 ratio from 12.1% pre
AQR to 7.8% post AQR (before adjustments to RWA, DTAs and Internal Ratings Based (IRB)
provisioning shortfall). Additionally, RWA was decreased by 8.8 billion, due to the additional
provisions and other appropriate adjustments were made to DTA and other deductions as well
as the IRB provisioning shortfall for banks that are IRB. Post the necessary adjustments
mentioned, the final post AQR CET1 ratio, which also represents the stress test starting point,
has been adjusted to 7.9%. The main drivers of the AQR impact can be seen in Figure 9 below
and are discussed in detail in Chapter 4.
Figure 9
Other adjustments
10,000
AQR adjustment/
capital impact ( million)
9,000
133
8,000
7,000
4,404
6,000
5,000
4,000
1,949
Projection of
findings
2,717
Individually 4,665
assessed during
credit file review
3,000
2,000
1,000
9,069
9,202
CVA
0
Individually assessed Collective provisioning
provisions
1. Additional DTAs (tax offsetting impact) as a result of additional AQR provisions have not been recognised. As a result, additional provisions
identified through the AQR exercise have a direct impact on AQR-adjusted CET1 ratio
The stress test base scenario capital impact across the four Greek significant institutions is 0.3
percentage points. This includes projected capital depletion of 1.0 billion, representing 6.3% of
total post AQR CET1 capital held by the banks at 30 June 2015, and a decrease in RWA of 5.3
billion. The stress test adverse scenario capital impact across the four Greek significant
institutions is 7.8 percentage points. This includes projected capital depletion of 16.0 billion,
representing 99.1% of total post AQR CET1 capital held by the banks at 30 June 2015, and a
decrease in RWA of 23.1 billion.
The main drivers of the stress test results can be seen in Figures 10 and 11 below and are
discussed in more detail in Chapter 5.
Figure 10
30,000
25,000
8,958
20,000
- 8,485
- 2,694
15,000
1,202
10,000
16,179
15,159
5,000
0
Jun-15 post- Cumulative pre- Cumulative
AQR capital provision profits impairments
position
(Jun-15 (Jun-15 Dec-17)
Dec-17)
Other
CRR deductions Dec-17 final
extraordinary
and other
capital position
items & tax1 capital effects 2
1. Other extraordinary items & tax is defined as adjustments of REO carrying values, impairments on goodwill,
profits of participations & methodological adjustments of participations carrying values, and tax
2. CRR deductions and other capital effects is defined as the values which must be deducted from the banks
tangible equity in order to arrive at CET1
Figure 11
25,000
20,000
3,735
15,000
- 17,118
10,000
15,159
16,179
5,000
- 1,926
-726
144
0
Jun-15 post- Cumulative pre- Cumulative
AQR capital provision profits impairments
position
(Jun-15 (Jun-15 Dec-17)
Dec-17)
Other
CRR deductions Dec-17 final
extraordinary
and other
capital position
items & tax 1 capital effects 2
1. Other extraordinary items & tax is defined as adjustments of REO carrying values, impairments on goodwill,
profits of participations & methodological adjustments of participations carrying values, and tax
2. CRR deductions and other capital effects is defined as the values which must be deducted from the banks
tangible equity in order to arrive at CET1
The detailed RWA impact decomposed into the impact from the AQR and the stress test is
shown in Figure 12 below.
Figure 12
250,000
- 8,835
-5,340
200,000
-17,739
150,000
100,000
213,717
204,882
181,802
50,000
0
Jun-15 preAQR RWA
position
AQR impact
Jun-15 postAQR
Stress test
(base) impact
Additional
stress test
(adverse)
impact
Dec-17 final
RWA position
The figures illustrated above are consolidated Group figures, including international
subsidiaries. In the figures below, a set of aggregate projections of key metrics is provided for
the Group and Greece, both for the baseline and adverse scenario. Note only Greek legal
entities were included in the AQR. Also, NPE levels are stated in terms of Basel EAD classified
as non-performing applying a simplified European Banking Authority (EBA) definition, divided
by total EAD (i.e. total loans and receivables and Hold-to-Maturity instruments, including
Sovereign exposures).
Table 2
Group baseline
H1 2015
H2 2015
FY 2016
FY 2017
3.3%
3.1%
3.1%
3.3%
CIR
55.2%
70.2%
58.6%
54.1%
158.3%
155.6%
134.4%
129.8%
H1 2015
H2 2015
FY 2016
FY 2017
2.8%
2.6%
2.5%
2.8%
CIR
56.3%
76.9%
62.9%
57.1%
3.1%
2.4%
1.6%
1.1%
38.4%
39.3%
40.8%
40.2%
51.1%
52.1%
52.7%
54.0%
168.3%
164.1%
136.2%
129.6%
Greece baseline
Table 3
Group adverse
H1 2015
H2 2015
FY 2016
FY 2017
3.3%
2.5%
2.5%
2.5%
CIR
55.2%
98.9%
73.4%
71.9%
158.3%
152.5%
134.9%
124.8%
H1 2015
H2 2015
FY 2016
FY 2017
2.8%
1.9%
1.7%
1.8%
CIR
56.3%
137.2%
88.5%
86.4%
3.1%
3.4%
2.7%
2.1%
38.4%
40.0%
44.1%
46.1%
51.1%
54.1%
53.9%
54.4%
168.3%
160.7%
138.0%
124.5%
Greece adverse
3.2
SHORTFALL IDENTIFIED
Each bank in the comprehensive assessment was required to maintain a 9.5 % CET1 ratio after
accounting for the effect of AQR results on their mid-year 2015 balance sheet. Each bank was
also required to maintain a 9.5% CET1 ratio at each year-end during the baseline stress test
scenario, and an 8% CET1 ratio at each year-end during the adverse stress test scenario.
