Draft Educational Note

IFRS 17 Risk Adjustment for Non-

Financial Risk for Property and
Casualty Insurance Contracts

Committee on Property and Casualty Insurance

Financial Reporting

May 2020
Document 220063

Ce document est disponible en français

© 2020 Canadian Institute of Actuaries

The actuary should be familiar with relevant educational notes. They do not constitute standards
of practice and are, therefore, not binding. They are, however, intended to illustrate the
application of the Standards of Practice, so there should be no conflict between them. The
actuary should note however that a practice that the educational notes describe for a situation
is not necessarily the only accepted practice for that situation and is not necessarily accepted
actuarial practice for a different situation. Responsibility for the manner of application of
standards of practice in specific circumstances remains that of the members. As standards of
practice evolve, an educational note may not reference the most current version of the
Standards of Practice; and as such, the actuary should cross-reference with current Standards.
To assist the actuary, the CIA website contains an up-to-date reference document of impending
changes to update educational notes.
MEMORANDUM
To: Members in the property and casualty insurance area
From: Steven W. Easson, Chair
Actuarial Guidance Council
Houston Cheng, Chair
Committee on Property and Casualty Insurance Financial Reporting
Date: May 8, 2020
Subject: Draft Educational Note: IFRS 17 Risk Adjustment for Non-Financial Risk for
Property and Casualty Insurance Contracts
The Committee on Property and Casualty Insurance Financial Reporting (PCFRC) has
prepared this draft educational note to provide practical application guidance on Canadian-
specific issues relating to the IFRS 17 risk adjustment for non-financial risk for property and
casualty (P&C) insurers. Guidance from this draft educational note would be considered
with the following CIA educational notes:
• CIA draft educational note Application of IFRS 17 Insurance Contracts; and
• PCFRC guidance on matters relating to IFRS 17, when such guidance is issued as draft
educational notes.
The PCFRC acknowledges the exposure draft status of IFRS 17, and the evolving nature of
international actuarial guidance. Nevertheless, the PCFRC believes this draft educational note
has sufficient content to be beneficial to Canadian actuaries implementing IFRS 17.
The purpose of this draft educational note is to provide the reader with possible interpretations
of the Standard, without advocating any particular approach. Each topic presented in this
document addresses the implications of the standard for risk adjustment for non-financial risk:
general considerations, quantile techniques, cost of capital method, margin method, combining
approaches and methods and quantification of the confidence level.
The draft educational note Compliance with IFRS 17 Applicable Guidance provides guidance to
actuaries when assessing compliance with IFRS 17. It is applicable to all educational notes
pertaining to IFRS 17 and members are encouraged to review it prior to reading any educational
note related to IFRS 17.
Various stakeholders were consulted prior to releasing this draft educational note: the CIA
Committee on Life Insurance Financial Reporting, the CIA Committee on the
Appointed/Valuation Actuary, the CIA Committee on Risk Management and Capital
Requirements, the Accounting Standards Board, the International Insurance Accounting

1740-360 Albert, Ottawa, ON K1R 7X7  613-236-8196  613-233-4552

head.office@cia-ica.ca / siege.social@cia-ica.ca cia-ica.ca
Committee, the Committee on Workers Compensation, and the Group Insurance Practice
Committee.
The creation of this cover letter and draft educational note has followed the Actuarial Guidance
Council’s protocol for the adoption of educational notes. In accordance with the Canadian
Institute of Actuaries’ Policy on Due Process for the Approval of Guidance Material other than
Standards of Practice and Research Documents, this draft educational note has been prepared
by the PCFRC and has received approval for distribution from the Actuarial Guidance Council on
April 14, 2020.
The actuary should be familiar with relevant educational notes. They do not constitute
standards of practice and are, therefore, not binding. They are, however, intended to illustrate
the application of the Standards of Practice, so there should be no conflict between them. The
actuary should note, however, that a practice that the educational notes describe for a
situation is not necessarily the only accepted practice for that situation and is not necessarily
accepted actuarial practice for a different situation. Responsibility for the manner of application
of standards of practice in specific circumstances remains that of the members. As standards of
practice evolve, an educational note may not reference the most current version of the
Standards of Practice; and as such, the actuary should cross-reference with current Standards.
To assist the actuary, the CIA website contains an up-to-date reference document of impending
changes to update educational notes.
Questions or comments regarding this draft educational note may be directed to Houston
Cheng at hhcheng@kpmg.ca or Veronika Molnar (chair of the subcommittee) at
veronika.molnar@aviva.com.

3
Draft Educational Note May 2020

Table of Contents

1. Introduction .................................................................................................5
2. Transition from IFRS 4 to IFRS 17............................................................................ 7
3. General Considerations ............................................................................................ 8
3.1 Measurement Approach ............................................................................................. 8
3.2 Diversification, Allocation, and Aggregation ............................................................ 10
3.3 Reinsurance Held ...................................................................................................... 13
3.4 Discount Rate ............................................................................................................ 14
3.5 Time Horizon ............................................................................................................. 14
3.6 Disclosure Requirements .......................................................................................... 15
3.7 RA under PAA ............................................................................................................ 16
4. Quantile Techniques .................................................................................................... 17
4.1 Introduction .............................................................................................................. 17
4.2 Generating a Distribution ......................................................................................... 18
4.3 Measuring Risk .......................................................................................................... 19
4.4 Aggregation and Allocation....................................................................................... 21
5. Cost of Capital Method ................................................................................................ 21
5.1 Introduction .............................................................................................................. 21
5.2 General Formula ....................................................................................................... 22
5.3 Capital (Ct) ................................................................................................................. 22
5.4 Cost of Capital Rate (rt) ............................................................................................. 22
5.5 Reinsurance Held ...................................................................................................... 23
6. Margin Method ............................................................................................................ 23
7. Combining Approaches and Methods ........................................................................... 24
7.1 Aggregate/Entity-Level Approach ............................................................................. 24
7.2 Hybrid Approach ....................................................................................................... 25
8. Quantification of the Confidence Level ......................................................................... 25
8.1 Quantile Technique as Primary Method ................................................................... 25
8.2 Quantile Technique as Secondary Method .............................................................. 26
8.3 Calibration Using MCT .............................................................................................. 27
Appendix 1: Margins – Brief Summary of IFRS 4 CIA Standards of Practice ......................... 31
Appendix 2: Simplified Calculation of RA based on CoC Method ........................................ 32

4
Draft Educational Note May 2020

1. Introduction
IFRS 17 Insurance Contracts (IFRS 17) establishes principles for the recognition, measurement,
presentation, and disclosure of insurance contracts. The purpose of this draft educational note
is to provide practical application guidance on Canadian-specific issues relating to the IFRS 17
risk adjustment for non-financial risk (RA) for property and casualty (P&C) insurers. References
to specific paragraphs of IFRS 17 are denoted by IFRS 17.XX, where XX represents the paragraph
number.
The requirement for the RA, which is a defined term in IFRS 17 Appendix A, is set forth in IFRS
17.37:
An entity shall adjust the estimate of the present value of the future cash flows to
reflect the compensation that the entity requires for bearing the uncertainty about
the amount and timing of the cash flows that arises from non-financial risk.
Further clarification is provided in IFRS 17.B86–B92. These paragraphs emphasize that the RA
relates only to non-financial risk. Insurance risk, lapse risk, and expense risk are listed as
examples of risks that are included, whereas operational risks and market risks are excluded.
IFRS 17.B91 clearly states that IFRS 17 does not prescribe the estimation technique(s) used to
determine the RA, and IFRS 17.B92 notes that “an entity shall apply judgement.”
IFRS 17.B91 states that the RA would have the following characteristics:
(a) risks with low frequency and high severity will result in higher risk adjustments
for non-financial risk than risks with high frequency and low severity;
(b) for similar risks, contracts with a longer duration will result in higher risk
adjustments for non-financial risk than contracts with a shorter duration;
(c) risks with a wider probability distribution will result in higher risk adjustments
for non-financial risk than risks with a narrower distribution;
(d) the less that is known about the current estimate and its trend, the higher will
be the risk adjustment for non-financial risk; and
(e) to the extent that emerging experience reduces uncertainty about the amount
and timing of cash flows, risk adjustments for non-financial risk will decrease
and vice versa.
The RA is explicitly included in the insurance contract liabilities and is disclosed per the
requirements of IFRS 17.100–107 and IFRS 17.119.
Chapter 4 of the CIA Draft Educational Note Application of IFRS 17 Insurance Contracts (Draft Ed
Note IFRS 17 Application) provides general guidance about the RA. The Draft Ed Note IFRS 17
Application adopts without modification the exposure draft of the proposed International
Actuarial Note (IAN) 100 – Application of IFRS 17 Insurance Contracts of the International
Actuarial Association (IAA).

