U.S. Securities and Exchange Commission
Dated: July 27, 2003
II - Discussion
A - Alternatives to the NRSRO Designation
B - Recognition Criteria
Existing substantive Criteria
C - Examination and Oversight of NRSROs
D - Conflicts of Interest
E - Alleged Anti-competitive, Abusive, and Unfair Practices
F- Information Flow
G - Other
III - Solicitation of Additional Comments
Limits & Concerns
Rating history before default, reliability and comparison
Concerns about NRSRO Recognition Committee
A-Alternatives to the NRSRO Designation
As the Commission may conserve the NRSRO status2 and because of a lack of knowledge relating to the "broker-dealers system", questions of Part A concerning the implementation of internal credit rating systems are not discussed. Nevertheless, the Commission must note that credit spreads do not seem to be an appropriate alternative. Indeed, beside the technical problems mentioned by the Commission (question 4), credit spreads are volatile on a single (issuer) and on a global3 (overall BBB credit spreads for example) basis. Moreover, according to academic research4, credit spreads do not necessarily reflect the effective default risk.
The requirement of minimum standards concerning methodologies and criteria used by rating agencies to estimate credit risk is not necessarily opportune. The review of the reliability and credibility of their ratings, conditioning NRSRO recognition and its keeping, appears to be sufficient. Indeed, the Commission should investigate the following points:
Nevertheless, it seems appropriate to require minimum information from rating agencies about their rating process and their performance. Moreover, few aspects and, especially, anti-competitive practices, unfair ones and potential conflicts of interest need exhaustive studies and full attention in order to prevent any distrust against the global rating system. Lastly, the definition of objective criteria, if it would not necessarily lead to more competition, would have the merit to determine the "rules of the game".
Existing substantive Criteria
Are the criteria currently used by Commission staff to determine whether a credit rating agency qualifies as an NRSRO appropriate?
The current criteria do not seem appropriate because of a lack of precision, objectivity and transparency of the decision's process. Indeed, how the Commission measures the national recognition of rating agencies' credibility and reliability? What are the appropriate determinants of an organizational structure and financial resources? In which measure the size and experience are ensuring efficiency? Which procedures could ensure rating's credibility?
If not, what are the appropriate criteria?
The appropriate criteria could be defined as follow:
- Criterion 1: Effective rating agency's reliability and credibility (see question 12).
- Criterion 2: Exhaustive rating's definition and key rating's process required.
- Criterion 3: Communication Requirements.
- Criterion 4: Organizational Structure.
- Criterion 5: Code of Ethics.
The market participants with Commission oversight could establish minimum standards. Proceedings could consider, among other things, the following points (see question 39):
- Criterion 6: An annual certification filing in which the rating agency confirms that criteria 2, 3, 4 and 5 are maintained on a continuing basis.
Furthermore, the rating agency should disclose, among other things:
Criterion 1 is a measure aiming at ensuring the global reliability of ratings. Criterion 2 requires certain elements that would ensure rating's reliability and prevent the development of "rating shopping" practices. Criterion 3 considers the role of rating agencies (reduction of asymmetry, dissemination of the rating and of the methodology). Criteria 4 and 5 are proposed in order to minimize potential conflicts of interest and to try to guarantee that the global rating system keeps its credibility. Criterion 6 defines a methodology to supervise rating business and to control potential conflicts of interest, anti-competitive practices...
How should a determination be made as to whether a credit rating agency has met each criterion?
Criterion 1 is based on the global methodology presented in question 12. Moreover, the Commission could review this criterion in order to ensure that rating agencies meet this original qualification criterion on a continuing basis. Criteria 2 and 3 could be disclosed by rating agencies. Criteria 4 and 5 could be communicated to the Commission. The criterion 6 would permit to review those criteria (2, 3, 4, 5). Furthermore, a ruling procedure could be implemented. (See question 25 for a presentation of the NRSRO Recognition Committee)
Is it appropriate to condition NRSRO recognition on a rating agency being widely accepted as an issuer of credible and reliable ratings by the predominant users of securities ratings in the United States (e.g., underwriters, dealers, banks, insurance companies, mutual funds, issuers)?
NRSRO recognition could be conditioned on a rating agency being widely accepted as an issuer of credible and reliable ratings by the predominant investors in the United States and by the regulatory authority. If an issuer seeks rating for regulatory purpose, it has no reason to choose non-NRSRO agency. Its first incentive is to choose a rating agency recognized by regulatory authorities.
Would this general acceptance be verifiable through the examples set forth above (e.g., requiring verification through attestations from, and interviews with, authorized officers of users of securities ratings, as well as using statistical data to demonstrate market reliance on an applicant's ratings)?
A global approach, divided in five points, could be defined and used in order to verify that the agency is an issuer of credible and reliable ratings:
As a more objective way of evidencing market reliance and credibility, should NRSRO recognition be conditioned on a credit rating agency documenting that it has been retained to rate securities issued by a broad group of well-capitalized firms?
Conditioning the recognition on a credit rating agency on this ground is not necessarily an objective way of evidencing market reliance and credibility. The use of NRSRO status in U.S. financial regulation is the first incentive to choose an agency. The fact that regulatory authorities define the credible rating agency prompts necessarily issuers to choose those qualified agencies. There is no reason for an issuer to choose an agency that is not qualified. Furthermore, well-capitalized firms can require multiple ratings because they can easily support their costs. Cantor and Packer (1995a)9 looked for explanatory factors of a third rating. More precisely, they asked why issuers request a third rating from Fitch or Duff & Phelps. It appears that the differences between Moody's and Standard & Poor's ratings are not an explanatory variable. In other words their results refute "convincingly the hypothesis that ratings uncertainty due to disagreements between Moody's and Standard & Poor's ratings increase the likelihood of obtaining a third rating to resolve the difference." In fact, they found that "the most important determinants to get an additional rating are a firm's size and age."
In 1997, Cantor and Packer10 tested the U.S. regulatory variable as an explanation of a third rating request. In order to explicit their hypothesis, it is possible to refer to the "Standardised Approach" proposed by the New Basel Accord11 in which the risk weights for corporate claims are defined as follow:
Table 1 - Risk Weight for corporate claims under the Standardised Approach
In case of multiple assessments, a methodology is specified:
- Rule 1: "if there is only one assessment by an ECAI chosen by a bank for a particular claim, that assessment should be used to determine the risk weight of the claim";
- Rule 2: "if there are two assessments by ECAIs chosen by a bank which map into different risk weights, the higher risk will be applied";
- Rule 3: "if there are three or more assessments with different risk weights, the assessments corresponding to the two lowest risk weights should be referred to and the higher of those two risk weights will be applied".
An issuer having one rating has no incentive to request an other one because of Rule 2. Consider now the following hypothesis (Table 2):
Case 1: An issuer having two similar ratings has no incentive to require a third one because of Rule 2.