As discussed in the previous chapter, the total projected change in CET1 from the
comprehensive assessment in the adverse scenario is 25.6 billion (including AQR
adjustments). Moreover, the RWA decreased by 31.9 billion in the adverse scenario to 2017
(as a result of deleveraging and provisioning, including also AQR adjustments), decreasing the
capital requirements. Offsetting this impact is the excess capital held by the participating banks.
Figure 13
16
Shortfall ( billion)
14
12
+1.0
10
8
14.4
6
4
2
4.4
3.4
0
Shortfall 9.5% post AQR
baseline (Jun-15)
Max. shortfall 9.5% over period Max. shortfall 8.0% over period
in ST baseline (incl. AQR)
in ST adverse (incl. AQR)
As shown above the total shortfall can be disaggregated into the main components of the
comprehensive assessment by identifying:
Shortfall from the AQR this is the aggregate shortfall due to the AQR adjustments
applied to the June 2015 capital positions of the banks
Shortfall from the stress test this is the aggregate shortfall (under the baseline and
adverse scenarios, measured against their respective thresholds) using the stress test
results, applied to the post AQR capital positions of the banks
AQR OUTCOMES
This chapter provides detail on the key results of the AQR and the AQR total impact for each
individual workblock and across the four Greek significant banks.
The AQR component of the comprehensive assessment involved performing a detailed assetlevel review of the in-scope portfolios, in line with current accounting and prudential regulation
set out in CRR / CRD IV capital rules. In some areas the ECBs methodology involved
additional prudential prescription to accounting concepts in order to achieve consistency and
adequate conservatism. AQR adjustments should therefore not be interpreted as breaches to
accounting rules.
As described in Chapter 1 and 2 in this report, the comprehensive assessment of the Greek
banking system in 2015 was primarily a response to the market environment in Greece upon the
3rd MoU between Greece and the Institutions. The results are of a prudential nature, in order to
assess the banks ability to withstand the weak economic environment.
Additionally, it should be noted that, while the banks reflected an important part of the AQR
adjustments from the comprehensive assessment 2014 in their accounts, the worsening of the
market conditions meant that additional AQR findings were not unexpected, particularly given
the prudential nature of the exercise.
The detailed outcomes of the 2015 AQR on the Greek significant banks is presented in the
following sub-chapter.
4.1
The total adjustment to the carrying amount of loan portfolios and fair values of derivatives
(CVA) was 9,202 million. This is shown by participating bank in the chart below:
Figure 14
4,000
3,500
5%
6%
5%
4%
4%
3,000
4%
3%
2,500
3%
2,000
3,213
1,500
1,000
2,300
2,337
1,906
1,746
2%
Total adjustment in
% of starting RWA
4,500
1%
500
Piraeus Bank
Alpha Bank
Eurobank
NBG
0%
Average
Note: Gross adjustment excludes any offsetting tax and risk protection effects. Adjustments include both additional provisions and
CVA adjustment
Table 4
Bank name
Pre-AQR
CET1
Collective
Provisioning
CVA
adjustments
Post AQR
CET1 before
adjustments
to RWA,
DTAs and
Tax/Risk IRB
Total AQR Protection provisioning
shortfall
adjustCET1
1
(Jun-15)
adjustments ments
in MM
in MM
in MM
in MM
in MM
in MM
Alpha Bank
6,792
-531
-290
-904
-22
Eurobank
5,389
-403
-286
-1,187
NBG
7,412
-692
-334
-1,311
Piraeus Bank
6,189
-1,091
-1,039
Totals
25,781
-2,717
-1,949
in MM
in MM
-1,746
5,045
-30
-1,906
3,483
-2,337
5,075
-1,002
-81
-3,213
2,976
-4,404
-133
-9,202
16,579
1. Offsetting impact is zero given no DTAs from AQR were allowed to be recognised
The following chapters provide further information on the drivers of AQR adjustments,
disaggregating the result into its three major components:
1.
Additional provisions resulting from the non-performing non-retail debtors from the
risk-based sample that were individually assessed and then projected to the rest of the
portfolio.
2.
3.
Additional adjustments of CET1 capital through the CVA challenger model impact.
Figure 15
10,000
Other adjustments
AQR adjustment/
capital impact ( million)
9,000
133
8,000
7,000
4,404
6,000
5,000
4,000
1,949
Projection of
findings
2,717
Individually 4,665
assessed during
credit file review
3,000
2,000
1,000
9,069
9,202
CVA
0
Individually assessed Collective provisioning
provisions
1. Additional DTAs (tax offsetting impact) as a result of additional AQR provisions have not been recognised. As a result, additional provisions
identified through the AQR exercise have a direct impact on AQR-adjusted CET 1 ratio
Each component of the AQR had several drivers which are discussed in more detail in the
following chapters first the individual specific provisioning assessment (credit file review,
collateral valuation and projection of findings) is discussed in Chapter 4.2, followed by the
collective provisioning assessment in Chapter 4.3 and the CVA challenger model adjustments in
Chapter 4.4.
As part of the credit file review the adjustment to provisions for non-performing debtors can be
assessed under two approaches: Gone concern which relies on a final sale value of collateral
exposures and going concern where operating cash flows are used to assess a prudent net
present value (NPV) of future cash flows. In the majority of cases the gone concern approach is
implemented given it is a more robust and prudent perspective. It is important to note that the
rationale for implementing these approaches is based on a means to measure a prudent
provisioning adjustment only and not a recommendation for bank strategy. Specifically, gone
concern represents a view that the prudent provision is based on the realisable collateral values,
but it may be that the bank shareholders continue to believe that better returns are available
through restructuring and subsequent sale of the business or curing of the loan. The percentage
of gone concern debtors should not be seen therefore as an estimate of the percentage of
companies or individuals to be liquidated but the percentage of companies or individuals who
should be provisioned down to a level based on the available collateral.
