
(AGENPARL) – gio 30 giugno 2022 Notes on
Financial
Stability and
Supervision
No. 30
June 2022
e papers published in the
Notes on Financial Stability
and Supervision series
reect the views of the authors
and do not involve
the responsibility
requirements and capital buers:
the case of Italian banks
Wanda Cornacchia – Giulio Guerra*
Overview
e term ‘buer usability’ refers to the possibility for banks
to use the capital buers included in the Combined Buer
Requirement (CBR) without simultaneously breaching any
other binding regulatory requirements; if instead some or all of
the capital set aside to meet the CBR is needed to satisfy other
the buer usability. In this work we investigate the mechanics of
the interaction between minimum requirements and buers by
developing a comprehensive methodology aimed at measuring
the actual usability of the CBR. Such a methodology takes into
account all four EU regulatory requirements simultaneously:
the risk-weighted one (RW), the leverage ratio (LR), the risk-
weighted MREL (MREL-RW) and the leverage-ratio-based
MREL (MREL-LR). is means that we consider not only the
placed on top of the risk-weighted MREL. According to our
methodology, the overlap between minimum requirements
and capital buers aects about one fourth of Italian banks and
reduces the CBR’s usability to 74 per cent of its theoretical value,
compared with a mere 27 per cent if the CBR placed on top of the
MREL-RW is not taken into account. When the CET1 absorbed
by the MREL-RW requirement is higher than the CET1 absorbed
by the RW one, the CBR may be more usable than is apparent
from the approach based solely on the RW requirement. is
explains why, by also considering the regulatory requirements
of the resolution framework, the usability of the CBR increases.
e issue of the overlap is being debated at international level
by the main nancial authorities and the standard setters.
anda Cornacchia, Directorate General for Economics, Statistics and Research
and Giulio Guerra, Directorate General for Financial Supervision and Regulation,
Table of contents
Introduction
and main conclusions
e overlaps and the stylized
mechanics of interaction
between CBR and
regulatory requirements
the RW and LR frameworks
Interaction between
the RW and MREL-LR
frameworks
Interaction between
the RW and MREL-RW
frameworks
A comprehensive approach
for measuring overlaps
Methodological Appendix
ISSN 2284-4198 (online)
Designed by
the Printing and Publishing Division of the Bank of Italy
Introduction and main conclusions
Capital buers are a key instrument of macroprudential policy
. ey are
included in the Combined Buer Requirement (CBR), equal to the sum of the Capital
Conservation Buer (CCoB), Countercyclical Capital Buer (CCyB), the buers for
global (G-SII) and other (O-SII) systemically important institutions, and the Systemic
Risk Buer (SRB). Buers were introduced after the Global Financial Crisis in order to
shield the nancial system, and consequently the real economy, from systemic risks.
Capital buers have two general purposes. First, they support the resilience of banks,
as they ask them to operate with an amount of capital signicantly above minimum
requirements in good times. Second, they can be drawn down following the indications
of authorities in bad times, when losses have to be absorbed, and are replenished
afterwards. Banks can therefore use capital buers instead of deleveraging or
de-risking their balance sheets. By using buers, credit ows to households and rms
can be substantially boosted in bad times, with benets for both macroeconomic and
nancial stability.
In the EU capital regulation for banks,
two main frameworks are applied
at the same time
: one for prudential purposes, the other for resolution purposes.
e rst one includes both a risk-weighted requirement (RW) and a leverage ratio
requirement (LR). Similarly, the resolution framework, which ensures that banks
have enough loss-absorbing and recapitalization capacity through a Minimum
Requirement of Eligible Liabilities (MREL), is based on two ratios that are to be met in
parallel: the MREL as a percentage of risk weighted assets (MREL-RW) and the MREL
as a percentage of the total exposure measure used for the purpose of the leverage
ratio (MREL-LR). According to the EU regulation, the CBR is only required on top of
the two risk-weighted requirements (RW and MREL-RW). is asymmetry implies
that the same capital can be used simultaneously to satisfy the CBR in one framework
and a minimum requirement in another framework. In such cases, overlaps occur
between the CBR and the minimum requirements, with the consequent impossibility
of using (in whole or in part) the CBR to absorb losses without violating a minimum
requirement.
