May 21, 2008 Document 208032 To: Katy Martin, Risk Margin Working Group, International Actuarial Association (IAA) re: March 24, 2008 Exposure Draft Measurement of Liabilities for Insurance Contracts: Current Estimates and Risk Margins The Canadian Institute of Actuaries (CIA) welcomes the opportunity to comment on the reexposure draft prepared by the IAA s Risk Margin Working Group (RMWG) on the measurement of liabilities for insurance contracts. We applaud the work that has been done. Though this paper was prepared in response to a request of the IAIS in the context of solvency valuations, we believe it will be equally valuable as input to the IASB in its deliberations on general purpose financial reporting for insurance contracts. The CIA strongly supports the efforts to achieve a consistent approach to the measurement of liabilities for regulatory and general purpose reporting. For over 15 years in Canada, we have worked effectively with a single set of accounts for both regulatory and general purpose reporting. We would be pleased to share the benefits of our experience in support of these efforts. The CIA hopes that its comments provided herein will be of value to the IAA. We appreciate the opportunity to contribute to the development of this important document, and are eager to provide any further explanation of the positions and views that we have taken in this response that might be of assistance to the drafting team. Respectfully submitted, James H. Murta President
1. In Section 4, key considerations for determining expected cash flows are described. a. Do you disagree with the description of any of them? Should any be left out or modified, or others added? Response: We have no comment at this time. b. In Section 4.3.1, reference is made to the relevance of the financial reporting context to the measurement and possible limitations and constraints that may be placed on the estimates. Do you believe this paper should discuss potential restrictions on current estimates or risk margins as is done in the current version of the paper or should this discussion be made in another research paper when more is known about the direction of specific constraints, e.g., later in the IASB Phase 2 project? If you believe that this paper should address these issues, please indicate the specific constraints that you believe should be addressed, and whether they should be discussed more extensively. Response: We found the discussion of potential restrictions to be a useful illustration of the concepts which encourages the reader to consider issues and implications specific to their circumstances. We believe it is appropriate for the paper to include this discussion. We believe that restrictions of this type would often be inconsistent with the determination of exit value. Two specific constraints that we believe should be addressed and discussed more extensively are: 1. Under current IFRS, the value for future income taxes is not discounted for the time value of money. However, when a block of insurance contracts is transferred from one entity to another, the time value of money for future taxes would be taken into account to the extent that it is material. To ignore the time value of money for future taxes in a liability calculation would, thus, be inconsistent with the concept of exit value as described in the IASB paper. 2. The current IASB proposal is that only those premiums necessary to maintain guaranteed insurability should be considered for measurement purposes. However, when a block of insurance contracts is transferred from one entity to another, all expected future premiums would be taken into account. To ignore expected future premiums would, thus, be inconsistent with exit value. c. Do you agree or disagree with the distinction provided in the selection of market based or non-market based assumptions? Why? Response: The distinction is reasonable to us. d. Are there other aspects of expected cash flows that should be addressed? Response: We have nothing to add at this time. 2. In Section 5, discounting is discussed. a. In Section 5.2 risk-free discount rates are discussed. Should these be the basis of discounting? If not, what should they be based on? Several approaches to 2
determining risk-free rates are described if risk-free rates are used, what should their basis be (e.g., spot rates, swap rates) and why? Response: The CIA provided its views on the question of appropriate discount rates in the response to Questions 2 and 14 of the IASB Discussion Paper: Preliminary Views on Insurance Contracts. They are repeated here for convenience: Question 2: On the topic of discount rates, we recommend that the IASB clarify whether it intends that risk-free discount rates be used. Many of our members interpret the Discussion Paper as requiring risk-free discount rates, a position that is supported by many of our Property & Casualty Insurance practitioners. Others interpret the Discussion Paper as allowing higher than risk-free discount rates to reflect the credit characteristics of the liability, a position that is supported by many of our Life Insurance practitioners. Question 14: We recognize that it will be difficult to achieve consensus in the debate about reflecting credit characteristics in the value of the liability. In general, we believe that it would be inappropriate to reflect changes in credit characteristics in the value of a liability, since it is paradoxical to allow an insurer to reduce the value of liabilities merely because it has become less likely that the insurer will honour