4 December 2018  
The solvency of European Union banks continues to improve  
Thomas Humblot  
The European Union's major banks have continued to  
strengthen their solvency. They have improved their ability  
to absorb a large-scale financial shock and to resist its  
economic consequences on their own.  
shock applied to them was more severe. This is because  
bank solvency ratios were significantly higher both at the  
starting point of the financial shock (Q1 2018) and at the end  
point (Q4 2020) compared with those of the previous tests.  
The CET1 ratios of the EU's major banks are higher than  
they were during the EBA's previous stress tests, both at  
the starting and end points of the shock, even though the  
shock itself was more severe.  
Like the 2016 tests, the 2018 tests were not a pass-fail  
exercise based on some hurdle rate. Observers are therefore  
free to apply to each bank their own capital thresholds which  
are influenced, among other things, by the size and/or the  
business model of the bank considered.  
There were significant discrepancies between the results of  
the various banks. The results highlighted the banks'  
differing sensitivity to the adverse scenario, related in  
particular to their business mix and/or their international  
The results of the 2018 EBA EU-wide stress tests will be  
incorporated into the Supervisory Review and Evaluation  
Process (SREP). This process leads in particular to estimate  
the amount of additional capital that some banks could be  
required to build up. The EBA also believes that the  
disclosure of granular data17,200 data points per bank—  
will contribute to market discipline.  
The results will be incorporated in the SREP process and  
will contribute to determine in part the amount of additional  
capital certain banks might have to build up.  
A similar methodology to that of the previous exercises  
The most recent stress tests coordinated by the European  
Banking Authority (EBA) were an opportunity for the major  
European banking groups to demonstrate their resilience in  
the face of a hypothetical financial shock and its economic  
consequences. Overall, the 48 major banks in the European  
Union and the European Economic Area (plus four Greek  
banks) continued to improve their ability to absorb a large-  
scale financial shock, at the highest level of consolidation.  
The EBA emphasized that the major EU banks have  
demonstrated a "continuous and significant" strengthening in  
their solvency since the first publication of its stress tests in  
The 52 banks subjected to the EBA's stress tests applied their  
internal models to the baseline scenario, supplied by the EBA,  
and to the adverse scenario, calibrated by the European  
Systemic Risk Board (ESRB).  
The stress tests covered 70% of the total assets of EU banks  
The EBA's stress tests covered around 70% of the total  
assets of the banking system of the entire European Union.  
This proportion remained stable compared with the 2016  
stress tests.  
The 48 major banking groups in the 2018 sample (vs. 51 in  
In 2018, the results obtained by the banks to the stress tests  
were better than those of 2014 and 2016, even though the  
016) were drawn from 14 European Union countries , plus  
Norway. Four Greek banks were also included: Alpha Bank  
EA, Eurobank Ergasias SA, National Bank of Greece SA  
The Supervisory Review and Evaluation Process (SREP) analyzes  
the risk profile of each bank which are under the direct supervision of  
the ECB, in the context of the Single Supervisory Mechanism. It is this  
process that results in additional capital requirements.  
Austria, Belgium, Denmark, Finland, France, Germany, Hungary,  
Ireland, Italy, the Netherlands, Poland, Spain, the United Kingdom  
and Sweden.  
(NBG) and Piraeus Bank SA. The results obtained by these  
The static balance sheet assumption  
banks to the stress tests were published earlier, on 5 May  
018, so that they would be available before the end of the  
European Stability Mechanism's third financial assistance  
programme for Greece on 20 August 2018.  
The EBA requires the banks to use a static balance sheet  
assumption. This means that the bank's assets and liabilities  
that mature during the stressed period (from Q1 2018 to Q4  
020 in the case of the 2018 tests) are replaced by similar  
financial instruments. Similarly, the bank's business mix is  
assumed to remain identical throughout the stressed period.  
The sample included only banks with total assets in excess of 30  
billion euros  
Each bank in the sample used by the EBA for its 2016 and  
Any actions carried out by the banks on their equity or  
assimilated instruments (new share or contingent convertible  
bond issues, conversion of hybrid securities, etc.) after 1  
January 2018 were not taken into account. Legally required  
regulatory changes were integrated into the stress tests, but  
expected changes were not, regardless of their probability of  
occurrence. Exchange rates were held constant as of 31  
December 2017. Finally, the banks were not authorized to  
change their internal models during the test period.  
