Transcript From G-SIB to D-SIB

Systemically important banks:
From G-SIB to D-SIB
identification and regulation
Jan Frait
Executive Director
Financial Stability Department
June 2013
1
Outline
I. Chronology of the G-SIB and D-SIB debate
II. Concept of systemic importance
III. How to identify SIBs in the Czech banking sector
IV. BCBS recommendations for D-SIB regulation
2
I.
The G-SIB and
D-SIB debate thus far...
3
Chronology of the G-SIB and G-SIFI debate
• 2009:
• April G20 meeting endorses plan to strengthen resilience of globally
operating banks (G-SIB)
• In October IMF, BIS, FSB issue “Guidance to assess the systemic
importance of financial institutions, markets and instruments”
• 2010:
• October G20 meeting approves FSB document “Reducing the moral
hazard posed by systemically important financial institutions”
• 2011:
• June G20 meeting confirms intention to increase capital requirements
for G-SIBs
• In November BCBS issues “Global systemically important banks:
Assessment methodology and the additional loss absorbency
requirement”: SIB surcharge 1–2.5% CET1
• November G20 meeting endorses FSB’s policy framework on
systemically important financial institutions (G-SIFIs); initial list of 29
G-SIBs published
4
From G-SIB to D-SIB
• 2011:
• At November G20 meeting FSB is asked to work, in
consultation with BCBS, on definition of modalities to extend
expeditiously G-SIB framework to domestically important banks
(D-SIBs)
• 2012:
• In October the FSB issued final version of framework for D-SIB
identification and regulation
• Some countries are at advanced stage of their own D-SIB framework
preparations (SE, NL, CH) with intention to implement them as soon
as possible (in some cases from 2013 onwards)
• CH – SIFIs will have 10% CET1 plus 9% contingent capital
• SE – SIFI surcharge up to 5% CET1
• NL – SIFI surcharge 1–3% CET1
5
CET1 capital for Swedish SIFIs
Countercyclical
buffer
0-2,5 %
Countercyclical
buffer
12 %
0-2,5 %
0-2,5 %
2.5%
2%
2%
4.5%
Minimum
requirement
Current requirement
Basel III
2.5%
2.5%
10 %
Countercyclical buffer
Capital conservation
buffer
7%
Capital conservation
buffer
Minimum
requirement
3.0 %
4.5%
Proposed 2013
SIFI-surcharge
Minimum
requirement
5.0 %
4.5%
Proposed 2015
6
CET1 capital for Nordic SIFIs
Source: ICMA
7
From G-SIB to D-SIB
• 2013:
• CRD IV/CRR – OSII buffer, systemic risk buffer
• Major deviation from BCBS recommendations
8
II.
Concept of systemic importance
9
Defining Systemically Important Financial Institution
• For the purposes of microprudential supervision a SIFI can be
defined as an institution whose failure would cause its creditors
and shareholders to suffer large losses in the form of direct costs.
• From a macroprudential perspective, SIFI is a system component
that contributes significantly to the accumulation of systemic risk
and/or whose failure would impose large losses on its surroundings
and would threaten the smooth functioning of the system as a
whole and have adverse knock-on effects on the real economy.
Komárková, Hausenblas and Frait (2012, CNB FSR 2011/2012)
• Macroprudential factor is more important, since it is the indirect
impacts on the SIFI’s surroundings which have the potential to
trigger a major and protracted crisis.
10
Defining Systemically Important Financial Institution
• A SIFI can be defined from the negative point of view as an
institution whose uncontrolled failure has the potential to seriously
damage the financial system, but it can also be defined from the
positive point of view as an institution whose viability is crucial for
the smooth functioning of the financial system and the real
economy.
• What do SIFIs entail in practice:
• positive externalities (economies from scale/scope, cross-border
allocation of resources…),
• moral hazard (SIFI enjoys its „too-important-to-fail“ statute – implicit
government guarantee distorts the markets),
• negative externalities (SIFI contributes to systemic risk which it is not
able to fully manage neither it bears the costs).
11
Imperfect markets and moral hazard
• Implicit government guarantees encouraged large banks
to take on more risks.
• Limited liability creates distorted incentives.
• Debt holders incorporate the guarantees into the price of
funding → debt is too cheap and the big get even bigger.
• Risk-taking, high-leverage → probability of a crisis is
higher.
• High burden on tax-payers + negative impact on
economic growth.
• The gap further widened after the outset of crises.
• The crises only increased the level of concentration when
guarantees became more explicit.