its obligations. One possible approach we could support would be to base the value of the liability on the credit characteristics of the portfolio within a standardized reference entity to which the portfolio is hypothetically being transferred. In other words, all insurers would measure the value of a portfolio using the same credit characteristics. We believe this would be consistent with the IASB s preliminary views as stated in Paragraphs 69 and 232 of the Discussion Paper. In Canada, the credit characteristics of insurance portfolios are influenced by a strong regulatory environment and industry guarantee associations. The current infrastructure and developments in the area of capital requirements suggest that the Canadian industry will be regulated to stay at a credit quality similar to an AA rating. Therefore, if the IASB adopts the position that credit characteristics should be reflected, the appropriate discount rates to use in the valuation of insurance liabilities in Canada would be current market rates available on other Canadian AA instruments that match the timing of liability cash flows. As further support for this position, we note that using AA discount rates is generally more consistent with industry pricing and appraisal practices than using risk-free rates would be, and thus is more consistent with the concept of current exit value. Due to the very long-term nature of some insurance liability cash flows, discount rates would be required for durations where markets are incomplete or supply/demand imbalances exist. Further research would be required to determine a complete set of discount rates consistent with the IASB s position. On the question of the basis for choosing risk-free rates, we note that the availability and reliability of various possible bases varies by country. Also, it may be possible in some cases to get to the same end result via different routes. In our view, the IAA should take the lead in articulating the principles underlying the selection of a risk-free yield curve. b. A discussion of liquidity is included in section 5.3. Should this be a factor to consider in the measurement of liabilities and/or should it be included in the final version of 3
this paper? If it should be included, is the discussion appropriate or do you have recommended modifications? Response: We believe that liquidity of the liability cash flows (or lack thereof) should be a consideration in the choice of discount rates. All else being equal, a relatively less liquid liability cash flow should have a relatively lower exit value, which is accomplished by increasing the discount rate accordingly. c. For cash flows that are not directly related (but are generally affected by the asset investment performance) to a designated set of assets (such as universal life or many, but not all, participating contracts for which competition also can play a role), should discount rates be based on the actual assets held or on a market-based set of yield rates or some other alternative? Response: When liability cash flows vary with the performance of a designated set of assets, a link between the value of the liability and the value of the corresponding assets is necessary to avoid accounting mismatch. However, providing this link through using asset-specific discount rates is only appropriate to the extent that asset investment performance is passed-though to the liability cash flows. In practice, this pass-through is often imperfect, for example, when minimum guarantees are provided. Therefore, for more complex products, our preferred approach would be to project liability cash flows under a sufficient number of scenarios, determine the present value of each liability cash flow stream using discount rates that are consistent with the asset returns in the particular scenario, and choose an appropriate liability based on the results. An approach whereby the weighted average of projected cash flows is discounted at rates appropriate for other businesses does not produce appropriate results in the situation where liability cash flows vary with changes in interest rates. d. For cash flows that are linked to the performance of a set of designated assets (e.g., certain participating or unit-linked/variable contracts), should the discount rate(s) be based on the expected performance of those assets or another basis? Response: See response to 2c above. Also, in section 5.4 (page 44), the description of the bottom-up approach indicates that an adjustment should be made for investment expenses. We assume this would be investment expenses related to risk-free assets, but would like to see this point verified. Also, the adjustment for undiversifiable asset-liability mismatch risk appears to be in the wrong direction. Additional risk should increase the value of the liability, which would mean a decrease in the discount rate. In the description of the top-down approach (page 45), the second adjustment should probably be asset default risk margin. Also, we think there should be an adjustment for investment expenses, which would be quite a bit higher than the corresponding adjustment in the bottom-up approach. It is possible that the adjustment for investment expenses is intended to be included in the total nominal expected investment rate, but if so, this should be stated. e. Do you believe that guidance is needed to develop market-consistent assumptions and why? If yes, who do you believe should provided it (e.g., regulator, the IAA, the 4
