which protection for bank liabilities?econ.sciences-po.fr/sites/default/files/merged_25...
TRANSCRIPT
Which Protection for Bank
Liabilities?
Mathias DewatripontNational Bank of Belgium
and
Université Libre de Bruxelles (ECARES & Solvay Brussels School)
Séminaire Sciences Po / Banque de France
Paris, March 25, 2015
2
Outline
1. A brief history of (de)regulation
2. On the cost of bailouts
3. Reregulation
4. The BRRD and financial stability
5. Trading off insurance and incentives
6. Conclusion
1. A brief history of (de)regulation
3
4
History of (de)regulation
• Banking is risky (maturity transformation).
• Almost century-old ‘cycling’ between 3 objectives:
productively efficient banking; financial stability (in
particular, no bank runs); fighting moral hazard (‘no
bailouts’).
• Until 1930’s: sacrifice financial stability, but many
bank runs, in particular in the Great Depression.
• From mid-1930’s to early 1970’s: sacrifice efficien-
cy, with strict limits on competition (on entry, size,
prices & activities); & introduce deposit insurance.
• No more bank runs & no bailouts but low product-
ive efficiency in banking (e.g. overbranching) +
development of nonbank competitors.
5
History of (de)regulation (2)
• As a result, gradual deregulation since 1970s, on
prices and entry, & on size and set of activities.
• But deposit insurance maintained (against financial
instability) and focus on (risk-based) bank solvency
(against moral hazard): Basel I and II capital ratios.
• Impact: since 70s, very few runs, but many banking
crises (147 worldwide (Laeven-Valencia, IMF, 2012)).
• Many linked to macro imbalances, but also to bank
behavior (moral hazard), especially when underca-
pitalized and ‘gamble for resurrection’ (easier for
banks than for regular firms: have access to ‘safe’
funding, just need to raise return on deposits).
6
History of (de)regulation (3)
• Interest rate and entry deregulation did benefit cus-
tomers, but at times at expense of financial stability.
• Mixed picture at best w.r.t. innovation (e.g. ATMs
versus very complex new financial products), and
w.r.t. size and scope (are big (universal) banks profits
and high management wages due to scale/ scope
economies or to market power and 'too-big-to-fail'
subsidy?).
• On the other hand, (Basel I/II) solvency (and liqui-
dity) in 2008 clearly insufficient.
• Problem of both capital ratio level and banks’ ability
to ‘manage’ it (internal models, securitization, … ).
7
Additional elements of the 2008 crisis(see Dewatripont-Rochet-Tirole, Fahri-Tirole)
• Household overindebtness (especially in the USA).
• Securitization and therefore complexification of financial products.
• Role (and conflict of interest) of rating agencies.
• Extreme illiquidity for some banks, with massive recourse to (very unstable) wholesale funding.
• Race for higher and higher ‘return on equity’.
• Role of globalisation as an incentive to deregulate ('race to the bottom').
8
Responses to the 2008 crisis
• Crisis significantly worsened after fall of Lehman : first big-bank bankruptcy, that triggered « move to another equilibrium » (à la Diamond-Dybvig, but for wholesale funding).
• Double response:
(i) « no more Lehmans », instead, significant rise of (retail) deposit insurance and massive bail-outs;
(ii) re-regulation.
2. On the cost of bailouts
9
10
Banking crisis outcomes(% of GDP; Source: Laeven-Valencia, 2012)
Area (Gross) fis- Increase Output
cal cost* in debt** loss***
Japan (1997) 14.0 42 45
Sweden (1991) 3.6 36 31
USA (1988) 3.7 11 0
USA (2007) 4.5 24 31
Euro area (2008) 3.9 20 23
*: committed funds, to date (but (almost) fully re-
paid in the case of Sweden, & USA 2007).
**: three years after the crisis; ***over 3 years,
relative to trend.
11
Some lessons
• Crises typically lead to very low growth (with po-
tential vicious circles). Exception: US S&L crisis
(more 'regional').
• Gross fiscal cost of bailout is only a fraction of
debt increase.
• Why was end outcome concerning this fiscal cost
so much worse in Japan but also in S&L crisis
than in the US in 2007 or in Sweden (where most
of it got reimbursed)?
12
The Japanese crisis (1992-?)
• Familiar starting point: burst of real estate and
stock market bubbles, then negative bank-real-
economy link (see Hoshi-Kashyap, 2004).
• Key problem: insufficient recapitalization led
banks to hide losses and favor loss-making
existing corporate customers rather than more
promising new borrowers. Such 'zombie lending'
led to collapse of productivity (see Caballero et
al., 2008).
