1

November 24, 2010

Sir David Tweedie Chairman International Accounting Standards Board 30 Cannon Street London EC4M 6XH United Kingdom

Re: July 30, 2010 Exposure Draft “Insurance Contracts” Dear Sir David, The Canadian Institute of Actuaries welcomes the opportunity to comment on the Insurance Accounting Standards Board Exposure Draft on Insurance Contracts. The Canadian Institute of Actuaries strongly supports the Board’s intention to achieve a consistent approach to the measurement of insurance contract liabilities worldwide. We congratulate the Board for addressing the obviously considerable challenges faced in grappling with the wide diversity of practices and products involved. Given these diversities, we also support the Board’s efforts to create a principles-based standard. We recognize that some topics would benefit from supplemental guidance, and we hope that the Board will support the global actuarial profession in its efforts to provide such guidance, perhaps by acknowledging the valuable contribution it would make to promoting consistency of practice around the world. Attached are our responses to all the questions posed in the Exposure Draft. Though we are generally supportive of the approach described in the Exposure Draft, we have a few key areas of concern, which are:

• the discount rate on contracts whose cash flows do not depend on the performance of specific assets (see question 3),

• the limitation to three specific methods for determining the risk adjustment (see question 5b), • the residual margin (see question 6), and • unbundling (see question 12).

The Canadian Institute of Actuaries hopes that its comments provided herein will be of value to the Board. We appreciate the opportunity to contribute to the development of this important document, and would be happy to provide any further explanation that might be of assistance to the Board in respect of the positions and views that we have taken in this response. Respectfully submitted,

Micheline Dionne, FCIA, FSA, MAAA President Document 2110004

2

1. Relevant information for users (paragraphs BC13-BC50) Do you think that the proposed measurement model will produce relevant information that will help users of an insurer’s financial statements to make economic decisions? Why or why not? If not, what changes do you recommend and why?

Response to 1:

We believe that the proposed measurement model will go a long way to eliminating inconsistencies that are currently prevalent due to the IFRS 4 reliance on local GAAP.

In general, we support the basic measurement model of expected present value of cash flows including an appropriate margin. Such a measurement model has been used in Canada for almost 20 years, and is generally considered to provide relevant and useful information. This is evidenced by the fact that reported liabilities also form the basis for the solvency measures of our insurance supervisor and are deemed to be suitable for tax purposes.

We also support the principles-based approach to setting financial reporting standards, leaving the detailed application to the professionals closest to the contracts being valued.

Though we support the general framework, and generally support the reasoning set out in the Basis for Conclusions, there are a number of areas where we believe the proposed approach could be improved. Our suggestions are discussed in our responses to the remaining questions below.

2. Fulfilment cash flows (paragraphs 17(a), 22-25, B37-B66 and BC51) 2a. Do you agree that the measurement of an insurance contract should include the expected

present value of the future cash outflows less future cash inflows that will arise as the insurer fulfils the insurance contract? Why or why not? If not, what do you recommend and why?

Response to 2a:

Yes, we agree. Fulfilment cash flows are a reasonable measure of the underlying obligations of the reporting entity. In particular, we agree with using explicit, current, unbiased, probability-weighted estimates of cash flows, and we agree that using the perspective of the entity is preferable to using the perspective of a buyer.

However, there are a few areas where the cash flow projections described in the Exposure Draft depart from the fulfilment concept. Therefore, we suggest that the IASB consider the following enhancements:

• Paragraph B61 limits the expenses to be included in fulfilment cash flows to “direct costs and systematic allocations of costs that relate directly to the insurance contracts or contract activities.” In our view, limiting allocated costs to those that relate directly to the insurance contracts would exclude many expenses that are clearly necessary to enable the insurer to fulfill its insurance contract obligations. These include costs of accounting, human resources, systems, etc. Moreover, insurance contract premiums clearly provide for an appropriate share of such

3

costs, so excluding them from the measurement of insurance contract would create an inappropriate front-ending of profits which is then arbitrarily offset through the residual margin, the amortization of which will bear no relation to the expenses themselves. This will create inappropriate volatility of income and provide less relevant information to users than would including the expenses related to the contract in the contract liability.

• In our view, all future taxes related to insurance contracts should be included in the projected cash flows as would any other expense that arises directly from the contract. Also, the time value of money should be applied to these cash flows as to any other. We understand the rationale in IAS 12 for not discounting these amounts is one of practicality and comparability but, nevertheless, we disagree. For long duration insurance contracts, the time value of money associated with future taxes can be a material item, and to ignore it can create a material misstatement of the value of the related asset or liability.

2b. Is the draft application guidance in Appendix B on estimates of future cash flows at the right level of detail? Do you have any comments on the guidance?

Response to 2b:

We believe the guidance provided in the Exposure Draft is at the right level of detail. It provides enough guidance to promote consistency of reported results, but is not so prescriptive as to restrict entities from applying an appropriate degree of judgement. However, this is a topic where supplemental guidance from the global actuarial profession would be helpful.

We note that concerns have been raised that the Exposure Draft implies that complex scenario testing is required for all types of risk. We interpret paragraphs B38 and B39 as allowing future cash flows to be projected using simpler approaches when any added precision of complex approaches would not produce materially improved results. If the Board agrees, we suggest that the intent be clarified.

3. Discount rate (paragraphs 30-34 and BC88-BC104) 3a. Do you agree that the discount rate used by the insurer for non-participating contracts

should reflect the characteristics of the insurance contract liability and not those of the assets backing that liability? Why or why not?

Response to 3a:

We agree that the discount rate should reflect the characteristics of the insurance contract liability. However, we do not agree that the discount rate should ignore the assets supporting the obligation.

The act of fulfilling an insurance contract involves the acquisition of assets whose cash flows will be sufficient to discharge the obligations across a wide range of possible future outcomes for both asset and liability cash flows in changing economic environments. The ongoing management of the asset/liability portfolio is the essence of an insurer’s business model for long-duration non-participating contracts.