5
Draft Educational Note May 2020

In this draft educational note, “approach” is used to denote an overall way of addressing the
RA, whereas “technique” and “method” refer to the detailed process (including calculations) to
determine and allocate (if necessary) the RA.
Within the IAA guidance, Question 4.3 of the Draft Ed Note IFRS 17 Application states (emphasis
added):
This general guidance means that there is no single right way for an entity to set
the risk adjustment. In general, there are other important considerations that will
be relevant to how an entity determines its approach to estimating the risk
adjustment:
• consistency with how the insurer assesses risk from a fulfilment
perspective;
• practicality of implementation and ongoing re-measurement; and
• translation of risk adjustment for disclosure of an equivalent confidence
level measure.
Therefore, a variety of methods are potentially available, although their ultimate
usage depends on the extent to which they meet the criteria above, given the
specific circumstances of the company. Potential methods include, but are not
limited to, quantile techniques such as confidence level or CTE [conditional tail
expectation], cost of capital techniques, or even potentially simple techniques such
as directly adding margins to assumptions or scenario modelling.
Regardless of the estimation technique, the actuary would ensure that the resulting RA
represents the compensation the entity requires for accepting uncertainty in the amount and
timing of the cash flows arising from non-financial risk (uncertainty related to non-financial
risk). This draft educational note provides specific application guidance, as well as background
and general information, to help inform Canadian actuaries when exercising judgment for
derivation of the RA.
Equally important to understanding the objective of this draft educational note is
understanding what the draft educational note is not intended for. Consistent with IFRS 17, this
draft educational note:
• does not prescribe which approach or method to use for the RA in the aggregate or for
the RA by portfolio of insurance contracts (portfolio) or group of insurance contracts
(group);
• does not include statistical detail of the methods included herein;
• does not include detailed descriptions of how any given approach or method would be
applied;
• does not contain an exhaustive list of the approaches or methods that may be
acceptable for deriving the RA. For additional detail (including underlying statistical
theory) regarding quantile techniques, the cost of capital method, internal models, and
diversification, all of which may be important for the actuary responsible for deriving

6
Draft Educational Note May 2020

the RA, the actuary is referred to the basic educational material of the actuarial societies
as well as the IAA monograph titled Risk Adjustment for Insurance Contracts under IFRS
17 (IAA Risk Adjustment monograph); and
• does not address the issue of the Appointed Actuary’s (AA’s) Expression of Opinion.
In writing this draft educational note, the PCFRC Risk Adjustment Subcommittee followed these
guiding principles:
• Consider Canadian-specific perspectives rather than simply repeating international
actuarial guidance;
• Develop application guidance that is consistent with IFRS 17 and applicable Canadian
actuarial standards of practice and educational notes without unnecessarily narrowing
the choices available in IFRS 17; and
• Consider practical implications associated with the implementation of potential
approaches and methods; in particular, ensure that due consideration is given to
options that do not require undue cost and effort to implement.
2. Transition from IFRS 4 to IFRS 17
Prior to the effective date of IFRS 17, insurance contract liabilities are subject to IFRS 4, as
guided by current CIA Standards of Practice and educational notes. As such, discussion in this
draft educational note about processes under IFRS 4 (such as the use of margins for adverse
deviations (MfADs)) pertain to Canadian accepted actuarial practice prior to the adoption of
IFRS 17.
In Measurement of Liabilities for Insurance Contracts: Current Estimates and Risk Margins, the
IAA states that risk margins can serve two distinct purposes:
1. As the reward for risk bearing, measured in terms of the inherent uncertainty in
the estimation of insurance liabilities and in the future financial return from the
contract; or
2. In a solvency context as the amount to cover adverse deviation that can be
expected in normal circumstances, with capital to cover adverse deviations in
more unusual circumstances. 1
The current MfADs used under IFRS 4 can be classified as meeting the second purpose above,
whereas the RA can be classified as meeting the first. Although the approach selected to derive
the RA may, in the end, be similar to the approach used under IFRS 4, IFRS 17 requires the RA to
reflect the compensation the entity requires for taking on risk as opposed to margins that cover
adverse deviations.
If the actuary uses IFRS 4 MfADs as the starting point for calculating the IFRS 17 RA, then the
actuary would assess the questions posed in Section 9.2 of the draft educational note
Comparison of IFRS 17 to Current CIA Standards of Practice:

1
International Actuarial Association, Measurement of Liabilities for Insurance Contracts: Current Estimates and Risk
Margins (2009), 2–3.

7
Draft Educational Note May 2020

• Is the current level of PfAD [provision for adverse deviations] consistent with
the compensation the entity requires for bearing uncertainty?
• Are the diversification benefits included in current PfADs consistent with those
that would be reflected in IFRS 17?
• How would the confidence level (to satisfy disclosure requirement of IFRS
17.B92) inherent in the current PfADs be determined?
• IFRS 17 requires reinsurance contracts held to be measured as separate
contracts. How would the PfAD appropriate to the net liability be split between
the direct and ceded contracts?
• Are any adjustments needed for pass-through features?
The CIA Standards of Practice relevant to IFRS 4 may provide insight for establishing margins
under IFRS 17. 2 In the margin-setting process for a given group, the actuary would look to the
risk exposure of the broader entity to consider whether there are potential diversification
benefits to reflect in the entity’s RA. (See Section 3.2.2.) As noted previously, IFRS 17 does not
“specify the estimation technique(s) used to determine” the RA. Some Canadian actuaries may
find it operationally efficient to continue to apply margins either to derive the total RA or to
allocate the RA between portfolios and/or groups. However, other considerations, such as the
suitability of the margins to reflect an entity’s requirement for compensation and the margins’
associated confidence level, which is required for disclosures, would also be considered. Use of
margins would be acceptable if the resulting RA satisfies the five characteristics defined in IFRS
17.B91. Note that existing Canadian IFRS 4 guidance for setting MfADs is based on similar
considerations.
In practice, most Canadian entities are unlikely to have previously identified a specific metric or
set of metrics that explicitly defines the compensation the entity requires for bearing non-
financial risk. Such metrics or articulation of risk appetite, if they exist, would likely consider all
risks including financial risks and thus not be directly comparable to the scope of the RA.
Therefore, the actuary would need to justify how the selected margins and/or the resulting
confidence level of the RA reflect the entity’s compensation required for the uncertainty
related to non-financial risk.
3. General Considerations
3.1 Measurement Approach
In supporting an insurer to achieve the requirements specified in IFRS 17.37, the actuary would
(1) understand the compensation required by the entity for the uncertainty related to non-
financial risk and (2) develop an RA that reflects such compensation. The compensation the
entity requires is a subjective assessment of an entity’s own risk appetite.
There is more than one way for an entity to develop a price for that risk. Questions 4.9 and 4.13
in the Draft Ed Note IFRS 17 Application provide further general guidance. The answers to these
questions refer to the entity’s pricing as a potential reference point for measuring the entity’s
2
Further detail contained in Appendix 1.

8
Draft Educational Note May 2020

risk aversion and/or compensation requirements. The actuary would consider whether the
compensation the entity requires reflects any pricing concessions due to competitive market
pressure and/or price discounting in pursuit of aggressive market positioning. One view is that
the actual pricing is market observable evidence of the compensation the entity requires. An
alternative view is that an entity may temporarily accept other than its theoretical steady-state
compensation requirements, and that the RA would reflect the latter.
It is necessary that the RA at each reporting date satisfies the overall requirements of IFRS 17
for measurement, presentation, and disclosure of insurance contracts. Measurement
requirements are based on the IFRS 17 unit of account (i.e., RA for a single contract or group),
whereas presentation and disclosure requirements tend to be at a higher level (RA for the
aggregation of portfolios or the confidence level for the entity per IFRS 17.119). In selecting a
particular approach, the actuary would consider the accounting measurement requirements for
the RA as well as the aggregated presentation and disclosure requirements. (Section 3.7
addresses RA requirements specific to the premium allocation approach (PAA).)
3.1.1 Measurement Requirements Related to the RA – Unit of Account
The unit of account for IFRS 17 is the group or the insurance contract, and the measurement
requirements (and some presentation and disclosure requirements) are applied at that level.
For the RA, the unit of account has the following implications:
• The RA is determined on initial recognition and at each reporting date and reported for
each group (IFRS 17.32 and IFRS 17.40);
• The RA for a group influences the measurement of the contractual service margin (CSM)
and/or the loss component for the group at initial recognition (IFRS 17.38) and
subsequent measurement (IFRS 17.B96(d)); and
• For contracts initially recognized in a period, the RA is required to satisfy the grouping
requirements of IFRS 17.16 (i.e., to identify onerous contracts) unless the PAA
measurement is used in which contracts are assumed to not be onerous unless facts and
circumstances indicate otherwise.
IFRS 17.24 allows the fulfilment cash flows (of which the RA is a part) to be determined at a
higher level of aggregation than the group and then allocated to the relevant groups, provided
that the allocations result in appropriate fulfilment cash flows in the measurement of the
group.
3.1.2 Disclosure Requirements Related to the RA – Aggregate/Entity Level
While the measurement requirements of IFRS 17 require an RA for each unit of account, most
of the presentation and disclosure requirements of IFRS 17.78–109 are typically met at a more
aggregated level, such as portfolio or entity level.
IFRS 17.117(c)(ii) specifically requires disclosure of the approach (which using the terminology
of this draft educational note would also include method) used to determine the RA, and IFRS
17.119 requires disclosure of the confidence level corresponding to the reported RA.
Depending on the approach and method used, the confidence level will be either an explicit
input to the RA calculation or an implicit result of the calculation.