Case 2: If an issuer has two ratings, e.g. A- and BBB+, it should be motivated to require a third one.
Case 2A: The obtaining of an other A- (or above) will lead to a risk weight of 50% because of Rule 3.
Case 2B: The obtaining of a third rating below A- will not change its risk weight because of Rule 3.
Table 2 - Regulation: explanatory variable of multiple rating?
This underlines that regulations could lead to multiple ratings (U.S. regulation could generate multiple ratings for different purposes, other than risk weight12). Nevertheless, despite the confirmation of their first conclusions (uncertainty and firm size), Cantor and Packer found that "there is no evidence that firms seek third rating specifically to help meet regulatory requirements."
My research on the explanatory variables of a third rating and, more precisely, of a Fitch rating in the U.S. market in 2001 confirms Cantor and Packer's conclusions about uncertainty (differences between Standard & Poor's and Moody's ratings) and firm's size. Moreover, new explanatory variables are identified. For example, it seems that regulations (variables used are not similar to those used by Cantor and Packer) are an explanatory variable of a third rating.
Lastly, the explanatory power of the regulation could, in fact, results from issuers' willingness to obtain the highest rating in order to reduce the cost of their debt. Nevertheless, according to my research, such strategy does not necessarily reach the expected effect.
Should the Commission condition NRSRO recognition on a rating agency developing and implementing procedures reasonably designed to ensure credible, reliable, and current ratings?
This does not seem to be a relevant criterion because it is a highly subjective one. The most important one is the effective reliability of ratings. If the Commission supervises periodically that rating agencies assign reliable ratings (see Recognition Process), it does not seem appropriate to require "procedures reasonably designed to ensure credible, reliable, and current ratings". Nonetheless, key elements of the agency's process should be required:
- Rating Committee;
- Surveillance tools: investors consider them as valuable tools (see question 15);
- "Meetings with issuer": agencies declare that issuers are "an important source of input"13. Moreover, if the Commission imposes the disclosure by issuers of particular information (rating triggers), "meetings with issuer" will be necessary.
At a minimum, should each NRSRO have rating procedures designed to ensure that a similar analysis is conducted for similarly situated issuers and that current information is used in the rating agency's analysis? What minimum standards should the Commission use to determine whether the agency's ratings are current?
The rating process and the surveillance tools used by current NRSROs seem to be an appropriate way to ensure this.
Should each NRSRO use uniform rating symbols, as a means of reducing the risk of marketplace confusion?
The use of uniform rating symbol by the NRSROs is not compulsory. The long term rating scales of the four NRSROs and agencies like Rating & Investment or Japan Credit Rating are quite similar. Nevertheless, the fact that a new agency is qualified with a significantly different rating scale (for example, a rating agency uses only 10 ratings) could create difficulties for investors to compare ratings given by different agencies. Nevertheless, the current rating scales do not appear to create confusion.
The main confusion is due to the fact that a rating A whether assigned by Fitch, Standard & Poor's, Moody's, Dominion Bond Rating Services or Rating & Investment.... seems to represent, broadly, the same credit risk. Some agencies refute this postulate. They have quoted academic studies during the proceedings about NRSRO status14, in order to demonstrate that identical ratings do not necessarily materialize an identical credit risk15 (Beattie and Searle, 199216; Cantor et Packer, 1995b17).
Based on observations selected in 2001 and 2003, I broadly obtain, despite some quite significant differences, similar results18. For example, the result of the comparison between Japanese agencies' ratings (Rating & Investment, R&I; Japan Credit Rating, JCR) and American ones19 [Standard & Poor's (S&P) and Moody's (M) ratings] are exposed below.
Table 3 - Distribution of Japanese and American agencies' ratings20
That is confirmed when we measure the differences in notch (a difference of one notch is a difference between A+ and A for example). The following graphic shows an example of those differences.
Graphic 1 - Differences in notch when R&I assigns higher ratings than Moody's
Is it possible to consider that Japanese agencies' rating is not a reliable information? This is not as simple as that and it is not argued that this study is conclusive. However it has the merit to present few methodological points that could be used by the Securities and Exchange Commission in order to estimate rating's reliability and credibility. For example, Packer and Reynolds (1997) and Packer (1999) affirmed that credit spreads are correlated to R&I ratings. Standard & Poor's (2002)23 indicated that similar criteria are broadly used to estimate Japanese and American corporate credit risk. In fact, it seems that Japanese agencies are taking into account the characteristics of the Japanese financial system24.
In order to appreciate more precisely the reliability of Japanese agencies' ratings, six defaulted companies25 have been selected. The relation between average rating history and default is calculated 1000 days prior the default26 (Graphic 2).
Graphic 2 - Average rating history before default
It appears that American agencies anticipate the default event before Japanese ones. This result is coherent with the observed distribution of agencies' ratings. Moody's and Standard & Poor's assign speculative grade ratings whereas Japanese agencies "grant" investment grade ratings. Then, between approximately - 780 and - 570 days, there is broadly a difference of two notches between Japanese and American agencies' ratings. From - 570 days, Japanese agencies' ratings converge, step-by-step, toward American agencies ratings. Even though this relation demonstrates that American agencies anticipate "correctly" the default event, the period - 570 days / - 370 days indicates that Japanese agencies forestall too. In other words, the Japanese rating downgrades during this period could be essential for investors.
Again, this result is not conclusive because of the lack of data, but it raises questions: are Japanese agencies' downgrades valuable information for investors? If yes, could it be sufficient to recognize them? (See Part III, Solicitation of Additional Comment - Limits & Concerns).
To conclude, one of the most important tasks for the Securities and Exchange Commission is to ensure that ratings given by different agencies identify similar credit risk. This brief study presents few methodological points that the Commission could use to assess rating's reliability and credibility.
When reviewing a rating agency's procedures for obtaining information on which to base a rating action, should the Commission establish minimum due diligence requirements for rating agencies? How could these minimum requirements be developed? By the Commission? By the industry, with Commission oversight?
Rating agencies should have the capacity to justify each rating action (upgrade, downgrade...). That is what they tend to do since 2002. "Minimum requirements" could be difficult to define. It seems appropriate that the rating industry with Commission oversight tries to establish them.
Should the extent of contacts with the management of issuers (including access to senior level management of issuers) be a criterion used to determine NRSRO status?
The "extent of contacts" could be a subjective criterion because "the extent" appears to be difficult to define. The majority of rating agencies has contacts with issuers. The latter are often disclosing non-public information, which are an important element of a rating's assessment. Meeting with issuer should be a criterion but it does not appear necessary to define "the extent". Question 15 refers to credit rating agencies that use solely computerized statistical model. The qualification of such agency does not seem to be a pertinent decision for two main reasons that will be developed in the appropriate question.
