Rethinking Bank Regulations Till angels govern but who are the angels

1
Rethinking Bank RegulationsTill angels
govern(but who are the angels?)

• Comments
• Roberto Rocha, FPDFS

2
Overview of Objectives, Methodology, and
Conclusions

• Main objective identify which approaches to bank
  regulation and supervision lead to better
  outcomes
• Outcomes defined by banking development,
  stability, efficiency, integrity
• Methodology Large questionnaire (275 questions)
  sent to 150 countries
• Answers combined to produce large number of
  indices
• Indices used as regressors
• Outcomes (dependent variables) measured by
  private credit/GDP, interest margin, crisis
  index, lending corruption index

3
Overview of Objectives, Methodology, and
Conclusions

• Main Conclusions
• Strong direct official supervision and capital
  standards (pillars 1 and 2) do not generate
  positive results
• In contrast, policies that facilitate private
  monitoring of banks (pillar 3) produce much
  better outcomes
• Boost bank development
• Improve bank efficiency
• Reduce corruption in lending
• Lower bank fragility

4
Overview of Objectives, Methodology, and
Conclusions

• Some of the arguments used to substantiate
  conclusions
• Capture of politicians/supervisors by control
  groups, (corruption) powerful supervision aimed
  at favoring the privileged. These actors are no
  angels.
• Regulators in high income countries take the view
  that their approach is best for other countries.
• Results show that policies that facilitate
  private sector monitoring work best it is
  essential to empower private actors through
• Disclosure of reliable, comprehensive, timely
  information
• Incentives to monitor
• Laws that strengthen the rights of private
  investors

5
General Comments

• General concept and objectives of the exercise
  are very attractive, could generate important
  policy lessons
• However, methodological problems are not trivial
• Potential differences between statutory
  parameters and the reality of supervision is
  acknowledged by the authors
• The questionnaire may not capture the reality of
  bank regulation and supervision, differences
  across countries
• Several questions are too general/vague
• No guidelines clarifying objectives of each
  question, and minimizing differences in
  interpretation
• Important questions missing
• Definitions/criteria used for building the
  indices from the answers are arbitrary, not clear
• Main problem measurement errors on the right
  hand side of the equation, but also other
  methodological issues

6
General Comments

• Scores and rankings (and the regression results)
  are very sensitive to definitions/criteria used
  to build the indices
• Example Correlation of BCL index of central bank
  independence, with another index by Arnone,
  Laurens and Segalotto (ALS) (IMF2007) is only 0.1
• ALS index is more comprehensive, capturing more
  aspects of central bank independence
• Some indices are inconsistent with evidence
  produced by the questionnaire itself
• Example CEE countries are the most open
  countries, as indicated by highest shares of
  foreign ownership yet most restrictive according
  to the entry restrictions index (almost all
  countries scored max of 8)
• Bona fide foreign investors could easily meet the
  8 requirements in the index In fact, correlation
  of index with share of foreign investors is
  positive (0.2)
• At the same time, real statutory restrictions not
  captured in the index

7
General Comments

• Country rankings are too counter-intuitive in
  many cases
• Examples Nigerian supervisor more independent
  than Chilean Brazil as open as New Zealand
  accounting/auditing standards very similar across
  countries
• Some answers reflect lack of guidelines
• Example Australia classified as multiple
  supervisor and UK as unified supervisor
• Questionnaire and regression results produce a
  main hero and a main villain
• Main hero Private Monitoring Index (pillar 3)
• Main villain Official Supervisory Power Index
  (pillar 2)
• More detailed analysis of these two indices is
  warranted

8
The Main Hero Private Monitoring Index

• What should the index capture
• The disclosure of reliable, comprehensive, timely
  information
• Incentives for private agents to monitor banks,
  both positive and negative
• How would the index capture these elements
• Quality of bank accounting
• Quality/integrity of external auditing
• Quality of disclosure rules
• Existence of subordinated debt
• Existence of effective rating industry
• Existence of de facto deposit insurance
• Legal liabilities of pillar 3 players directors,
  auditors, raters
• Existence of effective, clean judicial system

9
The Main Hero Private Monitoring Index

• Index probably does not reflect effectiveness of
  pillar 3 and differences across countries
• Assessment of the quality and integrity of
  external auditing is too shallow no differences
  across countries
• One single, trivial, aspect existence of
  certified bank audits
• Except for China, Italy, all countries got a 1
• Albania, Burkina, Bolivia, Central Afr. Republic,
  Honduras, Zimbabwe have the same auditing score
  as the US and the UK
• Some important audit information used in other
  indices but not in this one (e.g., scope of
  audit, relations with supervisor)
• Evidence from questionnaire on de facto absence
  of legal liability of auditors is ignored
• Key aspects of auditing function missing in the
  questionnaire (e.g., scope, independence,
  conflicts of interest, SRO)

10
The Main Hero Private Monitoring Index

• Assessment of quality of accounting too shallow,
  not capturing differences across countries
  either
• Question on whether accrued interest on NPLs
  enters the income statement disregards quality of
  loan classification
• Questions on elements of disclosure (off-balance
  sheet items, risk management) too vague,
  especially second - yes or no.
• Bank directors in all countries are liable for
  disclosing wrong information disregards evidence
  of lack of enforcement due to judicial corruption
  in ECs
• Answers on of 10 top banks rated are revealing
  some supervisors were not certain, no domestic
  raters in most countries
• Except for Germany, Tunisia, all countries got 3s
  and 4s (1-4)
• Quality of accounting in Egypt, Nigeria,
  Paraguay, Russia, Rwanda equal to US, UK, higher
  than France, Germany, Sweden

