The United Kingdom authorities response to the European Commission internal market and services consultation document on Credit Rating Agencies

Transcription

1 The United Kingdom authorities response to the European Commission internal market and services consultation document on Credit Rating Agencies 1.1 HM Treasury, the FSA, and the Bank of England (henceforth referred to as the UK authorities) welcome the opportunity to comment on the Commission Services consultation document focussing on further strengthening the EU regulatory framework for credit rating agencies (CRAs). 1.2 The Commission Services consults on a very broad and ambitious range of important issues, only shortly after EU Member States have agreed Regulation (EC) No 1060/2009 (CRA 1), and the transfer of regulatory responsibility for credit rating agencies to ESMA (CRA 2). These directives have not yet been fully implemented, and in our view a third directive should take full account of the progress that has been secured to date. 1.3 An effective reform package should be more narrowly focused on the key priorities that address the remaining causes of instability associated with credit ratings. It is therefore important that proposals for policy change are based on comprehensive and careful analysis of where exactly the problems lie, together with a full assessment of the potential impact of such changes. As a principle, changes should only be envisaged in those areas where it can be verified that problems have occurred. 1.4 In particular, the UK authorities share the Commission Services view that the hard-wiring and the mechanistic reliance on CRAs materially contributed to instability during the financial crisis and should be a key area of reform. The liquidity pressures associated with ratings-driven collateral calls on AIG are a case in point. 1.5 We would encourage the European Commission to have full regard to ongoing work in other international fora on reducing overreliance on credit ratings. In particular, the European Commission will be aware that the FSB Principles for Reducing Reliance on CRA Ratings, has set high-level guidance for reducing reliance on CRA ratings in standards, laws and regulations. 1 These principles were endorsed by the G20 at the meeting of Finance Ministers and Central Bank Governors in Gyeongju, Korea, October 23, Standard setters in the FSB are now working to translate the Principles into specific policy proposals. In addition, other jurisdictions, such as the United States, are considering actions to remove or replace references to CRA ratings in their laws and regulations. In the interest of ensuring an internationally consistent 1 Available at 2 Available at: 1

2 outcome, the Commission should seek to consider the findings emerging from this work, while maintaining the agreed approach and timeline for CRD Any changes must be carefully assessed and remain cognisant of the important contribution of CRAs to credit assessment in the international financial system. Notwithstanding concerns over the reliability of CRA ratings in specific forms of structured credit, and their response to new information in the sovereign debt markets, ratings have typically offered reasonably good estimates of credit quality among corporate issuers. Therefore alternatives to the use of CRA ratings must meet the test of delivering the desired diversity without compromising the accuracy with which exposures of different credit quality can be distinguished. We encourage a deeper analysis of the alternatives, which could serve to usefully complement ratings information The UK authorities also consider it critical that financial institutions improve their risk management processes and apply more judgement in the way in which they use rating information. All holders of securities particularly large institutions should understand the risks associated with the exposures they assume. They should therefore be subject to rigorous standards of due diligence and face financial and regulatory incentives to meet these standards. To ensure that sufficient information is available to investors to carry out such analysis, disclosure standards for securities issuers should also be enhanced as discussed in the FSB Principles. More generally, however, the UK authorities agree with the Commission that any measures taken in this area should be proportionate to the size of institutions. 1.8 The UK government agrees that competition in this industry should be improved. This should however be achieved without jeopardising the quality of ratings. CRA 1 and CRA 2 have already strengthened disclosure requirements and enable effective enforcement of those requirements by ESMA. In our view competition is therefore best enhanced by reducing the mechanistic reliance on ratings; enforcing transparency and disclosure which enables the comparability of different CRAs; and removing unnecessary barriers to entry. Transparency and disclosure are also key reforms particularly with regard to securitisations and covered bonds, where the complexity of the products and lack of transparency around underlying asset pools renders them difficult to deal with. 1.9 However, the UK authorities strongly oppose any issuance of credit ratings by a public European CRA, Central Banks, or public-private partnerships. The use of such ratings would potentially heighten scope for moral hazard and harm the independence of these institutions and their ability to fulfil other objectives (e.g. price stability). A public agency would also risk crowding out private sector ratings and inevitably stifle competition in the industry We note that as a result of CRA 2, ESMA will soon have a wide range of effective tools to enforce compliance with the enhanced requirements of CRA 1. In our view there is therefore no need to introduce civil liability, which would also be counterproductive to reducing the reliance on credit ratings, as well as risk weakening competition. 3 Given that CRD4 needs to be implemented before this ongoing work is completed; perhaps there is a scope to review CRD4 to ensure consistency with CRA 3 proposals after they have been agreed. The UK authorities would have grave concerns if the CRA3 process would jeopardise the timely and agreed implementation of CRD4. 4 In assessing the merits of different sources of information to rate credit, it could be instructive to distinguish between the three different roles credit ratings currently play in the financial markets: i) regulatory determinations, ii) certification (e.g., risk thresholds in investment mandates/policies, collateral agreements, etc.) and iii) information (reducing asymmetric information in investment decisions, delegated monitoring). These are distinct demands, for which market requirements are different. For instance, regulatory capital determinations typically require granular estimates of tail risks; certification requires much less granular objective measures that summarise risk appetite; and the information function requires granular estimates of relative expected loss-given-default. 2

