How to minimize incentives to rating-centered regulatory

Journal of Business and Policy Research
Vol. 7. No. 1. April 2012 Special Issue. Pp. 30 - 48
How to minimize incentives to rating-centered regulatory
arbitrage? The current Brazilian Approach to Minimum
Capital Requirements
Rodrigo Barbone Gonzalez*, Fernando Sotelino**
and José Roberto Ferreira Savoia***1
This theoretical paper explores the issue of regulatory dependence in credit
ratings, how it emerged and evolved exacerbating conflicts-of-interest
exposed in the Meltdown and leading to banking reforms in the U.S and in
Europe. It specifically examines The Dodd-Frank Act, the European
Commission proposal for Credit Rating Agencies (CRAs) and Basel II.5 and
III new trading book. After concluding that these three proposals do not fully
neutralize the incentives towards rating-centered regulatory arbitrage in
banking regulation, this paper points to the benefits of the Simplified
Standardized Approach of Basel II, the current Brazilian approach, and
demonstrates how it fully addresses such issue.
JEL Codes: G38
1. Introduction
The financial crisis of 2007-09 has generated considerable consternation about
distorted incentives on the part of key players in the financial markets, and the extent
to which these contributed to the crisis. Indeed, there are some who believe that the
crisis was in many ways a credit rating crisis. Financial securities emerging from
securitization, such as mortgage-backed securities, represented over $11 trillion
worth of outstanding U.S. debt (Thakor, 2010).
The Group of Twenty (G-20), the Financial Services Authority (FSA), the Financial
Stability Board (FSB), the International Monetary Fund (IMF) and the Organization
for Economic Cooperation and Development (OECD) have each presented
proposals to establish a level playing field in the CRA industry. All these proposals
combine recommendations on CRAs’ business models and the regulatory framework
surrounding them; but they still fail to properly address the issue of excessive market
reliance on rating-centered regulations.
This theoretical paper explores the issue of regulatory dependence in credit ratings,
how it evolved leading to some of the problems observed in the Meltdown and how
the Simplified Standard Approach (SSA), the current Brazilian Basel II approach,
prevents and could help to eliminate this type of regulatory arbitrage. It is divided in
three sessions: 1) The Dodd-Frank Act and the European Commission proposal for
*Rodrigo Barbone Gonzalez, Central Bank of Brazil, Sao Paulo, E-mail: [email protected]
The views expressed in this work are those of the authors and do not necessarily reflect those of the
Central Bank of Brazil or its members.
** Prof. Fernando Sotelino, School of International and Public Affairs, Columbia University, New York,
E-mail: [email protected]
*** Prof. Dr. Jose Roberto Ferreira Savoia, Department of Finance, School of Economics, Business
Administration and Accounting, University of Sao Paulo, Sao Paulo, E-mail: [email protected]
Gonzalez, Sotelino & Savoia
CRAs and their limitations; 2) Evolution of Credit Rating Agencies‟ (CRAs‟) role in
banking regulations from Basel I to Basel III; 3) The benefits of a Simplified
Approach of Basel II: exploring the Brazilian case; besides a conclusion and this
According to White (2010), following the 1930‟s market crash in the U.S., “bank
regulators were eager to encourage banks to invest only in safe bonds” and issued a
set of regulations culminating in a 1936 decree that “prohibited banks from investing
in „speculative investment securities‟ as determined by „recognized rating manuals‟.
With these regulations in place, banks were no longer free to use any other
judgments - not even their own - besides those of the publishers of these
“recognized rating manuals”. Before the Crash of 1929, Moody‟s, Poor‟s and Fitch
were dedicated to sell thick manuals of bonds‟ ratings to investors.
In the following decades, insurance state regulators followed a similar path, including
the establishment of a minimum capital requirement based on the ratings of the
bonds insurance companies invested in. In the 1970s, federal pension regulators
pursued a similar strategy. In 1975, the U.S. Securities and Exchange Commission
(SEC) created the Nationally Recognized Statistical Rating Organization (NRSRO)
designation and consequent protective entry barrier. Industry restructuring and
consolidation caused the leading NRSROs by the turn of the millennium (about the
time the market for structured finance, including for mortgage backed securities was
about to take-off in the U.S. and globally) to be the big three: Fitch, Moody‟s, and
Standard and Poor‟s. Formal detailed criteria for a firm to be designated a NRSRO
were never established by the SEC (Altman et al., 2010). This pattern of ratingcentered regulation and supervision was also followed in Japan and in the E.U.
(BCBS, 2009).
The business model of the large rating agencies began to shift from the “investor
pays” to “issuer pays” in the early 1970s. White (2010) points to the following
- A free-rider problem in the use of the rating manuals. Supposedly, the
widespread use of the high-speed photocopy machine imposed a fear that
investors at large would obtain copies from their friends tumbling their sales;
- Realization that the bonds‟ rating business, like other information industries
such as newspapers or magazines, is a “two-sided market”: payments could come
from one or both sides of the market, from those interested in the information as
well as those interested in the distribution of the information;
- Given that financial regulations demanded a “blessing” of one or more rating
agencies in order for bonds to fit their way into the portfolios of important
institutional investors, issuers should be even more willing to pay for this privilege.
Regardless of the reasons, the shift to the “issuer pays” business model (or at least
“mainly pays”) opened the door to potential conflicts of interest, an abundantly
studied moral hazard dilemma (e.g. Sy, 2009; Covitz and Harrison, 2003; Altman et
al., 2010; Cantor and Parker, 1994; Hull and White, 2009).