018 stress tests had total assets of at least 30 billion euros.  
This threshold is the same as the one the European Central  
Bank (ECB) uses to distinguish the so-called "significant"  
The national competent authorities (NCAs) had the option of  
including additional banks in the stress tests, at their  
discretion. In that case, they had to have total assets of at  
least 100 billion euros and/or be close to the end of a  
mandatory restructuring plan agreed by the European  
This static balance sheet assumption makes it easier to  
estimate the effects of the adverse scenario and to make  
comparisons between banks. That said, it amplifies and  
therefore overestimates the effects of the shock on the banks  
in that it does not allow them to adapt their balance sheet to  
the circumstances, whereas it seems likely that they would.  
The ECB supplied the baseline scenario, while the ESRB was in  
charge of the adverse scenario  
The banks' ability to absorb a large-scale financial shock was  
tested using an adverse scenario, the results of which were  
expressed as a deviation from a baseline scenario supplied  
by the ECB.  
From material threats to the model’s assumptions  
To calibrate the set of parameters of the adverse scenario,  
the ESRB identified four principal systemic threats to the  
stability of the EU's financial sector: i) a sudden and  
significant increase in the risk premia on financial markets, ii)  
an adverse feedback loop between weak growth and bank  
profitability, iii) uncertainty about the sustainability of public  
and private debt, and iv) liquidity risk in the non-bank  
compartment of the financial sector.  
The adverse scenario was hypothetical. It was not intended to  
cover all of the risks to which the banks might individually be  
exposed (these are handled by the SREP). The adverse  
scenario in the 2018 EBA EU-wide stress tests covered credit,  
market and operational risks, but pertained essentially to the  
first two. The effects of the shock on banks' net interest  
income, their P&L and their capital were also considered. The  
adverse scenario was ultimately validated by the ESRB, but it  
was calibrated in cooperation with the ECB, the NCAs, the  
EBA and the national authorities.  
These sources of risk would then give rise to a set of financial  
and economic sub-shocks.  
The adverse scenario would be triggered by a sudden and significant  
increase in the risk premia on financial markets outside the EU  
The banks then had to apply their own internal models to the  
common set of parameters. At this stage, the EBA's role was  
simply one of coordination between the various participants.  
The 2018 EBA EU-wide stress tests use a bottom-up  
approach, in contrast to the ECB's tests, which use a top-  
down approach. The ECB's tests also serve to validate the  
results communicated by the banks. They are also applied to  
the "significant entities" that are under the ECB direct  
The ESRB believes that an “abrupt and sizeable repricing” in  
the risk premia on financial markets outside the EU is the  
most significant risk for the stability of the EU's financial  
system, because it can trigger other systemic risks that would  
result in a set of macro-financial sub-shocks.  
The source of the first shock, occurring in Q1 2018, would be  
completely exogenous to the EU banking system. It would be  
due to a sharp revision of market participantsoutlook on  
economic policies outside the EU. For example, a restrictive  
trade policy in the United States. Heightened uncertainty  
among market participants would increase the risk premia of  
financial assets and reduce their valuation. US equities would  
see their market capitalization shrink by 41% on average in  
supervision in the Single Supervisory Mechanism (SSM )  
framework and that are not included in the EBA's scope.  
The banks' results were expressed as a proportion of their  
ratio of transitional or fully loaded Common Equity Tier 1  
CET1) . The results expressed in terms of fully loaded CET1  
ease comparisons between banks. Fully loaded CET1 is also  
more demanding than transitional CET1 and is therefore likely  
to provide a more accurate view of a bank loss-absorbing  
capacity under the most adverse conditions. Hence, fully  
loaded CET1 is the benchmark most generally used by  
analysts. Transitional CET1, on the other hand, is more widely  
used for prudential purposes and is therefore particularly  
useful for bank supervisors, in particular in defining additional  
capital requirements.  