The term ‘buffer usability’ refers to banks’ ability to use the CBR without
breaching any minimum requirements
. In the event of overlaps, a bank would
not use (all or part of) the CBR even when allowed to do so because such a use would
lead to a breach of a minimum requirement.
This issue, in turn, can undermine
the decision of macroprudential authorities to release part of the CBR,
i.e. to draw it down in order to allow banks to support the economy in
bad times.
Usability should not be confused with the unwillingness/reluctance
e ‘banking package’ comprises the Capital Requirements Directive and Regulation (CRDV/CRRII), the Bank Recovery
and Resolution Directive (BRRDII) and the Single Resolution Mechanism Regulation (SRMRII).
e CBR includes both releasable and non-releasable buers by macroprudential authorities: for example, the CCyB is
releasable while the CCoB is not. Anyway, all buers included in the CBR are usable.
of banks to use the buffers because of disincentives of various kinds (for instance,
stigma effects due to financial market reactions or maximum distributable amount
restrictions
non-risk weighted prudential and resolution requirements on actual buffer
usability for Italian banks.
In particular, we assess the overall CBR usability,
considering both the CBR stacked on top of the risk-weighted requirement, as does
the European Systemic Risk Board in a recent report, for instance,
and the CBR
placed on top of the risk-weighted MREL. This comprehensive approach results
in a significant increase in the CBR’s usability compared with an approach where
only the CBR stacked on prudential RW requirement is taken into account. Indeed,
we found that overlaps, which affect around one fourth of Italian banks,
compared with only 27 per cent if the CBR placed on top of the MREL-RW is not
taken into account. Indeed, when the CET1 absorbed by the MREL-RW requirement
is higher than the CET1 absorbed by the RW one, the CBR may be more usable than
is apparent from the approach based solely on the RW requirement. This explains
why, by considering the regulatory requirements of the resolution framework as
well, the usability of the CBR increases.
e issue of the overlaps is being debated at international level by the main nancial
authorities and the standard setters. At the centre of the debate, there is the nature
of the issue and the costs associated with a possible intervention. At the moment, no
decision has been taken on the opportunity to tackle the overlaps: further investigation
is needed to reach a common view among countries.
regulatory requirements
Interaction between the RW and LR frameworks
the latter. Based on these two requirements, the CBR’s usability is constrained
requirement.
Article.141 of CRD IV introduced the concept of the Maximum Distributable Amount (MDA), which requires supervisory
authorities to automatically restrict earnings distribution in the event of a CBR breach. Similarly, the MREL-MDA
(M-MDA) is imposed by resolution authorities, though with more discretion and no automaticity.
e ESRB approach was set out in the ‘
Report of the Analytical Task Force on the overlap between capital buers and
minimum requirements
’, December 2021.
Figure 1 compares the stacking order of
capital between the risk-weighted capital
framework (RW) and the LR framework
(LR), for a hypothetical bank.
We consider
the case in which the LR requirement
met with CET1 is higher than the RW
requirement met with CET1 (see dashed
line). In our example, an overlap of 2 per
cent of risk- weighted assets (RWAs) occurs,
equal to the dierence between the two pink
areas (8-6 per cent), and the CBR usability
is constrained to 1 per cent of RWAs.
e ‘distance from the overlap’ (see Figure 1)
measures the amount of capital (as a share
of RWAs) that can be used before the overlap
takes place (when there is an overlap). is
is given by the capital in excess of the CBR
or the capital employed to satisfy the CBR
which is not used by other requirements.
e overlap exists because the
regulation only requires the CBR on
top of the RW requirement and not also
above the LR requirement
(the same
applies to the overlap between RWs and
MREL-LR).
It should be noted that, other things being equal, for banks with higher risk
decreases and buer usability increases (i.e. the leverage ratio is not binding).
Indeed, if RWAs increase for the same total assets (other things being equal), the
minimum requirement covered with CET1 increases in the RW framework, while
nothing changes for the LR requirement; the overlap is consequently reduced.
Similarly, a higher level of AT1 instruments also reduces the overlap and increases
buer usability
(see Figure 1). While in the risk-based framework the minimum 6%
Tier 1 capital ratio must be satised at least with 4.5% of CET1, no similar proportion
applies for the purposes of the minimum 3% Leverage ratio (all other conditions for the
recognition of AT1 instruments being equal between the two frameworks). us, the
more eligible the AT1 instruments available, the lower the LR requirement that has to be
met with CET1 and the lower the overlap with the RW requirement.