Table 5
AQR component
Individually assessed provisions
Credit File Review
Adjustment
Section
4,665 million
4.2
2,717 million
4.2.1
232 million
770 million
1,585 million
130 million
Projection of Findings
1,949 million
4.2.2
4,404 million
4.3
3,516 million
888 million
CVA
133 million
4.4
133 million
9,202 million
4.2
This chapter provides a detailed view on additional provisions identified for individually
assessed non-retail debtors from the credit file review, collateral valuation and projection of
findings workblocks. The credit file review assessed the chosen samples of non-retail debtors
mainly in terms of (a) performing classification status and (b) need for additional provisions
(Chapter 4.2.1 below), with findings being projected to the remaining unsampled debtors per
portfolio according to pre-defined criteria (Chapter 4.2.3). Residential Real Estate facilities were
assessed for their performing status only as an input to collective provisioning (Chapter 4.3).
4.2.1
Following the review of all 2,528 sampled non-retail debtors across the Greek significant
institutions, the overall impact of the credit file review was an increase in provisions in the
sample of 2,717 million from 7,513 million to 10,229 million. Given that provisions relate
to NPE (existing or reclassified), all provisioning adjustments under the credit file review were
specific.
This increase in provisions by asset class is shown below.
5,000
20%
4,500
18%
4,000
16%
13%
3,500
14%
12%
3,000
10%
2,500
4,454
2,000
4,110
1,500
3,268
3,037
1,000
5%
4%
8%
6%
2%
500
605
930
603 738
4%
Additional provisions in
percent of cRWA
Figure 16
2%
0%
0
Large corporates
Large SME
Shipping
Pre-AQR provisions
Post-AQR provisions
Total provisions adjustments in % starting cRWA
Reclassified NPE
1,003 million of additional provisions in the credit file review was due to the reclassification of
240 debtors in the sample representing 16% of total non-retail debtors that were originally
classified as performing. As shown in Figure 17, across the Greek system, the participating
banks average proportion of non-retail reclassified debtors ranged from 8% to 22% (in the
sample). For residential real estate (RRE), the proportion of reclassified debtors ranged between
2% and 43% (in the sample). Please note that the sample has been selected on a risk-based
approach, i.e. simple averages across the sample are shown for illustrative purposes only but do
not necessarily reflect projection effects. See the relevant sections in this chapter for details.
Figure 17
50%
45%
40%
35%
30%
25%
43%
20%
15%
22% 22%
22%
10%
16%
5%
20%
12%
8%
11%
2%
0%
Average
Alpha Bank
Eurobank
Non-retail
RRE
NBG
Piraeus Bank
Table 6
Trigger
Times hit
% of reclassified debtors
169
70.4%
Forborne NPE
132
55.0%
Change in EBITDA
59
24.6%
38
15.8%
Change in equity
30
12.5%
Loan-to-Value
16
6.7%
Emergency funding
15
6.3%
2.9%
Probability of Default
Note: In a number of cases debtors will hit more than one trigger. The assessment whether a reclassification to NPE is required is based on a
holistic view of all triggers that impact a debtor simultaneously
Within the non-retail reclassified debtors, 63% of the debtors were treated under the gone
concern approach and 37% were assessed as going concern. Additionally, for the RRE portfolio
further reclassifications were primarily due to forbearance triggers.
The total change in provisions that were found for reclassified debtors is shown below by AQR
asset segment:
Figure 18
36%
34%
35%
500
30%
400
24%
25%
300
200
17%
424
20%
15%
431
10%
100
5%
60
86
Shipping
0%
0
Large corporates
Large SME
600
Existing NPE
For the 973 of debtors that remained NPE, the provisioning approach assessment determined
that 78% of debtors were to be treated under the gone concern approach and 22% under going
concern approach. The total adjustment amounted to 1,713 million.
Figure 19
55%
800
700
50%
45%
38%
37%
600
40%
500
30%
400
900
763
639
300
20%
200
10%
265
100
48
0
Large corporates
Large SME
0%
Shipping
Coverage ratio (%)
Note: Negative numbers imply a reduction provisions for given debtor(s) due to the AQR.
4.2.2
The collateral and real estate valuation workblock was run during the credit file review process
and was relevant for NPE gone concern debtors for which collateral liquidation was the more
likely workout strategy i.e. collateral valuation reductions led to increased provisions. In
addition, the workblock provided updated real estate property values and appraisal haircuts to be
used in the loss given loss (LGL) calculation for residential real estate portfolios in the
collective provisioning workblock.
Throughout the review, approximately 11,826 collateral items with a total value of 21,444
million were investigated. For 16% of retail collaterals and all collaterals in the shipping
portfolios, a full revaluation was carried out as part of the AQR.
Across the Greek banking system, collateral values were adjusted downwards by 2,567 million
representing an approximate 12% decrease compared to previous bank internal valuations. This
decrease was driven by a change in price indices as well as by changes due to AQR
revaluations. Figure 20 illustrates the impact of both effects on the aggregated collateral value.
For collateral items that required a reappraisal, the total adjustment can be split into a change in
property value due to indexation and an incremental change as a result of the revaluation.
Figure 20
22,000
-4%
21,500
-8%
21,000
20,500
20,000
19,500
21,444
20,568
19,000
18,500
18,877
18,000
17,500
Aggregate collateral value
pre-AQR
The collateral value reduction most severely affected commercial real estate (CRE) in absolute
terms while land and residential real estate in relative terms.
Figure 21
600
500
30%
25%
22%
400
300
200
20%
654
15%
9%
8%
364
8%
356
10%
291
100
5%
700
125
0
0%
CRE
Land
RRE
Shipping
Other RE
Note: Positive numbers imply a collateral value reduction. Numbers in this chart do not include other collateral outside of other RE
4.2.3
PROJECTION OF FINDINGS
The findings of the sample from the credit file review were extrapolated to the unsampled
population of each portfolio, which led to additional 1,949 million of provisions. Findings
were extrapolated both on NPE classification and required provisions. The effect of projection
to the provisioning adjustments made during the credit file review can be seen below.