018 compared with their 31 December 2017 valuation (the  
baseline scenario assumes their valuation would remain  
unchanged until 31 December 2020). The declining value of  
US equities would propagate to European equities, which  
would decline in turn by 30% in 2018. Nevertheless, the  
deviation from the baseline scenario would subside in 2019  
and 2020 (see Table 1). Given that the crisis would originate  
in the United States, equities traded in the EU would suffer  
less from the initial effects of the shock.  
Includes the ECB and the national competent authorities (NCAs) of  
the countries that participate in the SSM.  
Prudential filters were 100% removed.  
Selected transmission channels of 2018 EBA EU-wide stress tests’ financial shock  
Financial and economic  
Main financial stability risks  
Abrupt and sizeable repricing of risk  
Stock prices  
-29.9% -27.2% -21.5%  
0 bp  
230 bp  
330 bp  
330 bp  
Unemployment rate  
Residential real estate price  
Commercial real estate price  
1380 bp -1350 bp -440 bp -27,7%  
9,6% -9,8% -0,8% -19,1%  
1440 bp -1070 bp -580 bp -27,1%  
11,1% -7,5% -2,7% -20,0%  
350 bp -410 bp -110 bp -830 bp  
Adverse feeback loop between weak bank  
profitability and low growth  
40 bp  
-2,2% -2,7%  
-130 bp -190 bp -3,5%  
3 bp  
85 bp  
80 bp  
39 bp  
48 bp  
Long-term sovereign credit  
5 bp  
9 bp  
47 bp  
54 bp  
Public and private debt sustainability  
SWAP rates  
Retained data is focused on EU banks and is expressed as a deviation from the baseline scenario or in absolute level in the adverse scenario  
Table 1  
Sources : EBA, BNP Paribas  
The euro would appreciate against Central and Eastern  
European currencies. There would be temporary, adverse  
effects on trade between the two regions. More generally,  
overall demand for goods and services would contract, in  
particular those deriving from the EU. This decline in overall  
demand would be reinforced by a resurgence of protectionist  
measures in certain major industrialized countries (once  
again, typically in the United States). Commodities would also  
be negatively affected by the decline in worldwide demand.  
would decline to 1,350 bp in 2019, then more significantly to  
440 bp in 2020. During the test period, residential real estate  
prices in the EU as a whole would decline by a cumulative  
19.1%, with significant disparities, as prices would fall by 50%  
in Sweden but hardly at all in Hungary.  
Commercial real estate prices would move similarly to  
residential prices. In 2018, they would decline by 11.1%, i.e. a  
deviation of 1,440 bp compared with the baseline scenario  
(3.2%). Commercial real estate prices would post a  
Weak growth and bank profitability would result in an adverse  
feedback loop  
cumulative decline of 20% in 2020, but with relatively less  
disparity between countries than in residential real estate,  
despite more volatility from one year to another. This is in part  
because commercial real estate is more used as an  
investment than is residential real estate.  
As worldwide economic activity contracts, production would  
decline and so would the need for labour. The unemployment  
rate in the EU as a whole would rise from 7.6% in 2017 to  
.9% in 2018, on a weighted average basis, which would  
Demand for loansmortgage loans among themwould  
decline, and the recession would boost the default rate among  
non-financial corporations. These factors would have a  
negative influence on bank profitability. Not only would banks  
be affected by a decline in business activity and by an  
increase in their cost of risk, but they would also see a decline  
in quality on securities portfolio they might hold.  
represent a deviation of 70 basis points (bp) from the baseline  
scenario. The unemployment rate would continue to rise over  
the period, reaching 9.7% in 2020, or a maximum deviation of  
30 bp. This contrasts with the baseline scenario, in which the  
unemployment rate is expected to decrease steadily to 6.4%  
at the end of the test period.  
Residential real estate prices in the EU as a whole would  
decline by 9.6% in 2018. The baseline scenario projects a  
price of 4.1% in 2018, so the deviation would be 1,380 bp, the  
widest of any year during the stressed period. The deviation  
All of this would cause real GDP in the EU as a whole to  
contract by 1.2% in 2018, on an annualized basis,  
representing a deviation of 350 bp compared with the  
baseline scenario. The baseline scenario estimates real GDP  
growth at 2.2% in 2018, based on the projections of national  
central banks. The adverse scenario projects that the  
recession would worsen in 2019 (-2.2%), but that the  
economy would return to growth in 2020 (0.7%). Over the test  
period, the economy would contract by 2.7% under the  
Kingdom's withdrawal from the European Union in the  
adverse scenario.  