We assume that the CET1 ratio of this bank is 10 per cent and that the bank is subject to: an RW pillar 1 requirement of
8 per cent of RWAs; a pillar 2 requirement of 1.5 per cent, for simplicity to be met entirely with CET1 capital. e CBR is
set at 3 per cent. e bank meets the minimum requirement (p1+p2) with its CET1 (6 per cent), AT1 (1.5 per cent) and
T2 (2 per cent). e LR minimum requirement (3 per cent of the LR exposure) accounts for 9.5 per cent of RWAs and is
partially met with AT1 instruments (1.5 per cent).
the RW and LR frameworks
(amounts expressed as % of RWAs, also LR)
Figure 1 compares the stacking order of
capital
We assume that the CET1 ratio of this bank is 10
per cent
and that the bank is subject to: a
RW pillar 1 requirement
per cent
of RWAs; a pillar 2 requirement of 1.5
per cent
Figure 1: Styli
frameworks
amounts expressed as % of RWAs, also LR)
Note:
The notion of
stacking order defines the sequence
in which
different capital layers absorb losses.
In the risk
weighted capital
framework
(the RW bar) excess capital cover
losses first (green),
followed by the CBR (yellow) and
minimum requirement
Note: e notion of a stacking order denes the sequence in
which dierent capital layers absorb losses. In the risk-weighted
capital framework (the RW bar) excess capital covers losses
rst (green), followed by the CBR (yellow) and the minimum
requirement (pink).
Interaction between the RW and MREL-LR frameworks
occurs when the MREL-LR requirement
corresponding RW requirement
. In
Figure 2, a hypothetical bank
shows an
overlap of 2 per cent of RWAs (8-6 per cent;
see dotted line). e actual usable CBR is 1
per cent of RWAs (one third of the CBR).
This is the same interaction mechanism
observed for the overlap between the
RW and LR frameworks.
In this case
the CBR is not required on top of the
MREL-LR minimum requirement.
Hence, if a bank relies on the CET1
capital used to meet the CBR to meet the
MREL-LR as well, the buffer usability is
impaired.
The considerations on risk density and
on AT1 made for the RWs-LR case also
apply in this case, with one difference:
for this overlap, not only are the AT1
instruments relevant, but also all the
other liabilities eligible for the MREL.
When the MREL-LR is binding, an overlap
emerges – its size depends on the level of MREL eligible liabilities – and affects the
CBR’s usability and eventually the excess capital. To increase the usability of the
CBR, the bank could issue more liabilities computable in the MREL to reduce the
CET1 absorbed by the MREL-LR
is bank has the same characteristics as that shown in Figure 1. In addition, the MREL-LR is equal to 7.3 per cent of the
LR exposure (22 per cent of RWAs); the stock of MREL eligible liabilities accounts for 10.5 per cent of RWAs.
AT1/T2 and MREL eligible liabilities can be used without limits for compliance with both the MREL-LR and MREL-RW
minimum requirements. Medium-sized banks could have greater costs and market access problems when issuing eligible
liabilities than large banks.
the RW and MREL-LR frameworks
(amounts expressed as % of RWAs, also MREL-LR)
of the minimum 3% Leverage
ratio (all other conditions for the recognition of AT1 instruments
This bank has the same characteristics
that shown in Figure 1. In addition,
LR is equal to 7.3
per cent
of the LR exposure (22
per cent
of RWAs); the stock of MREL eligible liabilities accounts for 10.5
per cent
of RWAs.
Figure 2: Styli
and MREL
frameworks
amounts expressed as % of RWAs, also MREL
Note:
The notion of
stacking order defines the sequence in
which
different capital layers absorb losses.
In the risk
weighted
capital
framework
(the RW bar) excess capital cover losses first
(green), followed by the CBR (yellow) and minimum requirement
Note: e notion of a stacking order denes the sequence
in which dierent capital layers absorb losses. In the
risk-weighted capital framework (the RW bar) excess capital
cover losses rst (green), followed by the CBR (yellow) and
minimum requirement (pink).