Should the Commission limit the credit ratings that can be used for regulatory purposes to credit ratings that include access to senior management of an issuer? If so, why?
It is not necessary to limit credit ratings that can be used for regulatory purposes to those that include access to senior management of an issuer. Even unsolicited ratings could be used. Indeed, despite the fact that agencies are building those ratings on public information, they have knowledge of competitors, of the industry and, eventually, of the country risk. Thus, they can be reliable but they should be regulated (see question 48).
To the extent a credit rating agency uses computerized statistical models, what factors should be used to review the models?
It is not necessary to review computerized statistical models because major rating agencies "build" their ratings on quantitative and qualitative studies. They have incentives to improve their models in order to provide the most reliable and credible ratings. One way to judge rating agency's capacity to assign reliable and credible ratings could be the assessment of the five points proposed in question 12.
Could a credit rating agency that solely uses a computerized statistical model and no other qualitative inputs qualify as an NRSRO?
An agency that solely uses a computerized statistical model should not receive NRSRO qualification. As mentioned by Standard & Poor's and Moody's during the public hearings of November 15, 2002, non-public information can be essential in the rating's determination. Furthermore, qualitative elements cannot be necessarily modelized.
Especially, statistical models based, for example, on market data would generate volatile ratings. In 2002, rating agencies have sounded investors27. It appears that they do not want volatile ratings and that they are considering that the surveillance process defined by rating agencies (outlook and credit watch) is a valuable tool.
Should the size and quality of the credit rating agency's staff be considered when determining NRSRO status? Should the Commission condition NRSRO recognition on a rating agency adopting minimum standards for the training and qualifications of its credit analysts? If so, what entity should be responsible for oversight of qualifications and training? How could the Commission verify whether a member of a rating agency's staff is or was previously subject to disciplinary action by a financial (or other) regulatory authority?
Should the Commission condition NRSRO recognition on an entity's meeting standards for a minimum number of rating analysts or a maximum average number of issues covered per analyst? For example, should the Commission question whether a single analyst can credibly and reliably issue and keep current credit ratings on securities issued by hundreds of different issuers? Or would this level of scrutiny involve the Commission too deeply in the business practices of rating agencies?
Should the Commission consider a credit rating agency's financial resources as a factor in determining NRSRO status? If so, how? Should NRSRO recognition be conditioned on a rating agency meeting minimum capital or revenue requirements?
Questions 16, 17 and 19 are proposing subjective criteria, except those relating to training and disciplinary action. Indeed, how could the Commission define an optimal size and quality of the credit rating agency's staff? How could the Commission propose an efficient maximum number of issues covered per analyst? Would the criterion of a maximum number of issuers covered per analysts be a better request? Can minimum capital or revenue ensure an effective independence (see question 42)?
It does not seem efficient to impose such requirements because of the difficulty to define the appropriate number and the subjectivity that it would imply. Furthermore, those criteria would not necessarily guarantee reliable ratings (see Introduction). The Commission should not use such criteria and should mainly focus on the effective ratings' reliability.
Is a credit rating agency's organizational structure an appropriate factor to consider when evaluating a request for NRSRO status?
It seems an appropriate factor, as organizational structure can minimize potential conflicts of interest and can guarantee the independence of rating agencies (see question 11).
Should the agency that seeks recognition consent to limiting its business to issuing credit ratings or could it conduct other activities, such as rating advisory services?
Rating advisory services and ancillary services need a clear definition and specific proceedings (see question 47).
Should a rating agency that confines its activity to a limited sector of the debt market be considered for NRSRO recognition? Should a rating agency that confines its activity to a limited (or largely non-U.S.) geographic area also be considered?
An agency that gives reliable and credible ratings and that confines its activity to a limited sector of the debt market or to a limited geographic area could be considered for NRSRO recognition.
Furthermore, the Commission was used to qualify agencies that proposed special expertise (IBCA, Bank Watch) and it is not sure whether global agencies proposing expertise in major debt sectors could be qualified or not. The qualification for limited sector or geographic area would establish competition and then, assure the development of the new recognized agencies that would, then, be able to develop their capacity to assign reliable and credible ratings in other sectors of the debt market.
Should the Commission consider a provisional NRSRO status for rating agencies that comply with NRSRO recognition criteria but lack national recognition?
The provisional NRSRO status should not be required. If the Commission decides to control each NRSRO (questions Part C suggest the implementation of such procedures), it is not necessary to define a provisional status. However, if the Commission does not institute a process aiming at controlling the NRSROs, provisional status will be necessary because competition would generate "unfair practices". "Rating shopping" practices could be exacerbated. In other words, issuers would choose the agency, which give the highest rating. In order to gain market shares, agencies could erode their standards, by assigning ratings that do not necessarily reflect the effective credit risk. This point cannot be ignored because it may be one of the main issues that the Commission will have to deal with. Moreover, certain agencies' practices seem to validate this argument (see question 45).
Should the Commission develop supplemental criteria to evaluate ratings quality that would be applicable to both rating agencies performing traditional fundamental credit analysis and those primarily reliant on statistical models?
See Introduction and questions 16, 17, 19.
Should the Commission consider other criteria in making the NRSRO determination, such as the existence of effective procedures reasonably designed to prevent conflicts of interest and alleged anticompetitive, abusive, and unfair practices, and improve information flow surrounding the ratings process?
Yes (see Part E)
Should recognition of NRSROs occur through Commission action (rather than through staff no-action letters)? Should the Commission establish an appeal process if the staff remains responsible for the recognition of NRSROs?
The Commission decision could be based on the Rating Committee model. Such a Committee would be created to grant NRSRO recognition and its keeping. The NRSRO Recognition Committee could work as follow:
1-Study of the ratings' reliability of the applicant: the five first criteria exposed in question 11 should serve to establish rating agencies' reliability. The applicant would communicate the information required and the Committee would analyse those elements and the effective agency's reliability. If the criteria are objective and clearly specified, an appeal process is not required. (See questions 26 and 27).
2-The NRSRO Recognition Committee would maintain each NRSRO under review in order to control their reliability (see questions 28 to 31):
- Yearly, NRSROs would communicate their "certification filing" (criterion 6);
- A ruling procedure would be implemented (see question 29);
- Every three years or more, NRSROs' reliability would be studied thanks to the global approach defined in question 11.
If an agency loses its recognition, an appeal process could be used.
Should the Commission publicize applications for NRSRO recognition, and seek public comment on the credibility and reliability of the applicant's ratings?
The Commission should publicize applications for NRSRO recognition in order to institute transparency in the process. Public comments could be valuable if, for example, academics are studying the applicant ratings' credibility and reliability.