Figure 22
14,000
Provisions ( million)
12,000
1,949
10,000
2,717
8,000
12,178
6,000
4,000
7,513
2,000
0
Total provisions preAQR
Additional provisions
CFR
Projected additional
provisions CFR
The amount of additional provisions from the credit file review of the sample compared to the
projection of the sample varied widely by portfolio for two main reasons:
Sample coverage variation portfolios with a large number of debtors naturally had a
lower coverage rate (e.g. Large SME portfolios), and hence a proportionally larger
exposure on which to project
The relative increase of additional provisions due to projection of findings vs. credit file review
(sample) was relatively stable when compared with the 2014 CA (2.1 billion on the sample and
1.8 billion projected over the remaining portfolio in the 2014 CA vs. 2.7 billion and 1.95
billion respectively as per Figure 22 above).
The following shows the absolute contribution from the sample and projection by type:
Table 7
Asset segment
Sampling
exposure
coverage rate (% )
Additional
provisions CFR
( million)
Projected
additional
provisions CFR
( million)
Total additional
provisions
( million)
Large SME
20-40%
1,072
1,472
2,544
70-90%
326
122
448
Large corporates
70-90%
1,186
340
1,526
70-100%
132
15
148
2,717
1,949
4,665
Shipping
Total
Note: Sampling exposurecoverage rate is applied to the total exposure of all four banks
Figure 23
80%
70%
5%
5%
60%
50%
40%
6%
70%
3%
61%
30%
7%
20%
35%
33%
22%
10%
0%
Real estate related Large corporates
(non real estate)
4.3
Residential real
estate (RRE)
Shipping
If the bank team's estimate was 5-10% higher than the bank's estimate, the bank team investigated reasons across
data or methodology to explain the differences and applied discretion in selecting challenger model estimate vs.
banks estimate
Across all participating banks and portfolios, the collective provisioning workblock identified
the need for additional collective provisions of 4,404 million, 3,516 million of which
represent additional specific provisions for retail debtors and 888 million of which represent
additional IBNR. In relative terms, across the participating banks, this translates into an increase
in IBNR of roughly 47% and an increase in specific provisions of about 18%. The main drivers
of the increases due to cases of significant deviation are explained later in this chapter.
Figure 24
Specific provisions
IBNR
18%
3,516
Provisions ( million)
20,000
20,000
15,000
15,000
22,685
10,000
10,000
19,169
5,000
5,000
47%
888
Pre-AQR
specific
provisions
Additional
specific
provisions
Post-AQR
specific
provisions
2,779
1,892
Pre-AQR
IBNR
Additional
IBNR
Post-AQR
IBNR
Figure 25
4,000
12%
11%
3,500
10%
3,000
2,000
6%
3,087
1,500
4%
3%
1,000
2%
2%
500
0
Residential real estate
250
179
Retail SME
Other retail
Additional specific
provisions as % of cRWA
8%
2,500
0%
The challenger model provisions on non-performing exposures and, hence, additional specific
provisions revealed by this workblock are significantly impacted by a number of key drivers.
NPE identification: As outlined in Chapter 4.2.1, the stock of NPE was significantly
increased by the AQR, through the credit file review. Also, cure rates were adjusted
based on CFR findings. As a result, the basis for the calculation of specific provisions
on residential real estate increased, ultimately leading to an increase in specific
challenger model provisions and, in many cases, to additional specific provisions to be
recognised for the purpose of the AQR. As a result, the cure rate on the defaulted
population was adjusted down due to NPE reclassification.
Adjustment of RRE collateral: For residential real estate portfolios, property values
are an essential part of the loss given loss calculation, as the expected loss is calculated
using projected proceeds from a foreclosure of the underlying property. As part of the
collateral valuation workblock, properties were reappraised and indexed forward in
order to adjust the collective provisioning LGL directly. For properties that were not
reappraised, the average appraisal discount of the portfolio was applied. In addition,
expected proceeds from collateral were calculated based on assumptions representative
of the current conditions. Two key parameters used in the LGL calculation, sales ratio
and time to sale were harmonised across the banks in order to reflect current conditions
and to prudently account for obsolete / biased data. Specifically, these required the use
of the fall back value for the sales ratio of 75% - implying a haircut on collateral values
of 25%
Aggregate report on the Greek Comprehensive Assessment, October 2015
40
Parameters that are not point-in-time: In a number of cases, the banks models were
smoothing the impact of recent events, for example, through the use of long term
parameters. Due to its point-in-time nature, the challenger model reflects the current
conditions.
The IBNR increase is attributable to 16 of these 19 portfolios. For the remaining 3 portfolios, the insufficient
levels of IBNR were absorbed by a surplus identified in specific provisions for these portfolios
Figure 26
600
400
1.2%
0.9%
365
0.8%
300
0.8%
0.6%
232
0.6%
182
200
0.3%
0.4%
0.2%
100
57
21
20
Retail SME
Other retail
1.0%
Additional IBNR
as % of cRWA
1.4%
500
0.2%
0.0%
Large SME
Large
corporates
Additional IBNR
Residential
real estate
Shipping
The challenger model provisions on performing exposures and, hence, additional IBNR
revealed by this workblock is significantly impacted by a number of key drivers.
Credit file review: For all sampled debtors / facilities a classification review was
conducted during the credit file review. This review had an impact on the parameters
for both retail and non-retail portfolios.
-
Where debtors / facilities were stated to be performing in the banks loan tape were
found to have defaulted in the credit file review, this increased the probability of
impairment applied in the challenger model and additional NPE were projected;
also cure rates were adjusted for retail portfolios.
An increase in the loss given impairment (LGI) was applied to reflect increased
provisioning levels for debtors that had LGI is calculated as the coverage ratio of
the defaulted asset applied in the challenger model
Both of these effects led to an increase in the IBNR determined by this workblock.