Interaction between the RW and MREL-RW frameworks
CET1 capital used to satisfy the CBR on top
of the RW requirement cannot be used to
comply with the MREL-RW.
In this case,
the regulation is symmetric:
the CBR also
stacks on top of MREL-RW and remains
fully usable.
In Figure 3, the bank in the example
has an overlap of 1 per cent of RWAs, as
it uses 7 per cent of CET1 to comply with
the MREL-RW (compared with the RW
minimum requirement of 6 per cent). e
CET1 excess of the RW framework (1 per
cent) is therefore completely eliminated in
this case.
e overlap here only reduces the excess
capital and not the CBR, because there
CBR is required in both frameworks).
e reduction in excess capital is less of a
problem compared with a formal overlap.
e reduction of excess capital typically
involves banks with: i) high risk density:
other things being equal, the higher the
risk-density, the higher the MREL-RW;
and ii) lower credit rating and protability:
for these banks, the wholesale funding
(as for the MREL eligible liabilities) is
comparatively more expensive, so they are induced to save MREL eligible liabilities in
order not to worsen their protability.
See Article 128 of CRDV (specically, the reference therein to ‘risk based components’).
e CBR does not stack on top of MREL-RW in a particular case. Resolution authorities may increase the subordinated
MREL up to the value that corresponds to the prudential formula (that is two times the total minimum requirements
under Pillars 1 and 2 plus all buer requirements). e prudential formula may be satised with buer capital,
i.e. in such a case the double use of CET1 is allowed, and can be applied for a number of resolution entities (up
to 30% of certain categories of banks and under specic conditions). None of the Italian banks is subject to the
prudential formula. On the contrary, according to a
Sta Memo
of the Norges Bank, the largest Norwegian banks
will have to use a substantial share of their buer capital to satisfy MREL, i.e. subordinated MREL calculated using
the prudential formula, at least if the banks’ non-preferred debt issuance is small.
is bank has the same characteristics as that shown in Figure 1. In addition, the MREL-LR is equal to 21 per cent of
RWAs; the stock of MREL eligible liabilities accounts for 10.5 per cent of RWAs.
Imagine two banks showing the same total assets (100) but dierent risk-density: 60 per cent (bank 1) and 20 per cent
(bank 2). As a consequence, RWAs are 60 and 20 for bank 1 and 2, respectively. If the MREL-RW is equal to 20 per cent
of the RWAs for both banks, then bank 1 is subject to a MREL-RW of 12 (20 per cent x 60) while bank 2 is subject to a
MREL-RW of 4 (20 per cent x 20).
the RW and MREL-RW frameworks
(amounts expressed as % of RWAs, also MREL-RW)
could
issue more liabilities computable in the MREL to reduce the CET1 absorbed by the MREL
Interaction between the RW and MREL
RW frameworks
apital used
satisfy
the CBR
on top
of the RW
requirement
cannot be used
comply with
In thi
AT1/T2 and MREL eligible liabilities can be used without limits for compliance with both
LR and MREL
RW minimum requirements. Medium
sized banks could have greater costs and market access problems
issuing
eligible liabili
ties than large banks.
Article
CRDV (
specifically,
the reference therein to
risk based components
The CBR does not stack on top of MREL
RW in a
particular
case. Resolution authorities may increase the
subordinated MREL up to the value th
at corresponds to the prudential formula (that is two times the total minimum
requirements under Pillars 1 and 2 plus all buffer requirements). The prudential formula may be satisfied with buffer
capital
, i.e.
in such a case the double use of CET1 is
allowed
, and
can be applied for a number of resolution entities
to 30% of certain categories of banks and under specific conditions
. None of the Italian banks is subject to the
Figure 3: Styli
ed interaction
RW and MREL
RW frameworks
amounts expressed as % of RWAs, also MREL
Note:
The notion of
stacking order defines the sequence in
which different capital layers absorb losses.
In the risk
weighted
capital
framework
(the RW bar) excess capital cover
losses first
(green), followed by the CBR (yellow) and minimum requirement
Note: e notion of a stacking order denes the sequence
in which dierent capital layers absorb losses. In the
risk-weighted capital framework (the RW bar) excess capital
covers losses rst (green), followed by the CBR (yellow) and
minimum requirement (pink).