The results to the stress tests have continued to improve  
adverse scenario, representing  
compared with the baseline scenario. Nevertheless, the  
baseline scenario calls for growth to slow down in 2019 and  
a deviation of 830 bp  
Overall, the results of the 2018 EBA EU-wide stress tests  
were better than those obtained in 2016, even though the  
simulated shock was more severe. The banks' solvency ratios  
were higher at the starting point of this cyclical downturn than  
they were during the previous simulations. The same was true  
of the end of the test period and of its economic  
020 to 1.9% and 1.8%, respectively.  
Uncertainty would emerge about the sustainability of public and  
private debt  
The cyclical downturn would exacerbate the sudden rise in  
bond yields and risk premia on sovereign debt market  
following the initial shock. As a result, 10-year US sovereign  
debt yields would diverge by 235 bp from the baseline  
scenario in 2018.  
A question of framework  
The results of the stress tests cannot be fully interpreted  
simply by comparing the difference between regulatory capital  
ratios as of 31 December 2017 and 31 December 2020 in the  
adverse scenario. This is because the accounting framework  
Ten-year sovereign debt yields would spread to European  
Union member states, which would experience a 83 bp  
deviation compared with the baseline scenario in 2018. Over  
the entire period, member-state sovereign debt yields would  
rise, on a weighted average basis, to 2.6% in 2020 (vs. the  
actual average of 1.2% in 2017 and 1.8% estimated in 2020 in  
the baseline scenario). The deviation from the baseline  
scenario would narrow slightly at the end of the period. Yields  
would diverge further in countries whose debt is more subject  
to doubts about its sustainability.  
IAS 39 or IFRS 9) and, especially, the definition used for  
CET1 (fully loaded or transitional) have to be taken into  
account. For this reason, the banks' stress test results have  
been presented on the basis of fully loaded and transitional  
CET1 ratios for the baseline and adverse scenarios. Restated  
versions of these ratios were calculated as of 31 December  
017 so as to compare the results under a single accounting  
framework (IFRS 9 must be applied on 31 December 2020).  
The difference between a bank's fully loaded CET1 ratio in  
020 under the adverse scenario (CET1 ratio fully loaded  
We note that it would not be possible to attribute the rise in  
interest rates and bond yields to monetary policy, which is  
assumed to be identical in both the baseline and adverse  
adverse scenario 2020) and its restated fully loaded CET1  
ratio in 2017 (CET1 ratio fully loaded restated 2017) provides  
information on its ability to absorb the shock simulated in the  
018 EBA EU-wide stress tests. The smaller the difference,  
the more able the bank is to absorb the cyclical downturn. We  
think this measure is the most meaningful, because it eases  
comparison between banks and because the loss-absorbing  
capacity it reflects has the least uncertainty. It serves as the  
EBA's base of comparison. On the other hand, this measure  
gives an imperfect view of the additional capital the banks  
might be required to maintain. This is because the amount of  
additional capital is estimated on the basis of transitional and  
More globally, all rates would rise (bond yields, consumer  
loan rates, etc.). As such, the three-month interbank rate,  
Euribor, would diverge by 55 bp compared with the baseline  
scenario in 2018, reaching 0.24% (vs. -0.33% in 2017). This  
deviation would subside to 47 bp in 2019 and 39 bp in 2020,  
but the rate would continue to increase, reaching 0.54% at the  
end of the stressed period in the adverse scenario.  
not fully loaded CET1 (difference between CET1 ratio fully  
loaded adverse scenario 2020 and CET1 ratio transitional  
restated 2017).  
