This Comment Letter was sent by BDO Global Coordination B.V. on behalf of BDO International, to the International Accounting Standards Board in November, 2007:
Dear Mr Clarke,
Discussion Paper ‘Preliminary Views on Insurance Contracts’
We are pleased to have the opportunity to comment, on behalf of BDO International1, on the proposals of the International Accounting Standards Board (IASB) in the above Discussion Paper.
In general we are supportive of the Discussion Paper, as there is room for improvement in the way insurance contracts are accounted for. In particular, there is a critical need for consistency. The Discussion Paper represents the first step in the development of a comprehensive standard on the subject.
We discuss the questions to respondents set out in the Discussion Paper in the appendix to this letter. Our main comments are summarised below.
1. We consider that insurance accounting should be based upon accounting principles and accounting standards developed for general use. Rules and exceptions should not, to the extent possible, be made for this industry but rather guidance given on the application of those general principles and accounting standards. Thus, in our view, insurance accounting should be consistent with IAS1, IAS18, IAS37, IAS32 and IAS39 in relation to presentation, revenue, provisions and constructive obligations, the classification of debt versus equity and the recognition, derecognition and measurement of financial instruments respectively.
2. We are not persuaded that, in the absence of observable market data, entity specific data should be ignored in all circumstances or that hypothetical market data is superior to entity specific data.
3. We are supportive of the IASB’s objective of eliminating or minimising accounting mismatches through review of general principles. We are, however, less supportive of achieving this objective at the expense of consistency in the general principles of recognition and measurement or through the use of rules and exceptions, although we recognise that in certain cases there will be a need for this approach to be followed.
If you would like to discuss these comments, please contact Helen Thomson of BDO Global Coordination B.V. on +32 2 778 01 30.
Yours sincerely,
BDO Global Coordination B.V.
Appendix
Our comments on the specific questions are set out below.
Question 1
Should the recognition and derecognition requirements for insurance contracts be consistent with those in IAS 39 for financial instruments? Why or why not?
We would, in most respects, support the recognition and derecognition requirements for insurance contracts being consistent with those in IAS 39 as we consider that insurance accounting should be subject to accounting principles and standards developed for general use.
However, we note that IAS 39 contains a ‘bright line’ 10% derecognition test for financial liabilities. We consider that this implementation guidance should be revisited in the context of insurance contracts.
Question 2
Should an insurer measure all its insurance liabilities using the following three building blocks:
(a) explicit, unbiased, market-consistent, probability-weighted and current estimates of the contractual cash flows,
(b) current market discount rates that adjust the estimated future cash flows for the time value of money, and
(c) an explicit and unbiased estimate of the margin that market participants require for bearing risk (a risk margin) and for providing other services, if any (a service margin)?
If not, what approach do you propose, and why?
We are supportive of the three building block approach within an insurance accounting standard as implementation guidance for liabilities for which there is no quoted price, active market or observable similar market transactions.
Further, we are of the view that insurance accounting should be subject to the same accounting principles as developed for general use. We believe that the measurement of insurance liabilities should be at ‘fair value’ using the definitions and guidance in IAS39.9 and IAS39.48-49 and IAS39.AG69-82. In particular, with reference to IAS39.48A, where there is no quoted price, active market or observable similar market transactions the liability needs to be measured using a valuation technique that relies ‘as little as possible on entity specific inputs’.
In some instances entity specific data may be the most appropriate measure of fair value for insurance liabilities and should not be prohibited. However, consistent with the guidance in IAS39.AG76, where entity specific information is used, we do not believe a gain or loss on initial recognition should be recorded (see our response to question 4 below). In other words, on initial recognition the insurance liability should normally be measured at an amount equal to the transaction price.
Question 3
Is the draft guidance on cash flows (appendix E) and risk margins (appendix F) at the right level of detail? Should any of that guidance be modified, deleted or extended? Why or why not?
We believe that the level of detail in appendix F is appropriate.
Question 4
What role should the actual premium charged by the insurer play in the calibration of margins, and why? Please say which of the following alternatives you support.
(a) The insurer should calibrate the margin directly to the actual premium (less relevant acquisition costs), subject to a liability adequacy test. As a result, an insurer should never recognise a profit at the inception of an insurance contract.
(b) There should be a rebuttable presumption that the margin implied by the actual premium (less relevant acquisition costs) is consistent with the margin that market participants require. If you prefer this approach, what evidence should be needed to rebut the presumption?
(c) The premium (less relevant acquisition costs) may provide evidence of the margin that market participants would require, but has no higher status than other possible evidence. In most cases, insurance contracts are expected to provide a margin consistent with the requirements of market participants. Therefore, if a significant profit or loss appears to arise at inception, further investigation is needed. Nevertheless, if the insurer concludes, after further investigation, that the estimated market price for risk and service differs from the price implied by the premiums that it charges, the insurer would recognise a profit or loss at inception.
