The issue and its background Many actuaries consider discounting for the time value of money to be fundamental to the assessment and measurement of expected future cash flows, although they recognize the practical exception of cases in which its effect is not material. In phase 2 of the IASB s insurance contracts project, it appears that this principle will be extended to the major area in which it has often not explicitly applied before, that is, to the measurement of claims liabilities of a non-life insurer. The IASB has so far clearly stated in its view that, unless contractually tied (i.e., variable/unit-linked or certain participating contracts), the measurement of liabilities of financial instruments and related contracts such as insurance contracts, should be treated as being independent from assets. Under such standards, the discount rates applied should be de-linked from their corresponding assets that an insurer holds. This is in part due to the fact that assets are fungible, that is, can be traded without significantly affecting current value. In addition, also to date, the IASB has indicated that expected future investment income in excess of the risk-free rate (or in some cases, such as for pension plans in IAS 19, see below, in excess of a high grade corporate rate) should not be anticipated. Note that the phrase risk-free usually only includes default risk free, as even government securities carry other risks, e.g., market risk, sovereign risk, inflation risk, and in some cases liquidity risk. Of course, IFRS 4 as a standard only intended to be in effect as is for a limited period of time is currently an exception to this, in that for insurance contracts previous accounting policy maintaining such a practice is currently acceptable, although a change in accounting policy that introduces such a measurement approach is not permitted. One difficulty with the use of risk-free rates in the measurement of the liability for many life insurance contracts because, since most insurers anticipate that they will be able to earn something greater than government security rates, many such contracts (as described in the first IAA-ACLI research paper prepared two years ago) would result in a loss at issue. It is not clear how useful this information would be, as it would be very inconsistent with the insurer s expectations, although not necessarily inconsistent with IFRS. Although risk adjustment in the measurement of a liability can conceptually be incorporated in the measurement of cash flows or the discount rate, it is more common to incorporate risk together with the estimation of the cash flows, as the risks reflected are generally not simply a function of the passage of time, that is, are primarily related to the amount of the expected cash flows. The risks associated with future investment income are, at least to date, also not been anticipated as the interest to be earned in excess of the risk free rate are not anticipated. 1
However, except as given below, the IASB has not provided specific guidance regarding how to measure risk-free rates. As a result, especially in the case of countries in which an active and deep market in government securities of longer durations do not exist, some questions have been raised regarding how to determine risk-free rates. Current guidance and current IASB projects related to discounting Current guidance in IFRS or IFRS related material is provided in the Appendix. In addition, attached is a copy of a paper that the IAA prepared several years ago, entitled Valuation of Risk Adjusted Cash Flows and the Setting of Discount rates Theory and Practice that describes a method utilizing replicating portfolios. The IASB-related projects on Measurement (a research project being led by the Canadian Institute of Chartered Accountants (CICA)) and the FASB project on fair values may provide background or guidance to discounting. Alternative approaches that might be considered With the above discussion as background and assuming that the liability or asset should reflect the time value of money, the issues that would be appropriate for the IAA to discuss and recommend include the following: 1. What should the basis be for measurement of the discount rate used in the measurement of insurance contract (and claim) liabilities? Possible answers might include: a. Duration-specific government spot-rates. These might be measured from the normal financial market in which the central government fixed income securities are sold and traded or spot rate market, if there is a spot rate market in the country country. What are other alternatives, particularly if these aren t publicly available in a timely manner? b. Duration-specific high grade corporate rates, the level to either be dictated (e.g., A rating) or flexible depending on the circumstances / risks associated with the liability. c. LIBOR. d. An average discount rate corresponding to the expected duration of the cash flows to which the discount rates will apply. The disadvantage is whether this would create a bias if there is a steep yield curve (either decreasing or increasing) and that the duration would have to be constantly updated. e. Expected yield rate of the corresponding assets adjusted for expected default risk (if there is a specified set of assets so designated or if an expected mix of future investments can be specified), reflecting expected financial effect of asset defaults (plus credit fault risk?). Presumably this would be an expected portfolio rate, but how would it be affected by expected future investments? i. Issues with this approach include what should be the basis of such expected asset defaults (various studies of bonds in the U.S, have determined that somewhere between one-third and one-half the 2
nominal credit spread would be an expected level, but limited studies of the expected variance of this expectation. ii. How should the expected equity risk premium be measured if equities are used as that asset (or other non-fixed securities such as mortgage loans or property)? 2. How to derive risk-free rates when there is no active and deep market for applicable securities in a country? a. Countries with significant market-perceived (or central bank driven) sovereignty risk, e.g., currently in Brazil with 20% b. Beyond period in which actively traded securities are available 3. Is it ever appropriate to reflect an expected equity risk premium? 4. Is it ever appropriate to include the risk associated with the expected cash flows in the discount rate? 5. Is it ever appropriate to include a financial risk adjustment in the discount rate? 6. Should the insurer s own credit standing be reflected (possibly net of a third party s stand-ready guarantee obligation)? Although in non-linked cases, discount rates for liability measurement may, according to IFRS, have to be independent of the actual assets held, this constraint may not necessarily be applicable for liability adequacy testing or measurement of reinsurance assets, although it might take the form of a replicating portfolio of assets allocated to the liabilities. Recommendation and basis for recommendation (to be determined later) 3