Rating agencies‟ concerns about their reputations apparently kept conflicts-of-
Gonzalez, Sotelino & Savoia
interest in check in the first three decades of experience with the new business
model (e.g. Smith and Walter, 2002; Altman et al., 2010; White, 2010). But they
seemed to have been overwhelmed by the pressures from structured finance
products.i “The testimonies of former NRSROs‟ employees to the House of
Representatives in 2009 also helped clarifying the extent of the promiscuous
relationship between issuers and CRAs (Altman et al, 2010)”.
In the E.U., the big three received the especial status of External Credit Assessment
Institutions (ECAIs). The designation was created under the Capital Requirements
Directive (CRD) that implemented Basel II in the E.U. The ECAIs are the only
institutions whose opinions can be used by banks following the Basel II Standard
Approach (Utzig, 2010).
Calomiris (2009a) identifies two potential types of failures in credit ratings: inflated
ratings, defined as “purposeful under-estimation of default risk on rated debts” and
low-quality ratings resulting from flawed measures of underlying risk. “The recent
collapse of subprime-related securitizations revealed both problems in the extreme”.
Data provided by the IMF (2009, p. 89-93) confirms these views showing that 90% of
the AAA issuances were downgraded in 2008 and almost 60% of them below B
levels as of June 2009.
“The overall low quality of the ratings is even more problematic than the inflated
ratings itself (Calomiris, 2009b)”. It should be noticed that the securitization sponsor
(issuer side) is not necessarily the party who most benefits from the inflated ratings;
the buy side, whose capital charges are guided by the inflated ratings is the one that
benefits - hopefully to maturity or divestment - from the “extra yield” on the externally
rated security. Rating-constrained investors are attracted not only by the higher yield
of “structured securities”, when compared to corporate bond issues with the same
credit rating, but by the higher potential leverage. Moreover, some institutional
investors such as U.S. pension funds that are forbidden to held riskier securities
become entitled to buy these AAA securitized products.
Altman et al (2010) suggest that reform in CRA regulations should take into account
three aspects:
1. reduction of their role in the global capital marketsii and withdrawal of
regulation dependence on CRA opinions (particularly for capital
2 increase of the overall disclosure and quality of the information in CRAs‟ credit
reports (e.g. the inclusion of PD, LGD and stress tests information, as well as
disclaimers regarding liquidity and market risk; and
3 reduction of conflicts-of-interest, especially “issuer pay” faulty incentives.
As this paper demonstrates, the first aspect is almost neglected by current reforms
(session 2). The advances in this matter are minimal and they are bound to be
minimal, because the overreliance on CRAs´ opinions is deeply embedded in the
very nature of Basel II (session 3). Session 4 clarifies how Basel I and the Simplified
Standard Approach of Basel II address the issue.
Gonzalez, Sotelino & Savoia
2. The Dodd Frank Act and the European Commission Proposal
The Dodd-Frank Wall Street Reform and Consumer Protection Act, The Dodd-Frank
Act, considers these three problems iii. It calls for:
- improvement of NRSROs‟ internal controls and governance, including
complete separation between marketing and rating departments; creation of an
independent compliance department (Section 932a);
- independence of half of the members of a NRSRO board of directors from the
agency or its affiliates (Section 932t);
- empowerment of the SEC to have the right to suspend or revoke the NRSRO
permit to operate or issue reports in particular classes of securities (Section
932a) and also to deliberate on fines and penalties;
- removal of the Fair Disclosure (FD) rule, making the NRSROs accountable for
their reports, as it already applies to auditors, security analysts and investment
bankersiv (Section 939B);
- establishment and enforcement of look-back clauses when analysts leave
NRSROs to join rated companies (Section 932a);
- charging the SEC with the responsibility for training and testing NRSROs‟
- making public the information on an issuer‟s first rating, the methodology
utilized, limitations identified and explanations on the volatility of credit ratings
public (Section 932a);
making public statistical results (such as Probability of Default and Loss Given
Default), underlying assumptions and sensitivity analysis of the rating
obtained to changes in these underlying assumptions (Section 932s);
- making public the information on third party due diligence;
- studying the creation of an independent professional association for credit
rating analysts for the establishment of independent standards for governing
the profession of rating analysts and of a code of ethical conduct as well as
overseeing the profession of rating analysts (Section 939E).
- studying the creation of a standardized ratings terminology to be adopted by
the entire CRA industry “requiring a quantitative correspondence between
credit ratings and a range of default probabilities and loss expectations under
standardized conditions of economic stress (Section 939)”.
- examining the possibility of bringing together NRSROs and issuers of
structured finance products through an independent public private or selfregulated organization (Franken Amendment - Section 939F).
Gonzalez, Sotelino & Savoia
The Franken Amendment proposes a process to eliminate shopping around in the
initial rating of a structured finance product: an independent organization (to be
created) would take the request and a flat fee from the issuer, sponsor or underwriter
of the structured debt and assign this request to a CRA, either randomly or based on
the CRA‟s proven degree of accuracy in rating that specific product.
The merits of this process - minimization of conflict-of-interest and incentives for
better quality ratings - are acknowledge by Altman et al. (2010), especially if
accuracy and tracking error evaluation methodologies can be developed (e.g.
Calomiris, 2009a, 2009b).
Finally, The Dodd-Frank Act requires the “removal of statutory references to credit
ratings in federal and state law” and the replacement of the terms “non-investment
grade” for “does not meet standards of credit-worthiness”. Regulators at each
supervisory agency are commanded to develop their own creditworthiness standards
and scales and to review rating-centered regulations (Section 939a).