3 1.11 Imposing a new business model is a significant intervention which, if done without thorough cost benefit analysis, could have severe, unintended consequences for the wider financial system. We would therefore have concerns about endorsing such proposals before seeing the results of further internationally coordinated work in this area This remainder of this document sets out the joint response and addresses the main questions set out in the Commission Services consultation paper. It is organised as follows: I. Reducing Reliance on external credit ratings (Questions 1-15) II. Sovereign Debt (Questions 16-22) III. Competition in the Credit Rating Industry (Questions 23-30) IV. Civil Liability of Credit Rating Agencies (Questions 31-33) V. Potential Conflict of Interests from the Issuer pays model (Questions 34-37) I. Over-reliance on external credit ratings Reference to external ratings in regulatory capital frameworks for credit institutions, investment firms, insurance and reinsurance undertakings The Internal Ratings Based Approach as an alternative [Q 1,2] 1.13 The UK authorities support the core message of the FSB Principles that banks must not mechanistically rely on CRA ratings and should have the capability to conduct their own credit assessment. That is, in all cases, CRA ratings should be one of many inputs to the investment process, rather than a sole determinant. The Principles go on to say that larger, more sophisticated banks within each jurisdiction should be expected to assess the credit risk of everything they hold (either outright or as collateral), whether it is for investment or for trading purposes and encourage supervisors to incentivise banks to develop internal credit risk assessment capacity and to increase use of the internal-ratings-based approach under the Basel capital rules. We note however, that currently IRB models have some shortcomings and their extension would therefore need to be assessed carefully As the Commission s Consultation document rightly sets out, it is essential that standards of credit quality used in capital regulation deliver consistent high-quality, independent and objective estimates of credit risk. Therefore, supervisors need to devote very considerable resources to the oversight of banks risk-management and credit assessment processes. In particular, if the scope of the IRB approach were to be extended, it would be important for supervisors to enhance their capability to identify modelling deficiencies and rescind approval of a bank s internal model where deemed necessary. It would also be important to avoid an inconsistent approach in this area, as this could undermine the Single Market The wide and free availability of external ratings as a measure of risk has in all likelihood reduced banks private incentives to research and develop potential alternatives. Incentives could be provided within the capital framework itself for example, mandating that all large exposures be internally assessed, or applying penal capital charges where a bank cannot demonstrate that it has carried out adequate internal assessment. This would be an important change in emphasis within the regulatory framework, which in the implementation of Basel II, 3

