ANSWER TO IOSCO s CONSULTATION ON FINANCIAL BENCHMARKS FEBRUARY 2013

ANSWER TO IOSCO s CONSULTATION ON FINANCIAL BENCHMARKS FEBRUARY 2013 Amundi is a leading asset manager, ranking second in Europe and among the top ten in the world with assets under management above 710 billion euros at the end of September 2012. It is active in many different countries and serves a diversified clientele of retail, corporate and institutional investors through a large range of products and investment solutions. If Amundi is not a direct participant of the industry of indices it is a regular user of benchmarks produced by others. As any asset manager, Amundi relies on market indices and benchmarks in different circumstances and specifically when it invests in financial instruments referenced to standard indices or duplicates performances of a benchmark through an ETF (Exchange Traded Fund) More commonly when discussing with clients and potential clients and presenting its expertise, Amundi will make references and comparisons to indices. But this usage for reporting and marketing purposes is of a totally different nature and should be treated separately. Amundi thanks IOSCO s Task force for the quality of its work and for the opportunity it offers through the present consultation to express views on the possibility to regulate production and use of benchmarks. It shares the general objective to reduce the risk for manipulation of benchmarks and to promote higher standards in this field. Amundi will only answer to the questions on which it has a specific view and/or a direct experience as a representative of the buy-side. As a foreword Amundi wants to outline the following key messages: Amundi thinks that time is needed to let analysis and reflexion conducted by IOSCO (and others as the European Commission) lead to the mature and adequate regulation that is expected; Amundi is keen to participate to this consultation process in order to achieve a better regulation or conclude that self-regulation is more efficient; Amundi shares the opinion that transparency, responsibility, accountability are principles that should apply through all the steps of the benchmark-setting process and concern methodology, governance and supervision; in particular, the examples of interbank offered rates as Libor or Euribor evidence the need for a better organisation relying on principles that could easily be applied on a voluntary basis; Amundi is concerned with the economic consequences of new regulation introduced by ESMA s guidelines on ETF and other UCITS issues with respect to free access by investors to information on the methodology and the components of benchmarks. As subscribers to the services of the benchmark producer, asset managers have access within short delay to most of the required information. On the other hand, public at large can only get a restricted access (if any) free of charge. The case of investors in the funds that refer to benchmarks (like ETFs) has to be addressed as they should legitimately have - 1

a larger access than the public but cannot require the asset manager to be in breach of its license contract and communicate confidential information. A realistic compromise should be found or imposed by regulation on the level of necessary information accessible at a reasonable cost; The global responsibility of the administrator of the benchmark should be expressed as an overarching principle, irrespective of the fact that it delegates to third parties different functions (contribution, collection of data, calculation, publication ); The global architecture that Amundi envisions for the organisation of the industry of benchmarks includes high level principles expressed by IOSCO and applied by professional code endorsed by local authorities as recommendable but not mandatory practices and a general supervision by local or regional authorities. Amundi s responses to specific questions are given below, following the order of the consultation. Only those questions where Amundi has specific comments to add are addressed. Chapter 1 Scope 1. Do you agree with the scope of the report and intended audience? Are there other Benchmarks or stakeholders that have idiosyncrasies that should place them outside of the scope of the report? Please describe each Benchmark or stakeholder and the idiosyncrasies that you identify and the reasons why in your view the Benchmark or stakeholder should be placed outside of the scope of the report. The definition of benchmark as provided in Annex A differs from the definition expressed by European Commission in its consultation on benchmarks, last November. As benchmark industry is a worldwide business we advise different bodies to agree on a common definition. More precisely we agree with the European approach to concentrate on benchmarks that are used for financial transactions. The use of benchmarks by asset manager for reporting or marketing purposes is under the scope of distribution and clientele relationship regulations and should not be subject to any regulation of benchmarks which should concentrate on processing and publishing of benchmarks. Chapter 2 Benchmark design 2. Do you agree that the design of a Benchmark should clearly reflect the key characteristics of the underlying interest it seeks to measure? Quality and integrity of Methodologies 3. What measures should Administrators take to ensure the integrity of information used in Benchmarking-setting and that the data is bona fide? Please highlight any additional measures required where Benchmarks are survey based. Please also comment on each of the factors identified in the discussion on the vulnerability of data inputs such as voluntary submission, discretion exercised by Administrators. Are these measures adequately reflected in the discussion of roles and responsibilities of the Administrator discussed in section E? 4. What measures should Submitters implement to ensure the integrity of information provided to Administrators? Are these measures adequately reflected in the discussion of a code of conduct for Submitters discussed in section E? In particular, should Submitters submit all input data and not a - 2