4

The current Canadian valuation approach, which has served Canada well for almost 20 years, reflects this business model by coordinating the valuation of assets and liabilities and eliminating accounting mismatch. Though we understand the Board has rejected an approach that links the value of non-participating contract liabilities to the value of supporting assets, we would like to highlight a consequence of this decision that is often overlooked.

Only when the value of liabilities is linked to the supporting assets can that liability value reflect the degree to which asset/liability mismatch risk has been mitigated through the selection of assets with similar cash flow characteristics. However carefully insurers attempt to minimize risk by selecting assets whose cash flows match the liability cash flows in terms of timing, currency and liquidity, some degree of mismatch is unavoidable, especially in the case of long duration liabilities.

This risk is both fundamental and material to an insurer’s ability to fulfill long-duration non-participating insurance contract obligations, yet the degree to which it has been mitigated is entirely ignored in any measurement approach that ignores the assets supporting the liabilities. We understand that the Board is prepared to make this sacrifice on the grounds that this risk is market risk rather than insurance contract risk. We respectfully disagree, since the risk arises only when the insurer enters into the insurance contract and is an inevitable consequence thereof.

We note that the importance of the link between liabilities and their supporting assets appears to be acknowledged by the Board in the presentation requirement of paragraph 73. We support this disclosure, since we believe a proper understanding of the potentially serious accounting mismatches that would arise from applying the discount rates as proposed in the Exposure Draft would be essential to the proper understanding of the financial statement results.

3b. Do you agree with the proposal to consider the effect of liquidity and with the guidance on liquidity (see paragraphs 30(a), 31and 34)? Why or why not?

Response to 3b:

We agree that liquidity should be taken into account in determining the discount rate and support its inclusion in paragraphs 30(a), 31 and 34 in the Exposure Draft.

Our concern is that the lack of a generally accepted approach to measuring the adjustment for illiquidity will lead to financial results that are not useful for making economic decisions. For example, we have seen examples of companies reporting Embedded Values that use illiquidity adjustments ranging from 7 to 300 basis points.

We believe the illiquidity adjustment could be made more relevant and consistent among preparers, if it were determined indirectly, by subtracting a provision for credit risk from the total spread on assets.

First, we note that the spread of pure loan-like fixed income instruments that contain no prepayment or other options consists for all practical purposes of only two components: liquidity and credit risk, each including a provision for uncertainty.

Further, we contend that there exists far more objective evidence relative to the credit risk attaching to such fixed income assets than to the liquidity premium of those (and other) assets. Credit default swap

5

instruments may be available. The major rating firms publish such experience with respect to publicly traded bonds; other organizations compile similar data with respect to private placements or other loans (commercial or residential mortgages, personal loans etc), including experience as to their volatility over time and under different circumstances. Hence, not only expected credit losses, but also a margin for uncertainty could be estimated from such information.

Therefore, it would seem a more reliable approach to determine the illiquidity adjustment indirectly through the analysis of the credit experience of such instruments, subtracting a provision for credit risk from the total spread on assets.

We note that this approach is consistent with the principles in the November 2009 Exposure Draft “Financial Instruments: Amortized Cost and Impairment” for estimating the impact of future credit losses (see for example paragraphs B7-B9).

Another area where guidance will need to be developed is the discount rates to apply to liability cash flows that occur beyond the date for which instruments on which the discount rates are based are reliable. For many life insurance products, liability cash flows are projected 30, 40, 50 or more years in the future, which is longer than the yield curve horizon. One possible approach would be to use a multi-year moving average of long term risk-free rates to discount liability cash flows beyond the yield curve horizon back to the yield curve horizon. A similar approach has been used successfully for many years in Canada.

3c. Some have expressed concerns that the proposed discount rate may misrepresent the economic substance of some long-duration insurance contracts. Are those concerns valid? Why or why not? If they are valid, what approach do you suggest and why? For example, should the Board reconsider its conclusion that the present value of the fulfilment cash flows should not reflect the risk of non-performance by the insurer?

Response to 3c:

We are convinced that the proposed discount rate will, in many cases, misrepresent the economic substance of some long-duration insurance contracts. Long-duration insurance contracts need to be supported by long-duration assets. To do otherwise places the insurer at unsustainable risk. When those long duration assets are measured at fair value, changes in credit spreads on those assets will affect their fair value and hence income. If the measurement of liabilities does not incorporate changes in credit spreads of the assets, then there is an accounting mismatch and the amount of income will reflect this mismatch. The longer the liability duration, the more significant the impact will be.

Nevertheless, we do not believe the Board should reconsider its conclusion that the present value of the fulfilment cash flows should not reflect the risk of non-performance by the insurer. We agree that reporting a gain from a decline in an entity’s specific credit quality would be seriously misleading.

However, as we stated in our August 28, 2009 response to the June 2009 Discussion paper “Credit Risk in Liability Measurement”, just because an entity’s specific non-performance should not be reflected does not imply that no credit risk should be reflected. To reduce the accounting mismatch, we believe it is reasonable to take into account a price for credit risk that ignores the credit quality of the specific

6

reporting entity. This could be effected, for example, by using a discount rate based on the credit characteristics of a portfolio of high quality corporate debt obligations. In Canada, this would be corporate AA (or similarly rated) bonds.

Another rationale for this approach considers the reality that there is no deep liquid market for insurance contracts. A reasonable way to identify an appropriate discount rate for liability cash flows with particular characteristics as to timing, currency and liquidity, is to look to the asset market. If we can find assets with cash flow characteristics (timing, currency and liquidity) similar to those of the liability cash flows, then the market value of those assets could be used to derive the appropriate discount rate to use for the liability measurement.