4 allowed an overall reduction in regulatory capital for banks using the IRB approach to compensate for the additional costs incurred in building and maintaining credit risk models Steps could also be taken to encourage better disclosure by borrowers and securities issuers, so as to ensure access to adequate credit-relevant information to support banks internal assessment processes and overcome some of the existing limitations of the IRB approach (see also the response to Question 10). Currently, the ability of IRB models to deliver a demonstrably reliable measure of risk is often constrained by data shortcomings. In particular, they often lack the breadth of empirical experience to robustly predict how estimates and outcomes will vary with the cycle. As a result, to enhance their robustness, internal models are currently often linked back to CRA ratings or CRAs default histories as external independent measures of risk. 5 Further work should be encouraged to link internal rating models to a wider range of external measures of risk, or to combine different measures of risk (see also the response to Question 4) Even if some limitations of the IRB approach were overcome, the UK authorities caution against setting firm size-based thresholds for use of the IRB approach. The priority should be to establish a capital framework that ensures robust internal risk assessment, better oversight of such risk assessment, and improved access to credit-relevant information. Any proposed changes should of course be subject to rigorous impact assessments and comprehensive evidence base, with work under way internationally usefully complementing such efforts. The Use of Two External Ratings [Q3] 1.18 Requiring two external ratings may potentially improve the accuracy of some regulatory capital requirement calculations. For example, certain types of securitisation transactions and covered bonds are frequently only rated by one agency as other agencies have chosen not to rate the structures in question (e.g. certain re-remics). 7 In practice, the Basel II rules already imply some reference to more than one rating within the capital determination process, even though only one is ultimately used to set risk weights. 8 While this could conceivably be strengthened further, to do so would seem to contradict the message in the FSB Principles that references to ratings in regulations (including Basel II) should be removed or replaced wherever possible Furthermore, requiring two external ratings might reduce investor demand for some structures if it was not possible to get a second rating. This may be a desirable outcome in cases where the reluctance of other rating agencies is based on their concerns that the data available is inadequate to undertake credit analysis, but it could also have unintended consequences. These would need to be carefully considered before implementing such a rule. For example, a requirement for two external ratings could constrain issuance and innovation in some markets. It could also hinder the Commission s efforts to improve competition in the industry, especially 5 The FSA s experience in implementing the internal ratings based (IRB) approach under Basel 2 is that in practice banks internal models for wholesale exposures are usually closely linked to CRAs ratings and/or default histories. 6 For example, the FSA has observed banks using the commercially available Merton statistical default model as a mechanism that meets this standard By definition, however, the potential use of this approach is limited to quoted companies and even in this case it is questionable whether it can robustly be applied to all quoted companies. 7 Given the intricacies of structured products, it is often difficult for another CRA to provide a rating without having access to detailed information underlying the instrument. Disclosure requirements for the issuer may therefore also need to be improved to facilitate full access to information by both rating agencies. 8 If there are two assessments by ECAIs chosen by a bank that map to different risk weights, the higher risk weight will be applied. If there are three or more assessments, the determination of risk weights should refer to the assessments corresponding to the two lowest risk weights, with the higher of the two then applied. One exception is the qualifying category for securities issued by PSEs and multilateral development banks in the calculation of specific capital charges for interest rate risk in the standardised method of the Market Risk framework (paragraph 717), for which a requirement to use to external ratings already exists. 4

5 if it favours large incumbents that may be more likely to possess the resources to meet the demand for additional ratings More generally, different CRAs ratings tend to be very close together (as might be expected if methodologies converge over time). According to a 2006 report by the French AMF 9, even where ratings for a given issuer differ between CRAs, the difference is typically just one notch. Therefore, the practical effect of mandating multiple ratings may not be material anyway. Alternative measures of credit risk [Q4] 1.21 The UK authorities believe that alternative credit risk metrics constitute an important source of information that should be fully utilised to reach a much more informed investment decision. Although CRA ratings usually integrate such metrics into their ratings, using them to generate an alternative credit assessment, or as complementary information, would reduce the degree of reliance on external credit ratings However, we note that CRA ratings have offered reasonably reliable estimates of credit quality, at least amongst corporate issuers. Any alternatives to ratings for use in the regulatory framework must therefore meet the test of delivering the desired diversity of information (and hence removing hardwiring) without compromising the accuracy with which exposures of different credit quality can be distinguished This should be carefully considered as part of the evidence base that examines the use of alternatives. Similar to suggestions set out in the US banking regulators Advance Notice of Proposed Rule-making (ANPR), relevant criteria for assessing the viability of alternatives might include: objectivity; accuracy; transparency; timeliness; cost; incentives; stability; and market impact In the following, we present some high-level observations on market-based indicators, as well as three other possible alternative models for producing measures of credit quality for regulatory purposes: non-market-based indicators; credit assessments outsourced to a non-cra third party; and dual ratings Market-based indicators: These could potentially be useful measures of credit risk, since they do provide an arms length assessment. Banks have, however, to date been reluctant to incorporate these into their internal ratings. This is primarily because movements in market prices are driven by factors other than credit risk such as the depth and liquidity of the market and are prone to overshooting (procyclical effects). For instance, prior to the crisis, prime UK RMBS was trading at a spread of 9bps (implying an annual expected loss of at most 0.09%), but then during the crisis there was no market at all for many securitisation positions, including some of those that were most liquid before the crisis. Also, many securitisation positions are illiquid even in benign conditions. A point-in-time market price would also be inherently unstable. Therefore, if used for such a purpose, a moving average would be required. While this might address concerns around instability, it is not clear that the averaged measure could in practice distinguish between cyclical factors, changes in liquidity conditions, and changes in credit quality See the US banking regulators ANPR, available at: 5

