basle ii capital requirements and developing countries… docs/10.23.03 gdn conf/claessens... ·...
TRANSCRIPT
Draft, October 21, 2003
Basle II Capital Requirements and Developing Countries:
A Political Economy Perspective
By
Stijn Claessens Professor of International Finance Policy
Geoffrey R. D. Underhill Professor of International Governance
University of Amsterdam and Xiaoke Zhang
Lecturer University of Nottingham
Abstract The 1990s financial crises have triggered changes to the international financial system, the so-called international financial architecture. While much affected, developing countries have had very little influence on the changes, which the formulation of the new Basle capital accord (Basle II, B-II) illustrates. We show that B-II has largely been formulated to advance the interests of powerful market players, at the expense of those of developing economies. For these countries, B-II can raise the costs of and reduce the access to external financing. Importantly, B-II can exacerbate fluctuations in the availability of external financing, an unfortunate outcome, given that developing countries already suffer from volatile capital flows. For presentation at the workshop on Quantifying the Impact of Rich Countries’ Policies on Poor Countries, October 23-24, 2003, organized by the Center for Global Development and the Global Development Network. We would like Erik Feijen for excellent research assistance.
1
1. Introduction
The financial crises of the 1990s have led to a debate on the reform of the international
financial system, the so-called international financial architecture debate. In response to
the crises, international financial institutions, such as the Basle Committee, the OECD,
the Financial Stability Forum (FSF), the IMF and World Bank, have promulgated and
established several sets of international standards to regulate market behavior. These new
standards are being held up as models for developing and other countries to follow and
are being assessed through various mechanisms (Financial Sector Assessment Program
(FSAP), Report on Standards and Codes (ROSC), etc.). There is also a debate
concerning standards of corporate governance in emerging market economies, though
this has yielded less in terms of concrete results.1
Developing countries are likely to be affected most by these standards, in part as they
are institutionally further from the “norms” being promulgated and as the leverage that
international institutions and (major) developed countries have over them is larger than in
case of more developed countries.
At the same time, developing countries have had very little influence on the
formulation of these standards, potentially undermining their legitimacy and
effectiveness. While the formation of the G-20 and the Financial Stability Forum might
have rendered the international decision-making process more inclusive, the membership
and structural hierarchy of these forums have left no doubt that the global financial
system will continue to be run by the leading industrial nations. The G-20 and the FSF
still exclude the majority of developing countries. Even the emerging markets that are 1 See George Vojta and Marc Uzan, “The Private Sector, International Standards, and the Architecture of Global Finance,” in Geoffrey R.D. Underhill and Xiaoke Zhang (eds.), International Financial Governance under Stress (Cambridge: Cambridge University Press, 2003), pp 283-301.
2
included realize that they have different interests from and lack collective bargaining
power vis-à-vis the dominant members.
A core set of new international standards is the proposed new Basle capital accord
(Basle II), which is in its final stages of negotiation. This has been shown to have
significant implications for the costs of capital for developing countries and could reduce
their access to external financing. As with the other standards, Basle II has been
developed largely to the exclusion of developing countries’ input. This paper, using Basle
II as an illustration of the skewed process of the international financial architecture
reform, seeks to achieve two different yet interrelated analytical objectives.
In the first place, this paper aims to examine the process through which the new Basle
capital accord has been formulated and to explain how and why the accord has become
standardized. The core argument to be developed is that Basle II capital requirements are
formulated in relatively exclusionary and closed policy communities consisting of
regulators and private agencies from the leading industrial nations. In these networks,
private market interests find respondents in finance ministers and central bankers and
have thus been able to influence the global agenda for policy changes. The rules and
standards to be sanctified by Basle II thus tend to advance the interests of powerful
market players at the expense of those of developing and emerging market economies.
Furthermore, the paper attempts to examine the impact of the Basle II capital
requirements on the long-term development prospects of developing countries and to
suggest some methods by which the impact can be quantified. A central hypothetical
proposition developed here is that the implementation of the new capital requirements
embodied in Basle II is most likely to exacerbate fluctuations in the costs and availability
3
of external financing for many developing countries. This would be a very unfortunate
outcome within the context of the potential contribution of the new international financial
architecture in general and of the new Basle capital accord in particular, given that they
have been designed to serve the most vulnerable members of the international
community.
The analysis of the Basel accord is relevant in its own right, but only one of the many
elements under discussion in the new international financial architecture. Many of these
elements need to be considered jointly. For example, one of the reasons why the Basel
outcome does not favor terms for (sovereign) lending to developing countries is the lack
of an international bankruptcy system. Yet, progress on such a system has been stymied
by, among others, the unwillingness of creditor countries to consider changes. As such,
our analysis is more important as an example of what the process of the reform of the
international financial architecture has delivered so far and what it can be expected to
deliver in the future for developing countries. The political economy analysis suggests
that developing countries have had little stake in the negotiations. To assure more
sustainable changes to the system, reform of the governance of the international financial
system itself is needed.
2. The new international financial architecture, a short overview
During the 1980s, Mexico made dramatic and unilateral alterations to its approach to
the economic development process. The strategy shifted from one based on state
intervention, capital controls, and trade protection/import substitution to one dominated
by economic openness, fiscal discipline, and a greater degree of market-led adjustment,
4
in line with what came to be called, in an article by John Williamson, the ‘Washington
Consensus’.2 In a number of respects, Mexico and other countries went further than what
Williamson specified as ‘consensus’ to embrace more radical aspects of the ‘neo-liberal’
platform of rapid opening of the capital account, rapid financial system liberalization,
far-reaching privatization, and a reduction of welfare provision,3 and often pegged
exchange rates as well.
The approach was seen in historical terms as a corrective to earlier approaches to
development involving state intervention and protection based on ‘infant industry’
arguments and problems associated with declining terms of trade for natural resources
exporters.4 The sovereign debt crisis and extreme inflation of the 1970s had spelled the
end this sort of development policy rationale. The extraordinary growth of private capital
markets from the 1980s onwards demonstrated the need to think not only about official
sector transfers in development policy, but also the need to create conditions conducive to
private sector investment. Private investors prefer developed countries to the developing
world at least partially because developing countries constituted greater investment risks,
even though they should also promise greater returns if growth were to take off as in the
case of the Asian tigers. If developing countries could reduce the risks posed by
inflation, exchange rate volatility, indebtedness, reduce the arbitrariness of what were
perceived to be politically driven state intervention strategies, and reduce structural
2 See John Williamson, "What Washington Means by Policy Reform", in J. Williamson, ed., Latin American Adjustment: How Much Has Happened? (Washington: Institute for International Economics, 1990). Williamson claimed, in a cautious assessment, that there was consensus around was a series of market-oriented reforms in development policy for Latin American countries, and which pointedly did not include capital account liberalisation. 3 As Williamson himself points out, the term also came to represent a more radical ‘neo-liberal’ ideology of minimalist state involvement, greatly reduced welfare provision, monetarism, and radical privatisation (“Did the Washington Consensus Fail?” Speech to Centre for Strategic and International Studies, 6 November 2002, http://www.iie.com/publications/papers/williamson1102.htm). 4 Ideas associate with Raul Prebisch and the UN Economic Commission on Latin America in the 1950s.
5
rigidities impairing the adjustment process, then capital would be more likely to flow
towards emerging market countries. This would require a major adaptation of the legal,
institutional, and substantive policy framework within developing countries. Market-led
development would also lead to smoother adjustment processes, avoiding the pitfalls of
debt-financed state-led strategies.
By 1994 Mexico, along with Argentina and others who imperfectly embodied the
consensus, was regarded as a star pupil and had just entered the North American Free
Trade Agreement with Canada and the US and was on its way to admission to the OECD.