9
Draft Educational Note May 2020

3.1.3 Selection of a Measurement Approach

The actuary may view the aggregate entity level perspective as the primary basis for
determining the RA (perhaps driven by disclosure requirements or aligned at the level at which
the entity thinks about compensation). With an aggregate approach, the actuary would need to
allocate the total RA to the units of account to satisfy the IFRS 17 measurement requirements.
Some of the methods described in this draft educational note are more aligned with an
aggregate approach (e.g., quantile techniques) than with an approach focused on the unit of
account.
Alternatively, the actuary may develop the RA at the unit of account level to more directly
facilitate the measurement requirements of IFRS 17. The margin method can be used for a unit
of account approach. To the extent that the entity chooses to reflect the benefits of
diversification in its RA, the margins would be developed such that they reflect diversification
among the non-financial risks across the entity’s units of account. The sum of the RA calculated
at the unit of account level would be the entity’s aggregate RA.
3.2 Diversification, Allocation, and Aggregation
The entity’s perspective on diversification affects both the amount of the RA and the
assessment of the confidence level of the RA. Diversification may arise from the different types
of insurance risk (e.g., reserve, underwriting, and catastrophe), among portfolios, and among
related entities. The mechanics of how the actuary reflects diversification benefits may differ
depending on whether the actuary uses an entity level or unit of account approach.
The entity may consider the potential diversification among types of insurance risks when
calculating the RA for the liability for incurred claims (LIC) even if an explicit RA is not calculated
for the liability for remaining coverage (LRC) for contracts for which PAA is applied. In
determining the RA, the actuary would consider the non-financial risks associated with future
service (i.e., LRC) and past service (i.e., LIC).
3.2.1 Diversification and Allocation in an Aggregate Approach
To the extent that an entity level perspective is taken as the primary approach, the aggregate
risk distribution would reflect the entity’s perspective of the benefits of diversification among
its component risks. For example, the entity would assess the degree of diversification that it
expects arising from underwriting, reserve risk, and catastrophe risk to the extent facts and
circumstances warrant or management so chooses.
Incorporating diversification can be based upon statistical or empirical analyses, expert
judgment, or causal relationship. The more uncertain the diversification benefit, the less likely
such benefit would be fully reflected in the aggregate risk distribution. Two common methods
used by actuaries to quantify the effect of diversification are correlation matrices and copulas.
The Insurance Bureau of Canada’s Handbook for Economic Capital Modelling states the
following about correlation matrices:
Correlations are often used in explicitly modelling dependencies. Correlation is the
degree to which statistical distributions (and thus risks) are related to each other.
Correlation must take a value between -1 (perfect negative correlation) and +1

10
Draft Educational Note May 2020

(perfect positive correlation). A correlation matrix is simply a matrix in which the

correlations between pairs of data are specified. Correlation matrices must be
symmetric, which means that the correlation between risks A and B is the same as
the correlation between risks B and A in the correlation matrix. Correlation
matrices must also be positive semi-definite (PSD). For example, if a correlation of
+1 is chosen between risks A and B and risks A and C, a correlation of -1 between A
and C is not logical; this results in a non-PSD matrix. 3
If using correlation matrices, the actuary would consider the confidence level of the RA to
ensure that the correlation factors still apply at the selected confidence level. Furthermore, the
correlation factors would be considered in the context of the entity’s own circumstances; the
use of a “one size fits all” correlation matrix may not be appropriate.
The IAA Risk Adjustment monograph discusses copulas as follows:
The joint distribution of a set of random variables contains all the information
about their individual (marginal) distributions and dependence structure.
Dependence is a property of their copula. Copulas allow one to deal with the
dependence among random variables separately from their marginal distributions.
The estimation of the multivariate distribution is decoupled into estimation of the
marginal distributions, which is more robust, and the estimation of the
dependence relationship, which may have scarce data on which to rely. This
decoupling is achieved with a copula function. 4
For further information, see the IAA Risk Adjustment monograph.
The compensation the entity requires for non-financial risk would determine the confidence
level at which the entity chooses to set its RA. The benefits of diversification reflected in an
aggregate RA calculation are passed down to the unit of account via an allocation process.
The actuary may allocate the RA to groups directly (using a proportional or other method) or
indirectly (by calibrating margins such that a unit of account calculation aggregated across all
groups yields the same RA as the entity level calculation). In both direct and indirect allocations,
the sum of the RA for all units of account would be equal to the aggregate entity level RA.
IFRS 17 prescribes neither the aggregation nor allocation techniques. While this draft
educational note includes descriptions and examples of some approaches and methods, it is
beyond the scope of this draft educational note to provide an exhaustive list. The CIA published
a research paper on Risk Aggregation and Diversification in April 2016; and more generally, the
Enterprise Risk Management section of the CIA website contains additional resources on
aggregation and diversification.
3.2.2 Diversification and Aggregation in a Unit of Account Approach
When the RA is developed at the unit of account level, the entity’s aggregate RA is equal to the
sum of the RA for all units of account. The RA developed independently for one particular unit

3 The Insurance Bureau of Canada, Handbook for Economic Capital Modelling (2018), 153.
4 International Actuarial Association, Risk Adjustments for Insurance Contracts under IFRS 17 (2018), 89–90.

11
Draft Educational Note May 2020

of account may or may not reflect the benefits of diversification with other units of account of
the entity.
To the extent that diversification between different portfolios within an entity and/or
diversification between related entities are considered in pricing, there would be clear support
that reflecting similar diversification in the RA directly reflects the compensation the entity
requires. If pricing does not account for diversification between portfolios and/or entities, then
justification for including such diversification in the RA could prove more difficult and would
depend on the particular facts and circumstances of the entity. Ultimately, the level of the RA
for any given group is a matter of judgment, and the actuary would ensure that the resulting
aggregate RA reflects the compensation the entity requires for uncertainty related to non-
financial risk.
To the extent that the benefits of diversification are fully reflected in the assumed underlying
probability distribution but are not fully reflected in the calculation of the entity’s RA, the
resulting confidence level of the RA would be higher than had the full benefits of diversification
been passed down to the unit of account level. Expressed another way, the more conservative
the view an entity takes in applying diversification at the unit of account level, the higher will be
the resulting RA and its reported confidence level.
3.2.3 Diversification between Entities
Question 4.10 in the Draft Ed Note IFRS 17 Application presents two different perspectives on
diversification when a parent entity is composed of subsidiary entities.
One perspective is that each subsidiary entity would assess the compensation it requires for its
own non-financial risks independent of any potential diversification with risks across the
collective entities. The assumed probability distribution underlying the calculation of the
confidence level of the subsidiary entity’s RA would not reflect between-entity diversification.
The parent entity would either: (1) apply a diversification benefit at the parent entity level such
that the RA of the parent would be less than the sum of the RA of the subsidiaries or (2) simply
sum the RA of the subsidiary entities. The confidence level of the parent entity RA would be
higher in the second approach than the first.
Another perspective is that the diversification benefits of the parent entity would be reflected
at the subsidiary entity level. Thus, the assumed probability distribution underlying the
calculation of the confidence level of the subsidiary entity’s RA would reflect between-entity
diversification, and the degree of diversification credit reflected in the subsidiary’s RA
calculation would affect the confidence level of the subsidiary’s RA. The parent entity RA would
be the sum of the subsidiary entity RA.
With either perspective, the method used would be consistent from period to period and
reflect how the level of risk is considered and managed by the entity.
The Transition Resource Group for IFRS 17 (TRG) has discussed the topic of diversification
between entities in their May 2018 meeting; while TRG discussions are not official guidance
they do provide practical information and background on issues. The TRG meeting notes are