selection of such data so as to maximise the representation of the underlying market? Please comment on any practical issues that compliance with such an approach may give rise to. Answer to questions 3 and 4: Amundi totally supports the organization of responsibilities as described under A.5. For an end user of benchmarks like an asset manager, it is of prime importance to know that the person it has contracted with has full responsibility for the establishment and circulation of a benchmark. Thus, delegation should not lead to lower responsibilities of the Administrator vis à vis the bench mark users. The number of third parties involved into the process should not impact this overarching principle. Contrary to the Task force s opinion (expressed in 4.3 last ), Amundi does not favor the idea that benchmark administrator should communicate with the relevant authorities on suspicions regarding the submissions they receive. Twice in their recent history, under the Terror period in 1792 and during nazi invasion in 1940s, French institutions and people have heavily suffered of denunciation and cannot conceive not to demand more than suspicion before reporting to authorities. With reference to section E.1, Amundi is concerned with the high level of details required to be made public. Even if it shares the view that transparency is mandatory with a great level of details (see below) it feels that there should be a difference between information posted for access by the public and documents made available to professionals and/or authorities. It expects all details to be available in the framework of a due diligence, eventually with a confidentiality provision. But it thinks that a lesser degree of transparency should be made directly accessible. Transparency of Benchmark methodologies 5. What level of granularity with regard to the transparency of Methodologies would enable users to assess the credibility, representativeness, relevance and suitability of a Benchmark on an on-going basis and its limitations with respect to their intended use? Relevant factors could include; criteria and procedures used to develop the Methodology, type of data used, how data is collected, relative weighting of data used, how and when judgement is used, contingency measures (e.g., methods when transaction data is unavailable etc), publication of information supporting each Benchmark determination, etc. Please provide examples where you consider there are currently significant gaps in the provision of this information. Transparency of contingency provisions for episodes of market disruption, illiquidity or other issues 6. What steps should an Administrator take to disclose to Market Participants and other stakeholders the contingency measures it intends to use in conditions of market disruption, illiquidity or other stresses? Transparency over changes to the Methodology 7. What steps should an Administrator take to notify Market Participants of material changes to a Benchmark Methodology (including to Benchmark components) and to take their feedback into account? 8. How often should the Administrator review the design and definition of the Benchmark to ensure that it remains representative? Answer to questions 5 to 8: Transparency of methodology is key to a proper assessment of the quality of a benchmark. It is not sufficient to rely on the good reputation of an administrator to trust its benchmarks as, for illustration purpose, it proved to be more than dangerous to rely - 3

on credit rating agencies to negotiate reloading of some complex blind securitization schemes The point is to make a distinction between a total transparency with a high degree of precision on the general procedures, including the cases of emergency and market disruption, on one hand and the day to day management of the benchmark on the other hand. When acting on the basis of the disclosed procedures, the administrator of a benchmark should be granted to use flexibilities provided for without advance notice or prior consultation. Ex post information might be sufficient. However, advance notice is a requirement for the changes in components list or weighting as well as for any item necessary for a replication of the benchmark. Consultation of stakeholders might be appropriately organized within the Oversight Committee that is discussed later. We suggest that representatives of the buy-side participate to the Committee. Governance 9. The Consultation Report discusses a number of potential conflicts of interest that may arise at the level of the Submitters, between Submitters at different entities, and between Submitters, Administrators and other third parties. Are there other types of conflicts of interest that have not been mentioned that you consider may arise? If so, how best should these conflicts of interest be addressed? Are the measures discussed in the Consultation Report sufficient to address potential conflicts of interests at the level of the Submitters, between Submitters at different entities, and between Submitters, Administrators and other third parties? 10. Do you agree that the Administrator should establish an oversight committee or other body to provide independent scrutiny of all relevant activities and management of conflicts of interest? Please comment if and why any different approaches might be appropriate for different kinds of Benchmarks. What is the minimum level of independent representation this committee or body should include? Answer to questions 9 and 10: Amundi totally supports the idea of establishing an Oversight Committee to enhance transparency and credibility of the concerned benchmark. It feels that this will be a major step to a better governance and should be quite efficient in appropriately addressing conflicts of interests and other issues relating to benchmark administration. Amundi considers that representatives of the buy-side should be invited to join such committees as their advice as users should be considered. It acknowledges, however, that participation to the Committee may result in a new type of conflict of interests that could be taken care of by an appropriate code of good conduct applying to the members of the Committee. This code should at least insist on confidentiality and absence of use of the participation as a marketing tool. The buy-side could also be represented to the Committee by professional associations. Accountability 11. Should the Submitters establish accountability procedures to assess their compliance with operational standards and scrutiny of Benchmark submissions? 12. Are the measures discussed in the Consultation Report (e.g. Audit Trail, external audits and requirement for regulatory cooperation) sufficient to ensure the accountability of Submitters? Should additional mechanisms be considered? 13. How frequently should Submitters be subject to audits? Should these be internal or external audits? - 4