In late 1994 the world of the Washington Consensus was baffled as Mexico plunged into
a combination of exchange rate crisis and financial crisis that led to a controversial record
IMF loan package assembled under US leadership.5 At the Halifax G7 summit of June
1995 and subsequently at Lyons and Denver (1996-7), the summit placed considerable
emphasis on a re-examination of global financial and monetary governance.6 Although
the ‘Peso Crisis’ of 1994-5 threatened other emerging market economies with contagion,
and not only those in Latin America, it passed with little changed at the international
level despite a range of more and usually less radical proposals for change coming from
academics and official sources, and complacency reasserted itself.7 Proposals aimed at
improving supervision at the international level, improving the transparency of financial
markets, sensible macroeconomic policies and exchange rate regimes, better monitoring
5 It was of course always an option for more radical neo-liberals to claim that the real problem was that Mexico did not go far enough, and thus that the ‘consensus’ was in itself insufficiently radical. 6 See the Halifax Summit document on reform of financial governance at http://www.g7.utoronto.ca/summit/1995halifax/financial/index.html; re Lyons summit at http://www.g7.utoronto.ca/summit/1996lyon/finance.html; the final report of the G7 finance ministers concerning financial architecture was presented at the Denver summit of 1997 – see http://www.g7.utoronto.ca/summit/1997denver/finanrpt.htm. 7 Among the more radical proposals was that of an international bankruptcy court to facilitate sovereign debt workouts under more predictable conditions, and an international banking standard (see Morris Goldstein, The Case for an International Banking Standard, Institute for International Finance, 1997)
6
of macroeconomic performance by the IMF, and the like. A consensus had formed that
radical change was not necessary and that an incremental reform process would suffice.
The outbreak of the Asian crisis of 1997/98 once again caught officials and the
private sector by surprise. The Asian tigers, again star (if often misunderstood) pupils of
international development policy, were suddenly in the grip of financial and monetary
crises which appeared inexplicable. Explanations such as exchange rate rigidities, a lack
of transparency on macroeconomic policy, and poor financial supervision were of course
found ex post facto, among them the sin of ‘crony capitalism’ and a failure to develop
transparent market-based relationships in the corporate world. Though the shock of the
Asian crisis was quickly followed by troubles in Russia, Argentina, Turkey, and Brazil,
and indeed the LTCM incident nearly brought financial contagion and collapse to Wall
Street itself, the reform debate remained limited in scope and the process incremental in
nature. It is noteworthy that the one more radical proposal to originate from the official
sector, the Sovereign Debt Workout Mechanism (SDRM) proposed by Ann Krueger of
the IMF and which contained elements of a global bankruptcy procedure,8 was defeated
by a combination of developing and developed member states alike at the IMF meetings
of March 2003.
The goals of the financial architecture reform process were straightforward and in
keeping with the market-based approach to development signaled by the Washington
Consensus. Policies of economic openness, macroeconomic stability (including fiscal
discipline), and sound supervision and regulation were the aim, and these were to
8 Anne Krueger, A New Approach to Sovereign Debt Restructuring (Washington, DC: International Monetary Fund, 2002).
7
facilitate international financial stability and efficient allocation of global capital, spilling
into domestic financial sector stability and development.
In the 1980s and 1990s, many countries in and out of Latin America implemented
reform policies along the lines of the ‘Washington Consensus’ described by Williamson,
often going beyond it. The collapse of the Soviet Bloc and the emergence of ‘transition
economies’ contributed to the sense of a triumph of market-based approaches to
economic development. Some countries indeed pursued quite radical experiments.
When financial crises began regularly to punctuate this reform process, it was often
revealed that insufficient attention had been paid to the legal, institutional, and regulatory
aspects of reform. The debate on financial architecture focused on these questions at
both the international and national levels. There was no essential re-examination of the
market approach itself,9 but a serious reckoning with the way in which it had been
implemented in various settings, and how reform interacted with monetary and exchange
rate policies as well. An examination of the relevant G7 documents will serve as
guidance to the agenda; the Birmingham summit (1998) provides perhaps the clearest
statement of what it termed an “emerging consensus.”
The Birmingham document,10 developed at the height of the Asian Crisis, began with
‘transparency’: the provision of “accurate and timely” macroeconomic and financial
supervisory data, including the reserve positions of central banks and levels of national
public and private indebtedness. The IMF was charged with developing codes of
9 Although arguably the most successful examples of economic development, Japan and east/SE Asia, had based their strategies on state-led investment strategies, import substitution combined with trade protection (progressing in time to export promotion), financial repression, and strict local content rules relating to foreign direct investment. Such strategies were not insensitive to market forces and the need for competitiveness, they were anything but market-led. 10 See the report at http://www.g8.utoronto.ca/summit/1998B-Irmingham/g7heads.htm.
8
Special Data Dissemination Standards (SSDS) to which countries would adhere, and the
BIS would accelerate the publication of data on central bank reserve positions.
Policymaking should also be more open in government and at IFI’s, with a view to
enhancing the confidence of investors and rendering the investment climate predictable
and therefore easier to price on markets. Secondly, national economies needed better
preparation for international capital flows: weak financial systems needed to be reformed
through orderly capital account liberalization processes (sound macroeconomic,
regulatory and supervisory policies first, then capital account opening); foreign firms in
the financial sector should have full access to the markets of developing economies in
order to transfer skills and expertise. Thirdly, national financial systems needed
strengthening in relation to corporate governance practices and norms (something to
which a number of developed countries might have, in retrospect, paid more attention).
This could be achieved through the development and benchmarking of supervisory
practice and standards (Basle ‘Core Principles’), along with a system of multilateral
surveillance of supervisory practice. Fourthly, the private sector must take greater
responsibility for its lending decisions and the risks that they inherently involve. Implicit
or explicit government or IFI guarantees of ‘bailout’ in crisis must end as this leads to
moral hazard. The private sector must bear its share of the burdens in debt workouts,
continuing lending in times of crisis, so that public resources were not unduly
underpinning private gain. National policies should clarify bankruptcy procedures at the
national level, enhancing understanding of risks and the consequences of mistakes.
Fifthly, with the IMF as the specified lead institution, greater co-operation among IFIs
and a better relationship between multilateral and bilateral aspects of stabilization efforts
9
must be developed. Finally, global forums for a better dialogue between emerging
market and developed creditor countries should be developed; one should see the
establishment of the Financial Stability Forum/Institute (FSF/I) in Basle (1999) and the
G10-G20 consultation process as the principal results of this aspect of the reforms. 11
The reform measures largely assume that the causes of crises lay with the emerging
market economies themselves and their weak institutions and practices. In this sense,
global financial architecture was to be strengthened rather les that the individual
emerging markets economies within it. These proposals were thus not really proposals
for systemic architectural reform as such, except to the extent that they established the
FSF/I to enhance supervisory practice, and designated the IMF as the lead institution of
‘architecture’ such as it was. There was certainly no fundamental review, as there had
been at Bretton Woods, of the balance of international versus national policy obligations
in the face of market and other international systemic pressures in global monetary and
financial relations. Discretionary policymaking of developing countries was in essence to
be continuously constrained through enhanced monitoring and transparency/new
standards in order to ensure that the investment climate conformed to the expectations
and preferences of investors. Considerable empirical evidence was ignored that most
successful developing countries, including the development processes of European
countries, occurred under conditions of some financial repression, and not full openness
and market orientation.
It is also worth drawing attention to the way in which the debate about reform was
carried out: who were the key players, who controlled the agenda, and who responded to
11 For a broader account of standards and what was developed from the B-Birmingham principles, see the Financial Stability Forum web site “Compendium of Standards/12 Key Standards for Sound Financial Systems.” http://www.fsforum.org/compendium/key_standards_for_sound_financial_system.html.
10
and shaped proposals over time? Given that the emerging market economies were
identified as the primary obstacles to a smoothly operating global financial system, and
given the considerable diversity of financial systems and legal/policy making institutions
in the developing world, one might expect some considerable attention to the ‘one-size-
fits-all’ problem and therefore considerable consultation between those proposing the
reforms and those who must accept and implement them. Although there are those who
argue optimistically that emerging market participation in global financial governance
has significantly increased,12 the case, it may be argued, is a weak one. The G7 and G10
governments are in a commanding position relative to the IMF, the G7 process, the
design and establishment of the FSF/I, and other relevant institutions such as the OECD
or the Basle Committee on Banking Supervision or the broader ‘Basle Process’ based at
the BIS. With the exception of the IMF and the OECD, there is no emerging market
membership of any of these bodies,13 and in the IMF a considerable number of the
executive directors representing transition or developing countries are in fact from
developed countries (e.g., Belgium, The Netherlands, Spain, Italy, Canada, Iceland and
Australia together represent some 71 countries, or 64 in addition to themselves14). The
G7 finance ministers have developed the agenda and led the debate themselves. The
institutions which are central to implementation and further decision-making on the
process (the IMF, World Bank, the FSF/I, the Basle Committee, OECD) are again
dominated by G7/G10 membership, and these institutions were developed by them for
12 See Randall Germain, “Global Financial Governance and the Problem of Inclusion,” Global Governance, vol. 7 (2001), pp 411-26. 13 The establishment of the G20 group of emerging market economies as a consultative body to the G10/G7 process, including deliberations in the broader ‘Basle Process’, constitutes progress but not membership of key bodies. Hong Kong and Singapore are in fact members of the FSF/I, but they hardly qualify as developing countries under current circumstances, and they represent a tiny population. 14 See IMF web site, http://www.imf.org/external/np/sec/memdir/eds.htm.