12
Concept of systemic importance
• Thomson (2009):
• Size and concentration (too big to fail)
• Interconnectedness and risk of contagion ( to
interconnected to fail)
• Degree of similarity and correlation of exposures (too
many to fail)
• Context (economic and political conditions).
• Size is hardly the only factor.
• We should not reduce the problem to “too-big-to-fail” or
“too-important-to-fail”.
• SIFI is a complex component of the structural dimension
of systemic risk.
13
The role of macroprudential policy
• Identification of SIFIs.
• Measures to restrict the
• size, complexity and scope of SIFIs…
• Measures to reduce the probability of failure (PD):
• capital surcharge, intensive supervision,
• disclosure/transparency,
• bail-in approach.
• Measures to reduce potential costs of failure (LGD) via
• ring-fencing,
• effective resolution framework,
• mandatory resolution plans (goodwill).
14
III.
Identification of
Systemically Important Banks
in the Czech Banking Sector
15
Methods for identification of SIFI (1)
• Econometric models (PDxLGD approach)
• losses of the system as a whole analyzed, SIFIs are banks with largest
contributions to systemic events,
• CDS (or equity prices or any market indicator) covariance during extreme
events,
• bias in market prices not accounted for due to assumption of highly effective
markets (market participants may include bail-out in prices in reality).
• Computational simulations, stress-testing (LGD)
• simulation of whole spectrum of shocks
• free to implement any channel (usually defaults of one or more banks, firesales, liquidity shocks, …),
• for each bank a distribution of losses (resulting from its default) born by the
rest of the economy is estimated,
• shapley value, extreme value observations…
• generally rather academic.
17
Methods for identification of SIFI (2)
• Methods of quantitative and qualitative indicators (LGD)
• systemic importance derived from simple indicators,
• can be easily to implemented,
• make use of data from regular reporting (supervision, payment
system, Bankscope, …) – (off-)balance sheet exposures,
transactions
• considerable amount of expert judgment is necessary,
• problem of setting the threshold (do we really need it?)
18
Underlying paper
• CNB‘s Financial Stability Report 2011/2012 (June 2012)
• feature article „How to Identify Systemically Important
Financial Institutions”,
• The objective of the article is to show the results of the
SIFIs identification in the Czech banking sector using the
BCBS method developed for G-SIB identification.
19
Indicators of systemic importance
• BCBS (2011) method derives systemic importance from indicators
aggregated into five categories:
•
•
•
•
•
Size
Cross-border activity
Interconnectedness
Non-substitutability/uniqueness
Complexity.
• Only quantitative indicators‘ method applied, a set qualitative
judgment by experts and supervisors should come next.
• balance-sheet and off-balance-sheet indicators,
• network analysis based on data from interbank money market and
payment system.
20
Size and cross-border activity
• Size
• the most natural measure of SIFI,
• balance sheet size and business activity,
• indicators: total assets, credit exposure, revenues.
• Cross-border activity
• potential channel of contagion from abroad,
• includes exposures to foreign sovereign risk and links to
parent firms,
• indicators: liabilities and assets with foreign entities.
21
Interconnectedness
• Interconnectedness
• the problems of interconnected institutions directly
threaten the rest of system through mutual exposures,
• default of an interconnected bank entitles huge losses for
its counterparts on interbank market, for counterparts of
counterparts etc,
• domino effect increases systemic stress, may contract
interbank market and boost credit and liquidity risk.
• indicators: centrality in network of exposures, assets and
liabilities to credit institutions.
22
Limited substitutability and complexity
• Limited substitutability/uniqueness
• some institutions provide unique services for the rest of the
economy so that they cannot be easily and promptly replaced.
• it can be a small institution in absolute terms, but it form a market
on its own (CCP, stock exchange, insurance…).
• indicators: assets under custody, payments cleared and settled
through payment systems, values of underwritten transactions in
debt and equity markets (not implemented yet).
23
Limited substitutability and complexity
• Complexity
• increases risks of insufficient supervision, regulatory arbitrage
and costly crises resolution.
• indicators: trading book value and available for sale value, level
III assets (irrelevant in CZ), OTC derivatives notional value (not
implemented yet).
24
Indicator-based method – data
• Both supervision and payment system integrated in the
CNB → in-house data.
• detailed balance-sheet reporting,
• 15 largest exposures to credit institutions (to built the
interbank market network),
• payment system transactions (to built payment system
network),
• off-balance sheet items.
