A PUBLIC POLICY PRACTICE NOTE

Life Principle-Based Reserves (PBR)

Under VM-20

January 2019

Developed by the Life Principle-Based Approach Practice Note

Work Group of the Life Valuation Committee of the
American Academy of Actuaries
 VM-20 Practice Note

This practice note is not a promulgation of the Actuarial Standards Board, is not an
actuarial standard of practice, is not binding upon any actuary, and is not a definitive
statement as to what constitutes generally accepted practice in the area under discussion.
Events occurring subsequent to this publication of the practice note may make some of
the practices described in this practice note irrelevant or obsolete.

Life Principle-Based Approach Practice Note Work Group

Karen Rudolph, MAAA, FSA, Chairperson

Erik Anderson, MAAA, FSA Stephen Krupa, MAAA, FSA
Marc-Andre Brunet, MAAA, FSA Richard Nam, MAAA, FSA
Nadeem Chowdhury, MAAA, FSA David E. Neve, MAAA, FSA, CERA
Wei Ding, FSA, CERA Sean Pena, MAAA, FSA
Pamela Hutchins, MAAA, FSA Roland Rose, MAAA, FSA
Sheetal Kaura, MAAA, ASA, CERA Alan Routhenstein, MAAA, FSA
Jason Kehrberg, MAAA, FSA Lance Schulz, MAAA, FSA
Carrie Kelley, MAAA, FSA Amanda E. Young, MAAA, FSA, CPA

2018 Practice Note Update Work Group

Erik Anderson, MAAA, FSA Reanna Nicholsen, MAAA, FSA

Bryan Amburn, MAAA, FSA Gary Osterhout, MAAA, FSA
Elizabeth Caldwell, MAAA, FSA Roland Rose, MAAA, FSA
Nadeem Chowdhury, MAAA, FSA Karen Rudolph, MAAA, FSA
Jacqueline Keating, MAAA, FSA Grace Senat, MAAA, FSA
Jason Kehrberg, MAAA, FSA Benjamin Slutsker, MAAA, FSA
Stephen Krupa, MAAA, FSA Christopher Whitney, MAAA, FSA
Linda Lankowski, MAAA, FSA Amanda E. Young, MAAA, FSA, CPA
Michael McCarty, MAAA, FSA

1850 M Street NW, Suite 300

Washington, DC 20036-5805

© 2019 American Academy of Actuaries. All rights reserved.

 VM-20 Practice Note

TABLE OF CONTENTS
Introduction ......................................................................................................................... 1

PBR Calculation Schematic ................................................................................................ 2

1. Exemptions, Transition Rules, and Details on Products Covered .................................. 3

2. Available Information on Common Practice .................................................................. 6

3. VM-20 Calculation ......................................................................................................... 7

4. VM-20 Calculation Overview—Part A. Net Premium Reserve (NPR) ....................... 14

5: VM-20 Calculation Overview—Part B. Deterministic Reserve ................................... 23

6: VM-20 Calculation Overview—Part C. Stochastic Reserve ........................................ 28

7. Stochastic Exclusion Test ............................................................................................. 31

8: Deterministic Exclusion Test ........................................................................................ 37

9. Difference From Cash Flow Testing—Scenario Reserve Calculation ......................... 40

10. Considerations When Performing Reserve Calculations on Other Than the Valuation
Date ................................................................................................................................... 42

11. Details on Starting Assets and Asset Modeling .......................................................... 45

12. Details on Scenarios / Scenario Generators / Economic Assumptions ....................... 60

13. Setting Prudent Estimate and Anticipated Experience Assumptions ......................... 64

14. Setting Margins ........................................................................................................... 67

15. Setting Mortality Assumptions ................................................................................... 71

16. Setting Premium Assumptions.................................................................................... 81

17. Setting Policyholder Behavior Assumptions Other Than Premiums .......................... 83

18. Setting Expense Assumptions ..................................................................................... 87

19. Setting Non-Guaranteed Element Assumptions ......................................................... 89

20: Treatment of Reinsurance ........................................................................................... 92

21. Treatment of Hedging / Derivative Programs ............................................................. 97

© 2019 American Academy of Actuaries. All rights reserved.