11
the governance of monetary and financial relationships in the developed world. G7/G10
central banks and treasury ministries have close and long-standing relationships to their
respective private financial sectors and are responsive to their preferences.15 It is far
more likely that developed country private sector preferences were central to the
proposals than the preferences of either developing country states or the corresponding
financial institutions thereof. We now turn to an analysis of the Basle Committee on
Banking Supervision as an example of a crucial policy-making community shaped by the
integration of private sector financial institutions into the official decision-making
institutions of developed country economies.
3. Basel II specifically
The motivations for the new Accord (B-II), arise from a number of technical
weaknesses in Basel I (B-I), changes in financial services industries globally, changes in
the forms of financial services provision, and the corresponding emergence of new, and
changes in the pattern of old, risks. The main weakness of B-I was its limited number of
credit risk classifications and the rather crude or distorted way in which some types of
loans were weighted for capital adequacy purposes. One obvious distortion was the zero
weight given to loans to OECD government, irrespective of the riskiness of the country.
This led countries like Korea and Mexico to be treated for capital adequacy requirements
the same as developed countries. Also, little attention was given to the correlations
15 For a comparative analysis of state-financial sector relations under conditions of global integration in developing countries, see William D. Coleman, Financial Services, Globalization, and Domestic Policy Change, (Basingstoke: Macmillan 1996); and for an analysis of finance-government relationships relative to the negotiation of the EU single financial market, see J. Story and I. Walter, Political Economy of European Financial Integration (Manchester: Manchester University Press, 1998); for a classic characterization of state-financial sector relations in the UK, see Michael Moran, The Politics of Banking, 2nd edition, (London: Macmillan 1986).
12
among the various risks, which ignored the potential gains from diversification. More
generally, the risks assigned to bank loans did not attempt to distinguish between the real
default risks of different sorts of clients: there is a considerable difference between loans
to major, stable and recognized companies versus risky ventures with new technologies
or the uncertainty of speculative minerals exploration.16 Finally, the earlier accord had
not properly accounted for operational risk in loan and securities market portfolios of
banks.
These weaknesses led to the current round of negotiations on a new accord. Given
the many changes in the financial services industries and the growing difficulties
experienced by supervisors in keeping up with the complexity and changing nature of
risk in contemporary financial markets, the starting point was to emphasize the role of
market discipline in risk management. This led to the so-called three pillars consisting of
1. minimum capital requirements; 2. supervisory review of capital adequacy; and 3.
public disclosure.17 Under the three pillar system supervisors, bank risk managers, and
market forces combine in the supervisory process, meaning that bank supervisors will no
longer be exclusively responsible for the supervisory process and specifying levels of
capital adequacy.
Pillar one maintains the basic provisions of B-I but institutes important changes in the
way aspects of credit risk are to be calculated and expanding the range of risks to include
operational risks. Three different options are available to banks under the proposals. The
standardized approach for less sophisticated institutions is based on B-I but enhances risk
16 See A New Capital Adequacy Framework, (Basle: Bank for International Settlements, 1999), pp. 8-9, at http://www.bis.org/publ/bcbs50.pdf. 17 See Basel Committee, Overview of the New Basle Capital Accord, consultative document, (Basle: Bank for International Settlements, April 2003).
13
sensitivity by differentiating among exposure to different sorts of bank clients. It
includes differential ‘risk weightings’ for sovereign and corporate exposures, to be
calculated according to external credit assessments such as the OECD or commercial
ratings agencies (Standards and Poor, Moody’s etc). Option two is a simplified or
‘foundation’ version of the ‘Internal Ratings Based’ (IRB) approach for risk
management, making limited use of internal Value at Risk (VaR) models. And option
three is an ‘advanced’ IRB approach meant for the largest and most sophisticated
financial institutions. In the foundation version, only the probability of default is
calculated by the bank, and all other capital ratios specified by the supervisor. In the
advanced version, all aspects of credit risk are estimated by the bank itself, i.e., it is a
complete ‘self-supervision’ approach, except that the bank has to qualify and the models
it uses have to be approved by the supervisor. Collateral and loan guarantees are to be
taken into account as well. Pillar two is about compliance, and consists of an ongoing
review by the supervisor of the options under pillar one as they pertain to particular
financial institutions. Supervisors must initially approve and regularly assess (stress
testing) the internal application of VaR models in the risk management process. Pillar
three consists of ‘Market Discipline’ as a compliment to the first two pillars. Crucial to
discipline is public disclosure of bank risk profiles and capitalization. This approach is
based on claims by the industry itself18 that market discipline is the best guarantor of
sound risk management, and that supervisory oversight is essentially redundant in a
soundly functioning system of market disciplines. There is also a potential clash with
18 And therefore surely suspect as deriving from narrow self-interest, given the costs associated with supervisory capital. Recent corporate scandals must also surely cast doubt on public disclosure as an essential ingredient of the pillar.
14
national laws concerning the confidentiality of information developed in the supervisory
process.
The initial paper announcing B-II was published in 1999, and is now in its third
version.19 Officially, the proposal is still under review, to be finalized by end of 2003
with implementation by 2006, perhaps postponed to 2007. The long process of
developing the proposals reveals disagreements among four sets of actors: a).
disagreement among Basle Committee members themselves; b). disagreement between
Basle Committee members and other official interlocutors in for example developing
countries; c). disagreements between private sector consultation partners (principally the
Institute for International Finance or IIF), and Basle Committee members; d).
commentary from expert and academic circles. Disagreements among Basle committee
members principally concerns whether the proposal should apply to all banks in a country
or only to ‘internationally active’ banks, and concerning the extent to which it is
mandatory. In part this disagreement reflects political economy factors among major
developed countries, notably Germany and the US. Developing country criticism has
focused on problems of implementation for the less sophisticated financial institutions in
emerging market economies. These are almost by definition corralled into the
‘foundation’ approach of pillar one, which is by definition more costly and will therefore
affect the competitive playing field. Private sector criticisms have focused on the
complexity and stringency of the agreement, and on the extent of corporate disclosure
required.
19 See note 15 re the 1999 document; the second version was The New Basel Capital Accord, consultative document (Basle: Bank for International Settlements, January 2001) at http://www.bis.org/publ/bcbsca03.pdf; and the third version same title, April 2003, at http://www.bis.org/bcbs/cp3full.pdf.
15
Other criticisms may be classified in four categories, three of which are on the overall
approach and one on the specific risks measures being suggested.20
• Arguably, B-II misses the key issue: the importance of systemic risks and
systemic monitoring. Under B-II, supervisors will only designate prudential
standards for ‘systematically’ significant financial institutions operating
internationally. Other banks do not require special treatment in the form of
regulation and supervision and rather, as any non-financial firm, could be
overseen by their shareholders, creditors and other stakeholders. By requiring all
types of banks to adopt B-II, potentially additional and unnecessary costs would
be imposed on the banking system and borrowers. Furthermore, if B-II were
more generally applicable, supervisors might spread themselves too thinly and
thereby not reduce the systemic risks they are set out to manage in the first place.
• B-II relies extensively on market signals, in the form of both asset prices as well
as ratings. The new approach essentially assumes that aggregation of good
practice in individual ‘systemically important’ institutions leads to stability at the
systemic level. However, if a wide range of banks responds to similar perceived
risks in the market, their aggregate behaviour may lead to problems at the
systemic level. In particular, procyclicality is inherent in approaches based on
market prices, such as VaRs and to some extent also rating agencies (although the
latter claim to rate borrowers across business cycles on relative, not absolute
terms). B-II may thus be enhancing the procyclicality already inherent in
financial markets and bank lending. Procyclicality is further aggravated by
encouraging many banks to use the same tools: downturns and upturns are thus
potentially reinforced as banks may en masse downgrade or upgrade clients.
Given these what ought to be obvious difficulties inherent in the approach, it is
perhaps less than coincidental that B-II will cost the major financial institutions to
which it applies rather less in terms of supervisory capital.