We note that this approach supports the Board’s requirement that the present value of future cash flows be consistent with observable market prices (paragraphs 23(b), 34, B45, B47, and B52). It is also consistent with the Board’s decision to include credit risk in reflecting the time value of money for other long-duration liabilities, such as Leases, Provisions and Employee Benefits, and with the approach described in paragraph 32 for contracts where the amount, timing or uncertainty of the liability cash flows depend wholly or partly on the performance of specific assets.

Using a discount rate based on high quality corporate debt obligations will not fully address the concerns expressed about misrepresenting the economic substance of long-duration insurance contracts. In particular, the extent to which the insurer’s actual asset portfolio differs from a portfolio of high quality corporate debt obligations will result in volatility of reported income. We are aware of proposals that aim to reduce reported volatility by calculating a second liability using a discount rate based on the long-term rate the insurer expects to earn on its investments and reporting the difference between the two liability measures in other comprehensive income. We would not oppose such an approach as it could provide valuable information to users.

4. Risk adjustment versus composite margin (paragraph BC105-BC115) Do you support using a risk adjustment and a residual margin (as the IASB proposes), or do you prefer a single composite margin (as the FASB favours)? Please explain the reason(s) for your view.

Response to 4:

We support the concept of a risk adjustment as proposed by the Board for the reasons outlined in paragraph BC109, primarily because it conveys valuable information to users and because it provides a better reflection of the economic burden imposed by the presence of that risk.

5. Risk adjustment (paragraphs 35-37, B67-B103 and BC105-BC123) 5a. Do you agree that the risk adjustment should depict the maximum amount the insurer

would rationally pay to be relieved of the risk that the ultimate fulfilment cash flows exceed those expected? Why or why not? If not, what alternatives do you suggest and why?

7

Response to 5a:

We agree with the general concept, but believe the word “maximum” should be removed, since it can be interpreted as introducing a bias in the calculation.

The risk of uncertainty relates to the timing of cash flows as well as the amount thereof, and therefore the phrase “the ultimate fulfilment cash flows exceed those expected” should be “the ultimate fulfilment cash flows differ from those expected”. An alternative would be “the present value of fulfilment cash flows will exceed that expected”.

We note that paragraph IN11 implies a presumption that the risk adjustment will decline over time. This is normally, but not necessarily, the case.

5b. Paragraph B73 limits the choice of techniques for estimating risk adjustments to the confidence level, conditional tail expectation (CTE) and cost of capital techniques. Do you agree that these three techniques should be allowed and no others? Why or why not? If not, what do you suggest and why?

Response to 5b:

In our view, the Board should set out the objective of the risk adjustment and the principles for choosing it (as in paragraphs B68-B72), and then let reporting entities, with the assistance of professionals, determine how best to apply the principles. In particular, there are many situations for which a simple margin on assumption approach, or a bulk dollar amount, might be the most appropriate and practical means to include a risk adjustment. As well, improved approaches could be developed in future.

Also, permitting only three techniques might encourage the (inappropriate) practice of picking the method that gives the most advantageous result for each situation, perhaps even switching the approach from year to year, rather than applying the most appropriate method in each circumstance.

Moreover, the three specified techniques may be difficult to apply in some situations. The information that is required to develop confidence levels or conditional tail expectations may not be available, particularly for new products.

If the Board decides to specify only certain permissible approaches in the final standard then, in our view, applications of confidence levels are inappropriate in enough situations that their use should not be included as a permitted technique.

5c. Do you agree that the either the CTE or the cost of capital method is used, the insurer should disclose the confidence level to which the risk adjustment corresponds (see paragraph 90(b)(i))? Why or why not?

Response to 5c:

No, we do not believe this should be a general requirement. We agree that in some situations it might provide useful comparative information, but in other situations this disclosure may be inappropriate or

8

even misleading. Moreover, if the objective of the risk adjustment is clearly defined, such disclosure should not be necessary, particularly given the sensitivity information that is also disclosed.

The requirement would force entities to measure the confidence level in every situation, even those for which the confidence level approach is inappropriate.

Furthermore, requiring disclosure of results under any single particular approach will encourage entities to choose that approach (to avoid the work of determining the result under two different approaches) even where it is not the most appropriate to be used.

5d. Do you agree that an insurer should measure the risk adjustment at a portfolio level of aggregation (i.e., a group of contracts that are subject to similar risks and managed together as a pool)? Why or why not? If not, what alternative do you recommend and why?

Response to 5d:

Yes, we agree with measurement of the risk adjustment at the portfolio level. This is practical and generally consistent with our business model.

We note that for some contracts, most commonly property and casualty contracts, a single contract might properly belong to more than one portfolio. This could occur when a single contract covers two or more different risks that the insurer manages separately. In such a case, the risk adjustment for each of the portfolios would consider only the risk that belongs in that portfolio. The definition of portfolio of insurance contracts in Appendix A should be modified to recognize that a single contract might be divided between two or more portfolios.

5e. Is the application guidance in Appendix B on risk adjustments at the right level of detail? Do you have any comments on the guidance?

Response to 5e:

If the Board removes the limitation on the choice of techniques for determining risk adjustments, we believe the guidance provided in the Exposure Draft is at the right level of detail. It provides enough guidance to promote consistency of reported results, but is not so prescriptive as to restrict entities from applying an appropriate degree of judgement. However, this is a topic where supplemental guidance from the global actuarial profession would be helpful.

If the Board continues to limit the permitted risk adjustment techniques, then the Board should provide additional guidance or clarification in the following areas:

• Cost of capital – regarding both the amount of capital and the rate of cost of capital. The amount of capital in particular is subject to a wide variety of interpretations. Some would use economic capital (itself not well-defined), others regulatory capital, and others the larger of the two.

• Confidence level – if disclosure is required, regarding how to determine it in situations where no distribution is readily available.