In April 2009, the European Parliament approved the Regulation (EC) No. 1060/2009
of the European Parliament and of the Council on Credit Rating Agencies, a set of
rules addressing mainly disclosure and corporate governance guidelines for the CRA
industry (European Parliament, 2009). This initiative, as in the U.S., under the
influence of the IOSCO code, forced CRAs to establish compliance departments; to
appoint independent directors to their boards; to separate Sales and Rating
departments and to enforce “look back” clauses and procedures upon hire of their
professionals by third parties. The Committee of European Securities Regulators
(CESR) also prohibited CRAs from engaging in consulting services to rated entities,
in addition to requiring CRAs to disclose the methodology utilized and the underlying
assumptions of any credit rating issued (Utzig, 2010).
In January 2011, the European Security Markets Authority (ESMA) was established
with powers to register, supervise and cancel the registration of CRAs, surpassing
national authorities.v In June 2011, Regulation (EU) No 513/2011 strengthened
ESMA a little bit more giving the agency exclusive supervisory powers over CRAs.
Main issues now under discussion in the European Parliament regarding CRAs‟
regulation and supervision (European Commission Consultative Paper, 2010) vi and
the “Proposal for a Regulation of the European Parliament and of the Council:
amending Regulation (EC) No 1060/2009 on credit rating agencies” include the
greater transparency and competition among CRAs, including requirement for
issuers to make the information provided to the selected CRA available freely
to all interested parties;
how to facilitate and foster unsolicited ratings;
new standardized terminology for regulatory rating scales, i.e. the creation of
the European rating index (EURIX);
further means to mitigate the conflicts-of-interest inherent to the “issuer pays”
business model; which are mainly addressed with transparency requirements
Gonzalez, Sotelino & Savoia
such as amount paid by every issuer; the issuer need to change CRAs every
three years; and the need to hire two CRAs in the case of structured products.
the question of “whether there is a need to consider introducing a civil liability
regime in the E.U. regulatory framework for CRAs (European Commission,
2011; p. 24).”
creation of a public European CRA to compete with the private sector (Public
Utility model);
Market discipline and competition set the tone in the European Commission
Consultative Paper on CRAs. Controversial proposals, such as changing the CRAs
business model and creating a public CRA (Public Utility model) were put forward,
but ruled out on the last version of the document: European Commission (2011)
“Proposal for a Regulation of the European Parliament and of the Council: amending
Regulation (EC) No 1060/2009 on credit rating agencies”. The incentives in the
“investor pays” and in the Public Utility model were not found to be adequate. As
Calomiris (2009a)vii pointed out, the institutional rating-constrained investor (buy
side) may be the party benefiting most from an inflated rating in the form of lower
capital charges. Additionally, a Public CRA would be subjected to conflicts of interest
especially if rating sovereign debt (European Commission, 2011, p.12).
In short, material progress has been made regarding CRAs‟ internal governance
requirements and greater disclosure, both in the E.U. (since the enactment of the
Regulation (EC) No. 1060/2009 on Credit Rating Agencies) and in the U.S. (since
the Credit Rating Agency Reform Act of 2006). However, the objective of reducing
market reliance on rating-centered regulations remains not ignored but neglected on
both sides of the Atlantic and reduced to recommended changes in regulatory
terminology and rating scales. In other words, removing the linkage between
minimum capital requirements and NRSROs‟ or ECAIS‟ opinions is a fundamental
question both regulations skip.
3. Evaluating the Basel Accords: From Basel I to Basel III
Minimum Capital Requirements (MCR) are devised to withstand unexpected losses
(UL). Under Basel I, all loans were treated equally in terms of their potential
contribution to Unexpected Losses (UL) and a capital charge of 8% over total RiskWeighted-Assets (RWAs) was needed. In essence, all individual loans would be
charged a RWA of 100%. Banks‟ internal credit ratings could be utilized exclusively
to infer Expected Losses (EL).
Basel I did not allow the use of External Credit Ratings for determining capital
requirements. It also did not take into account differences in risk management and
loan recovery capabilities. In principal, credit exposures could generate the same
amount of losses, regardless of their nature, internal classifications, insurance or
collateral. At the end of the 1980s, such an approach seemed robust enough to curb
irresponsible risk taking by major banks and prevent traumatic experiences such as
the Latin American sovereign debt crisis.
By the mid-1990s, with the explosive growth of the global securities markets, it was
already clear that financial institutions‟ minimum capital requirements (MCR) had to
Gonzalez, Sotelino & Savoia
also account for not only credit risk, but also market risk. The financial crises in East
Asia and Russia revealed that financial institutions had to demonstrate much
stronger internal risk management processes and be forced into greater
transparency and disclosure. Once again, G-10 representatives came together to
discuss a better set of rules and Basel II was born.
Under Basel II, minimum capital requirements (Pillar I) had to not only be sufficient to
protect the bank against potential credit losses but also market and operational
losses; and detailed guidelines for internal risk management (Pillar II) and market
discipline (Pillar III) procedures were put in place.
Concurrent with this, however, was the consensus that it is not fair to treat large
sophisticated institutions - that make huge investments in technology and risk
management - and small less sophisticated banks in the same way. It was implicitly
being decided that the higher investments and superior capacity of certain banks
(now known as Systemically Important Financial Institutions or SIFIs) to “foresee the
unexpected” commanded a premium: the reward of potentially higher leverage.