20See, for example, Persaud, The Political Economy of Basle II, Inaugural Lecture As Mercer Chair in Commerce at Gresham College.
16
• The hallmark of B-I was its simplicity, even when flawed in some respects. That
of B-II maybe its complexity and stress on sophisticated use of market data. This
is creating unnecessary costs for (small) banks, favors large banks and erects
barriers to entry and competition (especially for financial institutions in emerging
markets). As mentioned above, the advanced approach of pillar is likely to result
in considerable savings in terms of supervisory capital, directly affecting the
transaction costs and thus the competitiveness of institutions. B-II also relies too
much on hard information and too little on ‘soft’ or local information. It is thus
biased against lending to SMEs, which is often more based on relationships. Its
stress on models and rules can furthermore generate a false sense of security and
possibly facilitates regulatory capture of supervisors as they supervisors ‘hide’
behind technical complexity.
• The differential risks weightings of B-II compared to B-I would lead to a
significant increase in capital requirements for loans to lower rated borrowers,
thereby increasing costs and reducing quantity of lending.21 These lower rated
borrowers of course tend to be those who need capital most, or may be those in
the most innovative sectors of the economy. For example, by some calculations
B-II proposals would require spreads for B-minus rated borrowers to rise by 1000
basis points or more (see further below). Furthermore, B-II may place insufficient
emphasis on the risk reduction effects of (international) diversification.
Many of these criticisms apply especially to the lending to firms and sovereigns in
developing countries. Borrowers in these countries are typically lower rated, there is
more limited data on which to base judgments concerning their credit risks, and they
already suffer from procyclical lending patterns. The shift in risk weights is particularly 21 The general agreement has been that B-II cannot lead to an increase in overall capital adequacy requirements compared to B-I; higher requirements for lower rated firms thus will be compensated by reduced requirements for lower rated firms.
17
significant for those developing countries, which are also OECD-members (Hungary,
Mexico, Korea), where under B-I lending is subject to a zero weight. Furthermore,
lending to clients in developing countries may provide risk diversification benefits, which
are not sufficiently recognized in B-II. As a consequence, developing countries could see
the cost of capital increase and lending volumes reduced, with a reinforcement of
procyclicality. We analyze these effects in more the next section, but first discuss the
specific political economy factors underlying the development of B-II.
4. The political economy of Basel II
The Basle Committee on Banking Supervision (initially ‘Basle Committee on
Banking Regulations and Supervisory Practices’) was founded in 1974. The Committee
was an initiative of the G10 central bank governors, who were spurred into action
following the twin collapse of the Franklin National Bank and the Bankhaus Herstatt in
eurocurrency trading, both of which nearly toppled the global financial system at the
time.22 The Committee thus reports to the G10 central bank governors, and membership
(currently in fact 13 countries23) consists of one representative of each national central
bank, and if this is not the banking supervisor, then in addition a representative of the
national supervisory agency (this does not add an extra ‘vote’ and the committee does not
vote anyway, operating on a consensus basis). The initial policy question under
consideration was one of supervisory responsibility for internationally active banking
institutions: who precisely was responsible for supervising bank branches and
22 For more on the history of the Basle Committee, see Duncan Wood, The Basle Committee and the Governance of the Global Financial System (Aldershot: Ashgate Publishing, forthcoming 2004). 23 Belgium, Canada, France, Germany, Italy, Japan, Luxemburg, Netherlands, Spain, Sweden, Switzerland, the UK, and the US.
18
subsidiaries across borders – home or host country? The result was the Basle Concordat
of 1975, which has since undergone numerous refinements and amendments.24
The committee quickly gained a reputation for ‘Olympian’ detachment as a guardian
of the public, essentially state, interest.25 The Committee operated under conditions of
strict secrecy and relative insulation from public and private institutions of government
and market. The institutional culture of its earlier years contributed to this impression:
global financial integration was in its early stages, and the strong ‘public domain’ of the
Bretton Woods post-war era underpinned the Committee’s role and decision-making
processes. The negotiation process leading to Basle I in 1988 was the crowning
achievement, and occurred with little formal consultation with ‘outside’ interests.
There is no doubt that up until the negotiation of the Market Risk Accord of 1996 (in
fact an amendment to the 1988 B-I agreement26), the Committee did operate in a
considerably more detached manner than is the case today. However, Olympian
detachment and insulation from the traditional politics of government lobbies covered up
a more prosaic reality. Financial policy-making has historically taken place in relatively
closed and exclusionary policy communities with central banks and autonomous
regulatory agencies at the core of the system. These policy communities have often been
characterized by ‘business corporatism’ and the delegation of public authority to private
agencies via self-regulation,27 which continues to be a primary instrument in the
regulatory process. This close relationship between regulatory/supervisory agencies and 24 See analysis in Geoffrey R.D. Underhill, “Private Markets and Public Responsibility in a Global System,” in Underhill (ed.), The New World Order in International Finance (Basingstoke: Macmillan, 1997), pp. 23-8. 25 See the state-centric account of Basle by Ethan B. Kapstein, Governing the Global Economy: International Finance and the State (Harvard University Press, 1994). 26 See Basle Committee, Amendment to the Capital Accord to Incorporate Market Risks (Basle: Bank for International Settlements, January 1996). 27 See Coleman, Moran, cited above.
19
their correspondents in the financial services industry is in fact enhanced by the
‘Olympian’ distance of central banks and other autonomous agencies with regulatory and
supervisory responsibilities from the rough and tumble of traditional policy-making in
democratic governments, as in the case of trade negotiations. Thus the politics of
financial governance the national and the transnational level is far removed from
traditional, democratically accountable policy-making, and takes place in relatively
closed communion between financial sector private interests and autonomous public
authorities who share skills and knowledge which enhance their power and effectiveness
in controlling the policy agenda and outcome.28 This is the case even in developing
countries with strong traditions of financial repression and state control of the credit
allocation process.29
The Basle Committee therefore might well deliberate in Olympian detachment, but
national central banks and financial supervisors never did. Supervision and regulation
was developed in close co-operation with a small community of private interests which
shared more with central banks and supervisors than with the other sectors of the
economy and the rest of society (central banks are, after all, banks, and many in Europe
were only nationalized some 60-70 years ago). The process of transnational financial
integration led to a need to renegotiate supervisory and regulatory bargains reached at the
national level in the post-depression/Bretton Woods era. Basle I was the first attempt to
achieve this in relation to capital adequacy, and the outcome of the agreement
redistributed the pattern of competitive advantages and disadvantages across national
28 These points are developed and supported empirically in Underhill, “Private Markets,” op. cit. 17-49; and Underhill, “Keeping Governments out of Politics: Transnational Securities Markets, Regulatory Co-operation, and Political Legitimacy,” Review of International Studies, vol. 21/3, July 1995, pp. 251-78. 29 See Xiaoke Zhang, The Changing Politics of Finance in Korea and Thailand: from deregulation to debacle (London: Routledge, 2003), pp 38-41.
20
sectors: some national banking sectors had to raise substantial amounts of new capital
following the conclusion of the accord, affecting transaction costs. Coupled with
criticisms voiced about B-I (see above), calls emerged for the Committee to consider
more closely the impact of its decisions on the sector. The result was the emergence of a
consultation process with the Institute for International Finance (IIF)30 based in
Washington, and this consultation process developed with Committee’s 1993 proposals
to amend B-I to include securities markets risks as applied to Banks.31
This at first informal and so far unprecedented consultation process began when the
IIF issued a paper sharply criticizing the 1993 Committee paper: the proposals “fail[ed]
to create sufficient regulatory incentives for banks to operate more sophisticated risk
measurement systems than those necessary to meet the regulatory minimum”,32 meaning
VaR models. A well-circulated and authoritative paper by Dimson and Marsh of the
London Business School, claiming to demonstrate that VaR models were more effective
than the Committee’s proposed approach, added to the pressure.33 The result was a new
consultative document from the Committee embracing the approach advocated by the
IIF.34 The pressure had worked, but the Committee’s new and soon to become formal
interlocutor was hardly representative of the range of interested parties which would be
affected by the new amended accord or its successor, B-II. There were no emerging 30 The IIF was originally formed as a consultative group of major US and European banks during the debt crisis of the 1980s, and became a more broadly based organisation representing some 350 member banks worldwide. See website for membership, http://www.iif.com/about/member_list.quagga. 31See Basle Committee, The Supervisory Treatment of Market Risks, consultative paper, (Basle: Bank for International Settlements, April 1993). 32Institute for International Finance, Report of the Working Group on Capital Adequacy (Washington: IIF, 1993), cited in Financial Regulation Report, December 1993, p. 3. 33 Elroy Dimson and Paul Marsh, The Debate on International Capital Requirements: Evidence on Equity Position Risk for UK Securities Firms, (London: City Research Project, London Business School, February 1994). 34 Basle Committee, Proposal to Issue a Supplement to the Basle Capital Accord to Cover Market Risks (April 1995), and — , An Internal Model-Based Approach to Market Risk Capital Requirements (April 1995).