(d) Other (please specify).
We would be supportive of (c) above if the valuation of the insurance liability was based on observable market, rather than entity specific information. This is consistent with the application guidance set out in IAS39.AG76 which notes that:
‘…The best evidence of the fair value of a financial instrument at initial recognition is the transaction price (i.e. the fair value of the consideration given or received) unless the fair value of that instrument is evidenced by comparison with other observable current market transactions in the same instrument (i.e. without modification or repackaging) or based on a valuation technique whose variables include only data from observable markets.’
However, some insurance products may not have associated observable market data. In such cases, the use of entity specific information should be permitted. If, in these cases, the use of entity specific data were to be permitted, our response to this question would be that we would support option (c) unless entity specific data was used to measure the liability in which case option (a) should be used in order that no gain or loss is recorded on initial recognition.
Question 5
This paper proposes that the measurement attribute for insurance liabilities should be the amount the insurer would expect to pay at the reporting date to transfer its remaining contractual rights and obligations immediately to another entity. The paper labels that measurement attribute ‘current exit value’.
(a) Is that measurement attribute appropriate for insurance liabilities? Why or why not? If not, which measurement attribute do you favour, and why?
(b) Is ‘current exit value’ the best label for that measurement attribute? Why or why not?
We are not persuaded that current exit value as defined in the Discussion Paper is the most appropriate measure of insurance liabilities and whilst we concur that market data for specific insurance contracts may not always be available, we would prefer the hierarchical approach to fair value measurement in IAS39 as it mandates the use of observable market data (or data linked to market data) where it is available. Where observable market data is not available, the use of entity specific data in measurement should not be prohibited as it may be the most appropriate measure of the liability where the only alternative is hypothetical market data which itself has no actual market observable data inputs to the model.
Question 6
In this paper, beneficial policyholder behaviour refers to a policyholder’s exercise of a contractual option in a way that generates net economic benefits for the insurer. For expected future cash flows resulting from beneficial policyholder behaviour, should an insurer:
(a incorporate them in the current exit value of a separately recognised customer relationship asset? Why or why not?
(b) incorporate them, as a reduction, in the current exit value of insurance liabilities? Why or why not?
(c) not recognise them? Why or why not?
We concur with the view expressed in paragraph 174 of the Discussion Paper, i.e. that if the insurer can not compel the payment of future premiums, then to recognise such future premiums as an asset or as a reduction in a insurance liability would be inconsistent with other IFRSs and therefore would not be appropriate. We are not persuaded by the rationale in paragraph 142 of the Discussion Paper that the existence of a contractual option to renew is a sufficiently robust conceptual argument for allowing the recognition of this internally generated customer relationship as IAS38 does not permit the recognition of other similar internally generated assets.
Question 7
A list follows of possible criteria to determine which cash flows an insurer should recognise relating to beneficial policyholder behaviour. Which criterion should the Board adopt, and why?
(a) Cash flows resulting from payments that policyholders must make to retain a right to guaranteed insurability (less additional benefit payments that result from those premiums). The Board favours this criterion, and defines guaranteed insurability as a right that permits continued coverage without reconfirmation of the policyholder’s risk profile and at a price that is contractually constrained.
(b) All cash flows that arise from existing contracts, regardless of whether the insurer can enforce those cash flows. If you favour this criterion, how would you distinguish existing contracts from new contracts?
(c) All cash flows that arise from those terms of existing contracts that have commercial substance (i.e. have a discernible effect on the economics of the contract by significantly modifying the risk, amount or timing of the cash flows).
(d) Cash flows resulting from payments that policyholders must make to retain a right to any guarantee that compels the insurer to stand ready, at a price that is contractually constrained, (i) to bear insurance risk or financial risk, or (ii) to provide other services. This criterion relates to all contractual guarantees, whereas the criterion described in (a) relates only to insurance risk.
(e) No cash flows that result from beneficial policyholder behaviour.
(f) Other (please specify).
Whilst our preference would be for consistency with IAS38 and thus that only contractually enforceable cash flows should be taken into account as per option (e), we believe that there is an argument, as per the IASB’s view in paragraph 173 of the Discussion Paper, that there is something of value, being the future premiums the policy holder must pay to retain guaranteed insurability and therefore we would also support option (a).
However, if this were to be recognised on balance sheet as an asset, this might result in the recognition of an asset which has attributes which are inconsistent with the definition of an asset in the Framework and (as noted above) the guidance in IAS 38. We note that the IASB is currently considering the definition of an asset in its project to amend the Framework and suggest that, as part of that review, consideration should be given to associated features of insurance contracts.
Question 8
Should an insurer recognise acquisition costs as an expense when incurred? Why or why not?
