This document is provided as the basis for discussion with the objective to develop an ACLI
position on the subject matter. These notes are based on staff’s analysis of tentative views of
the IASB and FASB, reference to various resource documents, and prior ACLI discussions and
position statements expressed in letters to accounting standard setters on the topic.
INFORMATION FOR DEPUTIES SUBGROUP
Project: Insurance Contracts
Topic: Discount rate
This paper discusses various approaches to setting and applying the discount rate to the expected cash
flows in measuring the insurance contract liabilities (for a portfolio of contracts) based upon the
building blocks approach set forth in the 2007 IASB Discussion Paper: Preliminary Views on
Insurance Contracts (DP). The proposed measurement approach would be a quasi-fair value model,
i.e., current value.
This paper concentrates on those methods having the greatest likelihood of meeting the measurement
objective. The American Academy of Actuaries (AAA) September 2009 White Paper1 titled
“Discussion on the use of Discount Rates in Accounting Present Value Estimates” is a key resource
used in developing the content of this paper. The Subgroup, however, should not rule out other
methods not described herein if such methods are determined to be a candidate. The following
methods have been identified (not in any order of significance or importance):
1. Asset earned rate
2. Crediting rate implicit in the contract
3. High quality corporate bonds
4. Risk-free rate adjusted for liquidity
While the discount rate is the topic of discussion, the Subgroup must be careful in its deliberations not
to make a decision prematurely, i.e., the Subgroup must consider the effects of the decision on other
elements of the measurement, e.g., explicit margins. The Subgroup has noted the significant
interrelationship of insurance contract features and options. Consequently, a decision without
consideration of the interrelationships could result in an inconsistent measurement approach.
Structure of the paper
b. Discount rate candidates
c. Existing IASB accounting guidance about discount rates
d. Questions for the Subgroup
e. Appendix A- Discount Rates for Insurance Contracts memo by Bill Schwegler and Carrie
f. Appendix B- ACLI Discount Rate Modeling Project – Fixed Deferred Annuities prepared
by Carrie Morton
The White Paper is an addendum to this staff paper and serves as additional material to be taken into account
in developing the ACLI position on discount rate.
The scope of the discussion should take into account all contract types, i.e., short-duration and long-
duration contracts, participating (including UL) and non-participating contracts, variable and fixed
contracts. In addition, the discussion should consider subsequent measurement of the discount rate
based on current market inputs. Any decision about the discount rate should not carve-out any
segment of the insurance business unnecessarily. This paper does not provide detailed research,
commentary or examples. Rather, the paper describes in a summarized way key points for and against
the various methods to assist the Subgroup in the decision making process.
Discount rate candidates
The IASB has tentatively decided that: a) the discount rate for insurance liabilities should
conceptually adjust estimated future cash flows for the time value of money in a way that captures the
characteristics of that liability rather than using a discount rate based on expected returns on actual
assets backing those liabilities b) the standard should not give detailed guidance on how to determine
the discount rate. The FASB has not yet arrived at a decision on this topic. The IASB’s tentative view,
while consistent with a principles-based approach, may not provide sufficient guidance for preparers.
In the ACLI February 6, 2009 letter to the IASB regarding “Should credit characteristics of insurance
liabilities affect their measurement?” the position expressed in the letter was:
“Consequently, we believe that the credit risk of the insurance contract at initial recognition
should reflect credit characteristics equivalent to a high quality debt instrument, e.g., AA -
AAA rating and that subsequent measurement should not reflect changes in credit standing.
Because of the capital requirements and guaranty funds, no change in the credit risk
assumption is necessary for subsequent measurement. The rationale is that policyholders and
beneficiaries are first in line in the event of insolvency of the insurer and that regulators
typically take action well in advance to significantly minimize additional loss in value of the
assets to ensure payment. Where there are not sufficient assets to pay claims as they come
due, the guaranty fund system provides additional resources.”
While the letter was specific to the issue of credit risk, it is relevant in the discussion on discount rate.
Critical to the discussion of the discount rate is the way that the provision for risk should be taken into
account. For example, should the discount rate reflect market-observable discount rates for cash
flow streams with similar timing and risk characteristics or should the cash flows be risk
Asset earned rate
Many industry responses to the DP commented on the discount rate urging the Board not to use a
risk-free rate since the result would likely mean a loss at issue on profitable business. A typical
response from US organizations was “we recommend using a discount rate that reflects the rate of
return that the reporting entity expects to earn on the portfolio of assets backing the liability.” While
the industry has often argued that the asset earned rate is fundamental in the pricing of insurance
contracts especially long-duration contracts, nevertheless, the IASB has tentatively rejected the use of
the asset earned rate in the measurement of insurance contracts. The junk bond example is often cited
as the scenario against using the asset earned rate noting that the earned rate of a portfolio of junk
bonds cannot serve as the discount rate for the insurance contract liabilities-especially non-par
While the asset earned rate may not be appropriate for all insurance contracts, it may be appropriate
for certain contracts such as participating contracts, UL contracts, and variable contracts, where the
policyholder assumes the investment risk.