21
market representations and the process did not extend beyond the traditionally close
relationship between banks and supervisors/regulators. At the transnational level, one
may argue, the emerging policy community was even further removed from traditional
lines of democratic accountability in the policy process.
Following the successful consultation process, the industry’s preferences as
articulated by the IIF were translated into Committee policy. The IIF-Basle Committee
relationship became regular practice as the Committee began to consider B-II in the face
of ongoing criticisms of B-I treatment of credit risk, which remained so far unchanged.
In fact, the private sector was playing an even stronger agenda-setting role than in the
past. The review of B-I began with a study group of the Group of thirty, a private think-
tank like body of members drawn from the public/official and private institutions in the
financial sector alike, many of whom had held prestigious appointments in both. The
group formed a study group and issued a report on systemic risk in the changing global
financial system in 1997.35 As Paul Volcker, chairman of the G30 stated in the
‘Foreword’ to the report,
The report concludes that an ambitious effort to produce an international framework to serve as a guide to the management, reporting and supervision of major financial institutions and markets is justified and even imperative, beginning with the global commercial and investment banks. A collaborative effort between financial institutions and their supervisors would be most likely to be effective and broadly acceptable over a wide range of institutions and countries.36
The report observed that management controls should play a central role in the
supervision of financial systems, and that ‘core’ financial institutions should take the
35 Group of Thirty, Global Institutions, National Supervision, and Systemic Risk, (Washington, DC: Group of Thirty/Study Group Report, 1997). The report includes the names of study group participants (pp. ix-x), and members of the G30 itself (pp. 47-8). 36 Ibid., p. ii.
22
initiative to develop a new system along with ‘international groupings of supervisors’.
The conclusions of the report imply that,
supervisors will be readier to rely on the institutions that they supervise, and that the institutions themselves will accept the responsibility to improve the structure of, and the discipline imposed by, their internal control functions.37
Here lie the origins of the market-based supervisory approach contained in the three
pillars of B-II. In 1998 the IIF issued its own report specifically urging the Basle
Committee to update B-I on the basis of banks’ market-based internal control
mechanisms.38 Although the Basle Committee invited consultations on its (now three)
sets of proposals for B-II, the IIF remains the principal interlocutor, and comments come
overwhelmingly from financial institutions, to a lesser extent official agencies, and a few
academics.39
Basle does appear to be opening up. The web site displays the comments of various
interested parties, and who made them. There is, according to Germain, the beginning of
a broader consultation process involving the G20 and banking supervisors in emerging
market countries.40 Yet this is only a beginning and the historically close and isolated
policy community will take time to change. It may be argued that the structural power of
private financial institutions and the knowledge and assumptions about the market shared
by public and private agencies alike constitute the strongest element of the consensus that
little needs to change in global financial architecture.
37 Ibid., p. 12. 38 Institute for International Finance, Recommendations for Revising the Regulatory Capital Rules for Credit Risk, Report of the Working Group on Capital Adequacy (Washington, DC: IIF, March 1998). 39 See Committee web site section on comments on proposals at http://www.bis.org/bcbs/cacomments.htm (comments on second proposal) and http://www.bis.org/bcbs/cp3comments.htm (comments on third consultative document); it is clear that few groups outside financial industry have responded. 40 See Germain, “Global Financial Governance,” op. cit.
23
The minimal claim which this section needs to support anyway is that it is far more
likely the Basel Committee and its member institutions will take into the account the
articulated preferences of private sector interlocutors in developed countries than the
interests of developing country supervisors and their corresponding financial sectors.41
The long-institutionalized relationship between regulators and the regulated in financial
supervision, which approximates conditions of capture, is now developing at the
transnational level. Basle makes decisions which affect global financial governance
without passing the floor of anyone’s legislature and reporting only to a central bankers’
caucus, the Basle Committee increasingly thinks like its private sector and increasingly
acts like it too. Policy outcomes derive directly from the positions of the private sector as
financial globalization renders the supervisory process increasingly difficult and beyond
the reach of national supervisors.
5. The impact of Basel II on developing countries.
If one may conclude from the analysis so far that B-II has largely been negotiated
with the interests of developed country financial systems and institutions in mind, it
remains for us to determine if the new accord might in fact have an adverse impact on the
interests of developing country economies and financial systems. Financial flows to
developing countries are particularly closely associated with the development prospects
of these economies, and any negative effect of B-II on already low capital flows to the
41 This claim is well supported in Geoffrey R.D. Underhill, “The Public Good versus Private Interests in the Global Monetary and Financial System,” International and Comparative Corporate Law Journal, vol. 2/3, 2000, pp 335-359; — , “States, Markets, and Governance for Emerging Market Economies: private interests, the public good, and the legitimacy of the development process,” International Affairs, vol. 79/4, July 2003, esp. pp. 771-774.
24
broad range of emerging markets is not to be welcomed. As noted, B-II may affect
capital flows to developing countries in two ways: through the cost of bank lending,
affecting in turn the sustainable supply of external financing; and through the
procyclicality of lending. Both must however be evaluated relative to the B-I regime, to
the extent the current regime is already binding on banks, or relative to what would have
constituted a first-best solution. We do not analyze the impact of B-II on developing
countries’ financial systems; there are many arguments to be made, however, that such
standards are not only well suited and may done more harm than good to developing
countries.42
Cost of external financing. Several papers have highlighted the increased costs of
external financing for many developing countries as a result of the new Basel accord.43
Some of these estimates are country specific, some for all developing countries, but we
can use nevertheless use them to demonstrate the potential impact of B-II on the cost of
capital. A recent, publicly accessible analysis is Weder and Wedow (2002). They show
that by simply applying B-II versus B-I using publicly available ratings from rating
agencies (on the basis of the proposal as of November 2001), spreads charged by banks
could change between a). –40 basis points for A-rated borrowers and 2000 basis points
for CCC-rated borrowers under the IRB approach and b). between –40 basis points for A-
rated borrowers and 350 basis points for CCC-rated borrowers under the standard
42 Standards applying at the country level raises issues of who sets the standards, whether standards are realistic, who enforces the standards, how to square the standards with countries’ sovereignty, and how to apply penalties for any non-compliance. See further Stijn Claessens, The International Financial Architecture: What is News(s)?, Inaugural Lecture, University of Amsterdam, October 24, 2002. 43 Stephany Griffith-Jones, Miguel Segoviano, and Stephen Spratt, Basel II and Developing Countries: Diversification and Portfolio Effects, mimeo, Institute of Development Studies, University of Sussex, Financial Markets Group, The Londosn School of Economics, December 2002; Helmut Reisen Will Basel II Contribute to Convergence in International Capital Flows?, OECD Development Centre, Paris, 2001; Weder and Wedow, Will Basel II Affect International Capital Flows to Emerging Markets?, OECD Development Centre Technical Paper 199, October 2002.
25
approach. These effects are significant. Countries rated less than BB- by S&P could see
their cost of capital go up under the IRB-approach, which as of 2001, were 10 out of 26
rated developing countries. But, for the standardized approach only borrowers rated
worse than B- would see their spreads increase, which, as of 2001, were only 3 out of 26
rated developing countries. Some, but not the majority of rated developing countries
would thus see an increase in spreads on the basis of a mechanical application of B-II.
The B-II accord allows sophisticated banks not only to use external ratings, but also
internal ratings. There can be a difference between external and internal ratings, which
might alter this conclusion if internal ratings provide a higher share of lower rated
borrowers. We have access to internal ratings of a large Dutch bank, which is
internationally very active and has actually a longer record than the rating agencies in
rating countries and also covers more countries. This bank thus covers many countries
which have not had (or sought) access to international bond markets and which have
consequently not (yet) been rated by S&P. Since these countries are typically the less
creditworthy countries, this bank’s internal ratings are on average lower than the S&P
ratings. This already suggests that the overall impact of B-II on developing countries can
be more adverse than previously noted using external ratings.44
We first map the ratings from the bank with those of S&P and Moody’s (Table 1).