9

6. Residual/composite margin (paragraphs 17 (b), 19-21, 50-53 and BC124-BC133) 6a. Do you agree that an insurer should not recognise any gain at initial recognition of an

insurance contract (such a gain arises when the expected present value of the future cash outflows plus the risk adjustment is less than the expected present value of the future cash inflows)? Why or why not?

Response to 6a:

No, we do not agree. In Canada, we have, for almost 20 years, successfully used an unconstrained valuation that permits day one gains. Day one gains have not been an issue in Canada, i) because companies are required to disclose the amount of day one gain or loss, and ii) because the actuarial profession has developed robust standards for projecting future cash flows, including margins for risk.

In our view, the residual margin as proposed is an artificial device that distorts the measurement of insurance contracts. It introduces an unnecessary bias in the calculation that reduces the comparability of results across reporting entities. For example, two contracts identical in every way except that one has a higher premium will be measured at the same initial value. We do not believe this provides faithfully representative information.

Similarly, as noted in our response to question 2, understating fulfilment cash outflows by limiting allocated cots to those related directly to the insurance contracts creates an inappropriate front-ending of profit. The Exposure Draft proposes to eliminate this front-ending of profit by creating an arbitrary residual margin with an amortization pattern unrelated to the expected cash outflows. This will create inappropriate volatility of income and provide an incentive for insurers to adopt possibly inappropriate administrative or reinsurance structures simply to allow more expenses to be included in the measurement of the insurance contract liability. In our view, it would be preferable to eliminate the need for a residual margin by including all fulfilment cash flows in the measurement basis.

Further, the corresponding constraint that permits day one reinsurance gains but not day one reinsurance losses (paragraphs 43-45) opens the door to manipulation of income through the selective reinsurance of business. Such opportunities for accounting arbitrage are a direct result of the artificial constraint on day one gains.

We see other opportunities for manipulation of income. For example, with profitable contracts there might be an incentive for insurers to use conservative assumptions at issue (minimizing the residual margin) and then revise those assumptions at the next reporting period to allow release of the profits that the residual margin was intended to defer.

Finally, we note that the question of which acquisition expenses should, and should not, be included in the initial measurement of an insurance contract liability would become moot in the absence of a residual margin. Acquisition expenses would be reported through profit or loss as incurred, and the measurement of the insurance contract liability at any point in time would include only future expenses in the fulfilment cash flows. In our view, this would be a much preferable result.

10

6b. Do you agree that the residual margin should not be less than zero, so that a loss at initial recognition of an insurance contract would be recognised immediately in profit or loss (such a loss arises when the expected present value of the future cash outflows plus the risk adjustment is more that the expected present value of future cash inflows)? Why or why not?

Response to 6b:

Yes, we agree it would be inappropriate to defer initial losses. This is consistent with our position that it is inappropriate to defer initial gains.

6c. Do you agree that an insurer should estimate the residual or composite margin at a level that aggregates insurance contracts into a portfolio of insurance contracts and, within a portfolio, by similar date of inception of the contract and by similar coverage period? Why or why not? If not, what do you recommend and why?

Response to 6c:

We believe that a residual or composite margin is totally inappropriate. However, if the Board persists in requiring a residual or composite margin, then it would appear that such a margin would theoretically be determined at the individual contract level. The Board’s choice of ‘cohort’ level is a more practical alternative, but we are nevertheless concerned with the complexity of the task compared to the value added. There are a number of areas where the determination of an appropriate ‘cohort’ within a portfolio is unclear, for example as interest rates change and as assumptions are refreshed. There is also some question of the extent to which contracts with day one losses might be allowed to offset contracts with day one gains before the residual margin is determined. We would recommend the Board provide more practical guidance around the definition of ‘cohort’.

Also, for insurance contracts where a portion of the risk is reinsured, we believe it would be more appropriate to determine the residual margin for the net liability rather than separately for the gross liability and the reinsurance ceded asset. For agreements where a proportion of the entire contract is ceded to the reinsurer, this produces the same result, but proportionate reinsurance is not always the case. For example, reinsurance premiums can have a pattern totally different from that of the gross premiums. Determining the residual margin separately for the gross and ceded could lead to a situation where the initial loss on reinsurance (which is deferred by the residual margin) is larger than the initial gain on the direct contract, resulting in the deferral of a net loss. Another possibility is that the initial gain on reinsurance (which is recognized) is greater than the loss on the direct contract, resulting in the immediate recognition of a net gain. Both of these possibilities seem contrary to the intent of the residual margin.

6d. Do you agree with the proposed method(s) of releasing the residual margin? Why or why not? If not, what do you suggest and why (see paragraphs 50 and BC125-BC129)?

Response to 6d:

We strongly disagree with the concept of a residual margin. However, if a residual margin is incorporated in the final standard, in our view the proposed method of releasing the residual margin is

11

unnecessarily complex and inconsistent with the forward-looking nature of the measurement of the liability.

We take particular exception to paragraph 53, which states that the residual margin would be reduced if there are fewer policies in force than expected at the beginning of the period, but not increased if there are more policies in force than expected at the beginning of the period. This introduces an unnecessary bias and complication into the measurement of the liability, but with no apparent improvement in the value or verity of the financial statements.

We also note that the pattern of release results in inconsistent treatment of otherwise identical contracts that differ only in whether they were issued before or after a change in experience assumptions.

An alternative that we could reluctantly support is to release the residual margin in proportion to the release of the risk adjustment. The residual margin is arbitrary and, consequently, we disagree with the suggestion in paragraph BC126 that the drivers of the release of the residual margin must be different from the drivers of the release of the risk adjustment. Tying the two together would at least be convenient and simplify the calculation.

6e. Do you agree with the proposed method(s) of releasing the composite margin, if the Board were to adopt the approach that includes such a margin (see Appendix to the Basis for Conclusions)? Why or why not?