Accountability of the Administrator 14. Are the measures discussed in the Consultation Report (e.g., complaints process, Audit Trail, external audits and requirement for regulatory cooperation) sufficient to ensure accountability of the Administrator? Should additional mechanisms be considered? 15. If recommended, how frequently should Administrators be subject to audits? Should these be internal or external audits? 16. Is public self-certification of compliance with industry standards or an industry code another useful measure to support accountability? This approach might also contemplate explanation of why compliance may not have occurred. If so, what self-certification requirements would make this approach most reliable and useful to support market integrity. Answer to question 16: Self-certification of compliance with industry code presents in Amundi s view the advantage of enabling a standard approach and a common level of quality worldwide. It gives users a framework for analyzing and scrutinizing on common principles the quality of benchmarks and benchmark administrators. In that respect it is more effective than a regulatory approach that might differ from one country or one region to another. This applies not only with a view to the relationship of the administrator with submitters (which should be contractually defined with enforceable value) but more generally to all the process of benchmark setting and publishing. Code of conduct for Submitters 17. The Consultation Report discusses elements of a code of conduct for Submitters. Are the measures discussed (e.g., adequate policies to verify submissions, record management policies that allow the Submitter to evidence how a particular submission was given, etc.) sufficient to address potential conflicts of interest identified or do you believe that other control framework principles should be added? Answer to question 17: Amundi supports the idea of a code of conduct for submitters, but thinks that the code will only state general principles and will not be sufficient to address all issues. Thus, it stresses that a contract should be signed between the administrator and each submitter that will include more detailed and better adapted provisions. 18. What would be the key differences in the code of conduct for Benchmarks based on different input types, for example transactions, committed quotes and/or expert judgement? Chapter 3 Approaches to enhanced oversight 19. What are the advantages and disadvantages of making Benchmark submissions a regulated activity? 20. What are the advantages and disadvantages of making Benchmark Administration a regulated activity? Answer to questions 19 and 20: On one hand, many submitters and administrators are already part of regulated institutions. It may create a discrepancy and develop an uneven playing field or an incentive to circumvent legislation not to subject the benchmark industry to direct regulation and oversight. In any case, it would not be sensible to regulate submitters only and not administrators. The proper way to proceed consists in regulating administrators and have them control submitters. That leads to the following suggestion: direct regulation of administrators and indirect regulation of submitters. - 5

On the other hand, to have a central data base of all benchmark and indices providers, however appealing, does not sound realistic. And requiring them to register is not appropriate as some official bodies, for example INSEE in France, produce with high professionalism indices that might no longer be used by lack of registration. Regulation may be counter-productive and divert participants to use freely accessible robust benchmarks. Clients best interest is definitely not to increase costs through excessive regulation. Amundi considers that the objective is to reach harmonized framework worldwide and suggests that a declination through professional code of conduct of international principles established by IOSCO could be a correct approach. Further down in granularity, contracts signed with the Administrator should deal with specific details. Freedom to continue relationship with nonparticipating administrators should be left open, on a case by case basis. Thus a non-exclusive regulatory framework should be considered as a safe harbor for a proper issuance and use of benchmarks. Participating administrators could be submitted to effective control of compliance with the code by local or regional authorities (provided they endorsed it). 21. Do you agree with the factors identified for drawing regulatory distinctions? What other factors should be considered in determining the appropriate degree of oversight of Benchmark activities (discussed in Chapter 3)? Please provide specific recommendations as to how the distinctions discussed in Chapter 3 should inform oversight mechanisms. 22. What distinctions, if any, should be made with regard to Benchmarks created by third parties and those created by regulated exchanges? Answer to questions 21 and 22: Amundi agrees with the suggestion that distinctions should be introduced among benchmarks. As an example, it draws IOSCO s attention to the fact that benchmarks used as underlyings for futures and options listed on regulated markets should be considered as standard products and should not require any due diligence on the part of the investor. Asset managers should be able to rely on the examination of the regulated exchange to consider the benchmark as adequate. As a further example, Amundi believes that benchmarks used in a transaction with retail investors should be more closely supervised than benchmarks specifically asked for by an institutional investor. 23. Assuming that some form of enhanced regulatory oversight will be applied to an asset class Benchmark, should such enhanced oversight be applied to the Submitters of data as well as the Administrator? 24. What are the considerations that should be taken into account if the Submitters to a Benchmark operate in an otherwise unregulated market (e.g., physical oil, gold or agricultural commodity markets) and are not otherwise under any obligation to submit data to an Administrator? 25. Do you believe that a code of conduct, either on its own or in conjunction with other measures outlined within the report, would provide sufficient oversight to mitigate the risks that have been identified in Chapter 2? What measures should be established in conjunction with a code of conduct? For which Benchmarks is this approach suitable? 26. What other measures outlined in the report, if any, should apply in addition to a code of conduct? If you believe a code of conduct, either on its own or in conjunction with other measures outlined within the report, would provide sufficient oversight to mitigate the risks that have been identified in Chapter 2, what type of code of conduct should apply (e.g., a voluntary code of conduct, an industry - 6

