Subject: Chief Risk Officer Forum Feedback on CEIOPS-CP-04/05

Transcription

1 30 September 2005 The Chief Risk Officer Forum Subject: Chief Risk Officer Forum Feedback on CEIOPS-CP-04/05 Henrik Bjerre-Nielsen Chairman Committee of European Insurance and Occupational Pension Supervisors Sebastian-Kneipp-Strasse 41 Frankfurt, Germany Dear Mr. Chairman, I am pleased to present to you in this letter The Chief Risk Officer Forum s (CRO Forum) feedback on CEIOPS-CP-04/05, Draft Answers to the European Commission on the second wave of Calls for Advice in the framework of the Solvency II project. 1 This letter re-iterates and clarifies those considerations previously presented to you and your colleagues by the CRO Forum at the 7 September CEIOPS public hearing. It is important to note that we have focused our feedback on those issues that we see as being key to development of the Solvency II framework (e.g. purpose and approach to liability valuation and solvency capital) while the advice presents details on many other necessary but less fundamental issues (e.g. specific data requirements) that will follow from the articulation of the key issues. Therefore, an absence of feedback from the CRO Forum on a given issue at this stage should not be misinterpreted as agreement, but instead a prioritisation of the key issues. We thank you again for your continuing engagement with industry, and in particular with our group of risk professionals from leading European insurers. The rest of this document is structured in the following sections: I. General overview A. Key areas of agreement B. Key areas of disagreement II. Detailed feedback on aspects of the draft answers to the Calls for Advice (CfA) 1. 1 For brevity, we will refer to CEIOPS-CP-04/05 as the advice or your advice and the individual calls for advice as CfA throughout the remainder of this document.

2 Page 2 In addition, we have provided as an Appendix the detailed responses from CRO Forum members to individual paragraphs in the calls for advice. Those comments have been summarised in this letter. I. General overview A. The CRO Forum strongly supports many of the concepts presented in the CEIOPS advice Our interpretation of the advice was, in general, very positive. There are several key concepts presented in the advice that The CRO Forum strongly supports: Harmonising supervisory approach across geographies (CfA 14) Moving towards market-consistent valuation for the regulatory balance sheet (CfA 7, CfA 8) Moving towards a truly risk-based approach for setting capital requirements (CfA 10, CfA 11) Incentivising good practice in risk management and continuous improvements (e.g. use of internal models) (CfA 11) Introducing Group Lead Supervisor concept (CfA 18) Embedding risk management in business decision-making by prescribing active involvement of senior management and board of directors (CfA 16) and the use test for internal models (CfA 11) These concepts are consistent with the current or developing risk management frameworks at CRO Forum member companies, who in turn are leading practitioners in risk management. It is our view that each of these is a critical component of the Solvency II framework. B. The CRO Forum is very much opposed to some concepts presented in the advice; these should be reconsidered The CRO Forum also wishes to bring to your attention what we perceive to be material inconsistencies of and/or shortcomings in the advice as written and the objectives of Solvency II; we believe that each of these threatens to undermine the positive aspects of the proposed framework and deserves further consideration:

3 Page 3 Solvency capital requirements are only partially aligned with risk Insolvency and solvency are not well defined; the SCR should be defined only when solvency (and confidence level for solvency) is clearly defined (CfA 7, CfA 8, CfA 10, CfA 11) Excessive prudence and detachment from true risk in setting technical provisions (CfA 7, CfA 8) SCR and risk margin concepts not presently linked (CfA 10) Risk mitigation tools not sufficiently recognised in MCR / Standard SCR (CfA 10, CfA 12, CfA 15) For liability valuation/setting technical provisions, swap curves should be used as the basis for discounting (CfA 7, 8) Safety levels are duplicative and possibly contradictory Excessively prescriptive asset restrictions duplicate the purpose of the SCR (CfA 9) Critical levels appear to be linked to Solvency I system rather than QIS results (CfA 15) Governance / supervision is insufficiently clear (CfA 18) Lead vs. local supervision does not distinguish between model review Group diversification Intervention Internal model review should be undertaken by lead supervisor only; information sharing on such validation may be shared amongst the supervisors (solo and lead) These concepts will influence the foundation for the Solvency II framework, and as such we believe it is imperative to resolve them. The CRO Forum is committed to working with CEIOPS to do so. II. Detailed responses, by CfA In this section we address each of the calls for advice (CfA), highlighting our key questions, concerns and suggestions for each. CfA 7 - Technical provisions in life assurance The proposed risk margin introduces excessive prudence The advice proposes an economic approach to the valuation of life insurance liabilities meaning where available, market prices are used. This concept is core and one on which we