Response to 6e:

If the Board were to persist in adopting the composite margin approach, we would suggest adopting a simple approach to its release, consistent with our response to question 6(d) above.

6f. Do you agree that interest should be accreted on the residual margin (see paragraphs 51 and BC131-BC133)? Why or why not? Would you reach the same conclusion for the composite margin? Why or why not?

Response to 6f:

We agree with the rationale outlined in paragraph BC131. However, we believe that the interest rate should not be locked in at inception. We think of the residual margin as an unallocated component of the risk adjustment and believe it would be appropriate and simpler to accrete interest on both using the same interest rate. This would also avoid adding to the ‘accounting mismatch’ described in paragraph BC172.

We would reach the same conclusion if a composite margin were adopted.

7. Acquisition costs (paragraphs 24, 39 and BC135-BC140) 7a. Do you agree that incremental acquisition costs for contracts issued should be included in

the initial measurement of the insurance contract as contract cash outflows and that all other acquisition costs should be recognised as expenses when incurred? Why or why not? If not, what do you recommend and why?

12

Response to 7a:

First, we note that the question of which acquisition costs to include in the initial measurement is only necessary when a residual margin is incorporated in that initial measurement. As noted in our response to question 6, we disagree with the concept of a residual margin.

However, if a residual margin is included in the initial measurement, we agree that acquisition costs should be included in the initial measurement of the insurance contract as contractual cash flows, but we disagree with the restriction to incremental acquisition costs measured at the contract level. We believe it would be more appropriate to include incremental acquisition costs measured at the portfolio level. All such costs are incurred clearly for the purpose of issuing new contracts and are directly relatable to such, even those costs not tied to a particular contract. Examples of such expenses include (1) the costs of direct marketing, (2) the costs of supporting a tied agency force, (3) bonuses payable to brokers.

Furthermore, separating acquisition costs related to contracts sold from acquisition costs for contracts not sold is information that may not be readily available since it has no bearing on how insurers manage their business.

As well, it will introduce issues of interpretation (hence consistency and comparability) and could provide an incentive for insurers to adopt possibly inappropriate administrative or reinsurance structures simply to allow more acquisition expenses to be treated as deferrable.

Furthermore, we believe that limiting acquisition costs to those that are incremental excludes many expenses that are clearly incurred for the purpose of selling new contracts. Excluding these expenses from the measurement of an insurance contract would provide less relevant information to users by forcing an artificial loss at issue.

8. Premium allocation approach (paragraphs 54-60 and BC145-BC148) 8a. Should the Board (i) require, (ii) permit but not require, or (iii) not introduce a modified

measurement approach for the pre-claims liabilities of some short-duration insurance contracts? Why or why not?

Response to 8a:

Our preference would be (iii). In a principles-based standard, reporting entities are always free to develop simplified approaches on the basis of materiality, and we can see no need for separate rules for short term contracts.

Moreover, having separate rules introduces inconsistent treatment of otherwise similar contracts, and could provide an incentive to structure contracts artificially simply to get one treatment or another.

If the IASB does not choose option (iii), then we would prefer (ii), where the premium allocation approach is permitted under certain circumstances, but not required. Since the premium allocation approach is simply an approximation of the result that would be generated under the standard approach, a reporting entity should not be precluded from using the standard approach.

13

8b. Do you agree with the proposed criteria for requiring that approach and with how to apply that approach? Why or why not? If not, what do you suggest and why?

Response to 8b:

We do not agree that the approach should be required, and note that if it is required, there will be difficulties with any demarcation line chosen. One year seems reasonable, but we have found many examples of contracts for which the one-year coverage period does not quite fit. One suggestion for the criteria was for ‘almost all’ of the business to be within a coverage period of 12 months or less.

It was also noted that the reinsurance of contracts requiring the premium allocation approach would not necessarily require the premium allocation approach. This is likely to cause confusion.

9. Contract boundary principle (paragraphs 26-29 and BC53-BC66) Do you agree with the proposed boundary principle and do you think insurers would be able to apply it consistently in practice? Why or why not? If not, what would you recommend and why?

Response to 9:

We agree with the proposed boundary principles and see them as a significant improvement over the approach in the discussion paper. The proposals are very similar to the approach currently in place in Canada, which has been applied successfully for many years.

There are a few clarifications we believe would be helpful:

• Paragraph 27(b) refers to the insurer’s ability to set a price that fully reflects the risk. For some insurance contracts, it is not the price that is adjusted, but rather the level of benefits paid, and we ask the Board to clarify that the contract boundary considerations would apply analogously in these situations.

• For insurance contracts where the adjustable element is the interest rate credited on an account value (such as some Universal Life contracts), we assume that the contract boundary considerations do not apply to that element. Otherwise, a contract with a continuously adjustable credited interest rate could be interpreted to have a contract boundary at the date of issue, which is clearly not the intention of the Board.

• Some contracts are subject to regulatory price controls, which can create an onerous obligation beyond the boundary of the contract. It is unclear to us whether such obligations should be recognized in the contract liability.

• It is possible for the boundary of an insurance contract to be different from the boundary of a related reinsurance contract. In our view, it would be inappropriate for the ceding company to calculate the gross contract liability using one boundary and the related reinsurance asset using a longer boundary.

• There are individual contracts of life or health insurance that, in accordance with a master agreement, may only be cancelled or reassessed at the level of the class or group, and not at the

14

level of the individual insured. We suggest that paragraph 27(a) be removed, and replaced with the words “either terminate coverage or” before “reassess the risk” in the first line of paragraph 27(b), and inserting the words “terminate or” before the word “reprice” in the first line of paragraph BC57.

10. Participating features (paragraphs 23, 62-66, BC67-BC75 and BC198-BC203) 10a. Do you agree that the measurement of insurance contracts should include participating

benefits on an expected present value basis? Why or why not? If not, what do you recommend and why?