Since the large and more sophisticated institutions were about to be allowed to make
use of their own Internal Rating Based (IRB) models to estimate credit risk, what to
do about the small and medium-sized banks lacking this same sophisticated risk
measurement and management apparatus? The Basel II Standard Approach was,
then, devised, mainly to meet the needs of the less sophisticated banks. As a
consequence, the 100% risk-weighted asset (RWA) category for corporate
exposures in Basel I gave place to a range of RWAs ranging from 20% to 150%
depending on the CRA rating of the borrower or the nature and amount of the
collateral (Table 1).
Table 1: RWA’s under the Basel II Standard Approach for Credit Exposures
Credit Exposure
Credit Rating given by an ECAI
BBB+ /
BB+ /
B- /and
A+ / ABBBBlower
BCBS, 2001, The new Basel Capital Accord: Consultative Paper, p. 10 (Adapted).
Option 1
Option 2
Short term
Surprisingly, the G-20‟s regulatory response to the credit crisis of 2007-2009 seems
to point out to a turnaround by regulators on the inherent trust on the self-imposed
risk discipline of financial institutions reflected in the quantitative and qualitative
guidelines of Basel II.
Credit Ratings began to affect minimum capital requirements for banks in 1994 with
the formal adoption of the Basel II Standard Approach (SA). External ratings, as
shown in Table 1, become as good as collateral (credit mitigators) allowing the use
of lower RWAs and reducing capital charges necessary to cushion Unexpected
Gonzalez, Sotelino & Savoia
Losses (UL). A credit ranked Investment Grade or higher, for instance, allows for a
major reduction in the capital required when compared to the NA column. On the
other hand, Table 1 clearly demonstrates that booking a riskier loan or a bond
demands higher capital charges, illustrating how beneficial an inflated loan is not
only to the borrower, but also to the bank itself. The same logic applies to the
securitization products (Table 2).
Under Basel I, an ordinary US$ 1 million mortgage (covered) claim (loan or
securitization) would require a capital charge of US$ 39,200 (assuming required
capital ratio of 8%, RWA of 50%, PD of 2%) viii. Under the Basel II Standard
Approach, this same mortgage would demand, if unrated, a bit less, US$ 27,440
(RWA of 35%); if transformed into an AAA asset-backed security (ABS) of
collateralized debt obligation (CDO) in the banking book ix, US$ 15,680 (RWA = 20%
- See Table 2); and if not in the banking but in the trading book, possibly US$ 2,500
(Specific Risk for a Qualifying asset is .25%) x. Under the proposed Basel II.5 xi
guidelines, the capital charge for this last type of exposure would be elevated to US$
16,000 (1.6% - See Table 3).
Table 2: RWAs under the Basel II Standard Approach for Securitization
Products (Banking Book)
Risk Weight
External Credit Assessment
A+ /
BBB+ /
BB+ /
B+ and
capital A Revised
BCBS, 2006, International Convergence of Capital Measurement and Capitalfrom
Framework (Comprehensive Version). Basel, June 2006, p. 127
Table 3: Capital Charges under the Standardized Approach for Securitization
Products in the trading book (Basel II.5)
Specific Risk Capital Charges under Standardized Approach
External Credit
AAA to
A+ to AA-2/P-2
BBB+ to
BB+ to
Below BB- and
below A-3/P-3
or unrated
BCBS, 2011, Revisions to the Basel II market risk framework. Updated as of 31 December 2010,
Basel, p. 6.
If a bank, as an investor not an originator, followed the Basel II IRB Approach, the
Rating Based Approach (RBA) would probably be usedxii and the amount charged
could be as low as US$ 5,448 (RWA = 7%) in the banking book (Table 4). In the
trading book, the capital charge would be internally calibrated by Value-at-Risk
(VaR) methodologies.
Gonzalez, Sotelino & Savoia
Table 4: Capital charges under the Basel II IRB Approach (RBA)
Long Term Weights:
Risk weights for
thick tranches
Base Risk Weights
backed by highly
granular pools
Risk Weights for
tranches backed by
non-granular pools
Below Ba3 or
ShortTerm Weights:
All other
ratings or
BCBS, 2005, International Convergence of Capital Measurement and Capital Standards: a Revised
Framework, November 25, p. 131-132.
Table 5 presents a comparison between RWAs and capital charges between AAA
ABS senior tranches and AAA-bonds throughout the different accords and
Approaches for different types of US$ 1 million nominal loans or securities. As can
be observed, it is practically impossible for an external party to validate the capital
charges for a bank following the IRB Approach.
Basel II.5 eliminated regulatory arbitrage between the trading and the banking book
for securitization products. In the example, if a bank follows the IRB, the RWA of 7%
(US$ 5,448) is compensated in the trading book by the flat charge of .56% (or US$
5,600)xiii. Under the standardized approach, banks are also required to hold an
amount similar to the RWA of 20% of the banking book, i.e. a flat charge of 1.6%
(US$ 16,000, Table 3).
Basel II.5 does not impose additional capital charges in relation to Basel II for the
banking book, but it imposes a significant increase in regulatory capital for the
trading book.
Gonzalez, Sotelino & Savoia
Table 5: Minimum Capital Charges (C) in different classes of US$ 1 MM claim
Basel I
AAA Corporate exposure
Basel II Standard
Basel II - IRB
Basel II.5 – IRB
or Standard
C (%)/
C (%)/
8% /
1.6% /
I.C* /
security in
banking book
8% /
1.6% /
security in
trading book
8% /
Senior tranches of AAA Residential Mortgage Backed Security (RMBS) or Senior AAA
ABS CDO tranches
covered loan
(1 family)
4% /
2.8% /
I.C* /
ABS banking
book (pool)
4% /
1.6% /
5,488 /
ABS trading
book (pool)
4% /
/ 1.6%
5,600 /
* I.C stands for Internally Calibrated.