IASB, the local actuarial association or regulator) and what form of guidance is needed? Response: In our view, the IAA should take the lead in articulating the principles underlying the selection of market-consistent assumptions. 3. In Section 6, risk margins are discussed. a. In 6.1, several possible objectives of risk margins are described. Do you believe that this discussion is reasonable? If not, what do you believe the objective of risk margins should be? Response: We believe the discussion is reasonable. b. Possible approaches to the quantification and qualification of risk margins are discussed in the remainder of Section 6. i. Are the desirable risk margin characteristics appropriately described? Response: The characteristics described by the IAIS/IASB are appropriately described. We wonder if the additional characteristics identified by the RMWG are necessary, and if so, some would benefit from clarification. For example, applies a consistent methodology for the lifetime of the contract could be interpreted to mean that the risk margin can never be changed. Also, what other financial contracts is the RMWG thinking of in #3? How is one to determine what are accepted economic and actuarial pricing methodologies, and what if those methodologies are inconsistent with each other? What is intended by facilitates disclosure of information useful to stakeholders? ii. Are the identified pros/cons of methods reasonably presented? If they are not, please describe those that you disagree with. Response: As discussed in our response to the original Exposure Draft, we believe that the explicit assumption approach, if calibrated appropriately, can satisfy the desirable risk margin characteristics in a relatively simple and practical manner. It has served us well in Canada for many years. We believe the paper, though acknowledging the limitations of the analysis, underestimates the benefits of the explicit assumption approach. We repeat our suggestion that the explicit assumption approach be considered as an appropriate approximation for complex risks and for use as a default approach for risks that do not lend themselves well to the more complex approaches favoured in the paper. iii. Should the IAA put forth a proposal for one or more preferred methods for the quantification of risk margins? If so, which method(s) should be recommended and should this be included in the final version of this paper or another work product? Response: We are in favour of the IAA proposing a preferred approach for the quantification of risk margins. This would narrow the range of practice, which supports the goal of comparability of reporting. We support the cost of capital approach, with recognition that an explicit assumption approach, appropriately calibrated, could be used as a reasonable approximation, or as a default approach for risks that do not lend themselves well to the cost of capital approach. 5
If the cost of capital approach is favoured by the IAA, more specific guidance will be required on the amount of capital and the cost of capital used to determine the risk margin. c. Practical issues of risk margin calculation methods are discussed. Are these issues appropriately identified and described? Response: We believe that the practical difficulties associated with the cost of capital approach are underestimated. The approach requires careful consideration of the cost of capital, the amount of capital, and the projected run-off of that capital, each of which involves a significant element of judgement. Additional guidance will help, but in any event we believe that simpler approximate approaches, appropriately calibrated, could be used effectively in many situations. d. Is the use of a reference entity or portfolio appropriate? Response: As noted in our response to the first Exposure Draft, we believe that the concept of a reference entity would promote consistency in the measurement of exit values. i. If so, what should the characteristics of the reference entity or portfolio be? If so, in what context should they be provided (e.g., in accounting standards, actuarial standards, by the regulator or developed by emerging practice)? Response: The characteristics of a reference entity should be consistent with the IASB/IAA s views on the definition of exit value. In particular, the reference entity should be: sufficiently large for process risk to be effectively eliminated described in a manner that clarifies the level at which it is appropriate to recognize diversification of risk used to clarify the appropriate level of expenses to be used in the measurement of exit value used to define the credit characteristics of the liabilities used in setting discount rates subject to typical industry taxes consistent with the desired measure of the cost and amount of capital when using the cost of capital approach to determine risk margins. It would be most helpful if these basic characteristics were included in the accounting standards, as this would improve consistency of practice as well as helping to clarify the concepts underlying the determination of exit values. This could also be done through IAA guidance. If left to regulators or emerging practice, inconsistencies are far more likely to arise. e. Should the size (with respect to process or random deviation uncertainty risk), diversification or other feature of the portfolio or entity be considered in the determination of risk margins? If so, what should the level be that they should be considered the portfolio or the entity? And if so, why should it be? Response: See response to 3d above. f. Do you agree with the assessment and comparisons included in Section 6.11? Response: See responses to 3b and 3c above. 6