The rising interest rates would pass through to the cost of  
loans. Bank loans would become more expensive, reducing  
borrower solvency correspondingly. In the end, uncertainties  
would emerge about the sustainability of private and public  
debt, especially in countries whose public deficit is already  
The difference between the fully loaded CET1 solvency ratio  
in 2020 under the adverse scenario (CET1 ratio fully loaded  
adverse scenario 2020) and the actual transitional CET1 ratio  
in 2017 (CET1 ratio transitional restated 2017) provides  
information on a bank's ability to absorb the shock while  
pursuing its transition to the new regulatory requirements and  
The financial sector as a whole could be impacted by worsening  
liquidity in the non-bank financial sector  
Non-bank financial institutions (pension funds, insurance  
companies, brokers, etc.) would be forced to sell a growing  
part of their securities portfolios as uncertainty rises and asset  
values decline, so as to uphold their liquidity commitments.  
Equity and bond prices would suffer a new decline, putting  
negative pressure on the value of bank portfolios. Moreover,  
non-bank financial institutions could withdraw funds deposited  
with banks, whose position would then be further eroded.  
accounting standard. This difference provides  
a more  
comprehensive view of what a bank might have to accomplish  
during a cyclical downturn. In a certain sense, it's the worst  
possible scenario in the case where the shock would be  
severe enough to make it difficult for the bank to finish  
transforming its balance sheet. In the same vein, it is also  
interesting to compare the fully loaded solvency ratio at its low  
point under the adverse scenario (which does not necessarily  
occur in 2020). The banks do not all suffer the effects of the  
shock simultaneously.  
Finally, Brexit was not taken explicitly into account.  
Nevertheless, the baseline scenario includes the average of  
its possible consequences on trade between the United  
Kingdom and the EU. In addition, according to the EBA,  
several items can be treated as effects of the United  
2018 EU-wide stress tests Results, p. 13 (EBA)  
Finally, the difference between the fully loaded CET1  
solvency ratio in 2020 under the adverse scenario (CET1 ratio  
fully loaded adverse scenario 2020) and the same ratio under  
the baseline scenario (CET1 ratio fully loaded baseline  
scenario 2020) provides information about the scale of the  
shock that could affect the bank. The smaller the difference,  
the less the bank is affected by the shock.  
2015) and at the end point (9.2% in 2018) of the previous  
exercise. Overall, therefore, the banks in the sample showed  
an improvement both in their solvency and in their loss-  
absorbing capacity between 2016 and 2018.  
Significant relative discrepancies  
The effects of the simulated financial shock in the 2018 EBA  
EU-wide stress tests differ significantly from one bank to  
another (see Graph 1). Certain German and British banks  
were more sensitive to the shock. The results of the Spanish,  
French and Italian banks were close to the average of all EU  
banks. Northern European banks were less affected by the  
adverse scenario. The Irish banks stood out by their greater  
use of the transition period for the application of the new  
regulatory requirements and the new accounting standard.  
This hierarchy is comparable to the one observed following  
the 2016 stress tests.  
The solvency ratios started and ended higher than they did during the  
previous exercises  
Under the adverse scenario, the aggregated fully loaded  
CET1 ratio for the full sample of banks declined by 395 bp  
compared with its 31 December 2017 restated level. It  
declined from 14.0% (14.51% on a transitional CET1 basis) to  
0.05% over the course of the stressed period. Under the  
baseline scenario, the aggregated fully loaded CET1 ratio for  
the full sample of banks is assumed to continue  
strengthening, reaching 15.3% as of 31 December 2020.  
There were also significant differences in the end-of-period  
results under the baseline and adverse scenarios.  
Nevertheless, a ranking of the banks under this view remains  
similar to a ranking of the differences between the banks'  
starting and end points under the adverse scenario.  
During the 2018 stress tests, the impact of the adverse  
scenario on the fully loaded CET1 ratio was greater than it  
was in 2016 (-340 bp). This means that the simulated shock in  
the 2018 exercise was more severe. But the bank solvency  
ratios were significantly lower at the starting point (12.6% in  
Stress test results showed significant discrepancies  
Graph 1  
Sources: EBA, BNP Paribas  
The decline in regulatory solvency ratios derives principally from an  
increase in credit losses  
an increase in credit risk on the portfolio, for which regulatory  
capital is calculated using an internal ratings-based approach.  
The decline in bank solvency ratios between 31 December  
There are significant differences in the sources of the decline  
in solvency ratios from one bank to another (see Graph 3).  