Probabilities of impairment (PI) that are not point-in-time: As for retail cure rates,
in a number of cases the banks' models use long term parameters whereas the challenger
model is fully point-in-time. Where banks exhibited a large number of defaults in the
period examined, this led to higher probabilities of impairment and, ultimately, in an
increase in IBNR.
Loss emergence period (LEP): In the majority of cases the LEP employed in the
challenger model exceeds the one applied by participating banks. This is partly driven
by the prudential nature of the exercise and partly by the fact that in many cases an
analysis of the bank's LEP revealed a high volatility of observed LEPs as well as loss
events, i.e. the beginning of the LEP, not being identified correctly. As raising the LEP
has a one-to-one impact on IBNR, this was an important driver of the additional IBNR
resulting from this workblock.
Adjustment of RRE collateral values: The reduction in residential real estate property
values due to forward indexing, sales ratio and time to sale raised LGL and led to an
increase in IBNR.
4.4
The four participating banks in the AQR were included in the CVA challenger model review
(one model per bank was reviewed), given the size of their derivative exposures (fair value:
2.725 million), 65% of which relates to sovereign counterparties. The process included a
template, which identified potential issues in the calculation methodology, parameters and
portfolio coverage.
The AQR impact from the CVA challenger model was 133 million, corresponding to a 32%
increase of an initial CVA amount of 412 million. A majority of this adjustment related to the
use of inappropriate PD and LGD parameters specifically for the Greek Sovereign, where banks
used lower LGD assumption and PDs that were not market implied. Additional adjustments
occurred due to exposures excluded by the banks from their CVA calculation.
The distribution of CVA adjustments across entity groups is shown in Figure 27.
Figure 27
250
126%
140%
120%
30
100%
150
80%
100
22
50
26%
87
81
165
60%
40%
96
18%
64
20%
0%
0%
0
Alpha Bank
Original bank CVA
Eurobank
NBG
Piraeus Bank
AQR adjustment
Relative adjustment
in % of initial CVA
CVA ( million)
200
STRESS TEST
This chapter discusses the methodology, underlying scenarios and outcomes of the stress test.
The stress test results are analysed in detail, mainly structured around the key drivers that
impacted the outcome.
5.1
This chapter captures the methodology for the 2015 stress test of Greek banks. The chapter lays
out the high level structure of the modelling approach and then drills down into all of the key
components. The drill downs cover the detailed approach; the key assumptions and their source.
5.1.1
CONTEXT
The 2015 stress test of Greek banks follows a centrally led top-down approach based on banks
data and considers baseline and adverse scenarios spanning from 30 June 2015 to 31 December
2017 for a total of 2.5 years. The stress test methodology combined system-level and bank-level
parameters to project balance sheet, profit and loss, and solvency position in annual increments
(6 months for 2015). The stress test involved a constrained dynamic balance sheet approach
that is to say the balance sheet evolves allowing for new lending, deposit evolution, asset sales
included in the DG Competition restructuring plans etc. Constraints were set by the ECB,
consistent with the assumptions underlying the wider Greek programme. The constraints were
differentiated between the baseline and adverse scenario.
The projections used to set the capital need were produced in a way that is consistent with the
prevailing accounting and prudential rules. This means that provisions were projected in a way
consistent with IAS 39; RWA projections reflected the standardised or IRB approach
whichever was used by the bank; capital projections were consistent with CRD IV, including
phase in rules; and other financial projections were consistent with prevailing accounting rules.
5.1.2
The key high level assumptions for the exercise were defined as follows:
Macroeconomic series for Greece and other relevant countries were set by the ECB and
are included in Chapter 5.2
The CRD IV CET1 capital definition has been used with phase-in rules. The
minimum capital hurdles for the baseline and adverse scenarios are 9.5% and
8%, respectively
Loan volume projections were defined at a system level by the ECB down to a banklevel using fixed market shares as per June 2015
Bank level deposits were projected based on system level projections set by the ECB
and allocated to banks based on fixed market shares as per June 2015
The eligibility of relevant collateral for Monetary Policy Operations (MPO) was
assumed to be restored from 31 December, 2015. Use of MPO was limited to the
amount of available collateral
Sovereign debt holdings were assumed to remain flat over the stress test horizon; T-Bill
holdings assumed to roll over at current levels
The valuation of participations was adjusted in line with the Greek market index
(excluding Greek banks)
5.1.3
DTAs: No new DTA formation was allowed in either stress test scenario
The high level modelling approach was as per the diagram below
Figure 28
P&L and
balance
sheet
projection
+ Balance
Sheet
projection
H1 2015
H2 2015
2016
2016
2017
2017
CET 1%
CET1
%
XX
XX
XX
XX
RWA
XX
XX
XX
XX
Balance Sheet
Assets
XX
XX
XX
XX
Liabilities
XX
XX
XX
XX
Equity
XX
XX
XX
XX
XX
XX
XX
XX
Expenses
XX
XX
XX
XX
Retained earnings
XX
XX
XX
XX
A projection of the P&L and balance sheet was produced based on the domestic business of the
Greek banks. The projection is consistent with the assumptions described above. A projection of
the marginal impact on the capital position of the foreign subsidiaries was also included,
reflecting current agreed restructuring plan commitments. The CET1 ratio projection considered
the roll forward of capital at the Group level in a mechanical manner, fully reflecting CRD IV
phase in and other assumptions defined by the ECB. Detailed drill down analysis was
performed on Net Interest Income, Other Operating Income and Expenses, RWA and
provisions. The drill down analysis fed the P&L and balance sheet projection.
In the following sub-chapters, detail is provided on the approach to projecting NII, volumes,
credit losses and RWA for the Greek portfolios.
5.1.4
The high level approach to projecting Net interest income involved the following steps:
1.
Project loans and deposits based on system level loan and deposit assumptions (as
described previously).
10
2.
Split lending into front book and back book and performing and non-performing
projections (NPE rates coming from the credit loss module).