Appendix -- IFRS and other references IAS 19 Employee Benefits provides guidance relating to the discounting of postemployment benefit obligations on some of the issues raised here. Paragraph 78 indicates the rate used to discount should be determined by reference to market yields at the balance sheet date on high quality corporate bonds. In countries where there is no deep market in such bonds, the market yields (at the balance sheet date) on government bonds should be used. The currency and term of the corporate bonds or government bonds should be consistent with the currency and estimated term of the post-employment benefit obligations. Paragraph 79 goes on to indicate that The discount rate reflects the time value of money but not the actuarial or investment risk. Furthermore the discount rate does not reflect the enterprise-specific credit risk borne by the enterprise s creditors, nor does it reflect the risk that future experience may differ from actuarial assumptions. Paragraph 80 describes a practical approach In practice, an enterprise often achieves this by applying a single weighted average discount rate that reflects the estimated timing and amount of benefit payments and the currency in which the benefits are to be paid. Paragraph 81 goes on to say In some cases, there may be no deep market in bonds with a sufficiently long maturity to match the estimated maturity of all the benefit payments. In such cases, an enterprise uses current market rates of the appropriate term to discount shorter term payments, and estimates the discount rate for longer maturities by extrapolating current market rates along the yield curve. IAS 36, Impairment of Assets, paragraphs 48-56 provides certain guidance regarding the discount rates applicable for impairment testing: 48. The discount rate (or rates) should be a pre-tax rate (or rates) that reflect(s) current market assessments of the time value of money and the risks specific to the asset. The discount rate(s) should not reflect risks for which future cash flow estimates have been adjusted. 49. A rate that reflects current market assessments of the time value of money and the risks specific to the asset is the return that investors would require if they were to choose an investment that would generate cash flows of amounts, timing and risk profile equivalent to those that the enterprise expects to derive from the asset. This rate is estimated from the rate implicit in current market transactions for similar assets or from the weighted average cost of capital of a listed enterprise that has a single asset (or a portfolio of assets) similar in terms of service potential and risks to the asset under review. 50. When an asset-specific rate is not directly available from the market, an enterprise uses surrogates to estimate the discount rate. The purpose is to estimate, as far as possible, a market assessment of: (a) the time value of money for the periods until the end of the asset s useful life; and (b) the risks that the future cash flows will differ in amount or timing from estimates. 4
51. As a starting point, the enterprise may take into account the following rates: (a) the enterprise s weighted average cost of capital determined using techniques such as the Capital Asset Pricing Model; (b) the enterprise s incremental borrowing rate; and (c) other market borrowing rates. 52. These rates are adjusted: (a) to reflect the way that the market would assess the specific risks associated with the projected cash flows; and (b) to exclude risks that are not relevant to the projected cash flows. Consideration is given to risks such as country risk, currency risk, price risk and cash flow risk. 53. To avoid double counting, the discount rate does not reflect risks for which future cash flow estimates have been adjusted. 54. The discount rate is independent of the enterprise s capital structure and the way the enterprise financed the purchase of the asset because the future cash flows expected to arise from an asset do not depend on the way in which the enterprise financed the purchase of the asset. 55. When the basis for the rate is post-tax, that basis is adjusted to reflect a pre-tax rate. 56. An enterprise normally uses a single discount rate for the estimate of an asset s value in use. However, an enterprise uses separate discount rates for different future periods where value in use is sensitive to a difference in risks for different periods or to the term structure of interest rates. IAS 37, Provisions, Contingent Liabilities and Contingent Assets. IAS 37 comes to some different conclusions than IAS 19, but is far briefer. Its paragraph 47 indicates The discount rate(s) should be a pre-tax rate(s) that reflect(s) current market assessments of the time value of money and the risks specific to the liability. The discount rate(s) should not reflect risks for which future cash flows estimates have been adjusted. IAS 39, Financial Instruments: Recognition and Measurement provides two descriptions of an appropriate discount rate to be used as a basis for the estimation of fair values (in contrast with measurement on the basis of observable transactions of identical or similar instruments). Paragraph 67 which considers initial measurement, indicates: If such market prices are not reliably determinable, the fair value of the consideration is estimated as the sum of all future cash payments or receipts, discounted, if the effect of doing so would be material, using the prevailing market rate(s) of interest for a similar instrument (similar as to currency, term, type of interest rate, and other factors) of an issue with a similar credit rating (see IAS 18 Revenue, paragraph 11). 5
IAS 39, paragraph 100, which addresses fair value measurement considerations, states: In other circumstances, as well as when a quoted market price is not available, estimation techniques may be used to determine fair value with sufficient reliability to satisfy the requirements of this Standard. Techniques that are well established in financial markets include reference to the current market value of another instrument that is substantially the same, discounted cash flow analysis, and option pricing models. In applying discounted cash flow analysis, an enterprise uses the discount rate(s) equal to the prevailing rate of return for financial instruments having substantially the same terms and characteristics, including the creditworthiness of the debtor, the remaining term over which the contractual interest rate is fixed, the remaining term to repayment of the principal, and the currency in which payments are to be made. The Joint Working Group of Standard Setters 1999 paper on Financial Instruments indicated in a discussion of the factors that would be reflected in the measurement of fair values through a model (in contrast to observations) on page 120 of its Application Supplement Measurement : The time value of money (that is, interest at the basic or risk-free rate). Basic interest rates usually can be derived from observable government bond prices and often are quoted in financial publications. These rates typically vary with the expected dates of the projected cash flows as a result of the yield curve effect. For practical reasons, an enterprise may use a well accepted and readily observable general rate, such as LIBOR/swap rate, as the benchmark rate. (Since a rate such as LIBOR is not the basic interest rate, the credit risk adjustment appropriate to the particular financial instrument would be determined on the basis of its credit risk in relation to the credit risk in this benchmark rate.) Ins some countries, the central government s bonds may carry a significant credit risk and may not provide a useful, stable benchmark basic interest rate for enterprises issuing financial statements in that reporting currency. Some enterprises in these countries may have better credit standings and lower borrowing rates than the central government. In such a case, basic interest rates may be more appropriately determined by reference to interest rates for the highest rated corporate bonds issued in the currency of that jurisdiction. 6