13
The Swedish crisis (1991-3)
• Fuelled by deregulation and real estate bubble.
• Dealt swiftly through nationalization of big banks
(with shareholders wiped out). So, no lingering
undercapitalization, thanks to availability of
public money (repaid in the end).
• GDP significantly helped by international growth
and depreciation of currency.
• See Jonung (2009).
14
Two contrasting US examples
• Savings and Loan crisis of the 1980s: much
smaller than recent one to start with, but
procrastination for many years. Accounting
gimmicks instead of recapitalization (FSLIC
without money at the time, Congress unwilling to
help ...), while losses mounted due to gambling
for resurrection by S&L's (see Dewatripont-
Tirole, 1994).
15
Two contrasting US examples (2)
• Current worldwide crisis originated in the US,
with subprime complex products.
• Still, US now in better shape than Europe.
• Key: TARP (Trouble Asset Relief Program) in
2009, at cost of $428 billion, but with net cost for
the taxpayer today of ... only $21 billion, i.e.
0.1% of US GDP ! (CBO estimate, May 2013).
16
Euro area
• Two different crises since 2008: (i) ‘subprime-
Lehman’ (trading book) crisis, which mainly
affected Northern Europe; (ii) Euro (and Spanish
housing) crisis, which mainly affects Southern
Europe.
• First one dealt with ‘US-style’ (see Beck et al,
Pisany-Ferry-Sapir), even if more gradually.
Question: enough or not?
• Second one still a ‘moving target’, depending on
GDP evolution.
17
Conclusion
• Procrastination really costly.
• Instead, swift intervention may pay for taxpayer
(even if ex-post net-cost computations fail to
take into account risk premia).
• Tradeoff current/future crisis: fighting moral
hazard good, but NOT worth delaying restruc-
turing, because lower GDP growth will raise final
cost for taxpayer !
3. Reregulation
18
19
Reregulation: busy reform agenda
• Mix of (i) continuity (with recalibration) and (ii) change: (iia) back to regulation of what a bank may/should be; (iib) introduction of 'system regulation'.
• (i) More and better capital (and an additional, simpler, leverage ratio).
• (iia) Liquidity ratios, recovery & resolution plans, large-bank surcharges, structural reforms. (Vickers, Volcker, Liikanen/Barnier/…).
• (iib) Macroprudential instruments (Counter-cyclical Capital Buffer, ...).
20
Assessment
• Reform agenda makes sense given previous crisis. Does involve a partial U-turn w.r.t. laisser-faire approach to banking activities.
• Impact of new approaches (liquidity, recovery & resolution, structural reforms, systemic approach to regulation) still untested.
• Debate continues on 'excessively low Basel III capital ratios' (e.g. Admati-Hellwig, 2013) vs 'difficulty of finding the money & risks to real-economy lending'.
• What to think about new trend: bail-in rather than bailout?
21
Bail-in
• Paradox of the crisis: (i) Basel III stresses quality of capital and micro/macroprudential distinction, while (ii) current « bailout fatigue » has now led to « bail-in fashion », with a desire to vastly enlarge set of bank claimholders meant to be « held respon-sible », and this even under systemic stress.
• Explanation: politicians and public at large do not feel that Basel III requires enough capital to protect taxpayers.
• Two concerns however: (i) cost of financial instabi-lity; (ii) who should bear risk?
• Relevant in particular in the EU, with BRRD (focus here, linked to FSB’s TLAC).
4. The BRRD and financial stability
22
23
Banking Recovery & Resolution Directive
“Other tools (than bail-in) can be used to the extent that they conform to the principles and objectives of resolution set out under the BRRD. In circum-stances of very extraordinary systemic stress, authorities may also provide public support instead of imposing losses in full on private creditors. The measures would nonetheless only become avail-able after the bank’s shareholders and creditors bear losses equivalent to 8% of the bank’s liabi-lities and would be subject to the applicable rules on State Aid.” (FAQs on BRRD)
24
Banking Recovery & Resolution Directive
“Bail-in will potentially apply to any liabilities of the institution not backed by assets or collateral. It will not apply to deposits protected by a deposit guaran-tee scheme, short-term inter-bank lending or claims of clearing houses and payment and settlement sys-tems (that have a remaining maturity of seven days), client assets, or liabilities such as salaries, pensions, or taxes. In exceptional circumstances, authorities can choose to exclude other liabilities on a case-by-case basis, if strictly necessary to ensure the conti-nuity of critical services or to prevent widespread and disruptive contagion to other parts of the financial system, or if they cannot be bailed in in a reasonable timeframe.” (FAQs on BRRD)
25
Banking Recovery & Resolution Directive
“The write down will follow the ordinary allocation of losses and ranking in insolvency. Equity has to absorb losses in full before any debt claim is sub-ject to write-down. After shares and other similar instruments, it will first, if necessary, impose losses evenly on holders of subordinated debt and then evenly on senior debt-holders.”