** VaR is the I.C 10 days´ value-at-risk (VaR) with an ɑ=0.01%.In Basel II.5, the
estimation of VaR becomes more robust. A multiplication factor of 3x and a stressed VaR
component (sVaR) are included in the IRB estimations.
Again, under the original 1988 Basel I Accord all corporate exposures were treated
the same and there was no regulatory arbitrage opportunity between loans and
securities as there was no trading book. Basel II, not only permitted notoriously lower
capital charges on the basis of external ratings, but also created a gap between
sophisticated (IRB) and non-sophisticated banks allowing the larger banks, most
likely the SIFIS, to hold much less capital for the same risk exposures than the
smaller financial institutions.
Basel II.5, notwithstanding all its increased rigor vis-à-vis Basel II (specifically its
mitigation of regulatory arbitrage between the trading and the banking books for
securitization products), still relies heavily on external ratings as a main driver for
regulatory capital (see Table 3). Basel III looms in the horizon imposing significantly
more restrictive definitions of capital, higher minimum capital adequacy ratios,
additional capital buffers for SIFIs, a counter-cyclical capital buffer and higher risk
weighted assets (RWA) for activities such as securities trading, securitizations,
securities lending and derivatives trading. It also imposes new quantitative minimum
requirements for liquidity and stability of funding, but is still mute to the over-reliance
in external ratings.
The Brazilian approach presented in the next session addresses precisely this issue.
However, it must be clearly stated that such approach necessarily sets the course
back to Basel I.
Gonzalez, Sotelino & Savoia
3.1 Was Basel I Better?
Basel I‟s simplicity and conservatism in the establishment of minimum capital
requirements gave place to Basel II‟s risk sensitiveness and comprehensiveness in
terms of minimum capital requirements (cushion for credit, market and operational
losses), but also greater complexity and flexibility in terms of capital charges. That
was done in exchange for expected high quality internal risk management,
supervision processes (Pillar II) and market discipline practices (Pillar III).
However, Pillars II and III are indeed weak because “what the market cannot
observe, it is unlikely to be able to exercise discipline over. And what the regulator
cannot verify, it is unlikely to be able to exercise supervision over.” The IRB
Approach is not exempt from this dilemma and it made “increasingly difficult for
regulators and market participants to vouch for the accuracy of reported capital
ratios. They are no longer easily verifiable or transparent (Haldane, 2011)”.
More importantly, by making minimum capital requirements (Pillar I) a function of
CRAs´ opinions, Basel II ended up creating, as discussed above, perverse
incentives for financial institutions and the CRAs.xiv
Ironically, in attempting to close the door to risk-insensitiveness also seeing as a
regulatory arbitrage opportunity in Basel I, Basel II created another one: the ratingcentered regulatory arbitrage with the compounding perverse effect of pro-cyclicality.
Under the Basel II Accord, models are meant to be used for the determination of riskweighted bank capital. Superficially, this represents a considerable improvement over the
situation in the Basel I Accord where the riskiness of underlying assets has only a cursory
connection with the level of capital, prompting undesirable behavior on behalf of banks and
undermining the guarantees provided by bank capital. Under Basel II, capital is to be directly
connected to the riskiness of the underlying assets. This however has a number of
undesirable consequences. Besides doubts about model quality, (…) it also tends to be procyclical especially in times of financial uncertainty, given rise to the notion of endogenous risk
(Beninkn et. al., 2009).
Figure 1 shows how Basel II introduced more pro-cyclicality under, both, the
Standardized and the IRB Approach by demonstrating how minimum capital
requirements are affected by ratings deterioration – always present in downturns
(Goodhart, 2005; Beninkn et al., 2009). In other words, Basel II reinforced
endogeneity. Under Basel I, the flat 100% RWA induced a stable 8% curve, forcing
banks to accumulate reserves during the good times to cushion losses when
slowdowns materialize (Figure 1). It is no wonder that Basel III now puts forward the
need for a counter-cyclical capital bufferxv.
Gonzalez, Sotelino & Savoia
Figure 1: Pro-cyclicality Evidence
3.2 The IRB
“The Basel II approach [to determine minimum capital requirements for credit risk
under the IRB] can be summarized in terms of a single „risk curve‟ which relates the
capital charge for any given loan to the risk attributes of that loan, such as its
probability of default (Beninkn et al., 2009)” (Figure 2).
Figure 2: Total Loss Function or IRB credit risk curve
As can be observed in Figure 2 above, the Total Loss (TL) curve, the density
function of a Vasicek distribution under the IRB, has three main components: the
expected loss (EL), the unexpected loss (UL), and the stress loss (large tail). The
stress loss is used to accommodate the statistical Type 1 error, usually .01%. The
Expected Loss (EL), representing the known costs of doing business per credit
exposure, is cushioned by loan losses‟ provisions plus pricing policies (defined exante as risk premia)xvi. EL is, by definition, Probability of Default (PD) x Severity of
the Loss (or Loss Given Default, LGD). On the other hand, the Unexpected Loss
(UL) is representing the unknown costs of doing business per credit exposure and
Gonzalez, Sotelino & Savoia
must be cushioned by Capital. It is against UL that a certain amount of minimum
capital should be held.