To increase the excess capital in the MREL-RW framework, the bank could issue more
MREL eligible liabilities in order to reduce the CET1 capital absorbed by the MREL-RW
minimum requirement.
A comprehensive approach for measuring overlaps
A comprehensive measure of the overlaps (and hence CBR usability) can only
be obtained by jointly comparing the use of the CET1 capital in each of the
regulatory frameworks in place.
In panel (a) of Figure 4, a hypothetical bank is characterized by a binding MREL-LR. In
this example, the MREL-LR is the requirement that absorbs most CET1, fully overlaps
with the CBR in the RW framework and partially overlaps with the CBR in the MREL-RW
framework. e CBR usability is therefore reduced to 50 per cent (1.5 per cent of
RWAs instead of 3.0 per cent, as shown by the last bar in panel (a) of Figure 4).
In this case, if we only consider the CBR stacked on top the RW framework, the buer
usability would be equal to 0 per cent (i.e. the CBR usability from the interaction
between the RW and MREL-LR).
and our comprehensive approach can also occur when a bank is characterized
with/using? CET1 is higher than the RW requirement.
In panel (b) of Figure
4, the CET1 absorbed by the MREL-RW and MREL-LR is higher than the CET1
Figure 4 – Final eects of the overlaps
(per cent of RWAs)
MREL-LR binding
MREL-RW binding
Table 1
highlights the contribution of
the comprehensive approach being proposed
by applying it
to actual data
Italian banks.
round one fourth of the Italian banks
constrained by one of
the leverage
based requirements (LR, MREL
LR, TLAC
LR): in such cases,
overlap occurs and
reduces
the CBR usability.
The table
difference
usability
Figure 4: Final effects of the overlaps
(per cent of RWAs)
LR binding
RW binding
Table 1
highlights the contribution of
the comprehensive approach being proposed
by applying it
to actual data
Italian banks.
round one fourth of the Italian banks
constrained by one of
the leverage
based requirements (LR, MREL
LR, TLAC
LR): in such cases,
overlap occurs and
reduces
the CBR usability.
The table
difference
usability
Figure 4: Final effects of the overlaps
(per cent of RWAs)
LR binding
RW binding
absorbed by the RW requirement. The CBR usability would be equal to 0 per cent
if we only compared the interaction between the RW and MREL-LR requirements.
With the full comparison proposed in this note, the CBR usability is instead 100
per cent, since the CBR in the MREL-RW framework does not overlap with any
other requirement.
Table 1 highlights the contribution of the comprehensive approach being proposed
by applying it to actual data for Italian banks. Around one fourth of the Italian
banks
are constrained by one of the leverage-based requirements (LR, MREL-LR,
TLAC-LR): in such cases, an overlap occurs and reduces the CBR usability. The table
shows the difference in the CBR’s usability between the proposed approach and
the RW approach. As already illustrated in the two previous examples, applying
the comprehensive approach reveals a significantly higher CBR usability:
the CBR usability increases from 26.7 to 73.6 per cent for the whole Italian
banking system.
This improvement is driven by the Italian banks subject to
MREL requirements (11 out of a total of 150 banks, which account for 80 per cent
of total system assets), whose CBR usability increases from 10.8 to 69.0 per cent.
Indeed, when the MREL-RW requirement met with CET1 is higher than the RW
one (as in panel (b) of Figure 4), the CBR is more usable than is apparent from the
approach based solely on the RW requirement. This explains why, by considering
the regulatory requirements of the resolution framework as well, the usability of
the CBR increases.
e dierence between the two approaches in measuring the CBR’s usability also emerges
from the distribution of the banking system RWAs by bucket of CBR usability (Figure 5).
In particular, according to the RW approach, almost 70 per cent of the banking system’s
RWAs belong to banks with a very limited CBR usability (between 0 and 25 per cent).
Conversely, according to our approach, 85 per cent of the banking system’s RWAs are
attributable to banks with a medium/high CBR usability (over 50 per cent).
From a theoretical point of view, another discrepancy between the RW framework approach and our comprehensive
approach could arise when the LR is binding, if the CET1 absorbed by the MREL-RW is slightly higher than the one
absorbed by the RW requirement. is particular case does not apply to Italian banks.