Response to 10a:

Yes, we agree that the measurement of insurance contracts should including participating benefits on an expected present value basis. We also agree with the approach to reflecting the time value of money on such contracts.

Paragraph BC70 describes the reasons we believe the measurement of insurance contracts should include all cash flows from those contracts, without limitation to those for which a legal or constructive obligation exists. This reflects the underlying nature of the contracts and will provide the most meaningful information to the users of the financial statements.

However, we have a concern with the complete removal of the limitation, because an element of circularity arises in its absence. If the policy benefits are discretionary, it becomes a question of what the insurer can afford, yet whatever is assumed in the measurement of the liability will determine what the insurer can afford. In order to avoid that circularity, in Canada we apply the principle that liabilities should provide for “policyholders’ reasonable expectations”, a concept fully described in our standards of practice that describes reasonable expectations of the insurer’s exercise of discretion when policy benefits are discretionary.

10b. Should financial instruments with discretionary participation features be within the scope of the IFRS on insurance contracts, or within the scope of the IASB’s financial instruments standards? Why?

Response to 10b:

In an ideal reporting system, it would not matter, since the two standards would apply consistent measurement principles and yield reasonably similar results.

These contract types are not prevalent in Canada. However, we agree that it makes sense to measure them according to the same general principles as insurance contracts, especially in the case where they participate in the same pool of assets, company, fund or other entity as insurance contracts.

15

10c. Do you agree with the proposed definition of a discretionary participation feature, including the proposed new condition that the investment contracts must participate with insurance contracts in the same pool of assets, company, fund or other entity? Why or why not? If not, what do you recommend and why?

Response to 10c:

No, we do not agree with the proposed new condition that investment contracts must participate with insurance contracts in the same pool of assets, company, fund or other entity. This would mean that otherwise identical investment contracts, one of which participates with insurance contracts and one of which does not, would be measured in different ways.

10d. Paragraphs 64 and 65 modify some measurement proposals to make them suitable for financial instruments with discretionary participation features. Do you agree with those modifications? Why or why not? If not, what would you propose and why? Are any other modifications needed for these contracts?

Response to 10d:

Subject to our disagreement in principle with the inclusion of a residual margin, we agree with the modifications in paragraphs 64 and 65 and the general guidance in paragraph 66.

11. Definition and scope (paragraphs 2-7, B2-B33 and BC188-BC209) 11a. Do you agree with the definition of an insurance contract and related guidance,

including the two changes summarised in paragraph BC191? If not, why?

Response to 11a:

The word “policyholder” is defined as “a party that has a right to compensation under an insurance contract if an insured event occurs” and yet the term is used in the definition of insurance contract as the party from whom insurance risk is accepted. For many types of insurance contracts, these are two different parties. The party who has a right to compensation is normally called the beneficiary of the contract. We believe that separating the two parties would improve clarity.

Regarding the two changes summarized in paragraph BC191, we agree with the change in paragraph B26, but we have a concern with the change in paragraph B25.

Paragraph B25 says that a contract does not transfer insurance risk if there is no scenario that has commercial substance in which the present value of net cash outflows paid by the insurer can exceed the present value of the premiums. This could be problematic for reinsurance contracts where a single contract covers an entire pool of risks, and as a result the risk of loss is significantly reduced. We would suggest the following wording be added to paragraph B25: “Where a reinsurance contract accepts the same risk of loss as the insurer on every insurance contract within the reinsurance contract, the reinsurance contract is considered an insurance contract.”

16

Paragraph B19(c) excludes self-insurance from the definition of insurance contracts on the basis that there is no agreement with another party. However, it is unclear whether this is intended to apply to an employer extending, say, long-term disability benefits to its employees. Will employers who provide the insurance themselves be required to report liabilities in accordance with the standard?

11b. Do you agree with the scope exclusions in paragraph 4? Why or why not? If no, what do you propose and why?

Response to 11b:

Yes, we agree.

11c. Do you agree that the contracts currently defined in IFRSs as financial guarantee contracts should be brought within the scope of the IFRS on insurance contracts? Why or why not?

Response to 11c:

Yes, we agree, for the reasons described in paragraph BC194-BC195.

12. Unbundling (paragraphs 8-12 and BC210-BC225) Do you think it is appropriate to unbundle some components of an insurance contract? Do you agree with the proposed criteria for when this is required? Why or why not? If no, what alternative do you recommend and why?

Response to 12:

Ideally, unbundling would be unnecessary as different types of contracts would be measured using consistent principles, even when measured under different parts of the accounting standards. However, given that this ideal has not been realized, we believe that unbundling should be used as little as possible. That is, it should only be required to counter a situation where an agreement that is unrelated to an insurance contract is brought into the insurance contract in order to benefit from insurance contract treatment. In other words, paragraph 8(c) should be the principle describing when unbundling is required.

In our view, unbundling components that might be treated as other types of contracts adds a significant amount of complexity for little, or no, additional benefit to users. Insurance contracts typically include a complex combination of insurance and investment elements, and attempting to split out the investment contract elements and treating them differently is unnecessary and does not provide relevant information. As long as all insurers are applying the same financial reporting rules and all the elements of contracts are measured in accordance with the Exposure Draft, the key objectives are met.

Moreover, splitting out the investment elements embedded in insurance contracts introduces an array of questions and interpretations that can only lead to reduced comparability of results.

17

In particular, the guidance on unbundling account balances could be problematic. In Canada, it would seem that most account balances under Universal Life contracts would satisfy the conditions in paragraph 8(a), but in many cases those account balances are merely a reference measure and are not directly available to the owner. Also, there is frequent interaction between the account balance and the host contract, making the projection of residual cash flows very difficult. Account values under most typical Universal Life contracts would appear to be ‘closely related’ to the host contract by any normal interpretation of that expression, and these should not be required to be unbundled. In our view, the example in paragraph 8(a) should be removed.