4 Page 4 conceptually agree. However the implementation of the approach, as suggested by the advice, falls short of this economic aim, introducing excessive prudence and raising several material issues. Several non-economic or excessively prudent valuation elements such as using government rates (7.18), artificial floors on valuations (7.36) and excessive one size fits all approach to setting risk margins ( ) Technical provisions should not contain prudence in excess of market consistent risk margins; that is role of solvency capital (7.31) Conversely, the CRO Forum defines economic value as, The present value of future cash flows, valued in such a way as to be consistent with current market prices where these are available. 2 This means that all liabilities that depend on market returns should be valued based on the arbitrage-free principles of derivative pricing theory. Moreover, all fixed cash flows should be valued using the current term structure of interest rates. For unhedgeable risks that cannot be fully diversified (such as certain large-loss insurance risks, or major parameter risks), a market value margin should be applied to best-estimate cash flows in order to ensure that their discounted value is consistent with the price at which the liabilities could be transferred to a willing, rational, diversified counterparty. Note, many companies have developed approaches using a market value margin as part of an integrated internal model approach, which ensures that there is no double counting of risk. For risks which are hedgeable, the hedge price should be used. No market value margin should be applied The CRO Forum believes that it is critical to reconcile the objective and implementation of these core components of the Pillar I solvency assessment. The form of and interaction between each of reserves, margins, solvency capital are to date unclear and in many ways appear to be in conflict with the aims of the Solvency II project. We strongly urge CEIOPS to prioritise these as development issues Principles for Regulatory Admissibility of Internal Models, The Chief Risk Officer Forum (10 June 2005)

5 Page 5 CfA 8 Technical provisions in non-life insurance Non-life technical provisions rely more heavily on mark-to-model approaches but the advice imposes many structural constraints The advice proposes for non-life insurance liability valuation an economic approach similar to that for life insurance liabilities (CfA 7). Our concerns with this advice are consistent with those raised for life insurance; namely that there are several non-economic aspects to the proposed approach which undermine its effectiveness: Several arbitrary approaches suggested for setting risk margins ( ) e.g. standard confidence intervals (e.g. 75%); adjustments to discount rates or the suggestion that liabilities may not be discounted at all (8.112) Several suggestions of non-economic or excessively prudent valuation elements such as use of government rates for discounting or floors to liability values e.g. present level of provisions ( ) In addition, the advice suggests certain structural requirements for model methods and parameters that we believe are impractical, for example: Requiring future claims inflation different from past experience, where differences may not be justified or feasible (8.61) For reserve calculation, requiring use of two approaches and imposing traditional methods where this is not appropriate for some business 3 (8.104) CfA 9 - Safety measures Prescriptive rules for assets do not promote sound risk management, inhibit innovation and put the insurance industry at a disadvantage The advice on safety measures presents what is in our opinion a very prescriptive regime. In particular, We believe that a prescriptive approach for assets duplicates the role of the SCR. That is, the amount of required capital should increase with increased risk rather than seeking to reduce the available capital by dis-allowing assets (9.91, ) 1. 3 For example, asbestos claims