6 1.26 Non-market-based measures: These might, for example, include financial ratios based on accounting data. The choice of such measures would differ by asset class and further analytical work would be required to fully establish which specific measures to use and how precisely they should be combined to distinguish effectively between different credit exposures. Use of such metrics might be more feasible for some asset classes (e.g., sovereigns) than others (e.g. structured finance assets). 11 For instance, in the case of sovereign ratings, ratios might include: total debt-to-gdp ratio; debt-to-export ratio; public debt-to-gdp; GDP growth; foreign reserves; and debt service commitments. Different indicators might also be applied to different groups of countries (e.g., low-income, middle-income, and high-income countries) Outsourcing (by product class) to non-cras that use a tailored approach to assessing credit quality: Such an approach was recently adopted by the National Association of Insurance Commissioners (NAIC) in the United States in relation to insurance companies RMBS exposures. NAIC appointed a third party to conduct a credit risk assessment exercise tailored to the capitaldetermination process. Such an approach could benefit from existing capabilities and niche skills. As long as conflicts of interest in the chosen third party could be avoided, such an approach should not create adverse incentives and should appropriately separate the capital determination process from other certification and delegated monitoring functions carried out in the financial markets. One approach might be to target investment and advisory firms, who already have a private incentive to invest in resources for credit assessment, but (i) will not be directly impacted by the capital charges (and so will not be conflicted), and (ii) do not otherwise sell their ratings in the marketplace. Regulators would, however, need to ensure appropriate governance of the process (e.g. Chinese walls) within the firm Dual ratings (an external rating complemented by an internal rating): For some capital determinations (e.g., the securitisation framework in Basel II), there may be scope for introducing a so-called dual ratings approach. Such an approach might involve taking the more conservative of a bank s internal rating (produced by a validated model) and the prevailing external rating. Of course, the challenges in delivering a robust internal rating as discussed in the response to Questions 1 and 2 must be acknowledged in this alternative. The calibration of risk weights applied where a bank did not have an internal rating might incorporate an incremental capital charge to create good incentives to produce an internal rating. Investment in securitisations [Q5] 1.29 Financial institutions should understand the risk of all assets that they hold and be able to demonstrate that they have carried out the requisite level of due diligence. Where an institution cannot demonstrate that it fully understands the risks associated with its investments, it should only be permitted to invest if a suitably conservative penal capital charge is applied to its exposures. A similar provision is to be introduced for securitisations in the amended Basel II framework. Although the principle of proportionality is sound, even smaller, less sophisticated financial institutions should not rely mechanistically on CRA ratings. Once again, the basic principle is that CRA ratings should be one of many inputs to the investment process, rather than a sole determinant. 11 For complex, bespoke assets such as securitisations and some covered bonds, it would be more difficult to identify robust market or non-market measures of credit quality. Equally, for such assets, even greater care would need to be taken with respect to reliance on internal models, due to the potential adverse incentives, the high cost of evaluation and the absence of comparables. The risks arising from greater use of internal models could perhaps be mitigated by applying a dual ratings approach as described in paragraph Alfonso, Gomes and Rother (2007) identify a set of core indicators that are relevant for the determination of countries credit ratings. 6

7 1.30 In some cases it may be disproportionate to require institutions to even attempt to perform a full credit analysis for all positions; e.g. for positions held in the trading book where the positions are held short term with trading intent. Introducing such a requirement might lead to institutions introducing minimum size restrictions to any investments/loans due to the cost of assessment (i.e. diseconomies of scale). For some products (e.g., securitisations), thorough risk assessment might preclude financial institutions assuming exposures in support of their market-making activities, which could in turn undermine market liquidity. This distinction is recognised in the due diligence guidelines developed for CRD 2 (Article 122a). This principle is also relevant in the context of the Implementing Measures developed under Solvency II As discussed in the response to Questions 1 and 2, data shortfalls may limit the possibility of a full internal credit analysis for all investments. While due diligence could be undertaken on a best-efforts basis using the information that is readily available in the market, full credit analysis is likely to have materially higher data requirements, including in respect of the underlying assets. Originators/sponsors should therefore be required to disclose more granular data to facilitate such credit analysis (as proposed by CRD2) see response to Question 10. Supervisory Formula and QIS approaches to measuring credit risk [Q6] 1.32 While it is of course appropriate that capital charges be based on an assessment of inherent risk, there may be circumstances in which highly risk-sensitive capital requirements can have unintended adverse consequences. Indeed, some modelling approaches may be unduly sensitive to perceived risk and fail to capture tail risks. For example, Basel II introduced a more risk-sensitive approach which led to many AAA securitisation assets being reduced from a 100% risk-weighting to 7% risk-weighting not long before securitisation default and loss rates rose sharply In particular, the supervisory formula approach for structured instruments can generate very low capital requirements on senior positions, which may not be appropriate. Its risk-capture could therefore be improved by making it less sensitive to PD and attachment, while not reducing risk-weights too much for junior exposures. However, it would require accurate risk measures for the underlying exposures, which is currently problematic, but could improve with better disclosure. Any change to the structure of capital requirements for securitisation exposures under the securitisation framework, including a change to the hierarchy of approaches to determining risk weights, should not lead to lower risk weights for exposures than if based on CRA ratings. Again, we would refer to ongoing work by the Basel Committee, which has duly considered these issues More generally, on the approach to calculating capital charges in securitised products, the mapping of CRA ratings to credit quality steps (CQS) in the securitisation framework is a fundamental issue. The different methodologies used by the major rating agencies (e.g. first dollar of loss versus expected loss) means the same rating from different agencies does not necessarily reflect the same potential for credit losses. Issuers therefore have an incentive to shop for the ratings from those rating agencies whose methodology would result in the lowest capital charge for a given level of risk. If references to CRA ratings are retained in the securitisation framework, it would be important to address this mapping issue to i) reduce the potential for ratings shopping (which is even more pertinent given efforts to improve 7