Crediting rate implicit in the contract
A modified version of the asset earned rate is the crediting rate, which is defined herein as the rate
used by the pricing actuary in setting the gross premium. For some contracts, such as short-duration
Discount Rates for Insurance Contracts
Prepared by Bill Schwegler (Aegon) and Carrie Morton (Principal Financial Group)
Date: September 23, 2009
IASB Agenda Paper 17D: Discounting describes several possible approaches for selecting a discount
rate for insurance contracts. These approaches can be summarized as follows:
1. Earned rate (paragraph 12)
2. Risk-free rate with adjustment for liquidity (paragraphs 17-21)
3. Observable market rates (paragraph 23)
a. High-quality corporate bonds
b. High-quality fixed-income debt instruments (this appears to be similar to 3.a.)
c. Risk-free rate
The purpose of this paper is to describe the advantages and disadvantages of each of the approaches
presented in the IASB Agenda Paper.
A few respondents to the DP argued that the discount rate should reflect the expected returns on the
assets backing the liabilities. However, these respondents were in the minority, and the IASB does
not support this approach.
o Consistent with common pricing practices for insurance contracts
o Would not cause an artificial loss to be reported at inception (unless other elements of the
accounting standard introduce excess conservatism)
o Measures liabilities and assets at a consistent discount rate, thus reducing “artificial volatility”
o If the liability is discounted at the risk-free rate, a change in credit spreads after inception
would cause a reduction in the value of the assets backing the liability, with no
corresponding reduction in the liability value. However, discounting at the earned rate
would eliminate this mismatch.
o May reduce comparability between companies
o Suppose that two companies sell products with identical product features, premiums and
cash flows, but Company A invests in treasury bonds, while Company B invests in BBB-
rated corporate bonds. Company B would use a higher discount rate than Company A,
and therefore, Company B would report a lower present value of future cash flows.
However, because the margin is calibrated to the initial premium, the two companies
would likely report the same total liability (provided that they charge the same premium).
o The earned rate is a characteristic of the assets, not the liability
o The IASB staff and the majority of respondents to the DP have expressed the view that
the liability measurement should reflect only the characteristics of the liability,
independent of the assets backing the liability.
o Could create perverse management incentives
o In a “current value” framework, an increase in the riskiness of the asset portfolio would
increase the discount rate and reduce the liabilities. Therefore a company could appear to
improve its financial standing by taking on additional asset risk.
Risk-free rate with adjustment for liquidity
In IASB Agenda Paper 17D: Discounting, the IASB staff expresses the view that the discount rate for
a liability should be based on the risk-free rate, plus a liquidity premium reflecting the liquidity
characteristics of the liability.
o Reflects only the characteristics of the liability (not the assets backing the liability)
o Uses an observable market value (the risk-free rate) as a starting point
o The liquidity premium generally is not a readily observable market value
o To estimate the liquidity premium, we could start by looking at the difference in yields
between highly liquid instruments (e.g. government bonds) and less liquid instruments
(e.g. corporate bonds). However, the difference in yield is due to differences in both
liquidity characteristics and credit quality. It may be difficult to determine precisely
which portion of the yield differential is due to a liquidity premium and which portion is
due to credit risk.
o Could lead to diversity in practice, as different companies may assign different values to the
liquidity premium for similar products
o May lead to a loss at issue for many products
o When pricing a product, insurers generally consider the yield that they expect to earn on
the assets backing the liability. This yield is composed of a risk-free rate, a liquidity
premium, and a credit spread. Excluding the credit spread from the discount rate may
result in a discount rate that is significantly lower than the earned rate assumed in pricing.
This may result in a loss at issue for many products, even those that are expected to be
o May create “artificial volatility” in the income statement, even if assets and liabilities are properly
o A change in credit spreads after inception would cause a reduction in the value of the
assets backing the liability, with no corresponding reduction in the liability value.
o We feel that additional clarification is needed regarding the context in which the risk-free rate
should be determined. For example suppose that a US-based company issues an insurance
contract in Latin America. Should the discount rate be based on the US risk-free rate or the risk-
free rate in the country in which the liability is issued?
o Agenda Paper 17D’s description of liquidity premium is not fully consistent with the liquidity
premium paper that Bill wrote for ACLI last year. IASB staff defines liquidity in the sense of a
contractual demand feature. Bill defined liquidity as predictability, i.e. the property that allows an
insurer to invest in illiquid assets to back a group of contracts.