12
Draft Educational Note May 2020

available as an IFRS® publication in Summary of the Transition Resource Group for IFRS 17
Insurance Contracts meeting held on 2 May 2018.
3.3 Reinsurance Held
Under IFRS 17, the RA on reinsurance held is reported as a positive asset. In effect, the
reinsurance RA represents the risk ceded to the reinsurer. Reinsurance credit risk is reflected
through a reduction in expected cash flows, not through the RA.
Under IFRS 17, insurance contract liabilities (including liabilities on reinsurance contracts
issued) are reported separately from liabilities on reinsurance contracts held. Where an explicit
disclosure of the RA is required, the RA is also reported separately. In this draft educational
note, “gross RA” refers to the RA included in insurance contract liabilities (including reinsurance
contracts issued) and “ceded RA” refers to the RA included in liabilities for reinsurance
contracts held. This concept is articulated in IFRS 17.64, which specifically requires an explicit
RA for ceded reinsurance contracts:
Instead of applying paragraph 37, an entity shall determine the risk adjustment for
non-financial risk so that it represents the amount of risk being transferred by the
holder of the group of reinsurance contracts to the issuer of those contracts.
This separation of gross and ceded RA may not always be intuitive. This issue is addressed in
Question 9.9 of the Draft Ed Note IFRS 17 Application:
A specific definition for the determination of the risk adjustment for reinsurance
contracts held is provided that replaces the general definition in paragraph 37 used
for insurance and reinsurance contracts issued in the standard. Under the
definition for reinsurance held, the quantum of the risk adjustment for non-
financial risk represents the amount of risk being transferred by the holder of a
group of reinsurance contracts to the issuer of those contracts (paragraph 64).
The risk adjustment for the reinsurance held can therefore conceptually be thought
of as the difference in the risk position of the entity with (i.e., net position) and
without (i.e., gross position) the reinsurance held. As a result, the appropriate risk
adjustment for the reinsurance held could be determined based on the difference
between these amounts.
For reinsurance held, because the risk adjustment for reinsurance held is defined
based on the amount of risk transferred to the reinsurer, the risk adjustment for
reinsurance held will normally create an asset. On this basis, where a reinsurance
contract held is reported as an asset the risk adjustment will have the effect of
increasing the value of the asset, and will decrease the liability value where the
reinsurance contract held is reported as a liability.
The RA reflects the compensation the entity requires for uncertainty related to non-financial
risk and would be apportioned to gross and ceded insurance contract liabilities. Ultimately, the
key concepts underlying the RA are:

13
Draft Educational Note May 2020

• the gross RA (i.e., pertaining to insurance contracts including reinsurance contracts

issued) represents the compensation for non-financial risk that the entity requires for
writing those contracts; and
• the ceded RA (i.e., pertaining to reinsurance contracts held) represents the non-financial
risk transferred from the entity to the reinsurer(s).
Any method that respects these concepts would generally be acceptable.
Reinsurance is a hedge against the risk in the insurance contract. Theoretically, where the price
of reinsurance is proportional to the level of risk being hedged (i.e., ceded) from the entity’s
perspective and where the majority of portfolios and years of claims reserves are subject to the
same ceded percentages, then the ceded RA may be proportional to the gross RA (depending
on the potential effect of diversification). The gross RA would be unaffected by the presence of
reinsurance unless the reinsurance hedge affects the level of compensation required on the
insurance contract.
In practice, an entity’s reinsurance portfolio will likely contain a mix of proportional contracts
(at potentially different ceding percentages by portfolio and/or by year) as well as excess of loss
or other forms of reinsurance contracts. From the entity’s perspective, when the price of
reinsurance is not proportional to the level of risk being hedged, the ceded RA may not be
proportional to the gross RA. The cost of the reinsurance may be viewed as evidence of the
price the entity is willing to pay to be relieved of risk and therefore indicative of the entity’s
compensation requirements related to the uncertainty of the risk being ceded.
3.4 Discount Rate
IFRS 17 provides no direction regarding the discounting of the RA. IFRS 17.B90 states: “[T]he
risk adjustment for non-financial risk is conceptually separate from the estimates of future cash
flows and the discount rates that adjust those cash flows.” Furthermore, IFRS 17.B92 states:
“[A]n entity shall apply judgement when determining an appropriate estimation technique for
the risk adjustment for non-financial risk.”
Consequently, the use (or absence) of discounting and the method of determining discount
rates, if applicable, are at the discretion of the entity. More than one discounting method is
possible. Regardless of the discounting method chosen, the actuary would maintain a
consistent method between reporting periods.
Changes in discount rates will affect the current value of the RA if the derivation of RA requires
the use of discounting. Under IFRS17.81, the entity is not required to bifurcate the change in RA
into its component pieces (i.e., change in undiscounted provision for non-financial risk vs.
change in effect of discounting). If not bifurcated, the entire change in RA is presented as part
of the insurance service result, and the entire change in RA related to future services adjusts
the CSM.
3.5 Time Horizon
The appropriate time horizon for calculating IFRS 17 RA is the lifetime of the uncertainty in the
insurance contract cash flows.

14
Draft Educational Note May 2020

Actuaries using an internal model 5 for determining the RA would be aware that there may be
no link between the confidence level corresponding to the RA and the confidence level
underlying the internal model. For example, the result of an internal model calibrated to cover
risks at a confidence level of value at risk (VaR) 99.5 percentile over a one-year horizon is
conceptually very different than an RA calculation calibrated to a lifetime horizon. The RA
would generally be calculated at a lower percentile over a longer time horizon than the
economical capital resulting from the internal model, and thus the two amounts would not
likely be comparable. For an actuary to use an internal model for determining the RA, the
internal model would need to be re-calibrated to reflect any differences in time horizon and
confidence level.
When using a cost of capital method, the capital amounts held over the lifetime of the
insurance contract cash flows would be estimated. Conceptually, there is some confidence level
that relates to the insurance contract lifetime cash flows selected for such modelling. The
difference between the selected contract cash flows and the best estimate cash flows
represents the amount of capital that the entity would use in calculating the cost of capital.
Actuaries using a margin method would consider the volatility of the cash flows over the
lifetime of the insurance contract runoff.
3.6 Disclosure Requirements
3.6.1 Disclosure of Reconciliations
General IFRS 17 disclosure requirements are outlined in IFRS 17.93 through IFRS 17.132.
Elements specific to the RA include the requirement to disclose a reconciliation of the
movement in the RA from the opening balance to the closing balance (IFRS 17.100 for PAA and
IFRS 17.101 for GMA) 6 and the requirement to disclose significant judgments and changes in
judgments used in the calculation of the RA (IFRS 17.117).
3.6.2 Disclosure of the Confidence Level
Disclosure requirements for the confidence level are noted in IFRS 17.119:
An entity shall disclose the confidence level used to determine the risk adjustment
for non-financial risk. If the entity uses a technique other than the confidence level
technique for determining the risk adjustment for non-financial risk, it shall disclose
the technique used and the confidence level corresponding to the results of that
technique.

5 The term “internal model” is often used interchangeably with “economic capital model.” The International
Association of Insurance Supervisors states: The term “internal model” refers to “a risk measurement system
developed by an insurer to analyse its overall risk position, to quantify risks and to determine the economic capital
required to meet those risks”. Internal models may also include partial models which capture a subset of the risks
borne by the insurer using an internally developed measurement system which is used in determining the insurer's
economic capital.
6
Disclosures required by OSFI are expected to be more granular than those required by IFRS 17. For example,
experience by coverage for automobile insurance policies is expected to be required by OSFI.

15
Draft Educational Note May 2020

It is reasonable to infer that IFRS17.119 refers to the entity’s aggregate RA, and it would be at
the discretion of the entity to disclose the confidence level of RA at levels lower than the entity.
With respect to determination of the confidence level, question 4.18 in the Draft Ed Note IFRS
17 Application states:
In order to determine confidence levels, it is necessary to be able to locate the
value of the fulfilment cash flow of a collection of insurance contracts on the
probability distribution of the present value of the cash flows for the contracts. If
that probability distribution is not explicitly derived as part of the valuation
process, some method or model might be needed to estimate the percentiles of
that combined portfolio distribution at the amount that reflects the risk
adjustment. The extent of the analysis needed for such estimation is likely to
require judgment.
Potential techniques for the determination of the confidence level range from full stochastic
modelling to a relatively simple assumption about the shape of the underlying probability
distribution.
Determining the confidence level corresponding to the RA may be operationally burdensome;
nevertheless, the confidence level is a required disclosure under IFRS 17. Therefore, the actuary
would need to assess the practicality, cost, and effort associated with the selected method. In
particular, it is possible that parameterization of a full stochastic model may require so many
assumptions that it could lead to spurious accuracy in the resulting calculation of the
confidence level. In many situations, a more simplified approximation technique may provide
an equally reasonable estimate of the confidence level at much less cost and effort. The degree
of rigour is an entity-specific decision subject to the judgment of the actuary and agreement of
the auditor.
Regardless of the technique selected, the actuary would be aware that the quantification of the
confidence level is an estimate, given the unobservable nature of the full probability
distribution of the present value of the cash flows. The actuary would make users of the
information aware that the quantification is based on certain methods and assumptions and
take care to apply those methods and assumptions consistently from period to period.
Disclosure requirements specific to PAA are described in Section 3.7.
3.7 RA under PAA
An estimate of the LRC calculated under the general measurement approach (GMA) includes
RA, whereas an estimate of the LRC calculated under PAA does not. Regardless of whether the
LRC is calculated under PAA or GMA, the LIC requires an explicit RA. The fact that the treatment
of the RA differs for the LRC and the LIC may complicate the calculation and/or disclosure of the
confidence level required by IFRS 17.119.
For an entity using PAA, an explicit RA calculation for the LRC is not required for financial
reporting purposes for groups that are not deemed onerous. To establish a trigger for onerous
contracts, an estimate of an RA would be determined. An explicit RA calculation is required to
calculate the loss component for groups of contracts that may be onerous. If the calculations