6 Page 6 On asset admissibility, we recommend a principles-based approach rather than a list of admissible assets (9.100). Were a list approach adopted, we prefer a negative list approach. With respect to concentration limits, overly prescriptive rules likely to lead to doublecounting or risk that is already captured in the SCR and penalise good risk management ( ). Instead, we propose a transparent approach to reporting exposures and sources of available capital. Suggestions that admissible assets should only be those traded in public, liquid or deep markets would restrict the investment universe, and could be interpreted to mean government bonds are the only acceptable assets ( ) We strongly believe that the MCR should not be excessively prudent so as to interfere with the SCR, so further work should be undertaken on calibration. ( ) CfA 10 - Solvency capital requirement - the standard formula (life and non-life) Risk margins and solvency capital do not appear to be aligned The CRO Forum recognises that both standard formulae and internal models are necessary components of the Solvency II framework. Our interest in the form of the standard model(s) stems from a desire to incentivise sound risk taking and orderly market pricing and competition. Moreover, we want to ensure that the form of the standard model does not inhibit the development of internal models, as there are some conceptual issues that arise in the advice for standard formulae which will also be linked to the internal models requirements, and as such these need to be resolved; namely SCR should be defined as the amount of capital necessary on top of technical provisions calculated on an economic basis to ensure that Total Assets in one year will be above technical provisions in one year in 99.5% of the cases. It is critical to give this definition to avoid double counting. The SCR and risk margin calibrations must be linked Danger of double-counting unless SCR is be determined relatively to discounted best estimates plus any risk (or market value) margin (10.27) Calibration confidence interval and the risk measure to use is not clear (10.26) Unclear as to links with proposed stress tests (10.111) There should be clearer incentives for risk management e.g. risk mitigation Premiums / reserve factors need to be net of reinsurance or other types of risk transfer e.g. securitisation ( )

7 Page 7 ALM risk capital requirements need to be net of hedging and other risk mitigation techniques ( ) Factor-based approaches (e.g. for ALM risk, embedded options and non-proportional reinsurance) do not capture the impact of risk mitigation ( , ) The standard model should incorporate risk diversification ( ) We strongly oppose an overarching framework that requires standard and internal models to be completely consistent, as this effectively removes incentives to advance internal modelling frameworks. CfA 11 - Solvency capital requirement - internal models (life and non-life) and their validation Promotion of internal models is fundamental to advanced practitioners but their benefits could be undermined by overly prescriptive structural requirements The development and use of internal models is, in our view, a critical component of a sound risk management framework; this is particularly true for insurance groups. With this in mind, the CRO Forum undertook a study of internal modelling frameworks of its member companies and proposed a set of admissibility criteria; we shared the results of the study with CEIOPS in June Our interpretation of the advice (11.24) is that it proposes to restrict internal models in a way that mirrors, and is indeed calibrated to, the standard formula. Although this is similar to the approach taken in the Basel 2 IRBA framework for banks, we do not support such an approach for insurance companies, where the risks are more varied and products less standardised than in banking. As described in our feedback to CfA 10, we believe such a prescriptive approach would stunt the positive developments/innovation in risk management and is untenable. Moreover, the validation of internal models for groups will function more efficiently if such validation is undertaken by the lead supervisor for the group (see CfA 18 for our view on the lead supervisor role). In addition to efficiency, the validation of the internal modelling framework will almost certainly be more robust if undertaken by a single supervisor, looking across the group, rather than by individual, local supervisors. We are strongly opposed to solo validation See Principles for Regulatory Admissibility of Internal Models, The Chief Risk Officer Forum (10 June 2005)

8 Page 8 Separately, we have concerns regarding the proposed Adjusted SCR. While we can appreciate the need for supervisory discretion which is indeed the basis for Pillar 2, and would in theory be supported by the introduction of the Adjusted SCR we believe there is a need for clearer guidance as to the circumstances under which this may be invoked. In essence, we want to be in a position to manage our risks in such a way as to avoid the imposition of such an additional capital constraint. Moreover, we believe there is a danger that such discretion could be applied differently in different Member States, and thus undermine the aim of supervisory harmonisation; clear supervisory guidelines should be developed to ensure that like risks are treated equally irrespective of the country of supervision. In summary, our key concerns regarding CfA 11 consist of the following: Approach to internal models is overly prescriptive, creating a variant of the standard formula (11.24, 11.68) Clarification desired on basis for supervisor applying Adjusted SCR (11.67) Clear European supervisory guidelines needed to ensure Adjusted SCR applied evenly ( , ) CfA 12 Reinsurance Reinsurance is well-considered in the advice but care must be taken not to discourage risk mitigation In general, it is our impression that the advice on reinsurance is a sound starting point, but that it requires further consideration in at least one important respect; it is an incomplete approach to the recognition of risk mitigation within the MCR and Standard SCR In summary, our key concerns regarding CfA 12 consist of the following: The extent to which risk mitigation is taken into account should not depend on the functional form of the MCR. It is imperative that a practical approach be found for covers that are capable of being readily incorporated (e.g. net quota share treaties) into the standard MCR and that entities are encouraged to apply a more risk based (internal models) approach (e.g. where MCR is equal to a percentage of SCR based on the internal model) where the standard formula does not appropriately capture the risks. (12.24) We observe that there is a risk that a double standard may be applied in the treatment of credit risk without full recognition of risk mitigation. If the credit risk associated with the