Observable market rates
The IASB staff noted that some observers support a prescribed discount rate. These prescribed rates
would not necessarily point the practitioner to the exact rate or set of rates to use, but they would
provide directional guidance. IASB staff provided the following examples from other accounting
o High quality corporate bonds (from IAS 19)
o High quality fixed-income debt instruments (from FAS 87 and FAS 106)
o Risk-free rate based on government bonds (from FAS 163)
In addition, the CFO Forum’s Market Consistent Embedded Value principles currently prescribe the
use of the swap yield curve.
o Prescribing an observable market rate would tend to reduce diversity in practice.
o Some alternatives (e.g. high-quality fixed income securities) would mitigate the perceived
problem of an “artificial” loss at inception.
o These alternatives would allow realization of some degree of asset credit risk in the
discount rate while avoiding arguably anomalous results of other approaches, i.e. a
reduced liability value resulting from an increase in asset risk or a decrease in the credit
standing of the liability.
o The use of swaps is justified on the assertion that, in many countries, the market for swaps is
deeper, longer and more liquid than the market for government bonds.
o The CFO Forum has, at various times, positioned the use of swaps as a proxy for risk-free
rates and as containing a provision for liquidity spread (although the premium is really for
credit risk). The Forum is currently exploring the addition of a liquidity premium.
o The prescribed rate or set of rates might not reflect the characteristics of the liability.
o For instance, the credit risk in a set of bonds underlying an index may be inconsistent
with the credit risk of the liability.
o A prescribed standard might need to be supplemented by a more principles-based standard.
o What if the prescribed market rate is not available or is deemed to be unreliable?
o Even with a “prescribed” rate, there is still likely to be diversity in practice.
o For instance, a high quality corporate bond rate could be based on numerous indices with
different definitions of “high quality.”
Other general comments
Agenda Paper 17D does not address the question of whether or how the measurement of insurance
liabilities should reflect non-performance risk. In order to perform a meaningful evaluation of the
various discount rate alternatives, we need to evaluate the discount rate in its entirety, including any
adjustment for non-performance risk. Without a discussion of non-performance risk, the discount rate
discussion in Agenda Paper 17D is incomplete.
ACLI Discount Rate Modeling Project – Fixed Deferred Annuities
Prepared by Carrie Morton, Principal Financial Group
In 2008, the Deputies’ Subgroup on Accounting Issues completed a discount rate modeling
project that illustrated the impact of discounting liability cash flows at the risk-free rate and the
asset earned rate.
In a February 2009 letter to the IASB, the ACLI recommended that insurance liabilities use a
discount rate equivalent to a highly rated corporate debt instrument.
This paper updates our prior analysis to reflect the ACLI’s discount rate recommendation.
The calculations are based on an exit value approach, with risk margins reflecting the cost of
bearing the risk. The risk margins are not calibrated to the premium.
Please see Attachment 1 for a description of the assumptions and methodology.
The following graph shows the annual pre-tax earnings for the first 10 years. Liabilities were
calculated using the risk free rate, the earned rate, corporate AAA bond yields, and Corporate AA
IFRS Income for SPDA w/ 5-year Guarantee
$10M initial premium, based on expected crediting rate
0 1 2 3 4 5 6 7 8 9 10
Corporate AA bond yields
Discounting at the AA bond yield produces a substantial gain at issue. The gain is significantly larger
than the gain that results from discounting at the asset earned rate. The difference in results is
primarily due to the slope of the yield curve.
Because the product has a 5-year interest rate guarantee, I assumed that a typical insurer would invest
the premium in 5-year bonds. Therefore, I set the asset earned rate equal to the 5-year corporate bond
yield, based on a blend of A and BBB rated bonds. This resulted in a net earned rate of 5.4%. The
resulting discount rate was applied to all cash flows, regardless of duration – i.e. for cash flows
occurring at time 10, the discount factor was (1 + net earned rate)-10.
When discounting at corporate bond yields, I used the entire corporate yield curve. I began by
obtaining yields for AA corporate bonds with maturities ranging from 2 to 25 years. The nominal
yields for AA bonds ranged from 3.01% at 2 years to 7.18% at 25 years. I then converted the nominal
yields to spot rates and constructed a yield curve. I used this yield curve to discount the liability cash
flows. The discount rates vary, depending on the duration of the cash flow – i.e. for cash flows
occurring at time 5, the discount factor is (1 + 5-year spot rate)-5, and for cash flows occurring at time
10, the discount factor is (1 + 10-year spot rate)-10. Thus, the cash flows occurring at later durations
are discounted at a higher discount rate than the cash flows occurring at earlier durations.