Should the Commission establish a time period to serve as a goal for action on applications for NRSRO recognition? If so, would an appropriate time period be 90 days after all required information has been received, or a shorter or longer period?
A time period could reinforce the transparency of the process. 90 to 120 days seem to be an appropriate timing to study agencies' reliability, receive public comments (60 to 75 days) and appreciate those elements (30 to 45 days). The information received should be disclosed to the public. The decision of the NRSRO Recognition Committee should be motivated and disclosed.
C-Examination and Oversight of NRSRO's
Should NRSRO recognition be conditioned on an NRSRO's meeting the original qualification criteria on a continuing basis? Should some other standard of revocation apply?
It is essential to condition this recognition on an NRSRO's meeting the original qualification criteria on a continuing basis. Certain agencies are worried about the fact that the development of competition would exacerbate "rating shopping" practices. As this argument is relevant, the NRSROs should be controlled.
If so, should a failure to meet the original qualification criteria lead to revocation of NRSRO recognition?
This question depends on the criteria defined by the Commission, but it seems coherent that the failure to produce reliable and credible ratings should lead to the revocation of NRSRO recognition. For example, if it appears that an agency assigns "favourable ratings", and that this practice generates an underestimation of the credit risk, this agency should lose its status. The five points described in question 12 could be the main tools to review the reliability of credit ratings.
Should some other standard of revocation apply?
The violation of some other criteria could lead to penalties and a scale could be elaborated. Penalty could be mainly used in order to convict unfair and anti-competitive practices and "default of specific quality" case. For example, the Enron case is a "default of specific quality" case. The Staff28 has reproached rating agencies, among other things, for their lack of inquisitiveness and the insufficient review of company materials. Consequently, this case could have lead to penalise rating agencies.
Nevertheless, this proposition would have to overcome two main difficulties:
- the Commission and the rating industry would have to define an exhaustive and clear penalty scale;
- the Commission cannot ignore that foreign regulatory authorities could also create their own scale (see question 32).
What would be the appropriate frequency and intensity of any ongoing Commission review of an NRSRO's continuing compliance with the original qualification criteria?
The NRSRO Rating Committee review could be divided in two parts.
The first part, on an annual basis, could control the global rating industry. In other words, NRSROs should communicate to the Committee the new procedures or services used in order to verify that they do not constitute unfair practices, potential conflicts of interest... This process is justified by the fact that, instead of having measures concerning rating agency's methodologies imposed by the Commission, the latter could supervise the existing and new procedures in order to ensure the flexibility of the rating process and to prevent the practices mentioned above. A "ruling procedure" could be implemented in order to allow rating agencies to address concerns to the Committee about potential conflicts of interest or unfair practices.
The second part, every three years or more, consists in reviewing the reliability and credibility of the ratings. In order to examine this principal aspect, four of the five methodologies presented in question 12 should be studied:
- the relation rating history-default;
- the annual and cumulative default rates by rating and rating transition matrix;
- the differences between ratings;
- interviews with authorized officers of users of securities ratings and public comment could complete this process.
Should NRSRO recognition be conditioned on a rating agency's filing annual certifications with the Commission that it continues to comply with all of the NRSRO criteria?
It could be conditioned on rating agencies filing annual certifications with the NRSRO Recognition Committee in which they indicate that they use the same procedure and/or communicate new elements that affect their global rating process (see criterion 6 and question 29).
Should the Commission solicit public comment on the performance of each NRSRO, including whether the NRSRO's ratings continue to be viewed as credible and reliable? If so, how frequently should public comment be solicited (e.g., annually)?
The Commission solicitation of public comment on the performance of each NRSRO may be relevant. The second part of the NRSRO Recognition Committee review, proposed in question 29, could be submitted to public comment.
Should NRSROs be subject to greater regulatory oversight? If so, what form should this additional oversight take? If necessary, should the Commission seek additional jurisdictional authority from Congress?
The term "greater regulatory oversight" is subject to interpretation. If it means that the Regulatory Authorities should have an important control of rating agencies, "greater regulatory oversight" would not be necessary. Indeed, if NRSRO criteria are "correctly" specified and reviews are conducted - more particularly, the reliability of their ratings which is the main criterion to obtain and keep the NRSRO status -, it does not seem necessary to develop "greater regulatory oversight". It is important to bear in mind that their reputation is their main incentive to assign reliable and credible rating. From a theoretical point of view, regulations (e.g. NRSRO status) are not necessarily required to assure market regulation.
Nonetheless, the rating industry is nowadays a specific market because of the recognition of the role of ratings and their use by investors and regulatory authorities. Consequently, the Commission should mainly focus on effective agencies' reliability and on certain practices (unfair, anti-competitive, ancillary services).
If "greater regulatory oversight" means the abolition of agencies First Amendment protection, my knowledge of U.S law is insufficient to discuss the opportunity to engage the agency's responsibility. This question merits exhaustive studies to define clearly the consequences in the United States and in other countries. If the Commission imposes it and that foreign regulatory authorities, which may also want to implement greater regulatory oversight, base their own regulation on the SEC model, the political independence of the rating agencies could be considerably eroded. That may create new potential conflicts of interest between rating agencies and Regulatory Authorities. Moreover, the penalty scale proposed could also be used by foreign regulatory authorities and could spoil the rating agency's independence and, in fine, rating agency's business.
Consequently, the Commission could wonder among other things:
- If it uses a penalty scale, how could it deal with foreign regulatory authorities?
- If it wants to abolish agencies First Amendment protection, what may be the consequences for the rating industry in the United States and in the world?
- Should the recognition depend on a world-wide authority?
Should NRSRO recognition be conditioned on recordkeeping requirements specifically tailored to the ratings business? Should NRSRO recognition be conditioned on a rating agency's maintaining records relating to the ratings business, including those relating to rating decisions?
Because rating agencies must be able to justify their decision process, the NRSRO recognition could be conditioned on a rating agency keeping records relating to rating decisions. Furthermore, this obligation would be used to identify "default of specific quality" case.
D-Conflicts of Interest
Should the Commission condition NRSRO recognition on an NRSRO's agreeing to document its procedures that address potential conflicts of interest in its business including, but not limited to, potential issuer and subscriber influence? If so, what other potential conflicts should these procedures address?
Current NRSROs seem to have effectively managed the potential conflicts of interest generated by fees. Furthermore, according to my research, the current fees' structure, used by a majority of NRSRO, appears to be an appropriate tool in order to limit "rating shopping" practices.
Nevertheless, despite the fact that rating industry is built on reputation, it seems adequate to specify few standards that would help to guarantee the reliability and credibility of the rating industry. For example:
- the Commission could supervise the potential conflicts of interest generated by "single issuer" case (see questions 11 and 43);
- the Commission should deal with ancillary services (see question 47);
- the Commission could condition NRSRO recognition to an annual certification filing in which rating agencies disclose potential conflicts of interest and could implement a "ruling procedure" (see questions 11 and 29);
- if the Commission worries about the subscriber influence, it could prohibit private contact between investors and analysts (see question 41) or at least require that rating agencies disclose "predominant subscribers", as proposed for issuer's potential conflicts of interest (see question 43).