8 competition in the industry), and ii) the potentially misleading comparison of ratings (and associated capital requirements) based on different methodologies As an overall point, if references to CRA ratings are retained in the regulatory framework, for example as one among a number of inputs to minimum capital requirements, they should not be permitted to reflect the potential for government support of the rated entity, especially in the case of financial institutions. Otherwise regulatory capital would be predicated on the assumption that government support would be forthcoming even though the overriding objective of prudential capital requirements is to minimise the chance that financial losses and risks are transferred from the private sector to the taxpayer. The principle that the credit assessment should reflect only an entity's own financial strength must be applied for the same reasons in internal models. Use of external ratings for internal risk management purposes Requirements of due diligence and internal risk management processes [Q7] 1.36 The UK authorities would strongly support the view that institutions should be required to perform and improve their due diligence. In fact, the expectation that all banks carry out a minimum level of due diligence is consistent with the FSB Principles which state that banks should have the capability to conduct their own assessment of the creditworthiness of, as well as other risks relating to, the financial instruments they are exposed to, and should satisfy supervisors of that capability The consultation paper rightly acknowledges that steps have already been taken, in the context of amendments to the Capital Requirements Directive (CRD 2), to enhance due diligence in relation to banks securitisation exposures. In particular, Article 122a of CRD 2 introduces due diligence requirements for credit institutions investing in securitisations. The Commission is encouraged to monitor banks responses to these changes and establish whether they are fully compliant, and consider whether similar requirements might be imposed on other institutions and in relation to other products. More generally, it should be noted though, that the due diligence required by CRD 2 does not explicitly require banks to undertake credit analysis of the type that would enable them to calculate a capital requirement Efforts to improve due diligence must work around two issues. First, even if a credit institution's due diligence produced a capital charge that is below that derived from ratings, under CRD, banks are required to use the capital charge derived from the external rating. So it may be instructive to consider the interaction of external rating generated risk weights and internal due diligence generated risk weights when determining capital charges. Second, internal risk management processes often make significant use of external ratings and it would not be appropriate to completely prevent such use, as long as it can be demonstrated ratings are merely one input of many Arguably a minimum level of due diligence could be specified, for example set by type of product/exposure type and potentially implemented via Pillar 2. Supervisors would assess banks risk-assessment processes for a range of specified exposure categories, perhaps defined as a function of risk and complexity. Banks that did not meet the threshold requirements might be 8

9 subject to higher capital charges, thereby providing an incentive to enhance their internal risk assessment. Disclosure of information to supervisors monitoring internal risk management processes [Q8] 1.40 Useful information, which should be disclosed to all stakeholders (and not only supervisors) by credit institutions should comprise the following: The institution's overall policy on use of external ratings; Classes of assets for which some reliance is placed on external ratings; Detailed explanation of the institution's approach to using external rating data and how it combines this data with its own data and looks to improve on the external rating. Current use of credit risk models for internal risk management [Q9] 1.41 The implementation of Basel 2 has extended the use of the IRB approach to credit risk, which is now used by almost all large banks in the UK. However, for wholesale borrowers these tend to contain indirect reliance on CRA ratings. Many sophisticated banks also use portfolio credit risk models for internal purposes, although there is variation in the degree to which they are embedded in practice. These models present even greater validation challenges For securitisation exposures many large banks use economic capital models to perform assessments of the risk transfer achieved via securitisation. This involves internal credit analysis of the underlying assets being securitised and the securitisation positions being originated. In practice, the UK s experience has been that banks' internal credit assessments will always have limitations Boards are involved in the review of the use of credit ratings in their investment policies and risk management processes to some degree. In practice, their knowledge and active involvement is relatively high-level, and they therefore place a high level of reliance on technical experts. However, as set out in the FSB Principles, senior management and Boards of institutional investors should ensure that internal assessments of credit and other risks associated with their investments are made, and that the investment managers they use have the skills to understand the instruments that they are investing in and exposures they face, and do not mechanistically rely on CRA ratings. Disclosure and Transparency [Q10] Achievements to Date 1.44 The UK notes the measures taken through CRA 1 significantly increase disclosure and transparency, ensuring a minimum standard for the quality and scope of information published by CRAs within the EU. They crucially enhance users ability to use ratings within their internal risk management in a more sophisticated way. EU registered CRAs are now required to provide: A clear view of their business model to users ; Periodic disclosures to users of the historical default rates of its rating categories as well as key clients; 9