3.
Project the composition of the remainder of the liability side (wholesale and Eurosystem
funding).
4.
5.
Calculate an adjustment for movements in income due to the structural interest rate riskposition.
5.1.4.2 Split the lending book into front and back book and performing
and non-performing
Next, the migration rates to NPE from the credit loss module (described in Sub-chapter 5.1.5)
were used to project the percentage of the portfolio that was back book and NPE. The total
exposure was also reduced in line with recent observed amortisation and prepayment.
The gap between the projected total exposure and the projected back book net of amortisation
and prepayment was filled with new lending. New lending was assumed to be Normal Risk
according to the AQR definition (as stated in the AQR Manual). NPE migration and
amortisation and prepayment for new lending were projected in line with what was observed for
Normal Risk customers on the back book.
A percentage of NPE was moved to write-off, so that in the baseline, income from nonperforming exposures was not overestimated.
Figure 29
Retail mortgages
Large SMEs
Total lending
New book
Performing
NPE
Total lending
Performing
New book
Performing
NPE
Performing
Back book
Back book
Non-performing
(NPE)
Non-performing
(NPE)
Write-offs
Write-offs
Jun-15
2015
2016
2017
Jun-15
2015
2016
2017
Table 8
Type of asset
Spread assumption
Performing
Credit
Front book: The lower of bank plans and the observed recent front book spreads
exposure
NPE
Type of asset
Spread assumption
mortgages (where moderate discount unwind is included on exposures with greater
than 6 months in arrears)
Back book: Spreads remain unchanged
Deposits
Front book: The lower of bank plans and the recent observed front book spreads
Wholesale
MPO
5.1.5
Fees and commissions, other income and administrative expenses were projected using recent
historic experience as the starting point. The starting point was projected forward, adjusting
levels based on relevant drivers that have a direct influence on the level of fees and
commissions, other income and expenses. For instance, expenses are clearly driven by full-time
equivalent (FTE) staffing numbers. Drivers were defined in the restructuring plan (e.g.
branches) or were calculated directly in the model elsewhere (e.g. new lending volumes). The
sensitivity of fees and commissions, other income and expenses to the drivers was set based on
expert judgment informed by historical experience of the banks and the banks plans.
5.1.6
The approach to projecting NPE volumes and impairment for most portfolios followed a 7 step
process:
1.
Observe recent historic NPE migration behavior and create migration matrices
by segment
2.
3.
4.
Further adjust migration matrices for the impact of capital controls and set the level of
this adjustment
5.
Project impairment rates for new NPE, conditioning for macroeconomic factors, capital
controls and other factors. Apply impairment rates to volume of new non-performing
loans to measure impairment flows
6.
7.
The key steps are described in further detail in the sections that follow.
There were two major exceptions to the approach described above:
For some concentrated portfolios (e.g. shipping and CRE), stable migration matrices
could not be applied. For these portfolios, a structural approach was used which
involved projecting the key risk drivers of the underlying debtors (e.g. loan to value
(LTV) and debt-service coverage ratio (DSCR)) and applying a default test to assess
whether the specific debtor would default in the scenario.
For sovereign portfolios in the loans and receivables and hold to maturity portfolios, in
the base case, no losses were assumed. In the adverse case, a reduction to the carrying
value was assumed in a manner that was consistent with the 2014 EBA exercise.
calibrate the projection. The analysis involved measuring the migration behaviour of each
exposure and using it to assess the probability of an exposure migrating in the future. The
analysis was differentiated by segment i.e. migration rates were analysed separately for retail
mortgages, corporate SME etc. Where appropriate, further segmentation was applied (e.g.
differentiating retail mortgages by LTV).
The output was a set of migration matrices that aligned with the risk classifications for the
AQR. This allowed for a pragmatic analysis (i.e. the granularity was not excessive). An
illustration of the output is shown below.
Figure 30
December 2012
1 Very low risk (excluded from AQR)
2 Normal risk
3 Normal risk (previously cured)
4 High risk
5 High risk (previously cured)
6 NPE (less than 6 months arrears
7 NPE (6-12 months arrears)
8 NPE (12-24 months arrears
9 NPE (24+ months arrears)
90%
8%
90%
1%
7%
70%
10%
19%
60%
5%
50%
2%
5%
3%
3%
4%
2%
5%
6
1%
2%
10%
25%
40%
45%
6%
4%
1%
1%
5%
5%
40%
15%
6%
10%
10%
60%
15%
15%
10%
70%
70%
December 2014
December 2013
Transforming the default rate metric into a credit quality indicator by taking the
normal inverse of the default rate
2.
3.
Comparing the models that are most predictive and exclude those which have
unintuitive relationships between the relevant macroeconomic factors and default rates
4.
Comparing the resulting set of models based on expert knowledge and existing
benchmarks
5.
Testing that the macroeconomic model results were consistent with the default rates
observed in the granular loan tape data between December 2012 and June 2015 and
adjust accordingly.
Adjustments were done based on the Merton model as illustrated in the diagram below:
Figure 31
Figure 32
Performing
exposure of
portfolio by
risk rating
(t-1)
Performing
exposure of
portfolio by
risk rating
(t)
Migration matrix
New
defaults (t)
LGI (t)
Impairment
flow
Table 9
Dec-15
Dec-16
Dec-17
PE
100
90
80
75
NPE
10
15
17
20
PD PIT
2%
4%
3%
2%
LGI PIT
45%
50%
45%
40%
IBNR stock
0.9
IBNR flow
1.8
1.08
0.6
0.9
-0.72
-0.48
Adjust NPE stock for write-offs (in order to feed the interest income from
NPE calculation)
As described above, income from NPE was calculated based on the NPE stock. Clearly, the
NPE stock reduced over time with write-offs. Therefore, it was important that the NPE stock
was reduced appropriately so that the income from non-performing exposures was reduced
appropriately.