Lastly, the Commission should consider the political independence status of the rating agencies (see question 32).
To what extent could concerns regarding potential conflicts of interest be addressed through the disclosure of existing and potential conflicts of interest when an NRSRO publishes ratings?
This "extent" will mainly depend on the Commission's answers concerning ancillary services.
Should NRSRO recognition be conditioned on an NRSRO prohibiting employees involved in the ratings process (e.g., rating analysts and rating committee members) from participating in the solicitation of new business and from fee negotiations? Would conditioning NRSRO recognition on a rating agency's establishing strict firewalls between employees in these areas and credit analysts address potential conflicts? Should the Commission also address the credit analyst compensation structure to minimize potential conflicts of interest?
It seems that few NRSROs have implemented such procedures. Nevertheless, prohibiting employees involved in the rating process from participating in the solicitation of new business and from fee negotiations could be a criterion of NRSRO recognition. Moreover, it appears essential that rating agencies and the Securities and Exchange Commission define, together, a code of Ethics that specifies minimum and explicit standards. The acceptance of the rules of this code should be a criterion of NRSRO recognition (see question 11).
Should NRSRO recognition be conditioned on an agreement by a rating agency not to offer consulting or other advisory services to entities it rates? Could concerns regarding conflicts of interest be addressed by limiting or restricting consulting or advisory services offered by rating agencies?
See question 47.
Should NRSRO recognition be conditioned on a prohibition on credit rating analysts employed by NRSROs from discussing rating actions with subscribers? If not prohibited, should the Commission adopt limits on contacts between analysts and subscribers? Or are existing remedies - antifraud, contractual, or otherwise - sufficient to deter inappropriate disclosures to subscribers?
It seems that current NRSROs allow subscribers to discuss with analysts. Even though rating agencies have imposed strict restrictions to their analysts, it appears coherent to prohibit private contacts between them in order to prevent inadvertent disclosure of issuers' confidential information and those regarding forthcoming rating changes (except if the Commission considers that private contacts constitute a "white noise"). If this proposition is not shared, it is possible to implement alternative procedures like public meeting between investors and analysts in order to ensure investors' equality.
Should NRSRO recognition be conditioned on a rating agency having adequate financial resources (e.g., net assets of at least $100,000, or annual gross revenues of at least $1,000,000) to reduce dependence on individual issuers or subscribers?
Consider two agencies:
- one with an annual gross revenues of $5,000,000 and a client that represents 20% of the total;
- one with an annual gross revenues of $1,000,000 and any client representing more than 5% of the total.
Those examples evidence that the definition of adequate financial resources, based on minimum revenues or assets, is not sufficient or is useless. Furthermore, a definition of adequate financial resources is not easy to establish and it could be subjective.
In fact, in order to ensure the independence of rating agencies, it seems more appropriate to require, among other things (see question 11):
- capital independence: no direct or indirect share capital link should exist between issuers and agencies;
- economic independence: issuers that represent a certain percentage of a rating agency's revenues should be disclosed (see question 43).
Should NRSRO recognition be conditioned on a rating agency not deriving more than a certain percentage of its revenues (e.g., 3%) from a single source to help assure that the NRSRO operates independently of economic pressures from individual customers?
Conditioning recognition on a rating agency not deriving more than a certain percentage of its revenue raises few methodological difficulties:
- if it is possible to impose to large rating agencies such a criterion, it is doubtful that small ones would be able to respect it;
- which percentage would ensure the independence of the NRSROs? 2%, 3%, 5%?
- how would this percentage be defined? As agencies (the majority of them) calculate their fees on a percentage of the par value of the security being sold, the importance of a client may vary widely from one year to another (e.g. 3% in 2001 and 0,5% in 2002).
For those reasons, it could be more efficient to require to rating agencies to disclose to the Securities and Exchange Commission, for example in the annual certification filing, issuers that exceed 3 to 5% of their annual revenues. Then, this will have the merit to draw the Commission (or NRSRO Recognition Committee) attention on the ratings of those "predominant issuers".
E-Alleged Anticompetitive, Abusive, and Unfair Practices
Should the Commission identify specific anti-competitive practices that NRSROs would agree to prohibit as a condition to NRSRO recognition? If so, what are those practices?
The lack of competition, pointed out by some participants during the Enron case hearings, could be explained by three factors:
- the Commission has not created an oligopoly because few rating agencies have been recognized. Fitch has contributed to reduce the offer. Nevertheless, the lack of transparency of the NRSRO recognition and of the criteria has contributed to dissuade potential competitors. The definition of new criteria and of a transparent recognition process will have the merit to define the "rules of the game".
- Nevertheless, this will probably not lead to the emergence of a new global credible agency. Indeed, as already stated, reputation is a key variable of the rating industry. Despite the fact that Fitch, Standard & Poor's and Moody's have been recognized at the same moment, the first one faces difficulties to develop its activity. Moreover, even with acquisitions of large agencies like Duff and Phelps Credit Rating, its market share is still thin.
- The fact that Moody's and Standard & Poor's are rating in the U.S. all public corporate debt issues29, has probably hindered the development of competition.
Nevertheless, it is essential that the Commission prevent anti-competitive and unfair practices in order to ensure the development of potential new NRSROs. Especially, the Commission should deal with those two practices:
- certain agencies are using in their rating process (rating process means the period starting from the first contact between the agency and the issuer to the decision of the rating committee) a methodology that may generate or suggest the existence of potential conflicts of interest. After the initial contact between the agency and the issuer and the communication of the appropriate information, few rating agencies (NRSROs and non-NRSROs) provide a preliminary rating, which can be comprised within a range (e.g., a preliminary rating A+ / A; in certain cases, a probability of realisation is indicated). If the issuer accepts this preliminary rating, the rating procedure is engaged; otherwise, it can abandon the process. This reinforce the influence of the issuer in the procedure and may lead to overestimate its credit quality;
- furthermore, in the United States, Standard & Poor's and Moody's are rating the majority of issues, with or without the issuer's request (e.g. Standard & Poor's admits that it rates 99,92% of the debt obligations and preferred stock issues publicly traded in the United States). This policy could lead to reinforce the relation between issuers and rating agencies that assign rating only under request. Indeed, if issuers have two ratings, which reasons would motivate the request of a third one? It represents a cost and, except for large firms, issuers could pay an other rating if the service is efficient. And, how could issuers estimate rating efficiency? Because of their relation with rating agencies or because of the rating they may have?
