Transcript SETTING MINIMUM CAPITALFOR INTERNATIONALLY ACTIVE …

Market Discipline -Effect on
Bank Risk Taking
Glenn Hoggarth
Patricia Jackson
Erlend Nier
6th June 2003
1
Market discipline
• Policy initiatives (eg Basel II) recognize
importance for financial stability
• Pillar III of Basel II attempts to strengthen
market discipline by requiring disclosure
• Greater disclosure is being resisted by
banks -argue costs outweigh benefits
• Hardly any evidence on effectiveness of
disclosure and market discipline
6th June 2003
2
Policy: Basel Committee
• Basel I - Created common metric for
measuring capital relative to risk - Risk
asset ratio - but some banks only publish
Tier1 plus Tier 2
• Basel II- Pillar III -minimum standards pf
disclosure -covering composition of capital
and risks
6th June 2003
3
Evidence that market discipline
may affect bank behavior
• Important to consider whether there would
be benefits to financial stability from
greater market discipline
• Or are banks right -and benefits not enough
to outweigh costs
• First need to consider conditions for
effective market discipline
6th June 2003
4
Concepts: Effective market
discipline
• Market must have information to
assess riskiness of banks
importance of disclosure
• Market participants must be at risk of
loss
importance of limited safety net
6th June 2003
5
A number of markets likely to
discipline banks- main ones
Equity
market
- cost and availability of
new capital
- takeover target
Affected by
• shareholders limited liability
- gambling for resurrection
• expectations of support
• sub-contract monitoring to regulators
6th June 2003
6
Interbank
market
Affected by
- cost and availability of
short-term funding
- ability to hedge risks in
OTC derivatives markets,
eg swaps, essential
- graduated reaction more
likely from wholesale
counterparties
• deposit protection arrangements
• too big to fail
6th June 2003
7
Assemble evidence related three
questions
• (1) Does market discipline affect the size of
bank capital buffers (resilience to shocks)
• (2) Does market discipline affect the
likelihood of crises
• (3) Does market discipline affect costs of
crisis resolution
6th June 2003
8
(1) Effect on banks’ capital
(resilience to shocks)
6th June 2003
9
• Bank of England research, “Market
Discipline, Disclosure and Moral Hazard in
Banking”, (Nier and Baumann) tested the
effect of disclosure and the safety net on
individual banks’ capital buffers
• cross country panel dataset
• 729 individual banks from 32 countries
• typically observations from 1993 to 2000
6th June 2003
10
Identified measures of the
strength of market discipline
6th June 2003
11
(1) Depositor protection
Index on existence and extent
Depins 2 = 1 or 0 - if schemes exist
Depins 3 = 1 or 0 - no co-insurance
Depins 4 = 1 or 0 - interbank deposits covered
Depins 5 = 1 or 0 - unlimited coverage
Depins = sum of depins 2, depins 3, depins 4,
depins 5
6th June 2003
12
(2) Government support
Fitch
Safety net 1 if public support rating = 1 or 2
0
6th June 2003
= 3, 4, 5
13
(3) Uninsured Deposits
Proportion of uninsured interbank deposits
6th June 2003
14
(4) Disclosure
Constructed an index on core
disclosure items from BankScope
18 categories covering following areas Bank’s risk profile
- interest rate risk
- credit risk
- liquidity risk
- market risk
Capital and reserves
6th June 2003
15
(5) US listing
NYSE, NASDAC or AMEX
6th June 2003
16
CAPit f(RISKit,MKDit,Zit ) υit
Risk
- components of weekly equity returns
- one period ahead loan loss provisions
Z
- control variables - return on equity
- log of total assets
- GDP growth
MKD - market disclosure/market discipline
variables
6th June 2003
17
Results- effect on capital
relative to risk
Deposit insurance and support:
negative
Interbank deposits:
positive
US listing and disclosure index:
positive
6th June 2003
18
Table 1[495]:
The effect of market discipline on bank capital
Dependent
variable
Constant
Provisions (t+1)
Beta
Idios. Risk
Logsize
Roe
GDP growth
Non-perf. Loans
Market share
Cap. Req.
Time trend
Dep. Insurance
Support
Bank deposits
Rating
Listing
Disclosure
No. of obs.
No. of banks
Goodness of fit
Log likelihood
(1)
Cap
(2)
Cap
(3)
Cap
-3.2609***
0.3657***
0.0044***
-0.1715***
-0.0043***
0.0535***
0.0058
-0.0970***
-0.0484***
0.0148***
0.0016***
-0.0023***
-0.0117***
-1.9414***
-0.1016***
0.0070***
-0.0537***
-0.0138***
0.0217***
-0.1154***
0.0016
0.0394***
0.0115***
0.0011***
-0.0065***
-3.0638***
-0.0320**
0.0068***
-0.0427***
-0.0147***
0.0247***
-0.1244***
0.0075
0.0282***
0.0122***
0.0016***
-0.0059***
0.0676***
0.0030***
0.0784***
0.0031***
0.0149***
0.0147***
728
199
0.46
2732
0.0098***
0.0157***
695
154
0.50
2424
726
199
0.46
2694
Source: Bank calculations.