In Figure 2, it is possible to observe that TL is represented as a curve with a right fat
tail. EL and UL are arbitrary points set by risk management models, but implicitly
based on the same data and assumptions. In other words, UL is only cushioning the
same risks measured in the very same way. In the case of credit risk, UL represents
the possible losses of a 3.29 standard deviations higher than EL. What if other latent
variables come into play in a deep recession, but were underestimated when EL was
Basel II does recommend that the historical data used to estimate PD be large
enough to cover an entire economic cycle (through-the-cycle perspective) and
therefore be less sensitive to recent cyclical information (BCBS, 2005; p. 101). It also
recommends that stress tests be used to estimate market risk (BCBS, 2005; p. 165),
counterparty risk (BCBS, 2005; p. 220) and to estimate capital adequacy. However,
methodology and scenario assumptions are entirely left to the discretion of banks.
Note that the same through-the-cycle perspective is supposedly used by CRAs‟
models and are not significantly different from what was just explained (e.g. Altman
et al., 2010)xvii.
The Standard Approach was mainly implemented in Europe combined with the IRB.
In the U.S., banking regulators were keen to implement internal advanced models at
their own pace. However, overreliance in external credit ratings played an important
role in the crisis, because other financial institutions especially pension funds and
money market funds were very rating dependent.
4. Addressing the problem: Simplified Standard Approach and the
Brazilian case
What was wrong with Basel I in the first place? Basel I did not allow banks to adjust
regulatory capital. In other words, it was too risk-insensitive in two ways: first,
regulatory capital was insensitive to borrower‟s default risk; second it was also blind
to the existence of collateral possibly encouraging banks to engage in riskier higher
yielding non-covered exposures.
Under Basel I, internal risk models could generate Probability of Default (PD) and
Loss Given Default (LGD) estimations, but these could be used only to estimate
provisions or risk-premium (ex-ante), i.e. an assessment of Expected Losses (EL)
that would affect the bank‟s capital only through the Income Statement. It was stated
that Basel II did not provide incentives to the development of more elaborated risk
management tools (which is not entirely true, because the models could estimate EL
or provisions).
Basel I was also a one-size-fits-all model: sophisticated and non-sophisticated banks
were equally treated, no premium for greater credit intelligence allowed. Additionally,
calculating the regulatory capital ratio “involved little more than half-a-dozen
calculations (…) They were transparent and verifiable. In that way, regulatory rules
(Pillar I) provided a solid platform for supervisory discretion by regulators (Pillar II)
Gonzalez, Sotelino & Savoia
and market discipline by investors (Pillar III). The Basel pillars were mutually
reinforcing (Haldane, 2011)”.
Conceptually, minimum capital requirements (MCR) are supposed to be the cushion
against Unexpected Losses (UL). By not recognizing that the existence of collateral
could affect UL and that ratings (external or internal) could help estimate UL, Basel I
fell short from leading to refined estimation of minimum capital requirements. As
stated in this paper, the first one is indeed a problem in Basel I, but the later is not
necessarily true.
How is it possible to eliminate the direct linkages between credit ratings and
minimum capital requirements, avoiding rating-centered regulatory arbitrage?
The Simplified Standard Approach addresses that problem calling for a central
concept of Basel I: a flat 100% RWA for non-covered credit risk. Collateral, already
permitted under the Standardized Approach, would be the necessary and the only
credit mitigator for a lower than 100% RWA. The Minimum capital requirements for a
bank‟s loan portfolio would be a function of the range of collaterals for the specific
exposures (Table 6).
In other words, the SSA is an intermediate approach between Basel I and the
Standardized Approach of Basel II, because it is sensitive to collateral but not to
Under the Simplified Standard Approach (SSA), a R$ 1 MM non-covered corporate
claim would require a capital charge of R$ 107,800 (capital ratio of 11%, RWA of
100%, PD of 2%). If the same claim has a guarantee (on the total claim) from either
a financial institution or a credit derivative, the capital charge will be R$ 53,900
(RWA of 50% - comparing the left to the right side of Table 6). If the claim has a
guarantee from a high level multilateral organization as the World Bank, there will be
no capital charge (RWA of 0%).
The SSA is the law of the land in Brazil since June 2008. xviii Credit ratings, external
or internal, cannot affect RWAs and only collateral can justify a RWA below 100%.
This seems to have been a wise choice, given the resilience shown by most of that
country‟s financial system during the worst moments of the recent global financial
Gonzalez, Sotelino & Savoia
Table 6: Simplified Standard Approach in Brazil
Non-covered Exposures
Credit Risk Mitigators (Collateral
Risk Mitigator (Collateral)
Type of Exposure
Cash items in local currency
Guarantee from the Treasury
or Central Bank of Brazil
Gold Investments
Guarantee from Multilateral
Development Organization
Cash Collateral
Treasury Bill, Bonds and
securities issued by Central Bank
of Brazil
Checking Accounts in local
Claims on banks
Claims on Central Banks and
Claims secured by property (in
Guarantee from Brazilian
Financial Institutions
Guarantee from Financial
Institutions from countries that
have not defaulted in the last
5 years
Certificates of Deposit (CDs)
or other Financial Institution
Credit Derivatives
Residential or Commercial property
- if is worth more than 200% of
Claims on Retail Portfolio
the claim
- if is worth more than 125%
Other loans
and less than 200% of the
- if is worth less than 125% of
Other claims
the claim
Based on BCB (Banco Central do Brasil - 2007). Carta Circular 3,360 / 2007.
The larger Brazilian banks are now in the process of validating their internal models
to obtain the necessary Central Bank blessing to migrate towards their IRB
approaches. Small and medium-sized banks will continue to use the SSA. The RBA
approach for securitization will also be different from the RBA presented in the Basel
II document International Convergence of Capital Measurement and Capital
Standards: A Revised Framework (Comprehensive Version) as it will not allow the
use of external credit ratings.xix
As previously stated, minimum capital requirements should be a cushion against UL.