Furthermore, if account balances are required to be unbundled, in many cases there will be cross-subsidies with other contract elements that would need to be extricated. This could mean, for example, establishing and maintaining a completely artificial ‘shadow account’ that represents what the account value would have been had it been set up without cross-subsidies. What purpose does this serve? Moreover, such an artificial ‘shadow account’ would not be available on transition to IFRS.

Finally, if the approach to unbundling remains unchanged, the final standard would need to include additional guidance on the meaning of ‘closely related’. There is some guidance in the current version of IAS 39, but it relates to embedded derivatives and it is unclear whether it applies to unbundling investment components of insurance contracts. Additional insurance contract examples would be helpful. For example, some have adopted the interpretation that the term certain period of a life annuity with a term certain should be unbundled, while others have presumed that it should not be unbundled.

13. Presentation (paragraphs 69-78 and BC150-BC183) 13a. Will the proposed summarised margin presentation be useful to users of financial

statements? Why or why not? If not, what would you recommend and why?

Response to 13a:

Insurers have always treated premiums as income and claims as expenses, and many users of financial statements might prefer to retain this widely understood treatment.

The statement of comprehensive income described in paragraph 72 is analogous to what we call “sources of earnings” in Canada, and which we have been reporting for about five years. We agree that it provides valuable information to users, but that can be achieved through disclosure rather than by changing the fundamental insurance contract presentation.

The requirement in paragraph 69 to show a single line item for each portfolio of insurance contracts provides an excessive amount of detail. “Portfolio of insurance contracts” is the level at which the risk adjustment is determined, of which there could be a very large number, especially for a large multi-line multi-national insurer. This requirement would impose considerable cost and inconvenience on reporting entities for little, if any, benefit to users.

13b. Do agree that an insurer should present all income and expense arising from insurance contracts in profit or loss? Why or why not? If not, what do you recommend and why?

18

Response to 13b:

Acquisition costs that are not incremental at the level of an individual contract should not be required to be included in the statement of comprehensive income (paragraph 72c). The Board has deemed these expenses to be unrelated to insurance contracts and, therefore, they should be presented as any other expenses that are unrelated to insurance contracts.

14. Disclosures (paragraphs 79-97, BC242 and BC243) 14a. Do you agree with the proposed disclosure principle? Why or why not? If not, what

would you recommend and why?

Response to 14a:

We generally support the disclosure principle in paragraph 79, but find that many of the specific requirements that follow would be far too detailed to be well understood or useful to the vast majority of users. It also imposes considerable cost and inconvenience upon reporting entities for no obvious benefit. If the objective is to provide more information about enterprise risk, we question whether the same standard is being applied to other types of entities.

We have heard that the IASB wishes to open what is sometimes called the insurance “black box”; however, providing excessive detail will not improve the situation and may, in fact, have the opposite effect of leaving users wondering what the insurer is trying to hide in the mass of information disclosed. In our opinion, the key is the selection of meaningful disclosures at the level of detail that is valuable to users.

14b. Do you think the proposed disclosure requirements will meet the proposed objectives? Why or why not?

Response to 14b:

As noted above, we think that the proposed disclosure requirements go too far and will end up being less useful to users of financial statements than more modest requirements would be. In our view, disclosure requirements should not be viewed as a checklist, but rather, the guidance should be clear that only disclosures providing useful information need to be reported. This could be accomplished by removing the word “additional” from paragraph 80.

14c. Are there any disclosures that have not been proposed that would be useful (or some proposed that are not)? If so, please describe those disclosures and explain why they would or would not be useful.

Response to 14c:

The reconciliations of contract balances in paragraphs 86 and 87 are far too detailed to be useful. Paragraph 86 requires reconciliations of 7 different balances, and paragraph 87 requires 10 items for each such reconciliation. Thus, there are 70 amounts reported for each reportable segment, of which

19

there could be a large number. This amount of detail would be so overwhelming as to be ignored by most users.

As noted above in our response to question 5c, we do not support the requirement in paragraph 90(b)(i) to disclose information about the confidence level to which the risk adjustment corresponds.

The disclosure requirements for measurement uncertainty in paragraph 90(d) are especially problematic for actuarial estimates. There are dozens of actuarial inputs for which more than one value might be considered ‘reasonable’. It would not be appropriate to disclose alternative measurements using different, yet still reasonable, inputs, since this would mislead users regarding the fundamental nature of the estimates involved. In our view, the following disclosures provide sufficient information to users about the effect of using alternative inputs:

• the risk adjustment, which makes explicit provision for the uncertainty of inputs,

• the quantification of changes in inputs (paragraph 90(c)),

• sensitivity to insurance risk variables (paragraph 92(e)(i)), and

• sensitivity to market risk variables (paragraph 96(a)).

The sensitivity disclosures in paragraph 92(e)(i) should be expanded to include all material measurement inputs rather than just insurance risk variables. For example, the rates at which policyholders surrender or lapse their policies and the future expenses assumed are not considered insurance risk variables, yet changes in those inputs can have a material effect on measurement. We would suggest that sensitivity analysis be required for at least those inputs covered by paragraph 90(a).

15. Unit-linked contracts (paragraphs 8(a)(i), 71 and 78, Appendix C, and paragraphs BC153-BC155 and BC184-BC187)

Do you agree with the proposals on unit-linked contracts? Why or why not? If not, what do you recommend and why?

Response to 15:

We agree whole-heartedly with the presentation of unit-linked contracts described in paragraphs 71 and 78 and discussed in BC184-BC187. This presentation is consistent with what we do in Canada today, and lines up well with the nature of the underlying contracts.

We also agree with the treatment of assets described in Appendix C and paragraphs BC153-BC155 because it provides more relevant information to users.