017 and 31 December 2020 under the adverse scenario is  
9% due to a reduction in CET1 capital (the numerator) and  
2% due to an increase in the risk exposure amount (the  
The stress test results will be used to estimate the amount of  
additional capital banks might be required to build  
denominator). This relatively lesser effect derived in part from  
the static balance sheet assumption.  
The Supervisory Review and Evaluation Process (SREP) is a  
component of pillar 2 of the CRD IV (minimum capital  
Credit losses represented the principal impact of the adverse  
scenario on all of the banks in the sample. They reduced the  
CET1 ratio by 424 bp between its starting point and 31  
December 2020 (see Graph 2). The impact was particularly  
noticeable in 2018 because of IFRS 9's prospective approach,  
under which the banks recognize the most of the loan loss  
provisions when the initial signs of a cyclical downturn appear.  
The potentially procyclical character of the new accounting  
standard cannot be fully appreciated through these stress  
tests, however, in part because the static balance sheet  
assumption prevents the banks from adapting their balance  
sheet to changes in the business cycle.  
requirements of the CRR). According to the ECB, the goal of  
the SREP is "to promote a resilient banking system as a  
prerequisite for a sustainable and sound financing of the  
economy." To do this, the supervisory authorities must  
examine the risks to which banks are exposed as well as the  
risks banks impose on the financial system.  
In this regard, the results obtained by the EU's major banks  
on the EBA's stress tests have been used since 2016 in the  
determination of additional capital required under pillar 2.  
The second most significant impact of the financial shock on  
the banks' aggregate solvency ratio between 31 December  
Overall, the major European banks have continued to improve  
their solvency and their ability to absorb large-scale economic  
and financial shocks. The stress tests can also be used to  
compare banks with one another and to compare test results  
at different periods of time for the same bank.  
017 and 31 December 2020 under the adverse scenario was  
market risk (-112 bp). Finally, operational risks led to a 100 bp  
reduction in the CET1 ratio .  
The increase in the risk exposure amount led to a decline in  
the CET1 ratio of 157 bp as of 31 December 2020 compared  
with its starting point. This effect is almost exclusively due to  
A few caveats apply, however, because stress tests are  
necessarily somewhat arbitrary. Since they became  
recurring part of the landscape following the 2007/08 crisis,  
Aggregated effects of the financial shock of the EBA's 2018 stress test  
Graph 2  
Sources: EBA, BNP Paribas  
The effects of the shock on operational risk are only available for the  
aggregate CET1 ratios. The fact that the 2018 stress tests focused  
essentially on credit and market risks might explain this. On an  
individual basis, the cost of banks' operational risk is deducted directly  
from pre-tax profit.  
Article 99 of Directive 2013/36/EU of the European Parliament and  
of the Council of 26 June 2013  
they have been the target of repeated criticism. Some critics  
maintain that the magnitude of the simulated shocks is too far  
removed from those most likely to occur, while others say the  
tests are too pragmatic and do not take into account certain  
highly improbable but potentially very costly events.  
The International Monetary Fund has recommended that the  
EBA make several improvements to its stress tests, following  
tests the IMF itself carried out. These tests measured the  
resilience of the major banks under the ECB's direct  
supervision, based on end-2017 data. The IMF also  
concluded that the banks' solvency and liquidity had  
improved, but that wide discrepancies remained. Specifically,  
it found that the large banks that are internationally diversified  
but relatively less exposed to the most complex financial  
products were the least sensitive to fluctuations in the  
business cycle. It is difficult to compare the IMF's results with  
those of the EBA, however, because the scenarios,  
methodology and more generally, the objectives were  
In the end, regulatory capital requirements are always  
intended to strike a balance between i) the economy's  
immediate need for financing, which necessarily implies risk  
exposure, and ii) the banking system's need for long-term  
stability, and therefore for more capital, whose cost remains  
Sources of the decline in solvency ratios  
Graph 3  
Sources: EBA, BNP Paribas  
QUI SOMMES-NOUS ? Trois équipes d'économistes (économies OCDE, économies émergentes et risque pays, économie bancaire) forment la Direction des Etudes Economiques de BNP Paribas.
Ce site présente leurs analyses.
Le site contient 2119 articles et 584 vidéos