Assuming that any new insurance standard requires insurance liabilities to be measured initially and subsequently at fair value through profit or loss, an insurer should recognise acquisition costs as an expense when incurred. We note that this approach would be consistent with that of IAS 39 when a financial asset or liability is measured at fair value through profit or loss.
In addition, we note that in respect of transaction costs, it can be difficult to demonstrate the control attribute which is necessary for asset recognition.
Question 9
Do you have any comments on the treatment of insurance contracts acquired in a business combination or portfolio transfer?
We have no comments, other than the approach should be consistent with that set out in IFRS 3.
Question 10
Do you have any comments on the measurement of assets held to back insurance liabilities?
In order to assist in eliminating or minimising accounting mismatches, it would be appropriate to include an IAS 39 style option to record, on initial recognition, all financial assets held to back insurance liabilities as at fair value through profit or loss.
We are supportive of the IASB’s objective to eliminate or minimise accounting mismatches and therefore would support the option in paragraph 280(a) of extending the IAS39 fair value option to other non financial assets.
We are not supportive of the option in paragraph 280(b) that the IASB should permit or require insurers to account for items such as treasury stock that do not qualify for recognition as an asset at fair value to profit or loss. We consider that insurance accounting should not introduce an inconsistency with IFRS, such that an entity’s own equity instruments are accounted for as assets.
Question 11
Should risk margins:
(a) be determined for a portfolio of insurance contracts? Why or why not? If yes, should the portfolio be defined as in IFRS 4 (a portfolio of contracts that are subject to broadly similar risks and managed together as a single portfolio)? Why or why not?
(b) reflect the benefits of diversification between (and negative correlation between) portfolios? Why or why not?
We concur with the view expressed in paragraphs 186-189 and 202(a) and (b), i.e. that the unit of account does not affect the expected present value of future cash flows and that risk margins should be determined for a portfolio of insurance contracts that are managed together and have similar risks but that diversification and negative correlation should not be taken into account, as this is consistent with measurement criteria in IAS39.
Question 12
(a) Should a cedant measure reinsurance assets at current exit value? Why or why not?
(b) Do you agree that the consequences of measuring reinsurance assets at current exit value include the following? Why or why not?
(i) A risk margin typically increases the measurement of the reinsurance asset, and equals the risk margin for the corresponding part of the underlying insurance contract.
(ii) An expected loss model would be used for defaults and disputes, not the incurred loss model required by IFRS 4 and IAS 39.
(iii) If the cedant has a contractual right to obtain reinsurance for contracts that it has not yet issued, the current exit value of the cedant’s reinsurance asset includes the current exit value of that right. However, the current exit value of that contractual right is not likely to be material if it relates to insurance contracts that will be priced at current exit value.
We consider that cedants should measure insurance liabilities and assets following the same general principles as those developed for other preparers including other insurers. Therefore we do not support the use of a different measurement model to that required by IAS39 as suggested in (b)(ii) of this question.
We note that reinsurance contracts and their associated cash flows can be complex, and may be difficult to measure. Consequently, we encourage the IASB to carry out appropriate field testing in this area before the requirements proposed in an exposure draft, or included in a new accounting standard, are finalised.
Question 13
If an insurance contract contains deposit or service components, should an insurer unbundle them? Why or why not?
We are of the view that, consistent with other standards, a single instrument or contract should be unbundled and its significant parts accounted for separately unless the recognition and measurement would be unchanged (in the same way as one of the current tests for the separation of embedded derivatives operates in IAS 39). In this respect, the appropriate approach will depend on the results of the IASB’s long term project on financial instruments.
It would be helpful if Discussion Paper could expand the discussion surrounding the more general principle i.e. that a ‘deposit’ element within a contract (insurance or any other contract) should be unbundled and accounted for as a separate financial instrument under IAS39 when it is in substance a ‘savings deposit’, whereas a refundable prepayment is not a financial instrument and therefore need not be unbundled and accounted for separately. In our view the Discussion Paper is not sufficiently clear that this is the general principle and that the guidance in paragraph 228 is only applicable when a ‘deposit’ that passes the definition of a financial instrument and thus should be unbundled has been identified.
We consider that service components within insurance contracts should be unbundled and accounted for as services are provided in accordance with IAS18.
Question 14
(a) Is the current exit value of a liability the price for a transfer that neither improves nor impairs its credit characteristics? Why or why not?
(b) Should the measurement of an insurance liability reflect (i) its credit characteristics at inception and (ii) subsequent changes in their effect? Why or why not?
We agree that the current exit value, as defined in the Discussion Paper, of a liability is the price for a transfer that neither improves nor impairs its credit characteristics. Please see above for our comments regarding the measurement of insurance contracts.
Question 15
Appendix B identifies some inconsistencies between the proposed treatment of insurance liabilities and the existing treatment under IAS 39 of financial liabilities. Should the Board consider changing the treatment of some or all financial liabilities to avoid those inconsistencies? If so, what changes should the Board consider, and why?