Consequently, if the Commission recognizes other NRSROs, "rating shopping" practices would be exacerbated despite the fact that reputation is the key variable of this business and, paradoxically, because of the lack of reputation of the new entrants. This could lead to a slight decrease of the standards of rating agencies and, in fine, of the rating industry. (See Part III, Solicitation of Additional Comment - Limits & Concerns).
In order to prevent this scenario, one should consider conditioning NRSRO recognition to the following rules:
- Rating agencies should not be allowed to produce preliminary rating;
- Rating agencies should rate issuers and issues only under their request, or under a special procedure for unsolicited ratings (see question 48).
The Commission could also analyse certain fee's policies. For example, can the fact that an issuer pays an "annual subscription fee based on the total short and long term debt and preferred stock of the firm that can be used as a credit against any future bond or other issues eligible for rating within the year"30 be considered as an anti-competitive practice?
Concerning the Fitch statement about structured finance, it seems necessary to study those practices. I would just mention that Dominion Bond Rating Service has specified that, in his DBRS's approach to rate CBOs and CLOs, "DBRS will in limited cases rely on ratings of other established rating agencies but only to the extent that DBRS has confidence in the rating approach of such agency and only after adjustment to reflect any comparative rating analysis that has been done. DBRS does not rely on specialized rating agencies (insurance, banking or otherwise) or rating agencies domiciled in only one country due to the inability to do any comparative rating analysis between DBRS and the agency in question. DBRS reserves the right not to accept any individual rating notwithstanding our reliance on an agency in other situations/transactions."31
Consequently, the Commission could condition NRSRO recognition to an annual certification filing in which each NRSRO would disclose new procedures and elements affecting its rating business. The Commission would examine those information in order to verify that they do not constitute anti-competitive or unfair practices. Furthermore, a "ruling procedure" could be implemented.
Would it be sufficient to condition NRSRO recognition on the adoption of procedures intended to prevent anticompetitive, abusive, and unfair practices from occurring?
It should be necessary to condition NRSRO recognition to the propositions expressed in question 45 (about preliminary rating, issuer request and unsolicited ratings). Annual certification filing and a "ruling procedure" should complete the review.
Should NRSRO recognition specifically be conditioned on an NRSRO's agreeing to forbear from requiring issuers to purchase ancillary services as a precondition for performance of the ratings service?
The first point the Commission has to deal with is the definition of ancillary services. Are they assessment services that estimate the impact of an issuer's action (merger, acquisition, debt restructuring...) or large management services which propose management, strategic, process risk...? In other words, what are the ancillary services proposed by the rating agencies? Those questions need clear answers.
Concerning rating assessment services, it seems coherent that issuers wonder what could be the rating consequences of an action (merger, debt restructuring...). Despite the fact that this kind of service does not currently represent an important source of revenue, it may become more and more essential and could generate conflicts of interest. Few alternatives could be considered:
- a clear separation between rating services and rating assessment services;
- rating assessment services could be offered only on a formalized issuer's request. It could be used only to assess potential effect on rating of determined issuers' actions.
- an explicit statement indicating that the rating assessment service, which estimates the impact of an issuer's action, does not mean that the effective rating will correspond to the estimated one. Indeed, for example, it is possible that between the assessment and the realisation of the issuer's action, its credit risk increases. Moreover, actions are not necessarily reaching the expected results.
Concerning others ancillary services like management, strategic risk..., it seems that rating agencies should not develop them in order to ensure that the rating industry keeps its credibility. Could strict "China walls" guarantee business units' independence? Recent events concerning other market "watchdogs" have showed that potential conflicts of interest can be difficult to manage. Should those ancillary services, despite "China Walls" and independence, use a similar denomination? How could the Commission ensure that rating agencies do not sell ancillary services? ... In any case, those services should be subject to full disclosure and specific hearings (market participants, the Commission, "experts" comments).
Should NRSRO recognition specifically be conditioned on an NRSRO's not engaging in specified practices with respect to unsolicited ratings (e.g., sending a bill for an unsolicited rating, sending a fee schedule and "encouraging" payment, indicating a rating might be improved with the cooperation of the issuer)?
As mentioned in question 14, unsolicited ratings are valuable information, which may contribute to limit "rating shopping" practices. Nevertheless, it should be regulated by a simple set of rules:
- Rule 1: Rating agencies are broadly explaining that unsolicited ratings are requested by investors. In order to legitimate their action, formalized requests from a significant threshold of investors should be required by the Commission.
- Rule 2: Issuers should be protected. In other words, they have to be able to estimate their risk when they choose a rating agency rather than another one. They should know that they could be subject to unsolicited ratings. Consequently, it could be pertinent to define a minimum level under which unsolicited ratings are disallowed. This threshold could be based on the issuer's outstanding total debt. This proposition would need proceedings between market participants and the Commission.
- Rule 3: after investors' request, the rating agency would be allowed to give an unsolicited rating without contacting issuer. Nevertheless, the latter should be informed. The rating disclosure should incorporate an issuer comment (if it wishes). Unsolicited ratings should be identified (see question 52).
Should the Commission address concerns about information flow from rating agencies? If so, should the Commission condition NRSRO recognition on a rating agency's agreeing to establish procedures to assure certain disclosures relating to its ratings business, such as those described above? Are there other disclosures that could be appropriate?
The fact that the Commission conditions NRSRO recognition to the propositions described above seems relevant. Besides, current NRSROs have broadly implemented those procedures (see questions 50 to 54).
Nevertheless, the Commission should not impose restrictive procedures. For example, Macey and DuPratt White (2002)32 proposed to rating agencies to disclose "whether a rating is being issued despite the fact that the rating agency lacks the information that a reasonable investor would consider relevant to the formulation of the rating." Reynolds (2002)33 states that rating agencies "disclose material risks that they find in the ratings review process even if the information has not been disclosed by management in financial filings." Those propositions do not seem adequate and the Commission and rating agencies should focus on the elements described below.
The Commission should try to improve the quality of information. As stated earlier, unsolicited ratings are reliable (thanks to agencies' knowledge of competitors, industry...). However, if we consider that public information is not sufficient to estimate a credit quality, this should not be sufficient for investors. In other words, public information would not be relevant for shareholders. Consequently, the Commission should focus on information's quality. For example, the Securities and Exchange Commission should impose to issuers to disclose "triggers" ("rating triggers" and conditional elements of material financial contract) because of their potential consequences on firm's credit quality. Moody's (2002)34 sounded U.S. corporate issuers in order to acknowledge their "triggers". The following interesting elements may be emphasized:
- Triggers are not publicly disclosed: based on issuers whose debt is rated Ba1 or higher, 22,5% of triggers are disclosed in issuer's SEC filings; based on issuers whose debt is rated Ba1 or higher, 22% of trigger effects are disclosed in issuer's SEC filings;
- Triggers are not necessarily disclosed to rating agencies: on 790 issuers with debt rated Ba1 or higher, 19 have not responded (2,4%); on 1077 issuers with debt rated below Ba1, 209 have not responded (19,4%).