10 Transparency reports providing information on, amongst other things, its legal structure and ownership and a description of the internal control mechanisms ensuring quality of its ratings These measures will also be further supported by the central repository (CEREP) which publishes data in a standardised form on the performance of ratings issued by CRAs registered within the EU. This will allow investors to make their own assessment about certain CRAs, thereby exerting more reputational pressure The UK authorities believe that disclosure and transparency are key proposals for securitisations and potentially also some covered bonds, and we note the progress made by CRA 1 and 2. We note that any registered CRA rating structured products must now comply with additional disclosure requirements on methodologies, models and key rating assumptions. It must also disclose, on an ongoing basis, information about all structured finance products submitted to it for their initial review or for preliminary rating. In addition, CRAs are now banned from giving any advisory services, an important step in reducing conflicts of interest. Steps forward 1.47 As yet we have not seen the final result of these measures as they will embed once the CRA registration process is complete. Importantly, the impact review carried out after they embed will highlight future areas where users need further disclosure or other changes considered in the Commission s consultation. The UK Authorities would be in favour of any transparency measures which increased investors awareness of CRAs activities, methodologies and governance processes More generally, the UK authorities reiterate that any approach that relies on enhanced internal risk assessment or due diligence should be accompanied by greater transparency and disclosure by securities issuers so as to facilitate rigorous internal risk assessment by potential investors. As an example of such measures, the Bank of England recently placed explicit information requirements on ABS issuers and covered bond issuers if their securities are to be eligible for the Bank s operations. 13 Similarly, the ECB has established loan-by-loan information requirements for asset-backed securities (ABSs) in the Eurosystem collateral framework. 14 This establishment of an EU-wide disclosure regime for securitisations is a welcome development We believe it is important that the relevant elements of this regime are also extended to covered bonds, in line with the Bank of England s approach. Covered bonds are of growing systemic importance to the European financial system. In the absence of clear disclosure of the assets backing covered bonds, the financial system will be exposed to the possibility of regulatory arbitrage, as risk could migrate unnoticed from other products. It is notable that the ECB has already expressed concerns about the possibility of risks migrating from ABS to covered bonds. 15 Of course, improving issuer-transparency is a complex undertaking, which should be pursued in a diligent manner. 13 See the Bank s Market Notice, 19 July: 14 See 15 OPINION OF THE EUROPEAN CENTRAL BANK of 6 August 2010 on a proposal for a directive of the European Parliament and of the Council amending Directives 2006/48/EC and 2006/49/EC as regards capital requirements for the trading book and for resecuritisations, and the supervisory review of remuneration policies (CON/2010/65) 10