25
Network analysis and the concept of centrality
• Position in a network (importance of individual
nodes/banks) is given by both number and size of
exposures and by the topology (shape) of the network.
• To avoid performing complex network simulations of
contagion we can reduce the analysis to measures of
centrality.
• Heavily used in other sciences (physics, information
theory and social sciences…)
• Are there some central nodes of the network (hubs) that
may be critical for performance of the whole system?
26
Centrality measures (1)
• Degree
• Number/total size of interbank claims
or liabilities
• Betweenness
• How does the bank interconnect all
the parts of the network,
• Better reflects the whole structure of
the network,
• Eigenvector centrality
• Connectedness to highly
interconnected banks
27
Centrality measures (2)
Cumulative degree distribution
100
• Financial networks
have scale-free
properties – efficient
and quite robust but
heavily dependent
on few of its
elements.
10
1
MIN
1.00E+06
MAX
1.00E+08
1.00E+07
Interbank market
CERTIS
Source: CNB, authors
28
Results of the network analysis (MM)
• Network analysis:
• Position in network
(importance of
individual
nodes/banks) is given
by both number and
size of exposures and
by the shape of the
network
• Interconnectedness
points to 5 potential
SIBs (other
categories lead to
different numbers)
• By size, interbank
exposures are not
dominant in banks’
balance-sheets.
high exposure
low centrality
low exposure
low centrality
high exposure
high centrality
Nodes with the fewest links are located at the edge; the
interconnectedness of the nodes increases towards the centre.
Account is taken of the number, size (indicated by the thickness of
the link) and frequency of the links.
29
Indicator-based method – results (1)
• All factors of systemic
importance have similar
distribution:
Estimated distribution of quantitative indicators
(x-axis: average score; y-axis: number of banks)
14
•
12
10
• low number of banks with
high values (high importance) 8
and large number of banks 6
4
with low value.
• Systemic importance is highly
concentrated in the Czech
banking system in all
dimensions.
Large number of
banks with low values
of indicators
2
0
0
6
12
18
24
Size
Interconnectedness
Complexity
30
Cross-border activity
Substitutability
Source: CNB, authors
•
High values of
indicators in case of a
few banks
30
Indicator-based method – results (2)
Complexity
Substitutability
Interconnectedness
• Indicators are unique (low overlap)
• Systemic importance cannot be
reduced only to size.
• However, the biggest banks are
also the most interconnected,
active in cross-border business
etc…
Cross-border
activity
Correlation matrix of indicators aggregated into categories
Size
Higher triangle does not cover
five on average most important
banks.
Size
Cross-border
activity
Interconnectedness
Substitutability
Complexity
Lower triangle of correlation
matrix computed on full sample.
0
Source: CNB, authors
1
31
Indicator-based method – results (3)
individual scores
• Four banks have composite indicator of significance above average
for whole sector – these may be labeled SIBs
• Other banks that have below-average indicators of importance (they
form a cluster below the line) Composite indicators for individual banks in all three variants
Weighted averages:
30
• Baseline BCBS: equal weights 25
• Alternative 1: accounted for
20
features of Czech banking
15
(lower weight to
10
interconnectedness and
Mean = 4.35%
5
complexity).
0
• Alternative 2: assumption of
60
40
20
0
70 80
cummulative sum of individual scores
potential increase in
Mean
BCBS method (2011b)
Alternative 2
Alternative 1
interconnectedness and
Source: CNB, authors
complexity in the future.
100
32
Possible improvements of the method
• Additional indicators?
• share of deposits insured by the Deposit Insurance Fund
on the total size of the Fund.
• interconnectedness to specific “systemically important”
corporations or industries.
• burden of potential “bridge bank” on the public budget.
33
Beware
• BCBS method can be used to sort a list of institutions according to
their importance and for identification of a group of banks with high
deviation from average importance.
• The approach cannot tell the banks capable to absorb systemic
risk from those that can spread it in the financial sector or to the
real economy.
• The approach is therefore not directly usable for setting the SIB
capital surcharges (SCS) for individual banks.
• Further analysis (especially of substitutability and complexity) and
additional methods (correlation of balance sheets) would be
necessary.
• No method can be applied mechanically, expert judgment is
necessary.
34
SCS setting
•
•
When setting SCSs it is vital to start by realising that the objective is primarily
macroprudential.
• The aim is not to stop SIFIs from getting into trouble or failing, but to at least
partially reduce the likelihood of systemic crises, and in particular (as in the
case of countercyclical capital buffers) to reduce the intensity of such crises
by absorbing the negative impacts of SIFI failures.