9 Page 9 risk mitigation instrument is firmly a part of Pillar I, then by definition the protection afforded by it must also be given explicit and unequivocal credit in Pillar I. (12.28) In terms of measuring the impact of risk mitigation, there should be allowance for use of either VaR or TailVaR risk measures 5, as each can be practically be used to evaluate many risk mitigation structures (12.29) Unless these issues are resolved they could serve to discourage risk mitigation and more generally sound risk management. CfA 13 - Quantitative impact study and data related issues The QIS is welcomed but the objectives must be clear and findings need to be considered in their entirety We wish to raise the following points with respect to the quantitative impact study, namely: We desire clarification on specific objectives of the various QIS; participation is not a small undertaking for any company and we will be in a better position of the objectives are agreed in advance We would discourage the drawing of incomplete observations from a single element of QIS results, without regard for the methods employed and outcomes of the over-arching study We support the participation of a range of companies, both small and large and in various lines of business CfA 14 - Powers of the supervisory authorities The scope of supervisory powers is reasonable but harmonisation should not be jeopardised by supplemental national objectives Our interpretation of the advice on supervisory powers is that these are largely consistent with practices in most of the Member States in which CRO Forum member companies are domiciled. To the extent the articulation and formalisation of these powers creates a consistent framework across Europe, we support the advice A given company would be expected to use a single risk measure, however we believe that either VaR or TailVaR are acceptable choices

10 Page 10 Nonetheless, there are a few specific areas of the advice on which we wish to provide feedback: The advice provides discretion to national authorities, allowing them to vest in their supervisors additional powers other than those outlined here; we accept this but do not wish for national objectives to jeopardise the Solvency II aim of harmonisation (14.19) We propose the additional role of technical arbitrator for those cases where the supervisor and the undertaking s experts cannot agree on technical grounds We believe that aspects of due process, rule of law and proportionality need further consideration ( ) We believe that hiring and firing should be responsibility of the Board rather than the supervisor (14.38) CfA 15 - Solvency control levels Control levels should adequately address risk mitigation Our general impression of the advice on control levels is that the ladders framework is reasonable, and that consequences appear appropriate for the conditions under which they are invoked. However, the advice also proposes specific implementation and calibration, elements of which in our opinion require further consideration. Namely: The role of risk mitigation should also be considered in ladders e.g. SCR on gross basis ignores risk mitigation (15.14) Critical levels should be calibrated to QIS results and concerns for policyholder protection rather than linked to Solvency I system (15.28) Control levels above the adjusted SCR represent an excessive layer of supervision (15.29) CfA 16 - Fit and proper Emphasis should be on composite, required make-up of professional skills and expertise rather than individual traits The advice on conditions for fit and proper is overly narrow in certain respects, namely:

11 Page 11 In general, the emphasis should be on composite, required make-up of professional skills and expertise rather than individual traits ( ) Board candidates with financial services knowledge should suffice, rather than restricting candidates to those with specific insurance experience; this is particularly restrictive for those groups that have both insurance and banking operations and also discourages the sharing of insights and management practices from other, analogous product and customer markets (16.28) The national language requirement is overly restrictive, and impractical for large groups operating in many countries (16.35) CfA 17 - Peer review Peer review amongst supervisors is welcomed as it promotes harmonisation of practice Our impression of the advice on supervisory peer review is that this is a welcome component of the Solvency II framework. We have two suggestions as to the implementation: For the permanent review panel, industry views should be taken into account as this will provider a user perspective and also supports collaboration between supervisors and industry ( ) Practically speaking, requiring peer reviews related to single entities operating a specific business in a national context may be of limited benefit to supervisors since mutual learning effect might be limited in these cases (17.28) CfA18 - Group and Cross Sectoral Issues Group Lead Supervisor concept is critical for internal model review The CRO Forum has dedicated significant effort and resource to informing the framework for group and cross sectoral issues 6 ; these issues are of vital interest and importance to us. We are pleased to see that the advice proposes some important, positive aspects (e.g. group lead supervisor, recognition of diversification); however, in our view it does so incompletely (e.g. does not recognise diversification across entities yet suggests that concentration penalties would apply across entities) See for example, A framework for incorporating diversification in the solvency assessment of insurers, The Chief Risk Officer Forum (10 June 2005).