11
The Main Hero Private Monitoring Index

• Assessment of the existence of moral hazard
  through provision of deposit insurance does not
  capture the reality of many countries
• Question relies on existence of explicit deposit
  insurance
• In many countries deposits are insured by the
  government, no depositor has lost a cent, despite
  absence of explicit insurance
• Examples Egypt, Kuwait, Saudi Arabia, other Gulf
  countries
• Egypt example Bank restructuring program has
  entailed exit of 20 weak banks (out of 50) in the
  past 3 years, through organized mergers and
  acquisitions.
• Question would need to be reformulated

12
The Main Hero Private Monitoring Index

• Final result compressed scores, ranging from 5
  to 11 average 7.8, SD 1.4 unlikely to reflect
  actual differences across countries
• Only 2 countries scored 11 Canada and Kuwait
• Egypt and Gulf countries (Kuwait, Oman, Saudi
  Arabia, Oman, UAE) got high scores, average
  10, reflecting inter alia, absence of explicit
  deposit insurance
• Different reality, strong (implicit but
  well-known) deposit insurance by Government
• Index should be revised, robustness of results
  tested
• Elements from the existing questionnaire should
  be included
• Several questions reformulated, some dropped
• Additional questions included
• Guidelines/instructions explaining objectives

13
The Main Villain Official (Direct) Supervisory
Power Index

• Index presumably measures statutory enforcement
  powers, based on 14 questions
• 3 questions relate to relations with auditor
  (presumably important for the effectiveness of
  audit function and pillar 3)
• 11 questions on enforcement, of which
• 3 deal essentially with the same supervisory
  action
• Whether supervisor can suspend dividends,
  bonuses, fees
• 5 questions on enforcement check/compare power of
  supervisor, other agencies, courts to enforce key
  actions
• Court involvement implies weaker powers, lower
  scores
• 1 question deals with a basic reporting
  requirement (whether off-balance sheet items are
  disclosed to supervisor)

14
The Main Villain Official (Direct) Supervisory
Power Index

• Very difficult to assess this index
• Index is not highly correlated with the
  supervisory independence index (corr. 0.2)
• Questionnaire does not capture regulatory powers
  (i.e. power to issue binding secondary
  regulation)
• Questionnaire does not deal with the supervisory
  process
• No information on basic numbers (e.g. numbers,
  salaries)
• No information on on-site and off-site procedures
• Index seems restricted to some specific statutory
  aspects of enforcement
• Therefore, index captures very limited aspects of
  pillar 2 and even these aspects may not be
  sufficiently addressed

15
The Main Villain Official (Direct) Supervisory
Power Index

• Not clear whether the three questions dealing
  with the auditor relate more to pillar 2 than
  pillar 3
• Many countries answered yes to whether supervisor
  can take legal action for auditor negligence but
  most of these also answered that this has not
  happened
• Either auditors are doing a very good job, or the
  supervisor is a lion without teeth
• Second information is not used at all in the
  exercise
• Absence of de facto legal liability could make
  all these three questions irrelevant

16
The Main Villain Official (Direct) Supervisory
Power Index

• Question on whether off-balance items are
  reported to supervisor is too basic all
  countries answered yes.
• Almost all countries also answered that
  supervisor have powers to order management to
  build provisions
• Final result Slightly wider differences across
  countries relative to private index (average
  10.9, SD 2.6), but is it capturing
  effectiveness of pillar 2?
• One difficult but critical issue not sufficiently
  examined the questions dealing with court
  involvement disregard the effectiveness of the
  judicial system

17
(No Transcript)
18
(No Transcript)
19
Other Methodological Issues

• Measurement errors in the dependent variables
• Variable measuring banking crisis what was the
  treatment of countries where banks were deeply
  insolvent but there was never an open crisis, and
  where large stock of NPLs was never shown before
  the main restructuring program?
• Specification problems
• Many, more important factors, affecting dependent
  variables such as stock of private credit/GDP
  inflation history, timing of recapitalization
  program with massive carve-outs
• Endogeneity
• Are there instruments to address response of
  regulatory and supervisory actions to shocks?

20
Summing Up

• Methodology
• Most serious problem measurement errors in the
  independent variables (indices built and used)
• Errors are probably of such a magnitude that
  cannot be addressed by any of the available
  techniques
• Specification problem in some of the regressions
• Overall, robustness of results is questionable

21
Summing Up

• Policy Messages
• Message that pillar 3 is important would find
  support
• Message that pillar 2 is always ineffective,
  leading to corruption, and that only hope is
  pillar 3 would not
• Directors, auditors, raters, courts are no angels
  either
• Exercise provides a very basic message, does not
  provide sufficient guidance to policy-makers
• No pillar 2 at all? Some pillar 2, ma non
  troppo? (Are there non-linearities?) If so, what
  type of pillar 2?
• What are the building blocks for an effective
  pillar 3?
• In particular, how to deal with judicial
  corruption?

22
Summing Up

• Proposal for next steps could include
• Revising the questionnaire, elaborating
  guidelines
• More inputs from bank supervisors,
  auditing/accounting experts
• Greater use of available material, ROSCs
• Objective build more reliable indices
• Combine cross-country regressions with analysis
  of individual cases
• Greater effort to identify policy lessons
• Work program on Pillar 3
• What are the pre-conditions/building blocks for
  an effective pillar 3?