 VM-20 Practice Note

Introduction

This practice note covers principle-based life insurance reserve valuation practices for individual life
insurance. Because the principle-based approach for life reserves is new, this practice note was not
developed from a survey of current actuarial practices. Rather, the practices here represent the views
of actuaries in industry, consulting, and public accounting firms that have been involved in the
development of the life reserving standards. The purpose of the practice note is to assist actuaries
with implementation of the principle-based life reserve valuation approach adopted by the NAIC as
detailed in the Requirements for Principle-Based Reserves for Life Products—2018 Edition of VM-20
dated November 22, 2017, describing the requirements for calculating minimum valuation standard
statutory reserves for individual life insurance products. It is important that the reader review any
differences that exist between the version of VM-20 on which this practice note was based and the
version that is viewed as applicable to current valuations because the Valuation Manual is a living
document and subject to change.

It is expected that actuarial practice for determining principle-based statutory reserves for life
insurance products will continue to emerge over time. As this practice note is an expectation of what
actuarial practice will emerge from the standards of the Valuation Manual, it is likely that additional
actuarial practice will be developed that is not contained in this practice note. The goal of this
practice note is twofold: to assist actuaries who are implementing VM-20 with the understanding of
what the requirements are and to provide an overview of industry practice. The work group
attempted to meet both goals as well as it could.

Additions and revisions to this practice note will likely be needed in the future as practices are further
developed and issues that are not anticipated below are addressed.

American Academy of Actuaries 1 www.actuary.org

 VM-20 Practice Note

PBR Calculation Schematic

For Term or ULSG: For Policies Other Than Term or ULSG:

Calculate Net Premium Calculate NPR Using VM-A and VM-C
Reserve (NPR) as Prescribed (Existing CRVM)
in VM-20

Clearly Defined Hedging Strategy?

No

Perform Stochastic Exclusion Test?

Yes Yes

Pass Stochastic Exclusion Test? No

Yes No

Term or ULSG Product? Calculate Stochastic

Reserve (SR)
Yes
No

Pass Deterministic Calculate Deterministic Calculate Deterministic

Exclusion Test No Reserve (DR) Reserve (DR)

Yes

Reserve Is NPR Reserve Is Reserve Is

NPR + Max [0, DR – NPR + Max [0, Max(DR, SR) –
(NPR - DDPA)] (NPR - DDPA)]

Chart assumes values are either consistently net of reinsurance or consistently direct (before reinsurance).
DDPA = Due and Deferred Premium Asset.
Each product group (ULSG; Term; Other) must follow this process separately.

American Academy of Actuaries 2 www.actuary.org

1. Exemptions, Transition Rules, and Details on Products Covered

Q 1.1: Which products are covered by VM-20?

A: According to Section II (Reserve Requirements) of the Valuation Manual, VM-20 applies to all individual
life insurance policies issued on or after the operative date of the Valuation Manual that fall within the scope
of VM-20. Policies subject to VM-20 include all individual life insurance policies whether directly written or
assumed through reinsurance:

• Universal life insurance policies;

• Variable life and variable universal life insurance policies;
• Term life insurance policies;
• Traditional whole life insurance policies;
• Indexed life and indexed universal life insurance policies; and
• Combination policies that include other benefits such as annuity benefits or long-term care benefits in
addition to life insurance benefits but are filed as individual life insurance policies.

Q 1.2: What products and reserves are not covered by VM-20, and where are these reserve requirements
listed?

A: The following shows the products and locations in the Valuation Manual of reserve requirements
according to Section II:

• Pre-need life insurance products are specifically excluded from VM-20. A definition of pre-need life
insurance can be found in VM-02 Section 3.B. Reserve requirements for pre-need life insurance
follow VM-A and VM-C.
• Industrial life insurance products are specifically excluded from VM-20. A definition of industrial life
insurance can be found in VM-02 Section 3.A. Reserve requirements for industrial life insurance
follow VM-A and VM-C.
• Annuity products—Reserve requirements are subject to VM-21 if variable annuity. For fixed annuity
contracts, the requirements are in VM-A and VM-C except for the minimum requirements for the
valuation interest rate for single premium immediate annuity contracts, and other similar contracts,
issued after Dec. 31, 2017. The maximum valuation interest rates for those contracts are defined in
VM-22. Section II of the Valuation Manual provides detail for establishing reserve requirements for
annuity products.
• Deposit-type products—Reserve requirements are subject to VM-A and VM-C. Section II of the
Valuation Manual provides detail for establishing reserve requirements for deposit-type products.
• Health insurance products—Reserve requirements are subject to VM-25 and VM-A and VM-C.
Section II of the Valuation Manual provides detail for establishing reserve requirements for health
insurance products.
• Credit life and disability products—Reserve requirements are subject to VM-26. Section II of the
Valuation Manual provides definitions of types of credit life and disability products and detail for
establishing reserve requirements for credit life and disability products.
• Claim reserves including waiver of premium claims are not subject to the PBR requirements of VM-
20, according to Section II of the Valuation Manual.