12 Page 12 In particular, we wish to highlight the following areas as requiring further consideration: Lead vs. local supervision should distinguish between activity ( ) The role of the group supervisor is critical to the efficient functioning of the framework, but the advice does not go far enough to mandate and empower such a role. Recognising the sensitivity of Member States to ceding authority to other Member States who might take on the role of group supervisor for a given insurance group, we think it is important to distinguish between activities Model review: should be responsibility of group supervisor Intervention: important role for local supervisor Pillar I should recognise diversification effects across the Group ( ) As shown in our diversification study, 7 to correctly assess the effects of risk concentration and diversification in an insurance Group, it is critical to have a Group level capital assessment that is not simply a sum-of-standalone capital assessments of each of the solo entities within the Group. Moreover, given the extent to which diversification effects can vary across different insurance Groups, based on their risk profile, business mix, geographical profile, corporate structure and risk management practices, it is difficult to define a set of standard factors that can appropriately capture the effects. Instead, the Group level capital assessment should aim to explicitly capture the effects of risk diversification and capital through a realistic, risk-based model. Pillar I should recognise diversification benefits in available capital to meet solo SCR ( ) In our diversification study 8 we presented a framework for recognising diversification in both the solo and group solvency assessments. One important component of that framework was that for the solo entity their SCR calculation would be unaffected by diversification effects in other entities (i.e. standalone insurers and solo entities that are part of groups would be treated equally). However, groupwide diversification benefits do have economic value, and in our framework we propose that a group could pledge these benefits as available capital to satisfy solo capital requirements. As described in our study, recognition of this type of capital support is already recognised in the setting of capital requirements for banking subsidiaries through Article 69 of the Capital Requirements Directive for Banks 9 although 1. 7 Ibid p Ibid pp Proposal for DIRECTIVES OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL Re-casting Directive 2000/12/EC of the European Parliament and of the Council of 20 March 2000 relating to the taking up and

13 Page 13 we believe it is important for this Directive to be extended, for both banks and insurers, to apply across Member States 10 and recognised territories outside the EU. Not recognising such instruments not only reduces the flexibility of insurance groups to use their full group balance sheet to support individual entities, but also at present places insurance groups at a competitive disadvantage to banking groups. There is a need for explicit, consistent and comprehensive recognition of internal and external risk transfer in terms of MCR and SCR determination at the Group level. (18.44) For validation of internal models, as mentioned in earlier feedback, we support internal model validation by the lead supervisor but oppose validation by each of the solo supervisors ( ) The suggestion that capital be distributed adequately within the group seems a redundant requirement, where the role of the MCR and SCR are to define the required capital and separately there are guidelines for admissibility of assets as available capital. Where these requirements are met, then by definition capital is adequately distributed (18.32) There is an apparent asymmetry in group issues e.g. concentration penalties and limits applied across group but diversification effects may be ignored (18.41) * * * * * We hope that these views are a useful contribution to the continuing dialogue on the Solvency II project. We look forward to continuing to working closely with CEIOPS towards the development of a prudential framework that aligns regulation with best practice developments in risk management. 2. pursuit of the business of credit institutions and Council Directive 93/6/EEC of 15 March 1993 on the capital adequacy of investment firms and credit institutions. Brussels, , COM(2004) 486 final 10 Article 69 currently applies in situations where both the subsidiary and the (parent) credit institution are subject to authorisation and supervision by the same Member State.

14 Page 14 Signed: Aegon N.V. Allianz AG Assicurazioni Generali S.p.A. Aviva PLC Axa Group Converium Ltd. Fortis ING Groep N.V. Munich Reinsurance Company Prudential PLC Swiss Reinsurance Company Winterthur Group Zurich Financial Services