11 1.50 The UK authorities argue that the measures proposed in Articles 8a and 8b which had been considered for inclusion in CRA 2 do not actually facilitate the reduction in reliance on ratings, as they lead to greater transparency only to other rating agencies rather than investors. So far there has been no compelling evidence base to support these specific measures as opposed to a broader transparency requirement in the areas of securitisations and covered bonds, which we would be strongly in favour of. In addition to this we are sceptical that these requirements would actually lead to greater levels of ratings of specific instruments given the risks and hurdles faced in using the data that would be available. In the event that these provisions are revisited in future legislation, there should be a full impact assessment, consultation exercise and examination of similar provisions such as the US SEC Rule 17g-5. Information disclosure for sovereign debt ratings and ability of financial institutions to assess credit risk of sovereign debt [Q11] 1.51 Freely available data undoubtedly form an important foundation of sovereign ratings. However, any sound analysis of credit risk, including that of agency ratings rests also on the use of judgemental inputs and a track record in carrying out ratings over a long period of time. We have concerns about the ability of some financial institutions to conduct internal analysis of a quality comparable to credit ratings. As explained before, less-sophisticated institutions may not have the resources to feed the relevant information into a robust analytical process at least not for all sovereigns More generally, ratings can be valuable as an input to internal analysis and as a benchmark. Indeed, this is borne out in evidence by the IMF. 16 Therefore, while all financial institutions should be encouraged to have access to a wider set of information and improve their internal analysis, we would currently question whether they all currently have the ability to produce a robust internal assessment of sovereign debt without some reference to CRA ratings. Use of external ratings in the mandates and investment policies of investment managers Flexibility clause in investment mandates [Q12] 1.53 As stated in the FSB principles, the UK authorities agree that investment managers and institutional investors must not mechanistically rely on CRA ratings for assessing the creditworthiness of assets. Some built-in discretion and flexibility is an important element to achieve this It is fairly common for UK investors to be flexible in responding to ratings changes which are relevant to their mandates. For example, certain types of institutional mandates state maximum and minimum exposures to particular credit ratings. These provide a degree of flexibility if limits are breached for technical reasons such as a downgrade, but the manager is expected to modify the portfolio over time to bring it back into line. Such flexibility also exists in some passive bond funds that track indices of corporate credit and are required to sell instruments that fall outside the index. Active managers, even where they track such an index, have generally some scope to deviate from the benchmark, and are likely to sell a holding only if they deem it to be in the medium- to long-term interests of the end-investor to do so. 16 Please see: 11

12 1.55 The UK authorities question the commercial viability of introducing additional flexibility through an explicit flexibility clause. Since the purpose of CRA ratings in investment mandates is to discipline fund managers, it will be in the interests of trustees to grant relatively little flexibility. Similarly, it may often be in the interests of fund managers to avoid excessive tracking error by responding relatively quickly to ratings changes Therefore, instead of introducing an explicit flexibility clause, due consideration should be given to identifying alternative indicators of threshold credit quality in investment mandates (see response to Question 14) and reference should be made to guidelines issued by trading associations to promote best practice among investors and investment consultancies. For example, the IMA, EFAMA, and ESF have usefully provided industry guidelines for sound investment practices with respect to credit rating agencies. 17 Limiting a proportion of portfolios that are solely reliant on external credit ratings [Q13] 1.57 The UK would oppose a rigid rule that prescribes a maximum share that is solely reliant. Ultimately, end-investors should justifiably expect their appointed investment managers to exercise judgement in their investments, rather than to respond mechanistically to ratings changes. Even a small fund manager with limited resources should be using a rating as an input rather than a sole determinant As stated in the FSB Principles, investment managers should conduct risk analysis commensurate with the complexity and other characteristics of the investment and the materiality of their exposure, or refrain from such investments. Investment managers should also be required to disclose their investment policies, including their approach to credit assessment, and submit such policies to regular Board review A proposal to consider requirements to have other factors set out in investment mandates may also merit further consideration. However, it is unclear whether it would be possible to impose such restrictions on an investment mandate that is a commercial agreement between two parties. In the event, due care would have to be taken in considering any rules in this area, as this could have a significant impact on smaller investors that do not have the resources to perform all their research internally. Alternative measures of credit risk to define the credit quality of a portfolio [Q 14] 1.60 Many of the issues considered in our response to Question 4 are also relevant to the assessment of alternative measures of credit quality for use in investment mandates. However, it may be easier to identify suitable alternative measures for this purpose, since the degree of granularity required will typically be less than for setting banks regulatory capital requirements. Investment mandates, or the performance benchmarks to which they refer, typically set broad investment-eligibility criteria e.g., investment grade versus sub-investment grade whereas bank capital requirements typically aim to achieve a much finer differentiation between credit exposures As noted in the response to Question 4, the main deficiencies of market-based criteria are that they encompass factors other than credit quality. Compared with credit ratings, there may also 17 These guidelines produced by The Investment Management Association (IMA), The European Fund and Asset Management Association (EFAMA), The European Securitisation Forum (ESF) can be found at: 12