We also provide in the Table the default probabilities as calculated by S&P and Moody’s
for equivalently rated corporate sector borrowers, so as to calculate needed capital
adequacy requirements and resulting spreads.
44 See further Claessens and Embrechts, Basel II, Sovereign Ratings and Transfer Risk External versus Internal Ratings, CEPR Working Paper, 2002 for a description of the data.
26
Table 1: Risk Mapping between internal and external ratings and default probabilities of S&P and Moody’s
S&P ratings Internal rating Default Moody's Default prob. S&P AAA 18 0 0 AA+ 17 0 0
AA 16 0 0 AA- 15 0,06 0,03 A+ 14 0 0,02 A 13 0 0,05 A- 12 0 0,05
BBB+ 11 0,07 0,12 BBB 10 0,06 0,22 BBB- 9 0,39 0,35 BB+ 8 0,64 0,44 BB 7 0,54 0,94 BB- 6 2,47 1,33 B+ 5 3,48 2,91 B 4 6,23 8,38 B- 3 11,88 10,32
CCC+ 2 18,85 21,32 CCC 2 18,85 21,32 CCC- 2 18,85 21,32 CC 2 18,85 21,32
Selective Default 1 18,85 21,32
Note: The risk mapping assumptions are based on Table 3 from Claessens and Embrechts (2002). The default probabilities are taken from Weder and Wedow (2002), Tabel II.2, with the modification that the C-category and SD are separately classified, altgoug they have the same default probability.
We next recalculate the results for the changes in spreads using these internal
ratings for the various credit classes instead of the usual external ratings. Table 2a and 2b
provide the results for Moody’s and S&P respectively. Our results show similar effects
ax for the external ratings, as the cost of bank financing could rise under B-II by up to
1700 to 1900 basis points compared to B-I. The better-rated countries, however, could
see their costs decline by up to some 150 to 180 basis points.45
45 There is again the assumption that the capital adequacy requirements are binding and that the required rates of return are determined in line with the observed spreads for each borrower.
27
Table 2a: Adjustments in spreads for equivalent rates of return under B-I and B-II Using S&P default probabilities
Internal rating
Assumed spread
Default S&P
BRW S&P
S&P cap. req./100$
S&P risk adj. Return (%)
S&P spread change (b.p.)
18 0 0,00 0,00 0,00NA 0,0017 0 0,00 0,00 0,00NA 0,0016 0 0,00 0,00 0,00NA 0,0015 0 0,03 15,72 1,26 0,00 0,0014 0,5 0,02 13,87 1,11 45,07 -43,0713 0,5 0,05 19,17 1,53 32,60 -40,4112 0,5 0,05 19,17 1,53 32,60 -40,4111 1 0,12 28,82 2,31 43,37 -71,1810 1 0,22 39,19 3,14 31,90 -60,81
9 1 0,35 49,62 3,97 25,19 -50,388 4 0,44 55,62 4,45 89,90 -177,547 4 0,94 79,34 6,35 63,02 -82,656 4 1,33 92,04 7,36 54,32 -31,845 7 2,91 126,89 10,15 68,96 188,234 7 8,38 215,47 17,24 40,61 808,263 7 10,32 242,79 19,42 36,04 999,512 7 21,32 362,43 28,99 24,14 1837,032 7 21,32 362,43 28,99 24,14 1837,032 7 21,32 362,43 28,99 24,14 1837,032 7 21,32 362,43 28,99 24,14 1837,031 7 21,32 362,43 28,99 24,14 1837,03
28
Table 2b: Adjustments in spreads for equivalent rates of return under B-I and B-II Using Moody’s default probabilities
Internal rating
Assumed spread
Default Moody's
BRW-Moody's
Moody cap. req./100$
Moody's risk adj. Return (%)
Moody's spreadchange (b.p.)
Notes: The table provides the decrease/increase in spreads such the rates of return under B-II is the same as under B-I. Note that the only differences between the two Tables a and b is that we report estimates based on S&P default probabilities (2001) or Moody default probabilities from 2000. The mapping of internal risk ratings to Moody’s and S&P ratings can be found in Table 1. Appendix 1 provides the calculations. BRW=benchmark risk weight.
18 0 0,00 0,00 0,00NA 0,0017 0 0,00 0,00 0,00NA 0,0016 0 0,00 0,00 0,00NA 0,0015 0 0,06 20,75 1,66 0,00 0,0014 0,5 0,00 0,00 0,00NA -50,0013 0,5 0,00 0,00 0,00NA -50,0012 0,5 0,00 0,00 0,00NA -50,0011 1 0,07 22,24 1,78 56,20 -77,7610 1 0,06 20,75 1,66 60,25 -79,259 1 0,39 52,39 4,19 23,86 -47,618 4 0,64 66,61 5,33 75,06 -133,567 4 0,54 61,44 4,92 81,37 -154,226 4 2,47 118,50 9,48 42,19 74,005 7 3,48 137,14 10,97 63,80 259,974 7 6,23 182,67 14,61 47,90 578,723 7 11,88 263,18 21,05 33,25 1142,272 7 18,85 339,80 27,18 25,75 1678,572 7 18,85 339,80 27,18 25,75 1678,572 7 18,85 339,80 27,18 25,75 1678,572 7 18,85 339,80 27,18 25,75 1678,571 7 18,85 339,80 27,18 25,75 1678,57
According to this calculation, and only considering those 40 countries for which we
have both external and internal ratings, the number of countries that would see their cost
of external financing increase is actually less than half on the basis of the internal ratings
as of end 2000 (Figure 1). The impact of Basel II can be therefore interpreted as generally
positive, as most middle-income countries have a rating higher than a scale of 6. If the
ratings were used as of early 1990s, then there would be more countries with a rise in
costs than countries with a drop. Essentially, the number of countries in lower rating
29
classes diminishes over time (Figure 2). The internal ratings may have improved over
time as the fundamentals of the countries improved or as the bank has learned more of the
countries or is able to better manage risks over time. Regardless, there remain a number
of countries for which B-II has adverse impacts on cost of external financing; since these
countries already have difficulty obtaining financing, they are further negatively affected
by the new accord.
Figure 1
Number of countries which have a positive, neutral, or negative spread change due to Basel II according to Internal ratings in Oct-90, Oct-96, and Apr-01 (total in sample=40)
0
5
10
15
20
25
30
35
positive spread change neutral spread change negative spread change
Oct-90Oct-96Apr-01
30
Figure 2
Number of countries in each internal rating class in Oct-90, Oct-96, and Apr-01 (sample=40 countries)
0
2
4
6
8
10
12
14
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Number of countries
Oct-90Oct-96Apr-01
The average impact of Basel II, based on these results, seems to be modest. Using the
internal ratings, however, the average increase still tends to be higher than for the
external ratings, since the bank also rates more countries, including lower creditworthy
countries. Figure 3 shows the average required spread increase for the complete sample
of developing countries for which we have either the internal or the external ratings.46
Under both ratings, the spread change is positive, and actually increasing, over the last
years. This confirms that under Basel II to keep the risk-adjusted returns the same as
under Basel I, spreads have to increase. For the internal ratings, the level of required
spread is larger for all periods as the bank rates more less creditworthy countries. The
earlier results that were based only on the externally rated borrowers did thus
46 Note that internal ratings includes almost all countries; only around 1997 does S&P has as much countries as the internal ratings cover. Since in the beginning of the period, S&P rated only the best, capital requirements based on external ratings are lower on average, so the graph before 1997 is biased.
31
underestimate the effects of the cost of capital as only the more creditworthy borrowers
are rated by S&P and Moody’s.
Figure 3
Average spread change in basis points under Basel II to produce risk adjusted return under Basel I
based on S&P and Internal ratings
-200
-100
0
100
200
300
400
500
600
700
okt-90
feb-91
jun-91
okt-91
feb-92
jun-92
okt-92
feb-93
jun-93
okt-93
feb-94
jun-94
okt-94
feb-95
jun-95
okt-95
feb-96
jun-96
okt-96
feb-97
jun-97
okt-97
feb-98
jun-98
okt-98
feb-99
jun-99
okt-99
feb-00
jun-00
okt-00
feb-01
avg. int . spread djII I i lAvg. spread adj. S&P
Nevertheless, as several papers have pointed out, there are some flaws in this form of
analysis and a number of factors might actually mitigate the impact of B-II. These
corrections include the fact that the simple analysis presumes that banks want to keep the
risk-adjusted rates of return the same under B-I compared to B-II. The ex-ante required
rates of return implied by the capital adequacy weights under B-I are quite high, however.