Table 1 – CBR usability of Italian banks
(per cent of CBR; data as of December 2020)
RW approach
Comprehensive approach
Banks with MREL requirements (11 banks)
Whole system (150 banks)
ource: Supervisory and resolution reporting.
ote: e assumption of closing the shortfalls applies in each column. In our calculations, the P2R is included in the minimum RW requirements.
Figure 5 – Distribution of Italian banking system RWAs according to CBR usability
(share of banking system RWAs by bucket of CBR usability; data as of December 2020)
Table 1: CBR usability of Italian banks
(per cent of CBR
; data as of December 2020
approach
Comprehensive
approach
Banks with MREL requirements (11
banks)
Whole system (150 banks)
Source: Supervisory and resolution reporting.
Note: The assumption of closing the shortfalls applies in each column. In our
calculations
2R is
included in the minimum RW requirements.
The difference between the two approaches
measuring the
usability
emerges from
distribution
banking system
by bucket of CBR usability
igure
In particular
according to the
approach
almost 70
per cent
banking system
RWAs
belong to banks
with
very
limited CBR usability (between 0 and
per cent
Conversely,
according to our
approach
per cent
of the banking system
RWAs
attributable to banks with a medium/high
CBR usability (
over
per cent
Figure 5
Distribution
Italian
banking system
according to
CBR usability
share
of banking system
RWAs
by bucket of CBR usability
data as of December 2020
Source: Supervisory and resolution reporting.
Source: Supervisory and resolution reporting.
The materiality of the overlaps and the related buffer usability of the Italian banks
is analysed based on bank-level data as of December 2020.
The analysis covers
150 banks, combining data from supervisory and resolution reporting (MREL).
We report the results based on consolidated data at the banking group level,
applicable.
When
analysing
the materiality of the overlaps, a steady state perspective
is assumed: it means that banks close their shortfalls and maintain an
excess capital above each regulatory requirement
. The excess capital is the
CET1 in excess that banks keep above all the capital requirements or the CBR when
applicable; we set the minimum excess capital at 1 per cent of the RWAs. These
adjustments ensure that the CBR’s usability is not limited simply because banks do
not fulfil future requirements yet (i.e. MREL).
Overlaps are in fact a structural
phenomenon: they are generated by the current design of the regulation
and not by temporary circumstances, i.e. they can also occur when banks
are compliant with all the applicable requirements.
We assume that possible
shortfalls are closed using the cheapest available eligible funding source: AT1
instruments, senior non-preferred bonds and senior unsecured bonds respectively
for LR, the MREL subordinated component and the MREL senior component; and
CET1 for RW.
We develop our analysis by assuming:
the expiration of the exemption of the central bank exposures from the
denominator of the Leverage Ratio;
the application of the intermediate MREL targets, binding starting from
1 January 2022; these targets have been set by NRAs as part of the 2020
resolution planning cycle and are compliant with the banking package;
the current capital ratios without any adjustments due to future implementation
of Basel III and the phase-in of IFRS9.
understand the limits of the empirical analysis
. First of all, we provide a static
balance sheet analysis: the actions that banks could take in the coming years to adjust
their balance sheet to comply with future requirements are not taken into account
MREL decisions are based on the 2020 resolution planning cycle (MREL policy 2020). It was the rst cycle based on the
Banking Package (Directive 2019/879, the Bank Recovery and Resolution Directive II).
e MREL eligible liabilities are computed by the hybrid approach in accordance with the MREL policy 2020. When
the resolution group perimeter diers from the prudential one, we use the former for MREL purposes and the latter
for RWs and LR.
e Bank Recovery and Resolution Directive II requires the NRAs to set a fully phased MREL target in 2024 (1 January);
the Directive also require NRAs to set an intermediate level of MREL to be met by 1 January 2022.
(for example, balance sheet composition policies like de-risking or credit crunch).
Second, the leverage ratio pillar 2 requirement is not included in our calculations. Third,
the calculations provided are point-in-time: data as of December 2020 are affected
by the COVID-19 crisis, which has probably reduced the risk-density of some banks;
in addition, MREL targets are calibrated every year and could be softened or
strengthened by the NRAs in the next resolution cycles.
As are other requirements, like the CCyB, OSII and so on, which are adjusted periodically (quarterly or annually).
B ’I
Notes on Financial Stability and Supervision No. 30 – June 2022