At later durations, the AA discount rates exceed the asset earned rate of 5.4%, despite the fact that the
earned rate is based on a slightly lower quality bond (a blend of A and BBB). This is because the
earned rate was based on a 5-year yield, while the AA rates reflect the entire yield curve. For
example, the 10-year AA yield is greater than the 5-year A yield. Therefore, discounting at the AA
yield curve resulted in a lower initial reserve – and hence a larger gain at issue – than discounting at
the earned rate.
The ACLI believes that high-quality (e.g. AA) corporate bond yields provide a reasonable basis for
setting the discount rate for insurance liabilities. However, due to the shape of the yield curve, this
approach may produce unrealistically high discount rates for cash flows occurring at later durations.
An insurer selling an SPDA with a 5-year guarantee would likely invest the premium in 5-year bonds,
and then continue to reinvest in 5-year bonds after the first bonds mature. Thus, the 10-year AA spot
rate (which is used to discount cash flows occurring at time 10) would likely exceed the rate that the
insurer would expect to earn on its invested assets. This duration mismatch leads to an
understatement of the reserve and a gain at issue. However, we could eliminate the gain at issue by
calibrating the margin to the premium.
Corporate AAA bond yields
The Corporate AAA bond yields were only slightly higher than the corresponding risk-free rates.
Discounting at the AAA bond yield produces a loss at issue, although the loss is slightly smaller than
the loss that results from discounting at the risk-free rate.
The discounting methodology for AAA bonds was the same as the methodology used for AA bonds.
Thus, when discounting at AAA rates, we still encounter the same duration mismatch that was
described above (i.e. duration 10 cash flows are discounted at the 10-year AAA spot rate, when in
reality, the insurer would likely invest in shorter assets). However, the AAA discount rate produced a
loss at issue, due to the lower level of the AAA yield curve. For example, the 10-year AAA yield is
still less than the 5-year A yield. Thus, this approach overstates the reserve, because the discount
rates are considerably lower than the expected asset earned rate.
Interaction of discount rates and risk margins
In Agenda Paper 5A, the IASB describes four possible approaches for setting risk margins for long-
(1) Current exit value as proposed in the Insurance Contracts Discussion Paper
(2) Current fulfillment value including a risk margin reflecting the cost of bearing risk (the IASB
staff has eliminated this method from consideration)
(3) Current fulfillment value as in candidate (2) plus an additional separate margin, calibrated at
inception to the premium
(4) Current fulfillment value including a single margin calibrated at inception to the premium
(similar to candidate (3), but with one overall margin, rather than two separate margins)
The calculations in the graph above are based on approach (1). The following paragraphs describe
how the various discount rates would work with each of the margin approaches.
Corporate AA bond yields
Under approach (1), the reserve is equal to the present value of cash flows, plus a risk margin. As
discussed above, this approach produces a substantial gain at issue. The ACLI believes that it is
inappropriate to recognize a gain at issue.
Under approach (3), we would start with the reserve from approach (1), but we would then add a
residual margin that is calibrated to the premium. Under approach (4), we would simply calculate a
single margin that is calibrated to the premium (rather than calculating a risk margin and then adding
a separate residual margin that is calibrated to the premium). Under both approach (3) and approach
(4), there is no gain at issue. We believe that this is a more appropriate result.
Corporate AAA bond yields
As noted above, discounting at the AAA bond yield produces a loss at issue under approach (1).
Under approach (3), we would start with the reserve from approach (1) and add a margin that is
calibrated to the premium. In this case, the initial premium is less than the sum of the initial reserve
from approach (1) and the acquisition expenses. Therefore, in order to break even at issue, the
residual margin would have to be negative. However, negative margins are not allowed, so the
residual margin would be set equal to 0, and a loss would be recognized at issue.
Under approach (4), we would calculate a single margin that is calibrated to the premium. However,
we would still need a negative margin in order to break even at issue. Because negative margins are
not allowed, we would once again recognize a loss at issue.
The results for the risk-free rate are similar to the results for the AAA bond yields, although the risk-
free rate produces a slightly larger loss at issue. Because negative margins are not allowed, all of the
margin approaches would produce a loss at issue.
The results for the earned rate are similar to the results for the AA bond yields, although the earned
rate produces a much smaller gain at issue under approach (1). The gain at issue would be eliminated
under approach (3) and approach (4).