For low rated borrowers, for example, the capital adequacy requirements combined with
the default probabilities of the corresponding rated class of corporations imply required a
three-fold increase in spreads (for B-rated assets). These very high required spreads for
lower rated borrowers are the result of applying the same ex-ante required rates for each
credit class under B-II as under B-I. Using a more realistic assumption that banks use a
fixed hurdle rate across all asset classes (of, say, 18 percent as suggested by Powell,
32
2001)47 would lower the increases in required spreads to between 100 and 200 basis
points for lower rated borrowers. Of course, this hurdle rate is ad-hoc and potentially
inconsistent with the principles of the risk-based approach, which requires different rates
as adjustments are made for risks, but it still shows some of the flaws.
Another mitigating factor is that developing countries receive funds from sources
other than banks, such as capital markets and non-bank financial institutions that are not
subject to capital adequacy requirements. This would reduce the impact of B-II; of
course, the access to capital markets and other financing may be more limited for
precisely these same lower rated countries, thus negating this effect. Also, banks subject
to the standard approach face lower capital requirements that those using the IRB
approach when lending to lower rated borrowers (specifically in the range below BB+).
Some clientele relationships may arise where banks using the IRB-approach choose the
safer borrowers and the banks using the standardized approach the riskier borrowers.48
This competition and clientele effects can mitigate some of the impact, although it cannot
be assumed that there are perfect substitutes available. There can, for example, be value
for borrowers from bank lending (from IRB-banks) which is not obtainable elsewhere.
For example, banks may better be able to assess, monitor and manage risk, and for those
reasons may be able to provide financing to countries relatively more cheaply. Without
these gains, the presence of perfect substitutes would raise the general question whether
mandatory capital adequacy requirements would ever be relevant as there always would
be some perfect substitute somewhere available.
47 Andrew Powell, A Capital Accord for Emerging Economies. Documento de Trabajo, No. 08/2001, Universidad Torcuato Di Tella, September. 48 While this may mitigate the effects on developing countries, it would go against the objectives of the new Basel accord in the first place as it introduces another distortion and may lead to risk-taking by those banks least qualified to assess risks.
33
The most important adjustment to the simple calculations is that banks may not be
constrained by the (new) capital adequacy requirements as they may already be adjusting
their economic capital in line with the risk associated with particular countries. Of
course, this argument can make B-II in a general sense irrelevant: if banks are already
doing what economic capital models require, there would not be any impact of capital
adequacy regulations properly based on such economic models. This goes against the
general trust of having an accord in the first place, so it is reasonable to assume there is
some binding effects of the accord. Consequently, there will have to be some effect on
spreads. Weder and Wedow (2002) investigate this issue as well by studying the
relationships between actual loan volumes to emerging markets and the capital charges as
would be required under B-II using the IRB. They find that the capital flows from BIS
reporting banks to 25 emerging markets over the period 1993-2001 are already affected
by the simulated B-II capital adequacy requirements, consistent with the interpretation
that banks have already largely adjusted their claims using a model anticipating the new
capital adequacy requirements. They do find that German banks may have been
constrained, but not the other countries.
Nevertheless, there might still adjustment necessary for some countries if the new
accord is not well calibrated; the simulation above suggest that some lower rated
countries may see their costs increase. Furthermore, internal ratings may differ from
external ratings, important as B-II allows greater use of internal ratings. Analysis by
Claessens and Embrechts (2002) suggests that the differences between the two types of
ratings are generally small (Figure 4). Still, on a country-by-country basis, the ratings are
not perfectly correlated over time (Figure 5). The average correlation for a sample of 40
34
developing countries over the 1997-2001 period between the two ratings is only 0.42.49
There has also been evidence reported that external ratings are slower to adjust to large
events, such as the East Asian countries’ financial crisis, than internal ratings are.
Figure 4
April 2001
0
2
4
6
8
10
12
14
16
18
20
0 2 4 6 8 10 12 14 16 18
Internal Rating
Exte
rnal
Rat
ing
Moody's S&P Linear (Moody's) Linear (S&P)
Note: internal and external ratings compared as of April 2001, using the conversion scale of Table 1.
35
49 The sample is small and short as few countries were rated in the early 1990s.
Figure 5
Correlation between S&P and Internal ratings in the period October 1997-April 2001
-0.2
0
0.2
0.4
0.6
0.8
1
Argentina (Republic of)
Botswana
Brazil (Federative Republic of)
Chile (Republic of)
China (People's Republic of)
Colombia(Republic of)
Cyprus (Republic of)
Ecuador (Republic of)
Egypt (Arab Republic of)
Greece (Hellenic Republic)
Guatemala(Republic of)
Hungary (Republic of)
Iceland (Republic of)
India (Republic of)
Indonesia (Republic of)
Israel(State of)
Jordan (Hashemite Kingdom of)
Kuwait (State of)
Malaysia (Federation of)
United Mexican States
Morocco (Kingdom of)
Oman (Sultanate of)
Pakistan (Islamic Republic
f)
Panama (Republic of)
Paraguay (Republic of)
Peru (Republic of)
Philippines (Republic of the)
Poland (Republic of)
Portugal (Republic of)
Qatar (State of)
Romania (Republic of)
Senegal
Singapore (Republic of)
South Africa (Republic of)
Taiwan (Republic of China)
Thailand (Kingdom of)
Tunisia (Republic of)
Turkey (Republic of)
Uruguay (Oriental Republic of)
Venezuela (Republic of)
Average total correlation: 0,42
Notes: Correlations refer to between internal and external ratings over period 1997-2001, on a quarterly basis. Some correlations are near zero due to the fact that at least one rating series has (near) zero variance, which makes for very low correlations.
One other aspect of the question that has been raised in relation to both B-I and B-II
is the (lack of) diversification effect of capital adequacy requirements. As risks are not
perfectly correlated, the sum of individual capital adequacy requirements as determined
by economic models applied to individual credits, does not add up to the overall need for
capital in respect of economic models applied to the overall credit portfolio. Banks not
only benefit from this diversification, but in fact in part exist as intermediaries for this
reason, and hold diversified portfolios to reduce their overall capital needs. B-II
acknowledges this diversification effect and allows banks to use an average asset
correlation of between 0.1 and 0.2, with lower correlations for assets more subject to
36
probable default (an increase in the asset default risk is argued to indicate a more
idiosyncratic nature of the asset, thus justifying a lower correlation). Still, as argued by
Griffith Jones et al. (2001), developing countries as a group exhibit a lower correlation
with developed countries than the correlations between most assets within countries or
from different developed countries. The potential diversification benefits from lending to
developing countries are therefore argued to be large, justifying lower capital adequacy
requirements. Griffith Jones et al. (2001) show that the chance of unexpectedly large
losses on a portfolio evenly distributed across developed and developing countries is
some 25 percent lower than that of a portfolio only distributed among developed
countries. Consequently, the capital adequacy charges should be set lower for a well-
balanced portfolio that includes developing countries.
This argument, however, also is only relevant if the capital adequacy requirements are
binding and not if banks already can, and do, allocate capital according to economic
criteria without regard to formal capital constraints. Furthermore, there is presumably
also a supply of assets within developed countries which also have low correlations with
other assets and that could also provide the diversification benefits sought.50 The issue of
low(-er) correlation for some specific assets also raises the question whether adjustments
should be allowed within the approach for specific assets or whether a generic approach
should be maintained.
In short, this section has demonstrated that on balance, the cost argument is not the
most important to the new accord from the point of view of most developing countries.
While there can be some impact for some borrowers, and especially for those with
50 A complete test would then also require comparing the div benefits from investing in emerging markets with those available from investing in all type of assets; this is done in the so called spanning literature.
37
limited access to market-based external financing, it need not be large on average for
developing countries as a group. At the same time, the analysis has shown that there is
little in the new accord that specifically addresses concerns of developing countries or
anything that could be attributed to developing countries’ specific inputs.