We would be in favour of eliminating or minimising differences between any new insurance standard and IAS39, and indeed the Framework, IAS37 and IAS18. We are not of the view that the principles governing insurance accounting should be different from principles developed for general use merely due to the type of industry involved.
In particular we are not convinced by arguments that measurement on initial recognition should be different from that which would be obtained from the application of IAS 39, as discussed in our response to questions 2 and 4 above. In particular, we consider that the recognition of a gain or loss on initial recognition should be rare.
However, any commentary on current inconsistencies which might arise may be made redundant in the context of the IASB’s long term project for financial instruments. We suggest that it may be appropriate for the IASB first to consider the responses to the Insurance Discussion Paper, and then consider how inconsistencies might be dealt with as both the insurance and financial instruments projects progress.
Question 16
(a) For participating contracts, should the cash flows for each scenario incorporate an unbiased estimate of the policyholder dividends payable in that scenario to satisfy a legal or constructive obligation that exists at the reporting date? Why or why not?
(b) An exposure draft of June 2005 proposed amendments to IAS 37 (see paragraphs 247–253 of this paper). Do those proposals give enough guidance for an insurer to determine when a participating contract gives rise to a legal or constructive obligation to pay policyholder dividends?
As noted above, we consider that the recognition and derecognition requirements of IAS 39 should (other than in certain respects, such as the 10% derecognition threshold for financial liabilities) apply to insurance accounting. Consequently the focus should be on contractual obligations, with these then being measured on the basis of expected cash flows.
If an additional amount to be paid is non contractual, but meets the definition of a constructive obligation in IAS 37, then we consider that the measurement of the liability should be based on probability weighted expected cash flows, and the policy holder’s share of underlying assets from which the discretionary distribution is to be made. However, in determining the amount to be recorded, no account should be taken of expected returns from future reinvestment activities.
Question 17
Should the Board do some or all of the following to eliminate accounting mismatches that could arise for unit-linked contracts? Why or why not?
(a) Permit or require insurers to recognise treasury shares as an asset if they are held to back a unit-linked liability (even though they do not meet the Framework’s definition of an asset).
(b) Permit or require insurers to recognise internally generated goodwill of a subsidiary if the investment in that subsidiary is held to back a unit-linked liability (even though IFRSs prohibit the recognition of internally generated goodwill in all other cases).
(c) Permit or require insurers to measure assets at fair value through profit or loss if they are held to back a unit-linked liability (even if IFRSs do not permit that treatment for identical assets held for another purpose).
(d) Exclude from the current exit value of a unit-linked liability any differences between the carrying amount of the assets held to back that liability and their fair value (even though some view this as conflicting with the definition of current exit value).
Please see our response to question 10 above. We would prefer general principles be re-examined and changed or refined to eliminate mismatches where appropriate rather than making rules and exceptions for a certain industry. However, if this is currently impractical we would be supportive of extending the fair value option for insurance companies to all assets that are held to back a unit-linked liability so long as the asset meets the definition of an asset in the Framework. Any mismatches that cannot be eliminated or minimised through the application of general principle or the extension of that general principle of a fair value option just for insurance companies, such as items that do not qualify for recognition as assets, should rather be disclosed and explained.
We do not agree that treasury shares should be recognised as an asset.
Question 18
Should an insurer present premiums as revenue or as deposits? Why?
We are of the view that the principles in IAS18 relating to revenue and unbundling of services should be applicable to insurance accounting. Thus in our view simple premiums should be accounted for in accordance with IAS18 as revenue as the insurance service is provided and any payments to policy holders should be accounted for as expenses. Further, in the case of contracts with investment elements (saving deposits) or discretionary participating features, such elements should be unbundled and accounted for, in accordance with IAS39, as financial instruments.
Question 19
Which items of income and expense should an insurer present separately on the face of its income statement? Why?
We prefer the principle based approach in existing IAS1, which requires additional line items if necessary for understanding an entity’s financial performance, with a list of example items that an entity might include on the face or in the notes similar to that given in IFRS4.IG26. We do not see why insurance accounting should be subject to different presentation principles than those developed for general use and therefore would not support industry specific disclosure requirements for items on the face of the income statement.
Question 20
Should the income statement include all income and expense arising from changes in insurance liabilities? Why or why not?
We would only agree with changes in insurance liabilities being recognised outside of the income statement in order to address accounting mismatches. However, our preference would be for all income and expenses arising from changes in insurance liabilities to be included in the income statement, and for accounting mismatches to be eliminated or minimised by the requirements or options contained in any new insurance standard.
Discussion Paper, Preliminary Views on Insurance Contracts, Comments due 30 November 2007:
Preliminary Views on Insurance Contracts-Part 1.pdf
Preliminary Views on Insurance Contracts-Part 2.pdf