5.1.7
Credit RWA was projected for all portfolios included in the AQR for the exercise reflecting key
dynamics:
EL parameters were updated and the IRB provisioning shortfall adjusted in line with
movements in provisions as described in the previous chapter
RWA for operational risk was projected so that it is proportional with the average of
gross income over the last 3 years
5.2
SCENARIOS
The following chapters provide details on the baseline and adverse scenarios underlying the
stress test in the Greek comprehensive assessment 2015, which were defined and / or agreed
under the third Financial Assistance Programme for Greece. Key macrofinancial variables for
the Greek economy as well as financial market shocks in Greece under the adverse scenario for
the years 2015 to 2017 that were used in the exercise are displayed in the tables below.
Table 10
Baseline scenario
Adverse scenario
Level
deviation
2015
2016
2017
2015
2016
2017
from
baseline
(2017)
-2.3
-1.3
2.7
-3.3
-3.9
0.3
-5.9
-0.4
1.5
0.9
-0.7
0.6
-1.0
-3.0
Unemployment rate
(end-of-year, %)
26.9
27.1
25.7
27.3
28.1
27.5
1.8
-7.5
-5.0
-1.0
-7.8
-8.8
-7.8
-10.9
Prime commercial
property price growth
(%)
-3.4
-1.2
1.1
-3.6
-3.4
-2.1
-5.5
Note: Level deviation from baseline (2017) for unemployment rate (end-of-year,%) is given in percentage points,
otherwise level deviation from baseline (2017) is given in percent relative to baseline.
Table 11
2015
2016
2017
40
80
80
204
390
170
-10.5
-20.1
-8.8
Note: Interest rate differentials are in basis points, stock prices are in percent
5.2.1
BASELINE SCENARIO
The baseline scenario is consistent with the assumptions of the third economic adjustment
programme for Greece. As such it is assumed to capture the expected impact of the recent
developments in Greece, in particular of the June-July 2015 bank holiday, the introduction of
the capital controls, and the fiscal measures introduced under the agreed programme.
5.2.2
ADVERSE SCENARIO
The adverse scenario is based on the premise that downside risks materialise in the external
environment of the Greek economy, as well as recent political and economic developments in
Greece triggering a shock to consumer and business confidence.
Specifically within the Greek economy, consumer and business confidence would sharply drop
in response to the imposition of deposit withdrawal and cross-border payment restrictions
(beyond expected impact already included in the baseline scenario), turbulent conclusion of the
financial assistance programme, as well as to the increased political uncertainty. This would
manifest itself through a reduction in private consumption and fixed investment. It is however
assumed that, due to timely programme implementation, domestic aggregate demand would
stabilise, at the lower level, in the final year of the scenario horizon. Impaired confidence would
propagate to the property market, leading to a reduction in demand for residential and
commercial property, which would drive the property prices lower with respect to the baseline
scenario.
As for the external factors, the scenarios are based on assumptions such as an exogenous global
shock originated in the US, leading to deteriorating economic conditions in the rest of the world
through credit spread increases (on sovereigns, financial and non-financial firms), and a stock
price shock in developed economies, and capital outflows in emerging markets. EU countries
other than Greece are affected via trade and financial market spill overs.
The resulting impact on the Greek economy would be sizeable. By 2017, GDP would be lower
by 5.9% compared to the baseline projection and unemployment rate would stand at 1.8
percentage points above the baseline level.
5.3
The aggregate impact of the stress test in terms of percentage point changes between June 2015
(post AQR) and year-end 2017 in the average CET1 ratio of participating banks is a decrease of
0.3 percentage points under the baseline and a decrease of 7.8 percentage points under the
adverse scenario.
Overall, the impact on CET1 capital over the two and a half years stress test horizon amounts to
a decrease of 1 billion under the baseline and of 16 billion under the adverse scenario.
The aggregate impact of the stress test by risk driver under the baseline scenario is shown in
Figure 33. In the baseline, the four Greek banks average CET1 ratio is projected to decrease
from 7.9% in the second quarter of 2015 (post AQR), to 7.6% by the end of 2017. The solvency
position under the baseline decreases mainly due to the projected loan losses (the 4.8 percentage
point total impairments effect shown in the chart below) and other operating expenses (6.6
percentage point contribution to the change in the CET1 ratio). The average development of
participating banks solvency positions, however, masks variations across individual
institutions.
Figure 33
Cumulative stress test impact on the CET1 % by risk driver under the
baseline scenario
12%
1.7%
10%
-6.6%
8%
6%
9.6%
4%
-4.8%
2%
0%
-0.9%
-2%
Net interest
income
Non-interest
income
Operating
expenses
Impairments
0.6%
-0.3%
0.2%
3
Other P&L
Impacts
Other CET1
elements 2
RWAs
Overall
cumulative
impact on CET1
ratio
-0.4%
-1.7%
-1.1%
-0.4%
-0.4%
2.5%
0.2%
-0.4%
0.4%
-0.1%
0.3%
0.3%
-0.1%
-0.8%
0.2%
-0.3%
Under the adverse scenario, participating banks' average CET1 ratio is projected to decrease
from 7.9% in the second quarter of 2015 (post AQR), to 0.1% by the end of 2017. This
corresponds to a decline of the average CET1 ratio between H1 2015 (post AQR) and year-end
2017 of 7.8 percentage points (see Figure 34).
The key driver of the CET1 ratio impact is the increase in loan losses (8.4 percentage point
contribution). Furthermore pre-provision profits are also lower compared to the baseline,
primarily driven by NII (contribution to CET1 ratio is reduced from 9.6 percentage points under
base to 7.1 percentage points under adverse). The Net Trading Income and 'Administrative and
other expenses' risk driver also impacts the overall results; however, they remain largely
unchanged between the baseline and adverse scenario. The mark to market of Sovereign
exposures in the available-for-sale (AFS) and fair value option (FVO) portfolios contributes to
the capital depletion, albeit the impact is less material than the impairment of held-to-maturity
Greek Sovereign exposures. Note in the adverse case, an adjustment to carry values of Greek
sovereign loans and receivables and hold to maturity exposures is included consistent with a
writedown of 14% over 2.5 years.