16
Draft Educational Note May 2020

confirm that the insurance contracts are onerous, the entity is required to separately disclose
the LRC excluding any loss component and the loss component; disclosure of an explicit RA
amount, however, is not required. If the calculations confirm that the contracts are not
onerous, the only disclosure required is the LRC excluding loss component under PAA.
As the LIC is always measured as the fulfilment cash flows relating to incurred claims7, the
calculation of RA is always required. Disclosure requirements for the LIC specify a split between
LIC excluding RA and RA at the entity level, because entities will have to consider what level of
disaggregation is appropriate in order to achieve the general disclosure objective in IFRS 17.93.
For entities where the primary (or only) measurement approach for the LRC will be PAA, the
actuary would likely seek a method to estimate the RA for testing onerous contracts that
maximizes operational efficiency. In situations where the RA is mainly driven by volatility in the
cash flows associated with claim activity and where the cash flows associated with premium
activity for the LRC are subject to volatility, the actuary may approximate the RA required for
the LRC of onerous groups by using the RA derived for the LIC. In practice, the actuary would
consider the volatility associated with the LRC, but facts and circumstances of an entity may be
such that this approximation has limited applicability.
The requirement to reflect diversification applies regardless of the entity’s selected
measurement approach (i.e., GMA vs. PAA). Thus, the considerations described in Section 3.2
apply for entities adopting PAA. Regardless, the calculations may be more challenging as the RA
may not be explicitly calculated for the LRC.

4. Quantile Techniques
As noted previously, this draft educational note does not contain an exhaustive list of methods
nor does it contain detailed statistical background and descriptions. The actuary is referred to
the IAA Risk Adjustment monograph, which was developed explicitly for purposes of IFRS 17.
4.1 Introduction
Quantile techniques, including VaR and CTE, may be used to assess the probability of the
adequacy of the fulfilment cash flows and thus to quantify the RA. One key advantage of a
quantile technique is that it directly satisfies the IFRS 17 disclosure requirements regarding
confidence level corresponding to the RA. The IAA Risk Adjustment monograph states: “A key
advantage of the quantile techniques is that the mathematics enable risks to be represented
graphically which creates ease and convenience in understanding the result. A disadvantage is
that if misrepresented, it may introduce spurious accuracy.”
Assessment of the confidence level corresponding to the RA would generally require underlying
assumptions of the risk distribution. Given a risk distribution, both VaR and CTE can be
calculated. It is important for the actuary to recognize that a VaR calculation may not capture
the risk for a particularly skewed distribution of cash flows, which are common for certain P&C

7
While the LIC is often described as always being measured under the GMA, for groups of contracts where the LRC
is measured under PAA there are some differences in the measurement and disclosure requirements for LIC as
well. See IFRS 17.59(b) and IFRS 17.97-109 for further details.

17
Draft Educational Note May 2020

risks, and thus may not be an appropriate method to use. For further discussion of the use of
VaR and CTE for RA, the actuary is referred to the IAA Risk Adjustment monograph.
This section provides a high-level overview of possible methods to generate a risk distribution
and describes how quantile techniques (including VaR and CTE) are applied to determine the
RA. Detailed theoretical background information and implementation guidance for quantile
techniques are beyond the scope of this draft educational note.
4.2 Generating a Distribution
To generate a distribution of the underlying future cash flows, different methods may be
considered:
• Fit future cash flows for non-financial risks to a suitably skewed probability distribution
(e.g., lognormal or gamma distribution);
• Monte Carlo simulation;
• Bootstrapping; and
• Scenario modelling.
Each of these are described briefly below.
4.2.1 Probability Distribution for Present Value of Cash Flows
Under IFRS 17, the actuary would estimate an unknown variable (i.e., fulfilment cash flows),
which conceptually is derived from an analysis of the full range of possible outcomes of the
contractual cash flows. In practice, however, it may be extremely difficult to observe the full
range of possible outcomes or the underlying probability distribution that defines the full range
of possible outcomes. The actuary may therefore assume the shape of the underlying
probability distribution. For example, the actuary may assume a lognormal or gamma
distribution, both of which exhibit skewness. There are many other distributions that may
appropriately represent the characteristics of an entity’s cash flows; however, it is beyond the
scope of this draft educational note to provide an exhaustive list.
4.2.2 Monte Carlo Simulation
Non-financial risks can be modelled stochastically. The Monte Carlo method may be used to
repeatedly simulate a random process for relevant risk variables of (such as reserving,
underwriting, and catastrophe risk) covering a wide range of possible situations. In general,
thousands of simulations are typically generated under the Monte Carlo method to reduce
sampling variability. The actuary is able to derive a probability distribution based on the
resulting simulations of the entity’s aggregate risks. This enables the RA to be set at the target
percentile level of the observed distribution.
In modelling insurance risk stochastically, the actuary would consider parameter risk, process
risk, and model risk. For further information, the actuary is referred to the IAA Risk Adjustment
monograph.

18
Draft Educational Note May 2020

4.2.3 Bootstrapping
The IAA Risk Adjustment monograph describes bootstrapping as follows:
This is a resampling technique where historical observations are used to create
stochastic scenarios. Rather than a hypothetical distribution, this technique relies
on historical information as potential future observations. As an example, to
estimate the variability of the sample mean in the original data set, sampling with
replacement to generate multiple future populations may create an appropriate
distribution of sample means. For non-life insurance reserves, this approach has
also been used to generate the probabilities of uncertain outcomes. However, in
many applications some sort of normalization would be appropriate to remove
such factors as seasonality, or adjust for exposure. This technique has merit
because it may more closely resemble what historical data has shown can happen.
This method also does not restrict the recognition of heavy tails or other
observations that depart from theoretical distributions. However, it may be a poor
approximation for small samples and it relies heavily on the fact that each sampled
variable is independent from another. Another disadvantage is that the variability
of outcomes for future cash flows may not be adequately represented by historical
observations in a particular data set, particularly for low frequency, high severity
outcomes or other unusual events.8
4.2.4 Scenario Modelling
Scenario modelling is mentioned as an alternative technique in Question 4.14 of the draft Ed
Note IFRS 17 Application for reflecting qualitative risk characteristics “provided suitable
extreme scenarios are included.” Instead of different assumptions applied to each risk, a
combination of assumptions or a scenario reflecting multiple non-financial risks may be applied
to the underlying insurance contracts. In practice, however, the actuary may have difficulty
calibrating appropriate scenarios for purpose of the RA.
Financial condition testing (FCT) 9 is one example of scenario modelling. FCT is a process of
analyzing and projecting trends in an insurer’s capital position given its current circumstances,
considering adverse scenarios that are severe but plausible. The materiality threshold for an
FCT analysis is generally higher than the materiality associated with a liability calculation for
financial reporting purposes. Therefore, the actuary would be cautious in applying the
techniques used to complete an FCT analysis for the determination of an RA.
4.3 Measuring Risk
Once a distribution is generated, both VaR and CTE can be calculated or observed.
4.3.1 VaR
The VaR technique can be summarized in the following three steps:

8International Actuarial Association, Risk Adjustments for Insurance Contracts under IFRS 17 (2018), 74.
9
Effective January 1, 2020, FCT replaced dynamic capital adequacy testing. See Section 2500 of the standards of
practice for further information.