“Deposits from SMEs and natural persons, includ-ing in excess of EUR 100,000, will be preferred over senior creditors.”
(FAQs on BRRD)
26
Banking Recovery & Resolution Directive
“By definition, this will depend on the systemic footprint of different institutions. Depending on their risk profile, complexity, size, interconnected-ness, etc., all banks should maintain (subject to on-going verification by authorities), a percentage of their liabilities in the form of shares, contingent capital and other unsecured liabilities not explicitly excluded from bail-in. The Commission, upon a review by EBA, could specify further criteria to ensure similar banks are subject to the same standards.” (FAQs on BRRD)
27
Comments
• BRRD insists on 8% bail-in even under systemic stress, as of January 1, 2016.
• Beyond secured liabilities, it exempts very short-term debt (up to 7 days).
• It gives priority to natural persons and SMEs.
• At this point, it does not impose hard targets for bail-inable securities (« GLAC », « MREL »).
• Suggestion: think of requiring a minimum of 8% of long-run junior liabilities (equity, hybrids and junior debt, or an « extended leverage ratio ») in order to foster financial stability.
28
Example of bank liabilities
Secured + very short-term liabilities 25
Retail deposits 40
Bail-inable senior liabilities 30
Junior liabilities 1.5
Capital 3.5
Total liabilities 100
•Losses for senior liabilities before a bailout can be considered: (8 – 3.5 – 1.5)/30 = 3/30 = 10%.
•Conclusion: to avoid bank runs (esp. with volatile wholesale deposits), better to increase junior liabilities to 4.5. Instead, including senior claims in MREL does NOT protect other claimholders !
29
Conclusion
• Aversion to bailouts understandable: taxpayer money, moral hazard, …
• Remember however the cost of financial instabi-lity: the costliest bank failure for taxpayers in last 10 years was Lehman, despite lack of bail-out, while TARP bailout has almost been fully repaid (more than 400 Billion $ out of 428).
• Remember also that « orderly » resolution will not prevent depositors from running if they can and feel their money is at risk.
• This requires sufficient long-term junior claims to absorb bail-in and reassure senior claimholders.
5. Trading off insurance and
incentives (Dewatripont-Tirole
1994a, 1994b, 2012)
30
31
Regulation as an incentive scheme
• Idea: when firm performance bad, risk for
management that control switches from (nicer)
equityholders to (tougher) debtholders.
• Representation hypothesis: in banks, debtholders
unable to exert control, so see bank regulation as
a way to replicate role of capital structure in
nonfinancial corporations.
• In a sense, Basel regulation does achieve this,
provided that control switch is credible (resolution
question),
32
Regulation as an incentive scheme (2)
• Key issue however: which performance?
• Answer: idiosyncratic performance, not perfor-
mance linked to aggregate shocks (Holmstrom) !
• This issue was ignored by Basel I and Basel II.
• Addressed to some extent by Basel III: counter-
cyclical capital buffer (similar to Spanish dynamic
provisioning).
• One problem though: this is only ‘self-insurance’,
which works provided bad shock ‘follows’ good
one, so that there is a buffer to be released !
33
Regulation as an incentive scheme (3)
• Better to introduce capital insurance (à la
Kashyap-Rajan-Stein), probably State-provided,
or automatic stabilizers (e.g. through deposit
insurance premia indexed on the business cycle).
• Based on the idea of the State as insurer of last
resort (classical in economics).
• Instead, BRRD seems to be based on ‘protecting
the taxpayer as much as possible’: OK for idio-
syncratic shocks, NOT for macro shocks !
34
Regulation as an incentive scheme (4)
• Private insurance of course potentially adequate
(provided it is credible: loss absorbency rather
than runs, and no resource constraints (AIG …).
• One way to make BRRD consistent with this
micro/macro distinction: have banks issue CoCos
whose triggers would distinguish between idio-
syncratic and macroeconomic events, so as to
appropriately discipline bank management.
• Not easy to design though. Why not complement
it with additional insurance mechanisms?