code of conduct submitted to and approved by the relevant Regulatory Authority, a code of conduct developed by IOSCO, etc.)? 27. Do you believe that the creation of a Self-Regulatory Organisation (.e.g., one that exercises delegated governmental powers) and itself subject to governmental oversight, whether or not in conjunction with industry codes is a viable alternative for sufficient oversight and enforcement to mitigate the risks that have been identified in Chapter 2? For which Benchmarks is this approach suitable? What if any complementary arrangements might be necessary, such as new statutory obligations or offences for Administrators and/or Submitters? 28. Do you believe that, for some Benchmarks, reliance upon the power of securities and derivatives regulators to evaluate products that reference a Benchmark or exercise their market abuse or false reporting powers creates sufficient incentives for the Administrator to ensure sure that Submitters comply with a code of conduct? Answer to question 28: Amundi believes that powers of enforcement of authorities and jurisdictions in case of market abuse or false reporting are very efficient to impose compliance to good practice principles. In the case of Euribor and Libor alleged manipulations, prosecution has certainly enhanced improvement of former habits. It shows that the area is not exempt of regulations and that abuses and fraud can be properly dealt with. There is then no urgency to act and time is necessary to achieve the best balance between risk, cost and innovation. 29. Do you believe that users of a Benchmark, specifically, the users who are regulated or under the supervision of a national competent authority should have a role in enhancing the quality of Benchmarks? Which form should this role take: on a voluntary basis (e.g. the user being issued a statement that will only use Benchmarks that follow IOSCO principles), or on a compulsory basis (e.g., the competent authority could request that users who are registered under their jurisdiction should only use Benchmarks that fulfil IOSCO principles)? Answer to question 29: regulated users such as asset managers do participate to the enhancement of the quality of benchmarks through the due diligence inquiries they conduct when using a benchmark. Furthermore in Europe ESMA published specific requirements in that respect when issuing Guidelines on ETFs and other UCITS issues. Amundi feels that benchmarks used as underlying for listed future and option contracts should be validated at the level of the regulated exchange or MTF that lists them. A voluntary approach is recommended for the asset management industry as there are many types of clients and different situations that cannot be encompassed by a mandatory approach. For example if Amundi supports the new European regulation on UCITS that are aimed at retail investors, it considers that some flexibility is necessary when dealing with institutional clients that may have very specific requirements in terms of benchmarks. Furthermore, Amundi stresses the fact that the use of benchmarks for reporting and marketing purposes should not be considered as an activity to be regulated otherwise than through clientele relationship and distribution regulations. European directives and regulations are quite explicit in that respect. Finally, the question of cost resulting from regulatory requirements appears to be quite significant when discussing with benchmark administrators. They seem very keen to charge extra fees for circulation to the holders of a fund of information relative to the composition, the methodology or the governance of a benchmark. A reasonable cost approach should be part of the principles applying to benchmark administrators, and transparency should also apply to their pricing policy. - 7