Indicates significance at the 10 %
6th*June 2003
**
***
level.
Indicates significance at the 5 % level.
Indicates significance at the 1 % level.
19
Effect on capital for given risk
Banks expected to have government support
have a capital ratio 1.2 percentage points lower
than those without.
Banks fully funded from uninsured interbank
deposits would have have a capital ratio 7
percentage points higher than a bank fully
funded from insured deposits
Banks disclosing none of the core information
measured have a capital ratio 1.5 percentage
points lower than those that do.
6th June 2003
20
• Findings lend weight to assertion that
market discipline could help strengthen the
financial system by increasing resilience to
shocks.
• But is there any more direct evidence?
6th June 2003
21
(2) Effect on the likelihood of
banking crises
6th June 2003
22
• Factors increasing market discipline disclosure- should reduce likelihood of crises
• Factors reducing market discipline
(government support, deposit protection
schemes) could have two opposing effects
•
-(a) reduce market discipline weakening
banking system but (b) prevent crisis from
materialising.
6th June 2003
23
Empirical approach
Baumann Nier Data-set
• 32 countries 1993-2000
• 7 banking crises starting /continuing after
1993
-Korea, Thailand, Indonesia, Malaysia ,
Japan, Turkey and Argentina.
6th June 2003
24
Market discipline variables
•
•
•
•
Deposit protection
Government support
Disclosure
US listing
6th June 2003
25
Crisis=f (MKD,Z)+e
Crisis = country dummy value 1 (crisis) 0 not
Simple OLS regressions of crisis dummy on
market discipline variables
Probit regressions of crisis dummy on market
discipline and control variables
6th June 2003
26
Results- effect on likelihood of
banking crisis
• Disclosure and US listing - weak negative
effect, appear to reduce likelihood of crisis
• government support - significantly negative
effect, clearly reduces likelihood of
systemic crises
• deposit insurance - weak positive effect,
appears to increase likelihood of crises
6th June 2003
27
Effect of components of deposit
insurance
• Existence of scheme -negative effect,
reduces likelihood of crisis
• interbank and coinsurane - no evidence
either way
• unlimited deposit protection - strong
positive effect, increases likelihood of crisis
6th June 2003
28
Probit regressions
With control variables
• - GDP per capita
• - GDP growth
market discipline variables retain sign.
• With current account deficit /surplus added
market discipline variables again retain
sign
6th June 2003
29
Caveat: preliminary work
• Small sample of crisis countries
• Will go on to look at effects at bank level fall in capital indicator of problems.
• But does indicate countries should question
role of unlimited deposit protection schemes
and should encourage greater disclosure.
6th June 2003
30
But deposit protection is there to
deal with crisis
• Countries concerned about future potential
crises will not change procedures if they
would damage ability to deal with banking
problems.
• Further question therefore - do unlimited
deposit protection schemes improve crisis
management ?
6th June 2003
31
(3) Effect on costs of crisis
resolution
6th June 2003
32
Effect on resolution costs
• Sample of 33 systemic banking crises
• Effect of blanket guarantees
• Effect of depositor protection
• 1 if limited scheme exists
0 if scheme is unlimited or does not exist
• regressions attempt to control for size of
shock, eg dummy for currency crisis
6th June 2003
33
Results- effect on resolution
cost
• Blanket Government guarantees appear to
increase resolution costs
• Limited deposit insurance schemes reduce
resolution costs - when compared to
unlimited or implicit schemes
6th June 2003
34
(4) Implications for public
policy
6th June 2003
35
Deposit Insurance
• Explicit deposit insurance may prevent
banking crises
• unlimited deposit protection schemes
could be harmful -affect bank behaviour
make crises more likely
6th June 2003
36
Implicit government support
• Support prevents crises from materialising
(if support is credible in fiscal terms)
• Support increases moral hazard and
reduces resilience of the banking system
• Where support arrangements substantial more onus on supervisors
6th June 2003
37
Disclosure
• More information disclosure has the
potential to strengthen the resilience of
the banking system
• Key is comparability of information
across banks
6th June 2003
38
Nature of disclosure
- comparable disclosure important
VaR
Lloyds
HSBC
Standard
Abbey Chartered Barclays
95%,
1 day
99%,
10 days
95%,
1 day
6th June 2003
97.5%,
> 1 day
98%,
1 day
39
Pillar III will be effective in
increasing amount of comparable
disclosure
Important for standardised and IRB
banks.
6th June 2003
40