Therefore, they should be derived from modeling (internal or external) that takes into
consideration stressed scenarios. Conceptually, if there is no reliable forecast of the
“unknown“, collateral should be the only risk mitigator. This would not only better
protect the financial system against the unexpected, but stimulate banks to price
more carefully underlying assets.
This approach should also push financial institutions to pursue contractual
arrangements and lending practices that allow for swift enforcement of collateral
obligations, and fast credit recovery. And last but not least, prudential supervision
could be made more concrete (as collateral valuation tends to be more tangible than
Gonzalez, Sotelino & Savoia
medium term projections of ability to repay) and on-site supervision more effective,
focused on the verification of existence and adequacy of collateral.
Basel III rules are expected to demand an additional cushion of capital from the
systemically important financial institutions (SIFIs), supposedly the ones adopting an
IRB approach. This would tend to better level the playing field, as it would mitigate
the potential leverage advantage of IRB adopters versus non-adopters. Still, it may
not resolve the fundamental issue of adequate cushioning against UL if the bigger
banks do not effectively present conservative and anti-cyclical models. Note that the
approach defended in this paper is at same time conservative and anti-cyclical. It is
only drawback is to be incredibly simple!
5. Conclusions
This paper defends the Simplified Standard Approach used, for example in Brazil, as
an alternative and more resilient Pillar 1 model than the current Basel II Standard
Approach, because it minimizes conflicts-of-interest in the CRAs´ business,
reinforces and incentivizes collateral pricing and is naturally anti-cyclical while
establishing a more conservative steady capital cushion.
It begins by examining the Dodd-Frank Act and the European Commission proposals
regarding CRAs‟ regulation and supervision. The analysis concludes that while
greater reliability on CRAs‟ opinions can be achieved through improved corporate
governance and greater transparency, two problems remain. The first is the difficulty
to satisfactorily address the conflict-of-interest posed by CRAs‟ dominant “issuer
pays” or “investor pays” business models, because inflated ratings are beneficial to
both parties. The second is the fact that, as long as credit ratings (whether internal or
external) can be utilized to determine minimum capital requirements, the incentive
towards ratings-centered regulatory arbitrage by financial institutions will remain in
As the investigation shifts to the evolution of the Basel accords, it becomes clear that
Basel II has created excessive reliance on ratings for the purpose of regulatory
capital, a problem Basel III continues to fail to address. In addition, Basel II
embodies an incentive to excessive leverage, especially for the most sophisticated
banks, the same SIFIs Basel III now wants to impose additional capital buffers to. In
short, under Basel I banks were more capitalized, less subject to pro-cyclicality, less
exposed to the dangers of rating-centered regulatory arbitrage and far easier to
Closer examination on how unexpected losses (UL) and capital charges are
calculated under the IRB Approach suggests that, while prevailing mathematical
models can be sound for determining expected losses (cushioned by banks through
provisions for loan losses and/or risk premia on the loans), they may not be reliable
enough as sole source of determination of the capital cushion necessary to
withstand credit deterioration under disruptive market conditions.
As a consequence, this paper proposal is towards a more conservative approach
while estimating the “unknown” or Unexpected Losses. A partial return to Basel I is
put forward: a flat 100% RWA for non-covered credit exposures. The optimization of
regulatory capital would only be available in the presence of collateral, a positive
Gonzalez, Sotelino & Savoia
aspect of the Basel II Standard Approach. This framework, the Simplified Standard
Approach (SSA), already exists but is being used mainly in Latin American Countries
or in smaller financial institutions worldwide.
Additionally, this approach should also have the benefits of pushing financial
institutions towards more careful pricing of underlying assets and towards reinforcing
their pursuit of contractual arrangements that allow for swift enforcement of collateral
obligations and fast credit recovery.
In this line of activity not only margins tend to be higher than in the straight corporate and sovereign
bonds‟ segment, but sponsors have the potential for a eries of issues of a similar nature over a short
period of time and underwriters (such as investment banks) have the opportunity to bundle similar
type of transactions from different issuers to entice the cooperation of the rating agency (Altman et al.,
The exacerbated reliance in the CRAs‟ opinions is extensively studied and most multilateral
organisms and Central Banks pointed out to the need of addressing this issue. See, for instance, FSB
(2010), BCBS (2009), COREMEC (2010) and Thakor (2010).
The Dodd-Frank Act, Title IX, Subtitle C (The Dodd-Frank Act, 2010, p. 497-515).
See Thakor (2010) on possible negative impacts of FD removal or increasing legal liability such as
reducing the information content of the reports, less responsiveness from the CRAs to new
information and downward-biased opinions.
Local authorities in the E.U. state members used to be in charge of supervising the CRAs in
coordination with the Committee of European Securities Regulators – CESR.
See Altman et al. (2010), European Commission (2010) and Utzig (2010)
See also Sy (2009) and IMF (2009, p.89 -93).
Under Basel I, securitizations did not receive any special treatment, i.e. the same RWA of 50% was
available for any specific mortgage exposure and for the securitization tranches.
A AAA ABS CDO is not exactly comparable to a mortgage loan. There are diversification benefits in
the tranching process that can make this type of security less risky than a highly rated mortgage.
The Minimum Capital Charge of 0.25% assumes a residual term of less than six months to final
maturity and no charges in the Maturity Method because coupons are no longer available (BCBS,
2006, p.167-171). Counterparty Credit Risk (CCR) is also assumed zero, as in the case of a Credit
Default Swap (CDS) on the whole exposition was also available (BCBS, 2006, p. 258).