6. Conclusion
35
36
• Search for optimal tradeoff between productive
efficiency, financial stability and fight against
moral hazard continues.
• At this point, ‘protecting taxpayers’ is given
priority.
• Don’t forget however the cost of financial
instability, while there have been successful
bailout experiences in case of macro crises.
• Therefore, do design bail-in a way that will not
trigger bank runs.
• Do complement it with capital insurance against
macro risks and/or automatic stabilizers.
37
References
• Admati, A. & M. Hellwig (2013), The bankers' new clothes:
What's wrong with banking & what to do about it, Princeton UP.
• Beck, T., D. Coyle, M. Dewatripont, X. Freixas & P. Seabright
(2010), Bailing out the banks: Reconciling stability &
competition, CEPR.
• Caballero, R., T. Hoshi & A. Kashyap (2008), “Zombie lending
and depressed restructuring in Japan”, American Economic
Review.
• Congressional Budget Office (2013), Report on the Trouble
Asset Relief Program - May 2013.
• Dewatripont, M. (2014a), “European banking: Bailout, bail-in
and State Aid control”, International Journal of Industrial
Organization.
38
References (2)
• Dewatripont, M. (2014b), “Banking regulation and lender-of-last-resort intervention”, European Central Bank, ECB Forum on Central Banking, Conference Proceedings: Monetary Policy in a Changing Financial Landscape, Sintra.
• Dewatripont, M., J.C. Rochet & J. Tirole (2010), Balanc-ing the banks: Global lessons from the financial crisis, Princeton UP.
• Dewatripont, M. & J. Tirole (1994a), The prudential regulation of banks, MIT Press
• Dewatripont, M. & J. Tirole (1994b), “A theory of debt and equity: Diversity of securities and manager-shareholder congruence”, Quarterly Journal of Economics.
39
References (3)
• Dewatripont, M. & J. Tirole (2012), “Macroeconomic shocks and banking regulation”, Journal of Money, Credit & Banking.
• European Commission (2014), “EU Bank Recovery and Resolution Directive (BRRD): Frequently Asked Questions”, available at http://europa.eu/rapid/press-release_MEMO-14-297_en.htm
• Fahri, E. and Tirole, J. (2012), “Collective Moral Hazard, Maturity Mismatch, and Systemic Bailouts”, American Economic Review.
• Holmstrom, B. (1979), “Moral hazard and observability”, Bell Journal of Economics.
40
References (4)
• Hoshi, T. & A. Kashyap (2004), "Japan's financial crisis and economic stagnation," Journal of Economic Perspectives.
• Jonung, L. (2009), "The Swedish model for resolving the banking crisis of 1991-93: Seven reasons why it was successful," European Economy Economic Paper 360.
• Kashyap, A., R. Rajan & J. Stein (2008), “Rethinking capital regulation”, FRB of Kansas Economic Symposium on Maintaining Stability in a Changing Financial System.
• Laeven, L. & F. Valencia (2012), "Systemic banking crises database: An update," IMF WP-12-163.
• Pisani-Ferry, J. & A. Sapir (2010), "Banking crisis management in the EU: An early assessment," Economic Policy.
Protecting bank liabilities
Sciences Po- Banque de France Seminar
March, 25, 2015
Vivien Levy-Garboua
What we used to think
Equity is important
Creditors of banks should be protected: it’s a question of monetary stability
Deposits must be insured
The « Home-Host » rule should prevail
2
What is the concern?
That:
A run on deposits might occur
This run may hurt non financial customers
The State may be called to the rescue of banks, at a cost for
taxpayers
The Central bank may have to be the Lender of Last Resort
Some moral hazard may be created, encouraging excessive
risk taking by banks
This attitude might be exacerbated by those banks that
feel « too big to fail »
3
What we say today
Equity is essential
Wholesale deposits are a real danger for stability
Retail depositors should be protected
The State and the taxpayer should not be liable of thebanks’ problems
A Resolution plan is necessary, and, in Europe, we need theequivalent of the FDIC in the US
A (local) regulator should control what he is accountablefor
4
But, we should not forget that…
All bank runs are the (direct or indirect) consequence of bad investments or bad loans
You cannot separate assets from liabilities in a bank’sbalance sheet
The liquidity ratios, and the structural reforms decided in European Directives and Regulations or, in the US, in the Dodd Frank Act, are an essential part of the protection of bank liabilities
5
Four layers of protection
1) Capital ratios: more or less double the ratio, +1300bn euros since 2009 in the eurozone. For the 9 largest, +246bn (withtotal equity of 613 bn at the end of 2014)
2) Deposit Insurance: 43 bn euros in the eurozone
3) Resolution Fund: 55bn in the eurozone (2018)
4) TLAC/MREL*: for the nine largest eurozone banks, 278 bneuros (assuming a 16% TLAC)
For a Global Systemically Important Bank (GSIB), with a back of the envelope calculation:
70+ 43+ 55+ 30= 198 bn euros.