• SCSs should therefore enhance the resilience of the system in terms of loss
absorbency.
The usual objection to SCSs is that capital is costly and SCSs force banks to
downsize their balance sheets.
• The advocates of SCSs usually respond to this objection by saying that
investors in bank shares are currently demanding a higher rate of return
than investors in non-financial corporations’ shares, because banks are
more leveraged and therefore more risky.
• If SCSs cause leverage to fall, banks will become less risky and investors
will not demand such high returns on their shares. And if SIFIs decide not to
provide some types of loans because of SCSs, they will be substituted by
smaller non-SIFI banks subject to lower capital requirements.
35
SCS setting
• Another argument against the introduction of SCSs is based on
economies of scale and states that the global financial system
needs SIFIs.
• This argument is undermined by the fact that some global banks have
expanded well beyond the point at which the economies of scale can
increase any further and which is needed to fund activities in the
global economy.
• Consequently, the main argument against identifying SIFIs and
imposing SCSs is still that labelling an institution as a SIFI leads to
growth in the moral hazard associated with it, since both it and the
public are being told de facto that it is “too important to fail”.
• The counter-argument is that market participants do not need a
formal label to know which institutions are systemically significant.
• In this situation, imposing capital surcharges on such an institution
can only reduce its riskiness, as they will cause individual and
economy-wide expectations of its profitability to converge. This is
because the surcharges will increase the institution’s funding costs,
which were previously held down because the market regarded it as
being too important to fail.
36
SCS setting
• In principle, SCSs should be set to reflect the difference between the
private and social profitability of the SIFI.
• Consequently, the size of the SCS should create an incentive for SIFIs to
stop expanding, to stop becoming more interconnected and to stop taking on
more risks if there are no related benefits for the economy as a whole.
• On the general level, the calibration of SCSs should be based on an
estimate of the potential impact of SIFI failures on the entire financial
system, which should be compared with the costs of additional regulation.
• This implies that the principal factors when setting SCSs should be those
which determine the potential impact of failures.
• If we accept the assumption that SIFI failures are the exception rather
than the rule since most SIFI resolution situations end in rescue, the
expected social costs of a SIFI rescue operation should be relevant for
determining the SCS.
• Those costs can be determined by the institution’s expected losses in a
worst case scenario. This is why the Basel Committee focused directly on
the additional capital needed to absorb SIFIs’ losses when formulating its
SCS recommendations.
37
SCS setting
• The models used by the BCBS to estimate this capital produced
results lying in a wide range of 1–8% of risk-weighted assets in
terms of CET1 equivalent, with a central tendency of around 2–4%.
• The final recommended calibration for G-SIBs of 1–2.5% was
therefore in the lower half of the range of estimates.
• This low SCS calibration assumes coverage by high-quality capital
capable of fully absorbing losses. If the risks of SIFIs are to be
covered also by lower-quality capital, the total SCSs should be set at
a higher level.
• SCSs are just one of the elements that can mitigate the systemic
risks associated with SIFIs. The resolution framework occupies first
place in the overall G-SIFI supervisory framework.
• The specific resolution rules should allow a SIFI to remain a going
concern during a crisis, and the shareholders must accept losses
associated with the institution’s poor financial condition.
38
IV.
BCBS recommendations for D-SIB regulation
(A framework for dealing with domestic systemically
important banks, October 2012)
39
BCBS principles for the D-SIB framework
• General approach:
• Framework has to be based on assessment by local authorities, who are
best placed to identify which banks are systemic within their borders
• G-SIB framework is overly prescriptive and application largely mechanical;
D-SIB framework should be more principle-based and include appropriate
degree of national discretion
• Necessary to avoid “double counting” of capital for D-SIBs that are part of GSIBs or D-SIB subsidiaries that are part of D-SIB parent groups
• A D-SIB framework is best understood as taking the
complementary perspective to the G-SIB regime by focusing on the
impact that the distress or failure of banks (including by
international banks) will have on the domestic economy.
• Home authorities should assess banks for their degree of systemic
importance at the consolidated group level, while host authorities
should assess subsidiaries in their jurisdictions, consolidated to
include any of their own downstream subsidiaries, for their degree
of systemic importance.
40
BCBS principles for the D-SIB framework – assessment
methodology
• The impact of a D-SIB’s failure on the domestic economy should, in principle, be
assessed having regard to bank-specific factors:
• (a) Size;
• (b) Interconnectedness;
• (c) Substitutability/financial institution infrastructure (including considerations
related to the concentrated nature of the banking sector); and
• (d) Complexity (including the additional complexities from cross-border activity).