19
Draft Educational Note May 2020

1. Entity determines the target confidence level at which it determines its compensation
required (e.g., xth percentile);
2. VaR is determined such that the probability of the present value of actual cash flows
being less than VaR is x%; and
3. RA is then determined as VaR at xth percentile less the mean of present value of
probability-weighted cash flows.
The VaR technique is similar to the technique frequently used for economic capital calculations
(such as own risk and solvency assessment, ORSA). An entity’s existing VaR techniques may be
applied to the calculation of RA. There are, however, important differences including:
• Risk profile – Economic capital typically includes all risks faced by the entity, whereas
the RA only reflects non-financial risk.
• Time horizon – Economic capital tends to be calculated over a one-year time horizon,
whereas the time horizon for the calculation of the confidence level of the RA would
reflect all cash flows within the contract boundaries (i.e., the lifetime horizon, where
lifetime is limited by the contract boundary). The entity may, if it so chooses based on its
own compensation requirements, determine the level of the RA based on one-year
shocks, but the associated confidence level would be calibrated against a lifetime
horizon.
• Comparability – Economic capital is often calculated at a higher percentile (e.g., 99.5%)
over a one-year time horizon. The confidence level of the RA would generally reflect a
lower percentile over a longer time horizon. As such, the two amounts are generally not
directly comparable.
4.3.2 CTE
The CTE method can be summarized in the following three steps:
1. Entity determines the target confidence level at which it determines its compensation
required (e.g., xth percentile).
2. From the probability distribution, an entity can determine:
A. Conditional mean of the present value of future cash flows beyond the target
percentile; and
B. Mean of the present value of probability-weighted cash flows.
3. RA is then determined as the difference between A and B.
Question 4.14 of the draft Ed Note IFRS 17 Application does not explicitly mention a CTE
technique. However, it mentions that “… it may be possible to incorporate allowance for
correlation and skewness effects.” To address skewness, a suitably skewed probability
distribution and/or CTE technique may be applied.

20
Draft Educational Note May 2020

4.4 Aggregation and Allocation

Once the aggregate percentile level and resulting aggregate RA are derived from a quantile
technique, the actuary would allocate the RA to the groups per the requirements of IFRS 17.24
and perhaps to more granular levels for the purpose of determining initial groups per IFRS
17.16 and IFRS 17.47. As noted previously, IFRS 17 does not prescribe the allocation method.
Possible solutions range from simple proportional allocation to more sophisticated weightings
based upon analyses of the component risks.
Alternatively, instead of producing a distribution of the future cash flows for the entire entity,
the VaR or CTE may be calculated for each non-financial risk and then aggregated using a
correlation matrix. See further details about allocation in Section 7.

5. Cost of Capital Method

5.1 Introduction
In a cost of capital method, the RA is based on the compensation that the entity requires to
meet a target return on capital. In this calculation, three elements are required:
1. Projected capital amounts, which are used to determine the level of non-financial risk 10
during the duration of the contract;
2. Cost of capital rate(s), which represent the relative compensation required by the entity
for holding this capital; and
3. Discount rates, which are used to obtain the present value of future compensation
required. The actuary may use similar discount rates for the RA calculation as are used
for other IFRS 17 calculations (such as discounting the LIC).
This method has the benefit of being conceptually close to the definition of the RA and
potentially allows allocation of the RA at a more granular level assuming a more granular
allocation method for capital amounts. On the other hand, the cost of capital (CoC) method
may be operationally complex, as the projection of capital requirements is an input to the
liability calculation.
Whereas the general formula for the CoC is simple, there are a variety of ways to determine its
components. A practical method to determine the compensation required by the entity is the
method used for pricing purposes (i.e., the way an entity determines compensation in its day-
to-day operations). Alternatively, an entity may prefer to define the compensation required on
a more theoretical basis. Both methods are discussed in this section.
In addition to the CoC calculations described below, there are simplified ways in which the CoC
concept could be applied to estimate the RA. One such example is presented in Appendix 2.

10
Capital applicable amount can usually be broken down between reserve risk and underwriting risk. In theory, there
could be other risks that would attract capital that would not fall under reserve risk and underwriting risk, including
market risk and general operational risk.

21
Draft Educational Note May 2020

5.2 General Formula

The general formula for the RA based on a cost of capital method is:
𝑟𝑟 𝑥𝑥 𝐶𝐶𝑡𝑡
RA = ∑𝑡𝑡 (1+𝑑𝑑
𝑡𝑡
𝑡𝑡 )^𝑡𝑡

where,
• Ct is the average capital amount for the period t;
• rt is the selected cost of capital rate for the period t;
• rt x Ct is the compensation required by the entity for the period t; and
• dt is the selected discount rate(s), reflecting a yield curve, if appropriate.
Considerations for defining Ct and rt are discussed in the following sections.
5.3 Capital (Ct)
As noted, a practical approach for a given group of insurance contracts is to determine the
capital requirement with the capital model used for pricing purposes. Other capital models,
such as the regulatory capital model (e.g., minimum capital test (MCT)) of the entity or an
internal model, may be used as long as such model is consistent with the view of the entity
regarding compensation. In selecting a capital model, the actuary would consider the entity’s
risk appetite with respect to capital, which may be expressed as an internal or operating target.
The actuary would use a regulatory capital model or internal model with caution as these
models may not be appropriate to calculate the entity’s capital requirement for RA purposes.
(See Sections 3.5 and 4.3.1 for further details.)
Furthermore, the capital requirement would be adjusted to reflect the following
considerations:
• Removal of the capital component(s) related to risks other than the non-financial risks in
scope of the RA (such as market risk or general operational risk);
• Diversification if not specifically addressed in the capital model being used; and
• Consideration of risk-sharing mechanisms (e.g., reinsurance and Facility Association)
reflected in the estimates of future cash flows.
The actuary would derive a method to allocate the capital requirement (initially determined by
considering the diversification at an aggregate level) to the most granular level. At a minimum,
the actuary would allocate the capital requirement by group to meet IFRS 17 requirements.
Literature includes other capital allocation methods, such as the pro rata, continuous/discrete
marginal, and the Shapley method, none of which are described in this draft educational note.
5.4 Cost of Capital Rate (rt)
The cost of capital rate is traditionally designed as the weighted average cost of capital for an
entity that considers all sources of capital minus the rate that could be earned on surplus.
Among the sources of capital, the cost of capital for common shareholders (or equivalent
stakeholders) is the most complex to define.

22
Draft Educational Note May 2020

A practical approach is to use target rates of return on capital by capital source and their
respective weights that are consistent with management’s view (i.e., used for pricing or as
corporate targets). Target rates of return on capital may vary by portfolio, product, etc. Even if
these rates of return are not supported by cost of capital theory, they may still represent the
compensation required by the entity.
Alternatively, theoretical cost of capital rates may be determined by the entity based on the
following considerations:
• The cost of capital would depend on the entity’s risk aversion.
• The amount of capital would reflect the level of risk (i.e., uncertainty). If the entity
requires different compensation for similar risks in different portfolios, the difference
would be reflected in the cost of capital rate rather than the amount of capital.
• The cost of capital rate may be defined as a rate that represents the profit required for a
given quantity of risk (risk perceived by the shareholders). Then, this rate is applied to
an amount of capital measured by a capital model. In theory, when the capital model
used measures perfectly the risks perceived by the shareholders, it would be reasonable
and practical to apply the same cost of capital rate to all lines of business, products and
risks, etc. In practice, however, capital amounts measured by models are generally
simplified measures of the underlying risks and it may be appropriate to adjust the cost
of capital to compensate for this.
• The risk-adjustment is a pre-tax item yet cost of capital requirements are often-stated
on an after-tax basis. The actuary would ensure that the calculations are internally
consistent.
5.5 Reinsurance Held
Section 3.3 of this draft educational note discusses general considerations with respect to
reinsurance held. A specific consideration in the cost of capital method is the need to develop
cost of capital rates on a gross of reinsurance basis. For this purpose, it may be practical to use
the cost of capital rate net of reinsurance. This is consistent with the considerations articulated
in Section 3.3. From a theoretical standpoint, the third bullet point in Section 5.4 suggests that
it is expected that the cost of capital remains unchanged when there is a change in the risk
profile (e.g., ignoring all reinsurance), unless the capital model inadequately captures the risk
perceived by the shareholders.

6. Margin Method
Under a unit of account approach with a margin method, the actuary would select margins that
reflect the compensation the entity requires for uncertainty related to non-financial risk. The
“compensation the entity requires” would be quantified through the margin-setting process,
which is not necessarily based on a specified confidence level.
For IFRS 17 disclosure purposes, the actuary would calculate the confidence level corresponding
to the resulting RA (i.e., sum of the indicated RA resulting from the selected margins). The
confidence level disclosure would be an output (not an input) of the process. To meet actuarial

23
Draft Educational Note May 2020

standards of practice for examining the reasonableness of a calculation’s result, the actuary
may choose to use a quantile technique to compare with the RA resulting from the margins,
taking into consideration the sufficiency and reliability of the data input and paying particular
attention to items such as the trend, mean, median, symmetry, skewness, and tails of
underlying distributions.