However, as discussed in our response to question 12, we do not support the unbundling of unit-linked contracts. Notwithstanding this, we would suggest to the Board that, if a unit-linked account value were unbundled under paragraph 8(a)(i), the presentation requirements and asset valuations of paragraphs 71 and 78 and Appendix C should apply equally to investment contracts subject to IFRS 9.

20

16. Reinsurance (paragraphs 43-46 and BC230-BC241) 16a. Do you support an expected loss model for reinsurance assets? Why or why not? If not,

what do you recommend and why?

Response to 16a:

Yes, we support an expected loss model for reinsurance assets, since it is consistent with the expected value approach used for the underlying cash flows. We would ask the Board to clarify that the development of expected losses should take account of contractually allowed offsets or other undertakings provided by the reinsurers, such as letters of credit or amounts on deposit.

16b. Do you have any other comments on the reinsurance proposals?

Response to 16b:

• Recognition – Reinsurance contracts typically remain open to accept new business beyond the financial reporting date. Paragraph 14 could be interpreted to imply that a reinsurer must recognize a liability for a contract not yet written by the direct writer. We ask that a clarifying statement be added.

• Contract boundary – It is possible for the boundary of an insurance contract to be different from the boundary of a related reinsurance contract. In our view, it would be inappropriate for the ceding company to calculate the gross contract liability using one boundary and the related reinsurance asset using a longer boundary. We would ask the Board to add a statement to clarify that the boundary of a reinsurance asset would be no longer than the boundary of the underlying insurance contract.

• Residual margin – As noted in our response to question 6c, we recommend that the residual margin, if the Board insists on imposing one, be determined on a net of reinsurance basis to avoid unintended anomalies. The resulting residual margin could be split between the gross liability and the reinsurance ceded asset for presentation purposes.

• Short duration contracts – It is not clear how reinsurance of short duration contracts should be handled. Often the reinsurance contract remains open for new business beyond the 12-month period. Similarly to the point above on recognition, we ask the Board to make it clear that a reinsurer would not recognize a liability for a contract not yet recognized by the direct writer.

• Paragraph B25 – As noted in our response to question 11a, the proposed change to the definition of insurance risk in paragraph B25 could be problematic for reinsurers.

17. Transition and effective date (paragraphs 98-102 and BC244-BC257) 17a. Do you agree with the proposed transition requirements? Why or why not? If not, what

would you recommend and why?

21

Response to 17a:

Notwithstanding our objection to the residual margin in general, if one is required, then allowing no residual margin at transition would have the unfortunate effect of requiring ‘pre-IFRS’ business to be forever reported differently (and hence separately) from ‘post-IFRS’ business. It would also mean a permanent reduction in present value of future income on pre-IFRS business. Where the effect is material, it will result in distortion of results and could immediately bring the new IFRS method into disrepute.

However, we appreciate the Board’s objective in choosing a practical approach to this difficult issue. The stated alternative of requiring entities to calculate what the residual margin would have been at issue for each cohort is untenable. We would expect that many insurers would not have the necessary historical information available to do an accurate calculation.

Therefore, we recommend that the Board allow entities to estimate the residual margin for the in-force block of business at transition and require disclosure of the amount, the approach used to derive the estimate, and the method that will be used to amortize the amount. One possible method of estimation would be the present value of future profits calculated using current assumptions.

Unbundling at transition could also be a concern. For example, we would not have the “shadow accounts” described in our response to question 12.

17b. If the Board were to adopt the composite margin approach favoured by the FASB, would you agree with the FASB’s tentative decision on transition (see the appendix to the Basis for Conclusions)?

Response to 17b:

The comments above would be exacerbated if a composite margin were to be adopted. In particular, we note that after IFRS is adopted, with FASB’s approach there would be no future expected profits on pre-IFRS business.

17c. Is it necessary for the effective date of the IFRS on insurance contracts to be aligned with that of IFRS 9? Why or why not?

Response to 17c:

Yes, it is necessary. IFRS 9 will govern financial assets and financial liabilities other than insurance contracts. For insurers, the cost and complication of two separate conversions would be much higher. Also, it may not always be evident what the optimal choices are if one standard is effective before the other. In particular, the ability to elect a fair value option to reduce a mismatch between the measurement of assets and the measurement of liabilities could be affected by the implementation (or not) of the other standard.

22

We note that the Board has indicated that reporting entities would be given the opportunity to redesignate financial assets upon adoption of the insurance contract standard. We appreciate this clarification, since it is vital to insurers.

17d. Please provide an estimate of how long insurers would require to adopt the proposed requirements.

Response to 17d:

The answer would vary widely for different insurers. Taking Canada’s recent conversion to IFRS (where the measurement of insurance contracts did not change) as a guide, a large multi-line multi-national insurer would require at least two years lead time before the first comparative balance sheet is required. If the standard is finalized in 2011, this would suggest an effective date of January 1, 2015 at the earliest.

18. Other comments Do you have any other comments on the proposals in the exposure draft?

Response to 18:

• Subsequent measurement – The conditions in paragraphs 48(a) and 48(b) could often be mutually exclusive. It is common to have an improved estimate at the end of a reporting period with no change in conditions during the period. Paragraph B54 gives further guidance, which we understand to mean that 48(a) takes precedence over 48(b). We believe the Exposure Draft should be revised to clarify that the best information should be used to set estimates, regardless of how the change in an estimate correlates with a change in conditions over the period.

• We note that the measurement of a portfolio of contracts acquired is different from the value of the same portfolio of contracts written directly. It is not clear what should be done in subsequent measurements, especially when the acquired contracts are similar in nature to directly written contracts and would be combined in a single portfolio.

19. Benefits and costs Do you agree with the Board’s assessments of the benefits and costs of the proposed accounting for insurance contracts? Why or why not? If feasible, please estimate the benefits and costs associated with the proposals.

Response to 19:

In our responses above, we have identified particular areas where we believe costs outweigh benefits, viz. residual margin, unbundling and disclosure.