13 be a higher risk of manipulation, and thresholds based on market prices could become important focal points, perhaps leading to gapping or cliff effects. If bond prices or CDS spreads were used as a proxy for assessing whether a bond should be removed from a portfolio or benchmark it would potentially encourage short-sellers or market-makers to drive down the price of a bond (or drive up the CDS spread) until the point at which forced selling was likely to occur Therefore, if used in investment mandates, market-based measures would again need to be used alongside, or in combination with other measures. Rolling averages of market prices would need to be constructed in a way that distinguishes between credit and liquidity or cyclical factors, and that also guards against manipulation. Although we are not aware of any mechanism that is able to do this in a robust way at present, this could be an area for further investigation. Other solutions limiting references to CRA ratings in investment policies and mandates [Q15] 1.63 One important area to target may be structured credit vehicles. These are significantly different from mainstream credit investment funds, and may be using investment mandates that require immediate realignment of a portfolio if a credit rating is changed. If this is case, care should however be taken to avoid imposing unnecessary costs on wider asset management firms. However, due account should also be taken of the fact that they constitute an important source of funding In addition, investment consultants, who typically advise clients on their investment mandates, could be encouraged to review the use of credit ratings within investment mandates. Direct regulation in this area may not be warranted but the identification of credible alternatives or the use of a wider set of ratings could be examined in promoting changes in this area. II. Sovereign Debt [Q.16-22] Our proposals on dealing with sovereign debt ratings [Q16, 22] 2.1 The UK authorities view the key reform in relation to sovereign debts to be supporting transparency and disclosure of CRA s processes and operations. This promotes CRAs to be more accountable for their ratings, improving comparability and consistency of sovereign credit ratings. More importantly, it gives investors an opportunity to ensure they understand the basis of a rating and further enables them to perform their own credit analysis, thereby reducing the mechanistic reliance on external ratings. We believe that the current EU Regulation will achieve this result (we refer to the existing transparency measures outlined in our response to Q10). 2.2 If along with transparency, the hardwiring of CRA sovereign ratings in laws and regulation (including Basel II) and their mechanistic use in private contracts is resolved, further specific changes to sovereign ratings processes may be unnecessary. The UK authorities would also recommend that further changes should ensure that sovereign ratings or alternative measures of sovereign credit quality allow prudential regulators to robustly apply a non-zero risk weight and distinguish appropriately between sovereign credits. 13

14 Proposals outlined in the Consultation [Q16-21] 2.3 Disclosure of research reports free of charge: Major CRAs currently make significant revenue from providing research to investors and the disclosure of full research reports on sovereign debt ratings could have a material commercial impact on CRAs and new entrants entering the market. If CRAs have to make ratings and research available free of charge it is quite likely that the loss of revenue resulting from this will affect the quality and quantity of staff they can employ, causing a reduction in the quality of their ratings. The reduced potential profits in the industry may also deter new entrants, harming competition. Furthermore, we would be sceptical of the marginal increase in the spread of information about the rationale for a particular rating decision, given the well-reported nature of key rating actions, including key drivers and rating rationale. 2.4 Transparency on the allocation of staff: Similarly, it is unclear how this would improve the quality of ratings. Although this might provide a degree of greater transparency around the institutional set-up of credit rating agencies, it would seem far more important to focus on increasing transparency around rating processes and methodologies. Focus on the number of analysts per rating could offer a misleading picture of the quality of sovereign debt ratings, and it is obviously the quality of the rating rather than the quantity of one input that ultimately matters to investors and debt-issuers alike. 2.5 More frequent review of sovereign debt ratings: As the Consultation notes, sovereign ratings also feed into other ratings, and because of the requirements of the Regulations, CRAs monitor sovereign debt ratings on an ongoing and continuous basis. As a result, CRAs already report sovereign outlooks/reviews/watches with a higher frequency than one year. Changes in such reviews often have a market impact even if actual rating doesn t change. Due to this, we do not view it as necessary for CRAs to be compelled to review sovereign ratings more often than already established under CRA Obliging CRAs to review sovereign debt more frequently interferes with CRAs processes and has a negative impact on the equal treatment of all ratings. We regard it is as crucial that CRAs have independent methodologies in order for them to issue independent credit risk assessments. Reviewing ratings every six months could discourage a longer, through-the-cycle type approach to assessing creditworthiness and exaccerbate the pro-cyclicality of credit ratings. 2.7 Notification period of 3 days: Concerns about market abuse and credibility are pertinent when it comes to sovereign ratings as they can have material implications for many other firms, and sovereigns are probably in a stronger position to put pressure on a CRA than a single issuer. An extended notification period to 3 days before publication of the rating seriously amplifies these concerns, and the UK authorities, therefore, oppose this proposal. It increases the risk of sovereigns seeking to influence the CRA to alter its opinion. Furthermore, a sovereign might be tempted to make use of the additional time to alter its debt issuance plans or take other action that pre-empts the rating announcement. More generally, it increases the risk of information leaks. 2.8 We note, however, that ensuring the accuracy of the facts used in the rating is extremely important, and that sovereigns are often in the best place to do that. Therefore sovereign 14