• In addition, national authorities can consider other measures/data that would
inform these bank-specific indicators within each of the above factors, such as
size of the domestic economy.
• National authorities should have national discretion as to the appropriate
relative weights they place on these factors depending on national
circumstances.
• National authorities should publicly disclose information that provides an outline
of the methodology employed to assess the systemic importance of banks in
their economy.
41
BCBS principles for the D-SIB framework – assessment
methodology
• National authorities may choose to include some country-specific
factors.
• A good example is the size of a bank relative to domestic GDP. If the
size of a bank is relatively large compared to the domestic GDP, it
would make sense for the national authority of the jurisdiction to
identify it as a D-SIB whereas a same-sized bank in another
jurisdiction, which is smaller relative to the GDP of that jurisdiction,
may not be identified as a D-SIB.
• See Skořepa, M.-Seidler, J. (2013): An Additional capital
requirements based on the domestic systemic importance of a
bank. Czech National Bank Financial Stability Report 2012/2013,
pp. 96-102
42
BCBS principles for the D-SIB framework – higher loss
absorbency (HKA, ie. SCS)
• National authorities should document the methodologies and considerations used
to calibrate the level of HLA that the framework would require for D-SIBs in their
jurisdiction.
• The HLA requirement imposed on a bank should be commensurate with the
degree of systemic importance.
• National authorities should ensure that the application of the G-SIB and D-SIB
frameworks is compatible within their jurisdictions.
• Home authorities should impose HLA requirements that they calibrate at the parent
and/or consolidated level, and host authorities should impose HLA requirements that
they calibrate at the sub-consolidated/subsidiary level.
• The home authority should test that the parent bank is adequately capitalised on a
stand-alone basis, including cases in which a D-SIB HLA requirement is applied at
the subsidiary level.
• The HLA requirement should be met fully by Common Equity Tier 1 (CET1). In
addition, national authorities should put in place any additional requirements and
other policy measures they consider to be appropriate to address the risks posed
by a D-SIB.
43
D-SIB regulation in the EU
• Discussions whether and how implement D-SIB surcharge in CRD IV.
• SIB identified on national level and allocated to 5 categories according to
systemic relevance.
• Surcharge 1% plus 0,5% for each subsequent category
• The general approach discussed in the final stage finally not containing
this concept.
• However, the systemic risk buffer may serve regulators as a
substitute.
• Some EU authorities still signal readiness to go for it as soon as
possible.
• The way of introducing the SIB surcharge difficult to predict
• On which level, who decides, what size, how to avoid duplicities
between G-SIB and D-SIB?
• Banking union consideration may have some impact.
44
Děkuji za pozornost
www.cnb.cz
Prof. Dr. Ing. Jan Frait
ředitel samostatného odboru finanční stability
[email protected]
Kontakt na samostatný odbor finanční stability ČNB:
E-mail: [email protected]
http://www.cnb.cz/cs/financni_stabilita/
KP v Basel III ve formě equity: 4,5 % CET1 + 2,5 % konzervační polštář + až 2,5 % proticyklický
polštář.
Celkový požadavek je: minimum Tier1 6% (CET1 4,5 % - other T1 2,5 %) + Tier2 2 % + 2,5%
konvervační polštář plus až 2,5 % proticyklický polštář
46
47
Basel III capital – GS 1
•
The starting point is Basel III new regulatory capitalization minimums
(cont.):
• Goldman Sachs view of Basel III (GS Global Investment Research):
1. CT1 ratio of 7% as the new regulatory minimum for banks of non-systemic
importance - for banks deemed to be of systemic importance, the CT1
minimum is subject to an additional surcharge and is therefore to be set at a
level above the 7% CT1 minimum.
2. A regulatory minimum is just that: a minimum. In practice, banks will aim to
exceed the minimum to be on the safe-side; and exceeded it further, before
capital return to shareholders is considered. 7% is not the “magic” number;
rather, it is a floor.
3. From an equity investor’s perspective, the relevant level of capital is not the
regulatory minimum but rather one above which all key parties—bank
managements, regulators, debt holders, “the market”—would not object to
capital being returned to shareholders. This level—the GS target
capitalization—will also differ among banks.
48
Basel III capital – GS 2
•
The starting point is Basel III new regulatory capitalization minimum
Goldman Sachs view of Basel III
49