7. Combining Approaches and Methods

The combination of multiple approaches and methods may take many forms. Question 4.23 in
the Application of IFRS 17 Insurance Contract states:
There is no requirement to use a single model or approach for all the business or all
the risks. An entity may use a mix or blend of methods to set risk adjustments
across different businesses provided such an approach makes appropriate
allowance for diversification and is done in a way that can be reasonably disclosed
and explained to external auditors and is relevant to users (which is likely the
biggest hurdle to a mixed model approach).
One possible way to combine methods under a unit of account approach is to use VaR for
groups with less skewed distribution and the cost of capital method or margins for groups with
highly skewed distributions, where the VaR does not provide a reasonable estimate of the RA.
In this example, the actuary would still need to determine the overall confidence level for
disclosure purposes. Moreover, the actuary would ensure that the aggregate RA from these
different methods achieves the entity’s compensation requirement for the uncertainty related
to non-financial risk.
7.1 Aggregate/Entity-Level Approach
Under an aggregate approach, the primary methods for calculating the aggregate RA are a
quantile technique and the cost of capital method. The margin method may be appropriate for
an aggregate RA if a single margin can be selected to reflect the compensation the entity
requires for bearing the risk associated with the underlying portfolios. In addition, margins may
be used to allocate the aggregate RA to the unit of account level.
7.1.1 Aggregate Approach Using a Quantile Technique
The actuary may allocate the aggregate RA using margins that are calibrated to ensure that the
sum of the RA calculated at the unit of account level is equal to the aggregate RA calculated via
a quantile technique. Other allocation methods are also possible. In choosing a reasonable
approach, the actuary has discretion to consider operational efficiency.
If using margins, the actuary would periodically review and recalibrate the margins. The actuary
may choose to limit change in the margins outside of the periodic review cycle (which may be
annually) only if the resulting confidence level corresponding to the RA drifts away from the
target confidence level by more than a pre-defined threshold.
7.1.2 Aggregate Approach Using Cost of Capital Method
Margins may be calibrated to replicate an aggregate RA derived from a cost of capital method.
These margins could be a practical alternative to a principles-based cost of capital calculation,

24
Draft Educational Note May 2020

given that the latter may be very difficult to execute in production within typical financial
reporting deadlines.
A cost of capital method may be a useful input into calibration of the level of the margins by
portfolio. Margins may be developed to produce RA by portfolio that are proportional, or
approximately proportional, to the capital requirements by portfolio. Actuarial judgment would
dictate whether a goal of proportionality is appropriate given the facts and circumstances
particular to the entity.
To comply with presentation and disclosure requirements, the confidence level corresponding
to the resulting RA would be calculated.
7.2 Hybrid Approach
There may be many different forms of hybrid approaches that incorporate the unit of account
and aggregate perspectives and various methods (e.g., quantile, cost of capital, and margins).
One possible hybrid approach is described in this section.
First, assume that the entity’s risk management policy specifies a target range for the
confidence level corresponding to the aggregate RA. This target range would represent the
aggregate compensation the entity requires for the uncertainty related to non-financial risk.
Next, assume that the actuary calculates a total RA and its associated confidence level using
margins established for each portfolio (or group) as a starting point, with adjustments for
diversification.
To the extent that the sum of the RA produced by the selected margins do not result in an
aggregate RA that is within the target range set out by policy, the margins would be re-
calibrated to ensure that the entity level RA was within the range.
Given the uncertainties associated with estimating confidence levels and the dispersion in
estimates of RA that may result from the use of different approaches and methods, this
particular example in which a range of target confidence level is established by the entity offers
an important operational advantage. Calibrating the RA within a sufficiently wide target range
may lessen some of the concerns with the precision (or lack thereof) for confidence level
calculations.
The actuary could follow a similar hybrid approach that incorporates the cost of capital method
or margin method instead of a quantile method. As such, the actuary may calculate the
aggregate RA based on a range of target cost of capital rates, and the margins would be
calibrated accordingly.

8. Quantification of the Confidence Level

8.1 Quantile Technique as Primary Method
Where a quantile technique is the primary method for determining the amount of the RA, there
is no need for a separate process to calculate the confidence level corresponding to the RA.
Given the requirement of a probability distribution to calculate the quantile technique RA, the
resulting confidence level of the selected RA would be directly available. Thus, a quantile

25
Draft Educational Note May 2020

technique that is used as the primary method for calculation of the RA directly satisfies the IFRS
17 disclosure requirements in IFRS 17.119.
8.2 Quantile Technique as Secondary Method
If the primary method for determination of the RA is the cost of capital, the margin method, or
some other method, then the actuary would need a secondary method to quantify the
confidence level corresponding to the RA to satisfy the disclosure requirement. As noted in
Question 4.18 of the Draft Ed Note IFRS 17 Application, this would usually require some
information about the underlying probability distribution of the present value of future cash
flows. The term “future cash flows” used throughout the remainder of this section is
understood to be the present value of future cash flows.
As noted previously, the distribution of future cash flows for P&C insurance is typically skewed.
In the following example, a lognormal distribution is assumed for presentation purposes only.
Lognormal distributions are commonly used in P&C insurance to model claim size, as the
distribution is positively skewed and the random variables take on only nonnegative values. The
purpose of the example is to illustrate how a quantile technique may be applied. In practice,
the actuary would select the distribution(s) that most adequately fits the entity’s cash flows.
A lognormal distribution can be defined by its parameters (μ, σ), where the parameters
represent the mean and standard deviation of the normally distributed variable log X and not
that of X. Any point on the distribution can be identified if these two parameters are known.
For lognormal distributions, the mean and standard deviation can be used to derive parameters
(μ, σ).
Random variable X has a lognormal distribution with parameters (μ,σ) if, and only if, log X is
normally distributed with mean μ and variance σ2. Therefore, the lognormal variable X can be
expressed as X = eσZ+μ, where Z is the standard normal random variable. The lognormal
cumulative distribution function is
0 if − ∞ < x ≤ 0
⎧
Fx (x) =
⎨ log x − μ
⎩Φ � � if 0 < x < ∞ (−∞ < μ < ∞, σ > 0)
σ
The continuous lognormal variable X has probability density function:

⎧0 if − ∞ < x ≤ 0
⎪
fx (x) =
⎨ 1 exp �− 1 (log
⎪ x − μ)2 /σ2 � if 0 < x < ∞
⎩ σ√2πx 2

The best estimate liability (BEL) represents the mean or central tendency of the distribution.
Ideally, the actuary would have a method to derive the standard deviation of the assumed
distribution of future cash flows, but in practice this may be difficult. The practical problem is
that it will likely be impossible to independently observe the standard deviation of the
distribution of future cash flows.

26
Draft Educational Note May 2020

One potentially reasonable approach is that the standard deviation of the distribution for
specific portfolios can be derived from the insurance risk factors in the MCT. (See Section 8.3.)
Using the standard deviation from the MCT and the BEL as the mean parameters, (μ, σ) can be
derived by using the formulas below.
The mean, variance, and skewness follow directly:
2 ⁄2
E[X] = eμ+σ

2 2
Var[X] = �eσ − 1�e2μ+σ

2
Sk[X] = �eσ + 2��eσ − 1
2

The actuary may also explore other approaches to define the standard deviation. For example,
the actuary may be able to turn to the entity’s internal model if such model is sufficiently
robust and can be recalibrated to reflect the time horizon and risk appetite required by the RA.
8.3 Calibration Using MCT
This section refers to OSFI’s MCT and the Branch Adequacy of Assets Test (BAAT). The
description is also applicable to the MCT of the Autorité des marchés financiers (AMF).
A practical advantage of using the MCT as a calibration point is operational efficiency to
leverage existing processes in the quantification of the confidence level. A potential
disadvantage is that the estimated confidence level may not be appropriate for a particular
entity.
The insurance risk factors in the MCT consider claim liabilities and premium liabilities and are
based on a review conducted in 2013. In the event that OSFI updates the MCT risk factors, the
considerations underlying the revised factors would potentially change the calculations.
Per OSFI,
To develop the new factors, OSFI undertook a variability analysis based on incurred
and paid data to assess the insurance premiums and claims risks. For unpaid claims,
OSFI performed a variability analysis between the estimated and the actual amount
of losses using two methods: lognormal and bootstrap. For premium liabilities,
OSFI’s variability analysis was built based on pure loss ratio data, assessing
variability in ultimate loss ratios by line of business for each accident year. A
correlation study between lines of business was also performed to determine the
level of diversification credit.11
The following are links to OSFI’s documentation of the variability analysis:

11Office of the Superintendent of Financial Institutions, Discussion Paper on OSFI’s Proposed Changes to the
Regulatory Capital Framework for Federally Regulated Property and Casualty Insurers (May 2013), 14.