Volatility. Basel II may also have a potential adverse effect through the possible
increase in volatility of access of borrowers to bank financing and increased volatility.
As noted, there is an element of procyclicality in the new accord as it encourages greater
use of models that rely more on market data, including asset prices, which are procyclical
to begin with. Furthermore, requiring the same type of model to be used by many banks
will induce more commonality among banks, thus increasing the risks of financial
contagion as banks react simultaneously to the same or similar signals. These tendencies
may be aggravated as the accord encourages greater use of external and internal ratings.
Both types of ratings are arguably volatile and probably procyclical (see Lowe, 2002).51
Since developing country assets are arguably already subject to more volatility and
procyclicality than other asset classes are, the introduction of B-II might be particularly
harmful for emerging markets.
This issue may be analyzed by comparing the volatility of developing country ratings
of with those of developed countries, beginning with a comparison of the behavior of the
internal ratings (IR) with that of the external ratings (ER). If there is more volatility in
the IR compared to the ER than there is some suggestive evidence that the new accord
will lead to more volatile lending. The argument has been made that while there is broad
similarity in migration probabilities between external and internal, external are less
51 Philip Lowe, Credit Risk Measurement and Procyclicality, Bank for International Settlements Working Paper, 116.
38
responsive than internal. External ratings tend to be slightly more stable and adjust
downward more step by step, whereas internal ratings adjust quicker, less ratchet and
more one-off effects in downgrades. Especially for in higher categories, can there be
sharp adjustment, often related to unwillingness to pay, of internal ratings.
We again use the ratings from a major internationally active Dutch bank and compare
these to S&P, converting both ratings to an ordinal scale from 1 to 20. We compare the
raw volatility for the same country. We find that the average (and median) volatility of
the internal ratings is higher than that of the external rating (Figure 6). The average
variance of the internal rating is 0.99, while the average external ratings is 0.48. Using a
F-test, we can show that the difference is statistically significant at the 1% level. Internal
ratings are thus much more variable than external ratings are, risking increases in the
volatility of capital flows.52
52 We should note that the distribution of both ratings is not normal, and as documented there is considerable rigidity in the ratings, followed by sudden adjustments. This can affect the power of the tests.
39
Figure 6
The problem with these simple comparisons is that correction needs to be made for
the
h the
ess of
Average variance of S&P and Internal ratings per country
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0
Argentina (Republic of)
Botswana
Brazil (Federative Republic of)
Chile (Republic of)
China (People's Republic of)
Colombia(Republic of)
Cyprus (Republic of)
Ecuador (Republic of)
Egypt (Arab Republic of)
Greece (Hellenic Republic)
Guatemala(Republic of)
Hungary (Republic of)
Iceland (Republic of)
India(Republic of)
Indonesia (Republic of)
Israel (State of)
Jordan (Hashemite Kingdom of)
Kuwait (State of)
Malaysia (Federation of)
United Mexican States
Morocco (Kingdom of)
Oman (Sultanate of)
Pakistan (Islamic Republic
f)
Panama (Republic of)
Paraguay (Republic of)
Peru (Republic of)
Philippines (Republic of the)
Poland (Republic of)
Portugal(Republic of)
Qatar (State of)
Romania (Republic of)
Senegal
Singapore (Republic of)
South Africa (Republic of)
Taiwan (Republic of China)
Thailand (Kingdom of)
Tunisia (Republic of)
Turkey (Republic of)
Uruguay (Oriental Republic of)
Venezuela(Republic of)
Variance Internal i iVariance S&P ratings
Total ave variance Internal ratings = 0,99 Total average variance S&P ratings = 0,48
underlying volatility of countries’ fundamentals. For example, ERs may not be
‘volatile enough’ if the external rating agencies do not adjust their ratings in line wit
changes in the underlying volatility. The higher volatility of internal ratings may then
more accurately reflect the higher volatility of the underlying fundamentals. This
nevertheless does not take into account measures of the fundamental creditworthin
borrowers. Measures such as secondary market prices for debt (or spreads) suffer from
the problem than the spreads are endogenous to the ratings themselves (although there is
some evidence that spreads are better predictions of country fundamentals than ratings
are).
40
One simple, suggestive test is to see whether the IR are more correlated with GDP
than the ER are, thus suggesting more procyclicality under B-II if many international
banks use internal models. We find that on this score the correlations between IR and
GDP are on average higher (0.40) than those between the ER and GDP (0.36) (see also
Table 3). Using a F-test, we can show that the difference is statistically significant at the
1% level (TBD). This is confirmed by Lowe (2002) review of studies suggesting that
capital adequacy requirements derived from S&P are less cyclical than those derived
from internal ratings. Whether GDP can be considered to the right measure of
fundamental values is left open.
Insert Table 3 with average correlations with GDP (to be done).
Under the assumption that the behavior of this international bank is representative
of the behavior of other international operating banks, greater use of internal ratings
could lead to an increase in the volatility and procyclicality
6. Conclusions
This paper has argued and offered evidence in support of the following points. Firstly, it
argued that the debate over the reform of financial architecture has been
disproportionately constrained relative to the frequency and depth of financial crisis in
emerging market countries. The system has not been seriously adapted to the needs of
developing and other countries, and specific proposals to stabilize the system during debt
workout processes following acute crises, such as the Fund’s Sovereign Debt
41
Restructuring Mechanism (SDRM), have been dismissed. The onus continues to be
placed on developing countries themselves to address internal weaknesses and strengthen
their position in the global financial system. Standards continue to be promulgated
largely by developed countries and compliance monitored through the very institutions of
global governance which they dominate. Proposals to attenuate the market-based
pressures of global financial integration and its consequences for the poor in the
development process do not find their way onto the reform agenda despite evidence that
these might bring benefits.53
Secondly, this paper analyzed the background to and substance of Basle II itself. The
paper looked directly at the political economy of the Basle process and how this policy
process yielded the current proposal. The evidence supported the claim that the Basle
process was dominated by developed country supervisors in a close relationship with
major developed country financial institutions, suggesting strong capture of the political
economy process. This provides a clear explanation as to why the needs of developing
countries might so poorly be taken into account by the Basle Committee, despite the fact
that the new accord has major implications for supervisory practices and costs in markets
around the globe.
Finally, the paper posed the question as to whether there is indeed evidence that the
new capital accord will have an adverse effect on the costs of capital and by implication
the volume of capital flows to developing countries. It argued that there were potentially
sound arguments for believing that the new accord will have an adverse impact on the
costs and volumes of capital flows to developing countries. It found some evidence that
53 See section 1 of Underhill and Zhang (eds.), International Financial Governance, op. cit., especially articles by Williamson (“Costs and Benefits of Financial Globalisation”) and by Cohen (“Capital Controls: the Neglected Option”).
42
the accord can increase the cost of capital for some classes of borrowers, although the
effects on average are small. Importantly, it found some evidence that the procyclicality
of capital flows to developing countries can increase with the use of internal ratings by
international active banks. The increase in fluctuations in the availability of external
financing would be a very unfortunate outcome, given that developing countries already
suffer from volatile capital flows.
43
44
Appendix: Calculations of required spreads and requirements The results for Table 2 used the following formulas, from Basel II newest modifications: http://www.bis.org/bcbs/qis/capotenmodif.pdf, page 5 Correlation (R) = 0.10 × (1 - EXP(-50 × PD)) / (1 - EXP(-50)) + 0.20 × [1 - (1 - EXP(-50 × PD))/(1 - EXP(-50))] Maturity factor (M) = 1 + 0.047 × ((1 - PD) / PD^0.44) Capital requirement (K) = LGD × M × N[(1 - R)^-0.5 × G(PD) + (R / (1 - R))^0.5 × G(0.999)] Risk-weighted assets = K * 12.50 We assume, like Weder and Wedow (2002), LGD=50 (see their note 6, “In the consultative document from January 2001, the Basel Committee expressed its belief that a LGD rate of 50 per cent for senior unsecured claims”) This yields the formula used: Risk-weighted assets= 625* N[(1 - R)^-0.5 × G(PD) + (R / (1 - R))^0.5 × G(0.999)] (1 + 0.047 × ((1 - PD) / PD^0.44)) For the table, we used the Libor spreads in Table III.1 of Weder and Wedow (2002), and the reported default probabilities of Moody’s and S&P in Table II.2 of Weder and Wedow (2002), respectively. The interpretation of the tables is similar to Table III.1 of Weder and Wedow (2002).