The term Basel II.5 is being used for the Basel III changes that were already implemented in
December 2010, mainly in the trading book. See BCBS (2010).
The Ratings-Based Approach (RBA) must be applied to securitization exposures that are rated. If
not rated, the Internal Assessment Approach (IAA) should be used if both the bank and the
securitization product meet certain requirements. If IAA could not be used, the Supervisory Formula
(SF) must be applied (BCBS, 2005; p. 130).
The small difference between these two values is because of a Probability of Default of 2% used in
this example. One should notice that the amount of the difference is a provision that should be found
on the Income Statement. Basel II.5 made regulatory arbitrage between the trading and banking book
indeed impossible for the securitization tranches.
Note from the examples that in both the Standardized Approach and the IRB (RBA) there is
reliance on CRAs‟ opinions. The RBA is the most common approach for all banks following the IRB,
according to BCBS (2005). As a consequence, all banks (buy side) are captured at least when it
comes to the securitization products.
See the Basel Committee on Banking Supervision Consultative Document Countercyclical Capital
Buffer Proposal from July 2010 for more information.
See Goodhart (2005) for more details.
CRA ratings “are not properly considering the volatility of the underlying assets and possible
disruptions of the cycle (Altman et al., 2010). Despite being supposedly resilient, stable and
determined in a “through the cycle” perspective, empirical studies previously mentioned show different
evidences about external ratings of MBSs (e.g. Hull and White, 2009)
The decision of not accepting a CRA dependent model in the Brazilian banking system was clearly
expressed in the Resolution of the National Monetary Council 2.099/2004 (Brazil, 2004) and in other
financial institutions (COREMEC, 2010).
Gonzalez, Sotelino & Savoia
More information on: “Edital de Audiência Pública 37”. Available (only in Portuguese) at:
Altman, EI, Oncu, TS, Richardson, AS & White, L, 2010, “Regulation of Rating
Agencies”, in: Acharya, V, Cooley, TF, Richardson, MP & Walter, I (eds),
Regulating Wall Street, John Wiley & Sons Inc, Hoboken, NJ, pp 443-467.
Basel Committee on Banking Supervision – BCBS, 2001, The new Basel Capital
Accord Second Consultative Paper, January, Basel.
Basel Committee on Banking Supervision – BCBS, 2005, International Convergence
of Capital Measurement and Capital Standards: a Revised Framework,
November, Basel.
Basel Committee on Banking Supervision – BCBS, 2006 International Convergence
of Capital Measurement and Capital Standards: A Revised Framework
(Comprehensive Version), June, Basel
Basel Committee on Banking Supervision – BCBS & The Joint Forum, 2009,
Stocktaking in the use of credit ratings, June, Basel.
Basel Committee on Banking Supervision – BCBS, 2010, Guidance for national
authorities operating the countercyclical capital buffer, Basel.
Basel Committee on Banking Supervision – BCBS, 2011, Revisions to the Basel II
market risk framework, Updated as of 31 December 2010, Basel.
Beninkn, H; Daníelsson, J & Jónson, A, 2008, „On the Role of Regulatory Banking
Capital‟, Financial markets, institutions & instruments, vol. 17, no. 1, pp. 85-96
Calomiris, C, 2009a, „A recipe for ratings reform‟, Economist’s Voice
Calomiris, C, 2009b, „Prudential Bank Regulation: What‟s Broke and How To Fix It?‟
Graduate School of Business Columbia University Working Paper, New York
Covitz, D & Harrison, P, 2003, „Testing Conflicts of Interest at bond rating agencies
with market anticipation: Evidence that reputation incentives dominate.‟ Federal
Reserve Board Working paper, Washington, DC
European Central Bank – ECB, 2011, European Commission’s public consultation
on Credit Rating Agencies – Eurosystem reply, Special Report, Feb, pp.2
European Commission, 2010, Public Consultation on Credit Rating agencies, 5 Nov,
European Commission, 2011, Proposal for a Regulation of the European Parliament
and of the council: amending Regulation (EC) No 1060/2009 on credit rating
agencies, 15 Nov,
European Parliament, 2009, Regulation (EC) No. 1060/2009 of the European
Parliament and of the Council on Credit Rating Agencies, 16 Sep 2009
European Parliament, 2011, Regulation (EC) No. 513/2011 of the European
Parliament and of the Council on Credit Rating Agencies, 01 June 2011
Financial Stability Board – FSB, 2010, Principles for Reducing reliance in CRA´s
ratings. Report, Oct 10. <>
International Monetary Fund – IMF, 2009, Global Financial Stability Report,
Washington, DC: IMF, October. <>
Haldane, 2011, „Capital Discipline‟, Bank of England Paper based on a Speech given
at the American Economic Association, 09 Jan 2011, Denver, USA,
Gonzalez, Sotelino & Savoia
Smith, RC & Walter, I, 2002, „Rating Agencies: Is There an Agency Issue?‟ In:
Levich, RM, Majnoni, G & Reinhart, CM. (eds.), Ratings, Rating Agencies and
the Global Financial System, Boston, Kluwer, p. 289-318.
Sy, A, 2009, The Systemic Regulation of Credit Rating Agencies and Rated Markets,
IMF Working Paper, <>
Thakor, A, 2010, The Economic consequences of proposed regulation on NRSRO
regulation. Working Paper.
House of Representatives – HR, 2010, The Dodd-Frank Wall Street Reform and
Consumer Protection Act of 2010, HR 4173, 2nd Cong
White, LJ, 2010, „Markets: The Credit Rating Agencies‟, Journal of Economic
Perspectives, vol. 24, pp 211-226,