As we can notice, speculative grade issuers are more reticent to disclose such information despite Moody's request. Consequently, the Securities and Exchange Commission should impose this disclosure but it would have to deal with two elements:
- the Commission should not give to issuers an important "margin of appreciation". Then, the Commission and market participants should define together an exhaustive list of triggers that should be disclosed;
- the Commission should determine if such disclosure would be public (SEC filings) or only addressed to rating agencies. Because of their potential consequences (e.g. default in certain circumstances), the first proposition seems adequate. Nevertheless, triggers disclosure would probably infringe confidentiality clauses.
Concerning rating agencies, they should not disclose material risks because their role consists in reducing informational asymmetry. Indeed, such disclosure could increase it because of issuers' reluctance to communicate private information that could become public. In fine, rating's reliability could be eroded.
In fact, rating agencies should try to improve issuer's disclosure. Downgrades or withdraws are not necessarily adequate because they can lead to potential conflicts with issuers. Indeed, which objective criteria could allow rating agencies to downgrade issuers because they estimate that the latter do not communicate sufficient information? How could rating agencies explain downgrades to investors? If rating agencies argue that issuer's downgrades are justified by the issuer's reluctance to disclose information, what could be the impact on investors' confidence? Could it affect issuers' credit quality and, in fine, lead to default because of investors' distrust?
Consequently, such procedures do not seem relevant. Rating agencies should try to implement procedures that enhance issuer's disclosure. Few alternatives are existing but they require exhaustive and objective definition.
Specifically, should NRSRO recognition be conditioned on a rating agency disclosing the key bases of, and assumptions underlying its rating decisions? If so, should these disclosures be made pursuant to standards developed by the industry, or otherwise?
It seems that current NRSROs have recently implemented such procedures in order to justify and to motivate their rating decision. It could be appropriated to define standards developed by market participants with Commission oversight.
Would it be advisable for the Commission to condition NRSRO recognition on a rating agency's agreeing to disclose performance information periodically? If so, what type of performance information would be most useful? How often should it be disclosed?
NRSRO recognition should be conditioned on a rating agency's agreeing to disclose performance information, general rating methodology... annually. Annual and cumulative default rates, rating transition matrix should also be disclosed (see questions 11 and 12).
Should NRSRO recognition be conditioned on a rating agency's disclosing whether or not an issuer participated in the rating process? Or, could issuers be required to make such disclosures?
Unsolicited ratings should be permanently distinguished (not only during the initial assignation, but also during the "rating's life"). As stated by certain rating agencies, issuers are an "important source of input"35. Consequently, unsolicited ratings, although reliable (see question 14), have not necessarily the same value than solicited ones. Each agency should differentiate those two categories of ratings (certain agencies use an initial p, pi...).
Concerns have been raised that certain credit rating agencies make their credit ratings available only to paid subscribers, and that it would be inappropriate to require users of credit ratings to subscribe for a fee to an NRSRO's services to obtain credit ratings for regulatory purposes. What steps, if any, should the Commission take to address these concerns? For example, should NRSRO recognition be conditioned on a rating agency's agreeing to public dissemination of its ratings on a widespread basis at no cost, as is currently the case?
Current NRSROs appear to offer a "free large public dissemination". It should be a criterion in order to ensure a similar disclosure to investors (see question 11).
Should NRSRO recognition be conditioned on a rating agency's implementing procedures to assure public notification when it ceases rating/following an issuer. If so, what form of public notification would be appropriate?
Current NRSROs seem to have implemented a procedure of notification in order to inform that they cease the rating/following of an issuer. Nevertheless, the reason could be stated through an initial (for example, WRc could mean Withdraw Rating because the issue has been "Called").
What steps, if any, can the Commission take to improve the extent and quality of disclosure by issuers to rating agencies or to the public generally, and in particular, regarding: (a) ratings triggers in financial covenants tied to downgrades; (b) conditional elements of material financial contracts; (c) short-term credit facilities; (d) special purpose entities; and (e) material future liabilities.
See Question 49.
III - Solicitation of Additional Comments
During the proceedings of the Enron case, ratings' reliability has been debated. An approach, essential to appreciate ratings' reliability of a particular agency, consists in analysing the rating's history of the issuer before default. The following graphic represents different approaches of the rating history of issuers, 1500 days before the default event:
- Moody's historical (1983-2001) average rating before default36;
- Standard & Poor's and Moody's average ratings before default (2001), according to my research;
- Standard & Poor's and Moody's average ratings before Enron default.
Graphic 3 - Issuers' rating history before default37
This demonstrates rating's reliability because it shows that average rating is well below the speculative grade threshold four years before the default event. It seems relevant to make two observations:
- 1500 days before default (more than four years), Moody's historical average rating is a speculative grade rating (Ba3). The speculative grade category means that credit risk is relatively high and that issuer creditworthiness could be affected for example by their environment [financial and economic conditions according to numerous studies (Fons, 1991; Jonsson and Fridson, 1996; Helwege and Kleiman, 1997, ..., and my research];
- Defaults registered in 2001 produce similar results, except for Enron. 1500 days before default, the average rating is B+/B1;
Consequently, could the Enron case (and few others) evidence the non-reliability of ratings? Is it necessary to impose rating's procedures as historical and recent studies have demonstrated rating's reliability? Could future requirements affect agencies' flexibility and ratings' reliability? Could the definition of rating's procedures improve predictions of default whereas some credit pricing models38 consider that the "default event is random"?
Limits & Concerns
Rating history before default, reliability and comparison
The NRSRO Recognition Committee should deal with the following question: which threshold could allow applicants' qualification? In other words, if two applicants present ratings' history curves that are respectively one and two notches above the "worst" NRSRO, could the Committee recognize the first and reject the second?
Consider three current NRSROs and three applicants. In order to analyse applicants' reliability, the NRSRO Recognition Committee could use its first tool: the rating's history before default. The six rating agencies present the following curves39 (Graphic 5).
Graphic 4 - Agencies rating's history curves before default
It seems obvious that Applicant 1 would not be qualified. Nevertheless, which threshold could allow Applicant 2 to be recognized? Could the Commission measure this threshold in notch or use one or two standard deviations? The simplest situation will be when the applicant (Applicant 3) is comprised within current NRSROs' margins. Nevertheless, such case seems unlikely. Consequently, the Committee would have to define a threshold.