Chapter 4 Data sufficiency 30. Do you agree that a Benchmark should be anchored by observable transactions entered into at arm s length between buyers and sellers in order for it to function as a credible indicator of prices, rates or index values? How should Benchmarks that are otherwise anchored by bona-fide transactions deal with periods of illiquidity due to market stress or long-term disruption? 31. Are there specific Benchmarks for which you consider that observable transactional data is not an appropriate criterion or the sole criterion? If so, please provide a description of such Benchmarks and what value you think such Benchmarks provide? 32. What do you consider the limitations or value in Benchmarks referencing asset classes and underlying interests where there is limited liquidity? Please describe the uses and value of such Benchmarks in the financial markets. 33. Do you agree that the greatest weight should be given to transactions in the construction of a Benchmark and that non-transactional information should be used as an adjunct (e.r., as a supplement) to transactions? Answer to question 30 to 34: reference to effective transaction prices should be preferred when available but cannot be considered as the only valid source of data. There are many instances where other data are more appropriate: bond prices are better reflected on bid/ask quotations continuously updated than on transactions prices; emerging market securities are another example where quotes are more significant than price. It is true that on highly liquid markets with many participants there is no justified alternative to transaction price, but it does not mean that benchmarks should be limited to that kind of markets. Benchmarks are efficient ways for investors to judge of the diversification of their investments and to properly adapt them. Innovation in the field of benchmarks may lead to an alternative kind of investment, be it through an ETF or through direct investment, for the benefit of the investor. More than liquidity the accuracy of the price discovery system as mentioned p 40 is the key reference : it will enable benchmarks to develop on less liquid markets. However these benchmarks should be submitted to specific care as they may be more easily subject to manipulation or conflicting interests. 34. What factors and how often should Administrators (or others) consider in determining whether the market for a current Benchmark s underlying interest is no longer sufficiently robust? What effective methods of review could aid in determining the insufficiency of trading activity within the market for a Benchmark s underlying interest? Transition 35. What precautions by Benchmark Administrators, Submitters, and users can aid Benchmark resiliency during periods of market stress, mitigating the potential need for market transition? 36. What elements of a Benchmark living will, drafted by a Benchmark Administrator, should be prioritised? 37. By what process, and in consultation with what bodies, should alternatives be determined for Benchmark replacement? - 8

38. What characteristics should be considered when determining an appropriate alternate Benchmark? (Examples below) Should any of these factors be prioritised? o Level and Type of Market Activity o Diversity/Number of Benchmark Submitters o Length of historical price series for the Benchmark alternative o Benchmark Methodology o Existing regulatory oversight o Existing enforcement authority o Volume, tenors and contract structure of the legacy trades 39. What conditions are necessary to ensure a smooth transition between market Benchmarks? 40. What considerations should be made for legacy contracts which reference a Benchmark in transition? To what extent does a substantive legacy book preclude transition away from a Benchmark? What provisions can be included in [new and existing] contract specifications which would mitigate concerns if and when a Benchmark transitions occurs? 41. How should a timeframe be determined for market movement between a Benchmark and its replacement? What considerations should be made for: o Altered regulatory oversight? o Infrastructure development/modification? o Revisions to currently established contracts referencing the previous Benchmark? o Revisions to the Benchmark Administrator? o Risk to contract frustration Answer to questions 34 to 41: Amundi shares the view that benchmarks and indices are brought to light one day and may die another day. It is as important to organize for innovative creation of new benchmarks as to make sure they are still representative. Stress resilience and market representation of the index or benchmark are essential characteristics and this last section of the consultation is not the least interesting. For an asset manager it is on one hand very difficult, except for benchmarks used for reporting purpose as reference to compare performances or define an investment universe, to ask investors to change reference. It can effectively occur mainly when the previous benchmark is no longer published. On the other hand, the asset manager can easily measure benchmarks that are representative of market segments or investment strategies of interest to investors at a given time: new subscriptions are directed to funds that refer to them. And it may change. Thus asset managers assess more often the interest of a fund within a product range than the interest of a benchmark. But at the end of the day it comes down to the same conclusion: products (funds or benchmarks) have to be monitored and their adequacy regularly reassessed. Usually commercial and legal considerations help to make that choice, but defective production of the benchmark is a case that cannot be argued against. Regulation should again limit itself to general principles in this field. - 9

Contact at AMUNDI : Frédéric BOMPAIRE Public Affairs 90, boulevard Pasteur 75015 PARIS 33 (0) 1 7637 9144 frederic.bompaire@amundi.com Siège social : 90, boulevard Pasteur - 75015 Paris - France Adresse postale : 90, boulevard Pasteur - 75730 Paris Cedex 15 France Tel. : +33 (0)1 76 33 32 00 - Fax: +33 (0)1 76 33 68 00 - amundi.com Société Anonyme au capital de 578 002 350 euros - 437 574 452 RCS Paris Société de Gestion de Portefeuille agréée par l'amf (Autorité des Marchés Financiers) sous le n GP 04000036-10