Figure 34
Cumulative stress test impact on CET1% by risk driver under the
adverse scenario
12%
10%
8%
1.4%
6%
4%
- 6.6%
7.1%
2%
0%
-2%
-9.1%
-4%
-7.8%
-6%
0.0%
-8%
-10%
Net interest
income
Non-interest
income
Operating
expenses
Impairments
-0.3%
-0.4%
Other P&L
Impacts
Other CET1
elements 2
RWAs
0.2%
-0.9%
-0.5%
0.0%
-1.4%
1.0%
-0.7%
-0.6%
0.2%
0.0%
0.2%
-0.3%
Overall
cumulative
impact on CET1
ratio
-7.5%
-8.4%
-7.3%
-7.8%
APPENDICES
6.1
DETAILED RESULTS
AQR adj.
H1 2015
Post CA 2015
Baseline
Post CA 2016
Baseline
Post CA 2017
Baseline
Post CA 2015
Adverse
Post CA 2016
Adverse
Post CA 2017
Adverse
Table 12
Alpha Bank
12.7%
9.6%
9.0%
9.3%
9.6%
5.1%
3.8%
2.1%
Eurobank
13.7%
8.6%
8.8%
9.1%
8.7%
5.7%
4.1%
1.3%
NBG
11.6%
8.1%
6.8%
7.6%
7.3%
2.6%
1.3%
-0.2%
Piraeus Bank
10.8%
5.5%
5.6%
5.3%
5.2%
2.3%
-0.1%
-2.3%
System wide
12.1%
7.9%
7.4%
7.7%
7.6%
3.8%
2.1%
0.1%
Name
AQR adj.
H1 2015
Post CA 2015
Baseline
Post CA 2016
Baseline
Post CA 2017
Baseline
Post CA 2015
Adverse
Post CA 2016
Adverse
Post CA 2017
Adverse
Table 13
Alpha Bank
1,708
73
-263
-113
31
-1,408
-1,980
-2,743
Eurobank
1,663
-339
-253
-158
-270
-796
-1,279
-2,122
NBG
1,344
-831
-1,576
-1,069
-1,344
-2,932
-3,449
-4,602
763
-2,188
-2,034
-2,133
-2,213
-2,853
-3,937
-4,933
Name
Piraeus Bank
Table 14
Residential
real estate
Other retail
Corporates
CVA
Total AQR
adjustment
816
n/a1
908
22
1,746
271
700
200
705
30
1,906
NBG
966
1371
2,337
Piraeus Bank
787
2346
81
3,213
271
3269
200
5330
133
9202
Name
Alpha Bank
Eurobank
Total
6.2
Description
A
AQR
AFS
Available-for-sale
B
bps
Basis points
C
CA
Comprehensive assessment
CET1
CFR
CIR
CPMO
CRD
CRE
CRR
cRWA
CSA
CVA
D
DIV
DSCR
Term
Description
DTA
E
EAD
Exposure at default
EBA
ECB
EL
Expected loss
ELA
ESM
ESRB
F
FAQ
FTE
Full-time equivalent
FVO
FX
Foreign exchange
G
GDP
H
HICP
I
IAS
IBNR
IFRS
IRB
ISA
J
JST
L
LGD
LGI
LGI PIT
LGL
LTD
Loan-to-deposit ratio
LTV
Loan to value
Term
Description
M
MoU
Memorandum of Understanding
MPO
N
NCA
NII
NIM
NPE
Non-performing exposure
NPV
P
P&L
PD
PD PIT
PE
Performing
PI
Probability of impairment
PP&A
Q
QA
Quality assurance
R
RRE
RWA
Risk-weighted assets
S
SME
SSM
ST
Stress test
6.3
BIBLIOGRAPHY
Source
Reference
http://www.ecb.europa.eu/pub/pdf/other/assetqualityreviewphase2manual
201403en.pdf
http://www.ecb.europa.eu/pub/pdf/other/castmanual201408en.pdf
CRR / CRD IV
http://ec.europa.eu/internal_market/bank/regcapital/legislation-inforce/index_en.htm
6.4
Name
Definition
Administrative expenses
capital
transitional arrangements
ratio
arrangements
period)
Credit RWA including off-balance sheet items (in accordance with CRD
Assets
IV / CRR)
Credit Valuation
Adjustment
Ratio
Debtor
adjustments to Common
Equity Tier 1
Dividend income
Expected loss
Forbearance
The length of time between the specific event of loss for an exposure
(eg, a retail client losing their job, a corporate losing a large customer) and
the banks observation of loss.
The level of loss (after discounted recoveries) that can be expected if the
facility does not cure. Term used for retail exposures in the context of the
AQR collective provisioning challenger model. See Chapter 7.7 of the
AQR Phase 2 Manual for details
Net interest income is equal to total interest income minus total interest
expense
Net interest income (annualised if for half year) divided by the average
annualised)
Net trading income
Net trading income is equal to the total gains or losses made through
trading activities
Pre-provision profits
Probability of impairment
Return on equity (based on Return on equity is equal to net income attributable to owners of the
RWA * 9.5%) (average over parent net of estimated dividends excluding impairments on financial
period, annualised)
assets and other income and expenses (annualised if for half year) divided
by 9.5% of the average volume of Risk-Weighted Assets over the time
period
A shortfall occurs if the banks CET1% falls below the CET1% hurdle
Shortfall
rate at any point in the scenario. The capital shortfall is equal to the
difference between the banks CET1 capital and the amount of CET1
capital required (CET1% hurdle rate * Risk-Weighted Assets). The year
where the maximum shortfall appears is used
Specific provisions /
The ratio of specific provisions over the non performing exposure (EAD),
period)