This threshold will be legitimate as current NRSROs will be under review (see Recognition Process and Examination and Oversight of NRSROs). In other words, the Committee could ensure that old and new NRSROs do not modify their standards. Consider that the Committee recognize two applicants (two current NRSROs exist). The Committee may control agencies' reliability in order to ensure that a particular agency or that the rating industry maintain a high level of reliability. Graphic 6 shows the "worst case scenario" in which NRSROs diminish their reliability (dashed lines represent for example agencies' reliability three years latter).
Graphic 5 - Agencies rating's history curves before default (under review)
Consequently, the Committee should control rating agencies' reliability. Nonetheless, two "technical problems" exist:
- the Committee should define a threshold (standard deviation);
- the Committee should find a relevant basis of comparison. For example, it is useless to compare the Moody's and Fitch averages rating history because Fitch have rated only few speculative grade issuers. Nevertheless, a comparison of rating history of issuers rated by both could be relevant.
The Committee should take into account that economic and financial environment can have an impact on curves.
Concerns about NRSRO Recognition Committee
How would the independence of the Committee be ensured? How many persons should constitute the Committee?
Is it necessary to integrate "experts" and market participants? If yes, how ensure that they are independent?
Could public comment improve the Committee process? Or could it generate difficulties?
Would it be opportune to integrate foreign regulatory authorities' officials? Would it be realistic to imagine that a world-wide authority will be in charge of agencies' recognition?
I would like to thank the Commission for having given the opportunity to participate to this important and interesting discussion. I look forward to continuing dialogue with the Commission and I am staying at its members' disposal if they have any questions or queries regarding this comment.
The following information can be publicly disclosed.
I am achieving a Ph.D. in Economics. My thesis is mainly dealing with rating agencies (their reliability, credibility; ratings and credit spreads), regulations (consequences, NRSRO criteria) and default and recovery rates (explanatory variables). I am working in an asset management structure as an economist. Nevertheless, those comments and propositions result exclusively from the thesis.
Basel Committee on Banking Supervision, 2003, "Overview of the New Basel Capital Accord", Consultative Document, Bank for International Settlements, April.
Beattie V., Searle S., 1992, "Bond Ratings and Inter-Rater Agreement", Journal of International Securities Markets, Summer.
Cantor R., Packer F., 1995a, "Multiple Ratings and Credit Standards: Differences of Opinion in the Credit Rating Industry", Federal Reserve Bank of New York, Research Paper No.9527, December.
Cantor R., Packer F., 1995b, "The Credit Rating Industry", The Journal of Fixed Income, December.
Cantor R., Packer F. and Cole K., 1996, "Split Ratings and the Pricing of Credit Risk", Federal Reserve Bank of New York, Research Paper No.9711, March.
Cantor R., Packer F., 1997, "Differences of Opinion and Selection Bias in the Credit Rating Industry", Journal of Banking and Finance, 21.
Delianedis G., Geske R., 2001, "The components of Corporate Credit Spreads : Default, Recovery, Tax, Jumps, Liquidity, and Market Factors", Working Paper, The Andersen School at UCLA, December.
Dominion Bond Rating Service, "CBO/CLO Rating Approach".
Ercisson J. and Renault O., 2001, "Liquidity and Credit Risk", Working Paper, London School of Economics, August.
Fitch, 2001, "Fitch Corporate Bond Default Study, A Decade in Review", Corporate Finance, November 8.
Fons J., 1991, "An Approach to Forecasting Default Rates", Moody's Special Report, August.
Helwege J., Kleiman P., 1996, "Understanding Aggregate Default Rates of High Yield Bonds", Federal Reserve Bank of New York, Current Issues in Economics and Finance, Vol 2, n°6, p1-6.
Hearing before the Committee on Governmental Affairs, 2002, "Rating the Raters: Enron and the Credit Rating Agencies", United States Senate, One Hundred Seventh Congress, Second Session, March 20.
Jewell J., Livingston M., 1999, "A Comparison of Bond Ratings from Moody's S&P and Fitch", Financial Markets, Institutions & Instruments, New York University Salomon Center, Vol 8, No 4.
Jonsson J., Fridson M., 1996, "Forecasting Default Rates on High Yield Bonds", The Journal of Fixed Income, June, p69-77.
Leftwich R., 2002, "Comment noter les agences de notation ?", L'Art de l'Entreprise Globale, LesEchos.
Letter from Leo C. O'Neill, President of Standard & Poor's, to Congressmen Tauzin, Dingell, Greenwood & Deutsch, March 19, 2002.
Letter from Moody's Investors Service to the U.S. Securities and Exchange Commission, December 5, 1994.
Lubochinsky C., 2002, "Quel Crédit Accorder aux Spreads de Crédit ?", SEMASFI.
Moody's Investors Service, 2002, "The Bond Rating Process in a Changing Environment", Global Credit Research, January.
Moody's Investors Service, 2002, "Default and Recovery Rates of Corporate Bond Issuers: 2001", Special Comment, Global Credit Research, February.
Moody's Investors Service, 2002, "The Bond Rating Process: a Progress Report", Global Credit Research, February.
Moody's Investors Service, 2002, "Understanding Moody's Corporate Bond Ratings and Rating Process", Global Credit Research, May.
Moody's Investors Service, 2002, "Moody's Analysis of US Corporate Rating Triggers Heightens Need For Increased Disclosure", Global Credit Research, Special Comment, July.
Packer F., Reynolds E., 1997, "The Samurai Bond Market", Federal Reserve Bank, Current Issues in Economics and Finance, June, Vol 3, No 8.
Packer F., 1999, "Credit Risk in Japan's Corporate Bond Market", Federal Reserve Bank, Current Issues in Economics and Finance, November, Vol 5, No 15.
Report of the Staff to the Senate Committee on Governmental Affairs, 2002, "Financial Oversight of Enron: The SEC and Private-Sector Watchdogs", October 8.
Standard & Poor's, 2001, "Corporate Rating Criteria".
Standard & Poor's, 2002, "Survey reflects a Divergence of Opinion on Time Horizons for Ratings", RatingsDirect, May 7.
Standard & Poor's, 2002, "Japan CreditStats", December.
Standard & Poor's, 2003, "Corporate Defaults Peak in 2002 Amid Record Amounts of Defaults and Declining Credit Quality - Hazards Remain", Risk Solutions, January 23, Charts 8 and 9.
Statement of Fitch Ratings, 2002, U.S. Securities and Exchange Commission, November 12.
Statement of Standard & Poor's Rating Services, 2002, U.S. Securities and Exchange Commission, Public Hearing, November 15.
Written Statement of McDaniel R., 2002, President, Moody's Investors Service, Before the United States Securities and Exchange Commission, November 21.
Written Statement of Rating and Investment Information, Inc., 2002, Submitted to the U.S. Securities and Exchange Commission In Connection with the November 15 and 21, 2002 Hearings on Credit Rating Agencies (File No. 4-467), November 214.