4. Section 7 deals with risk mitigation approaches. a. Do you disagree with the treatment of any of the approaches indicated? If so, please provide your preferred approach? b. Do you believe that diversification should be reflected in the measurement of liabilities (either within a portfolio or inter-portfolio)? Why or why not? Are the approaches given in Section 7 regarding diversification or in Appendix C appropriate? If not, do you have any suggestions as to how the effect of diversification should be measured? Response: We have nothing to add at this time. 5. Section 8 deals with several miscellaneous topics. Are these treated sufficiently and do you disagree with the description of any of them? If not, or if additional topics should be addressed here, please indicate them and why you believe they are relevant to this paper. Response: We have nothing to add at this time. 6. Are the appendices included useful and sufficient for the purpose indicated? If one or more are not, please indicate which and why. Response: Generally speaking, we found the appendices useful and sufficient. 7. Appendix D indicates that the operating expenses of a portfolio or the entity should be used rather than market-based expenses. Which basis for determination of expenses do you believe would be appropriate to be used? Should it vary depending on the application? If market-based expenses should be used, what approach(es) should be considered for use in determining them? Should the size of the entity or portfolio matter? Response: We believe that the reference entity concept should be used to define the appropriate level of operating expenses to be used. However, in practice, the best source of information about expenses will be entity-specific. If entity-specific expenses are used, appropriate disclosure of any material differences will be required. 8. Should any deviations or special considerations be made in the case of a jurisdiction in which the actuarial profession is not sufficiently developed or is emerging and the experience and skills of the actuaries involved are not at a level that the approaches described in the paper require? If so, please indicate in what way and what other approaches should be taken or guidance provided? a. Should there be any modifications of the methods described in the paper be made to apply either smaller companies, less sophisticated markets or products, or new coverages? If so, what are they and how should they be addressed? b. Should the IAA take any steps to address this area? If so, please provide suggestions. Response: We believe that considerations of this type are best handled through general actuarial guidance on the application of professional judgement and the appropriate use of approximations. We do not support an approach where standards of practice are watered down for various specific circumstances. 7
9. Please provide additional references (to those already included in the reference section of the paper) or glossary definitions that you believe would be helpful to the reader of the final version of this report. Response: We have nothing to add at this time. 10. Do you have any additional comments relating to the measurement of liabilities for insurance contracts that are relevant to this paper? Please mention the section number and offer proposed drafting as appropriate. Response: As discussed in response to question 3.b.iii., we support the use of the cost of capital method for risk margins. We believe it would be helpful if the paper expanded further on the approach to cost of capital. One issue the paper might address is the difference in cost of capital perceived to be appropriate in North America vs. that suggested in the RMWG paper. This difference is illustrated by the discussion on page 65 of the RMWG paper, which suggests a cost of capital of 6% for a BBB company and a cost of capital of 4% for an AA company. In contrast, the Society of Actuaries recently published a report of findings from a research project on Financial Reporting for Insurance Contracts under Possible Future International Accounting Standards that illustrates a cost of capital of 12% and 18%. With a difference so large, we believe that it would be beneficial to discuss this issue in the RMWG report. 11. The possibility of including a risk margin in the measurement of the liability for post-retirement benefits, including pensions has recently been discussed. Do you believe that this would be appropriate? Please explain the reason for your response? Response: We do not believe that the inclusion of a risk margin is appropriate for the financial reporting of employer-sponsored post-retirement benefits. Employer-sponsored post-retirement benefits differ considerably from insurance contracts in that the benefits provided are not subject to the same degree of guarantee and are often not fully funded. As such, post-retirement benefits should be valued and reported in a manner consistent with the other assets and obligations of the organization. The inclusion of a risk margin for the financial reporting of post-retirement benefits would not enhance benefit security and would be inconsistent with the measurement of the organization's other obligations. Due to the subjective nature of determining the margins, comparability between organizations and between fiscal years would become more difficult. The accounting rules in Canada currently require "best estimate" assumptions, which would not include a risk margin. 8