6* TLAC= Total Loss Absorbing Capacity; MREL= Minimum Requirements for Eligible Liabilities
Some stylized facts*
Over the 2007-2013 period, on a sample of 157 banks,cumulated losses have reached:
- more than 8% of total balance sheet for 13 banks,
- more than 9.5% of their total Risk Weighed Assets (RWA) for 21 banks
The losses are concentrated on small banks:
- 85% of banks with cumulated losses > 9.5% of RWA have a balance sheet < 250bn euros
- No cumulated loss > 8% of total balance sheet for banks > 250bn
*data are from Laurent Quignon and BNP Paribas Research
7
Some stylized facts (2)
Of the 13 loss-making GSIB, no-one lost more than 8% of total balance
sheet and 2 (UBS (9.3) and RBS (13.1)) lost more than 5.5% RWA
During the crisis, in the EU, only one small bank would have justified an
intervention of the Resolution Fund
No large bank with a balance sheet > 700bn has benefited from injection of
funds larger than the 7% RWA
Ex post numbers, maybe too optimistic
So, are the four layers:Not enough? Too much? Just right?
8
Theory (1): Modigliani Miller
An influential view (Admati, Hellwig (2014)) is that the MM theorem applies to banks. It implies that there is no social cost to increasing the capital requirement to 20 or 30%. It would not raise the cost of funds to banks.
Two counterarguments:
- Holmström-Tirole (1997) and Dewatripont-Tirole (1993) build models with inside capital or that insist on the monitoring role of lenders, that contradicts MM, even in theirframework
- My point is that deposits and loans are linked and that the cash flows of banks X is not independent of leverage. Hencethe arbitrage proof of MM does not go through
9
Theory (2): Bail in vs Bail out
A simple model for a large bank (Levy-Garboua-Maarek(2015))
P is the loss of the bank, P(1+k) the cost of the Bail out a is thefraction of Home lenders to the bank, (1-a) the fraction of theForeign lenders
C(.) is the cost to the lenders in case of bail in
10
Home Foreign
Bail in C(aP) C((1-a)P)
Bail out P(1+k) 0
Theory (2): Bail in vs Bail out
For the Home country, BI<BO iff:
P(1+k)< C(aP)
But for the Foreign Country, the cost is an externality.
Can it do something about it?
11
Home Foreign
Bail in C(aP) C((1-a)P)
Bail out P(1+k)-T T
BI<BO now iff:For the Home country : P(1+k)-T< C(aP) For the Foreign country: T< C((1-a)P)
Theory (2): Bail in vs Bail out
12
Theory (2): Bail in vs Bail out
13
There is a zone (the core of the game) where BO is preferable
Europe has decided to choose Bail in in all cases: it is not rational
Theory (3): a straitjacket?
A simplified balance sheet:
9 variables, 7 constraints, 2 balance sheet accountingequalities,
14
Assets Liabilities
Securities
Loans
Reserves
Equity
TLACMarket debt
Repo’s
Deposits
Refinancing
Summing up
Probably too much protection for individual banks
The issue is « contagion »
Once you introduce contagion, the limit becomes unknown
The regulators should concentrate on stress tests for the financial system as a whole, not for individual banks
Nobody knows where the next crisis will come from
A unescapable trade-off between growth and stability
15
Seven Ideas
1) From a micro perspective, the concern is with the protection of retail depositors2) From a macro-perspective, it’s a different story3) The macro perspective used to focus on monetary stability. Now, it looks at the protection of both retail depositors and taxpayers4) The theoretical arguments used to justify capital neutrality and bail in are not convincing5) The four layers of protection are « too much » from a micro perspective6) Contagion is the characteristic of a big financial crisis, and then, no one knows7) Regulators should focus on network effects rather than building what could appear to be « Maginot lines » around banks in the nextcrisis
16
Conclusion
It’s an impossible task to be a good regulator
You should be afraid of everything and, at the same time,…
If you are on the hard side, you are wrong 99% of the time, and you may be right once in a century
If you are pragmatic, you are sure to take the blame.
So, in the end, « High risk, low return »
Thanks to the ACPR .
17