27
Draft Educational Note May 2020

• Discussion Paper on OSFI’s Proposed Changes to the Regulatory Capital Framework for
Federally Regulated Property and Casualty Insurers (May 2013)
• Disclosure on OSFI’s Review of Insurance Risk Factors (December 2013)
• Presentation: The Next Generation of the Minimum Capital Test - A Canadian Regulatory
Capital Framework (login required) (June 2013)
The risk factors were established at a confidence level of VaR 99.5% with an explicit adjustment
for diversification. The factors were reduced by approximately 45% for claims liabilities and 11%
for premium liabilities to account for risk diversification. Per OSFI, the “correlation study
demonstrated that premium liabilities by lines of business are more correlated compared to
claims liabilities; therefore a lower diversification credit was applied.”
Use of the MCT risk factor as the second point on the distribution requires the following
considerations:
• The appropriate level of diversification when aggregating multiple lines and potentially
LIC and LRC; the actuary would consider the entity’s mix and volume of business.
• Adjustment for volatility due to size and other considerations relative to the “average”
entity included in the OSFI review. For example, smaller entities tend to exhibit greater
relative volatility than larger entities due to increased process and parameter risk, all
else being equal.
With the assumption of a lognormal distribution and removing diversification based on the MCT
factors, the following table shows the indicated standard deviation by line of business for the
LIC and the LRC. The standard deviations should correspond reasonably well with unpaid claims
and premium liabilities. The MCT risk factors were scaled to the average of the four largest
entities included in the OSFI review. The risk factors were also reduced by OSFI’s estimate of
the average MfAD for each line of business.

28
Draft Educational Note May 2020

LIC LRC

Percentile Percentile
Standard Standard
Deviation 65th 75th 85th Deviation 65th 75th 85th
Category

Personal Property 12% 4% 8% 13% 10% 3% 6% 10%

Commercial Property 10% 3% 6% 10% 10% 3% 6% 10%

Aviation 18% 5% 11% 18% 13% 4% 8% 14%

Auto Liability – BI 12% 4% 8% 13% 10% 3% 6% 10%

Auto – Pers. Acc. 12% 4% 8% 13% 10% 3% 6% 10%

Auto – Other 12% 4% 8% 13% 10% 3% 6% 10%

Boiler & Machinery 15% 5% 9% 15% 12% 4% 7% 12%

Credit 18% 5% 11% 18% 13% 4% 8% 14%

Credit Protection 15% 5% 9% 15% 12% 4% 7% 12%

Fidelity 18% 5% 11% 18% 13% 4% 8% 14%

Legal Expense 20% 6% 12% 20% 13% 4% 8% 14%

Liability 20% 6% 12% 20% 13% 4% 8% 14%

Other Approved Products 18% 6% 12% 20% 13% 4% 8% 14%

Surety 18% 5% 11% 18% 13% 4% 8% 14%

Title 15% 5% 11% 18% 12% 4% 8% 14%

Marine 18% 5% 9% 15% 13% 4% 7% 12%

In its variability analysis, OSFI determined that a portion of the volatility depended inversely on
the size of the entity and the remaining portion of the volatility was not dependent on the size
of the entity, with the proportions varying by line of business. These proportions are not
disclosed in the OSFI analysis. To the extent that individual entity characteristics differ from the
average of the four largest entities contained in the 2013 OSFI analysis, the actuary would
adjust the volatility accordingly. Diversification would be included based on entity-specific
considerations.
These calculations represent rough approximations of a lifetime 65th, 75th, and 85th percentiles
(selected as examples not recommendations) based on the MCT. The findings in the preceding
table represent an approximation in a context where an entity has no better information to

29
Draft Educational Note May 2020

derive a second percentile point on the distribution of the present value of future cash flows
over a lifetime horizon and excludes the effect of diversification across lines of business and
between the LIC and the LRC. It is important to note that the calibration of the MCT factors
reflects a large entity’s relative volatility. To the extent that these parameters are different in a
particular entity’s RA calculation, the actuary would adjust the percentile factors accordingly.
Significant differences are possible. The actuary would take care to check the reasonability of
the standard deviations based on the MCT factors considering the facts and circumstances of
the entity.

30
Draft Educational Note May 2020

Appendix 1: Margins – Brief Summary of IFRS 4 CIA Standards of Practice

Subsections 2250 through 2270 of the CIA Standards of Practice 12 provided guidance to
actuaries in setting margins for adverse deviations prior to the effective date of IFRS 17. While
no longer binding after the effective date of IFRS 17, this guidance might be helpful to actuaries
in quantifying the degree of uncertainty in non-financial assumptions, and by extension
quantifying the compensation for non-financial risk that the entity might require.
Under Subsections 2250 through 2270, the range of margins for claims development was
between 2.5% and 20% of the best-estimate assumption. Selections above this range would be
appropriate in situations such as:
• Unusually high uncertainty; and
• Unusually low best estimate resulting in an unreasonably low dollar provision for
adverse deviations.
Selections below this range would be appropriate in situations such as:
• Coverage that is reserved at the stop loss limit.
Considerations for placement in the ranges would have been similar to those noted in IFRS
17.B91.

12
Canadian Institute of Actuaries, Standards of Practice – Insurance (January 1, 2020), 2023–2026.

31
Draft Educational Note May 2020

Appendix 2: Simplified Calculation of RA based on CoC Method

Simplified CoC calculations, such as the example presented in this appendix, could be an
alternative way to estimate the RA that allows insurers to use the general CoC concept. The
example included in this appendix may provide a more intuitive way to estimate the RA for
insurers that have a profit margin or combined ratio target instead of a target return on equity
(ROE).
The basic concept is that the target profit margin is allocated between reserve risk,
underwriting risk, and other risks that are not relevant to the RA.
The profit margin could be directly determined, in the case of an entity with a target profit
margin or combined ratio, or calculated for an entity with a target ROE and premium to surplus
ratio. Standard formula can be used to convert a target ROE and a premium to surplus ratio to a
target profit margin. A simple formula using ROE, corporate income tax, investment income on
surplus, and premium to surplus ratio is:
Profit Margin on Premium =
Target ROE
� (1-Tax)
- Investment Income on Surplus� /[Premium to Surplus Ratio]

Next, the total profit margin is split between underwriting risk, reserve risk, and other risks that
are not relevant to the RA, based on the proportion of the capital allocated to each of these
risks. The actuary may rely on ORSA or other processes used to allocate capital to reserve risk,
underwriting risk, and other risks.
Using these types of calculations, the actuary would recognize that underwriting risk disappears
once the coverage is expired and reserve risk diminishes over time as claims are settled. Thus,
using amounts derived from the profit margin on premium, to estimate the RA:
• The LRC is assigned both the profit margin associated with underwriting risk and reserve
risk; and
• Given that the underwriting risk does not exist for the LIC, the LIC is assigned only the
profit margin associated with reserve risk.
The RA amount associated with the LIC (i.e., premium multiplied by the profit margin
associated with reserve risk) would wind-down in an appropriate manner to reflect the
settlement of claims. Assuming that the reserve risk is correlated to the amount of claims that
are outstanding and unreported, the actuary could calculate the present value of future cash
flows at the beginning of each time period (i.e., the expected reserves) and then determine the
present value of this stream of cash flows. As a result, the reserve risk profit margin unwinds as
the present value of the stream of present values unwind. This is comparable to the rate at
which the RA decreases in the traditional CoC calculations. The applicable profit margin is the
RA at that point in time for the LIC.

32
Draft Educational Note May 2020

With these calculations, the confidence level for the RA is based on the distribution of the
present value of cash flows, which could be shaped significantly different for LIC versus LRC for
some P&C coverages.
Some potential limitations of this approach:
• The proportion of capital allocated to each risk may vary by portfolio or group;
• The profit margin may vary by portfolio or group (different ROE targets and/or different
premium to surplus ratios);
• The approach still requires a confidence level to be determined for disclosure purposes,
which requires a distribution of the present value of cash flows;
• The approach requires the projection of cash flows for the unwinding of the reserve risk;
• Changes in the allocation of capital by portfolio or group, or by risk, over time, which
may result as a change in mix or volume of business, could result in changes in indicated
RA; and
• Changes in profit margins objectives could result in changes in RA for prior policy years.
Illustrative Example
Assume that an insurer has only one line of business that it prices with a 10% profit margin.
Further assume that a robust ORSA model indicates that capital is allocated 50% for
underwriting, 30% for reserve, and 20% for other risks.
The profit margins associated with the different risk categories are 5% for underwriting risk, 3%
for reserve risk, and 2% for other risks.
The LRC RA would then be calculated as 8% of premium (5% for underwriting risk plus 3% for
reserve risk).
The LIC RA for a given policy year would start off at 3% of expired premium and decrease over
time, which as a percentage of the present value of future cash flows could be a higher or lower
value than the LRC RA.

33