American Academy of Actuaries 3 www.actuary.org

Q 1.3: Are riders and supplemental benefits that are attached to life insurance policies subject to VM-20
reserve requirements?

A: According to VM-20 Section 2.H, reserves for riders and supplemental benefits are calculated as described
in Section II of the Valuation Manual. According to Section II, if the base policy is subject to VM-20 and the
rider or supplemental benefit has a separate premium or charge, then the rider reserve treatment varies. For
term life insurance riders on persons other than the named insured(s) on the base policy, the reserve may be
computed as if it were a stand-alone policy. For term life insurance riders on the named insured(s) of the base
policy, the reserves must be valued as part of the base policy. For riders such as secondary guarantee riders
that enhance or modify the terms of the base contract, those reserves must be valued as part of the base policy.

Generally, if the rider or supplemental benefit does not have a separate premium or charge, then all cash flows
associated with the rider must be included in the reserve calculation of the base policy. If the rider or
supplemental benefit has a separate premium or charge, then reserves may be calculated on a stand-alone
basis following the reserve requirements for that benefit.

Q 1.4: Are return of premium (ROP) riders attached to life policies treated as a stand-alone policy for
purposes of VM-20?

A: Consistent with Actuarial Guideline XLV, as referenced in VM-C, some actuaries might wish to combine
ROP benefits with the base policy benefits for the purposes of VM-20, whether there is a separate premium or
not.

Q 1.5: Will VM-20 apply to all inforce policies as of the operative date?

A: VM-20 applies only to policies issued on or after the operative date of VM-20. It does not apply to
business in force prior to the operative date.

Q 1.6: Are there any transition rules at the operative date of the Valuation Manual?

A: Section II of the Valuation Manual states that a company may elect to establish minimum reserves using
VM-A and VM-C for business otherwise subject to VM-20 during the first three years following the operative
date of the Valuation Manual.

Q 1.7: Do changes to a policy issued prior to the operative date of the Valuation Manual (e.g., the addition
of a rider) make it subject to the requirements of the Valuation Manual?

A: Section II of the Valuation Manual states that the minimum reserve requirements in the Valuation Manual
apply to contracts issued on or after the operative date. Therefore, only if a new contract is issued would the
requirements of the Valuation Manual, and VM-20 if applicable, apply. Many actuaries would conclude that
when a new policy number is issued for a contract after the operative date of the Valuation Manual, VM-20
would apply.

Q 1.8: What is the Life PBR Exemption?

A: Subject to commissioner approval, a Life PBR Exemption (informally referred to as the “small company
exemption”) is allowed for companies that meet the following conditions (Valuation Manual, Section II):

American Academy of Actuaries 4 www.actuary.org

 • The company has less than $300 million of ordinary life premium and, if the company is a member of
a group of life insurers, the group has combined ordinary life premiums of less than $600 million.
• The company has reported total adjusted capital of at least 450 percent of the authorized control level
risk-based capital (RBC) as reported in the prior calendar year annual financial statement, or has less
than $50 million of ordinary life premiums.
• The appointed actuary has provided an unqualified opinion on the reserves for the prior calendar year.
• Every ULSG policy issued or assumed by the company with an issue date on or after Jan. 1, 2020, and
in force on the company’s annual financial statement for the current calendar year-end valuation date
only has secondary guarantees that meet the VM-02 definition of “non-material secondary guarantee.”

Q 1.9: Is the Life PBR Exemption optional?

A: The exemption is optional. If a company decides to use the exemption, it must elect the exemption each
year prior to July 1 by filing a statement with its commissioner certifying the company meets the exemption
criteria. That statement must also be included with the NAIC second-quarter filing.

Q 1.10: If a company uses the Life PBR Exemption for a time period and then no longer qualifies or the
exemption is disallowed, are reserves recomputed from the effective date of PBR or going forward from the
date the exemption is no longer allowed?

A: The computation method in place at policy issue should be retained throughout the life of the policy.

Q 1.11: What is the process for electing the Life PBR Exemption?

A: Section II.D in the Valuation Manual provides that a company meeting all of the stated conditions may file
a statement of exemption for the current calendar year with its domestic commissioner prior to July 1 of that
year. The statement of exemption must state that the appointed actuary has issued an unqualified opinion on
reserves, that the exemption conditions are met based on premiums and other values from the prior calendar
year financial statements, and that any ULSG business issued since the operative date of the Valuation
Manual meets the definition for non-material secondary guarantee. The statement of exemption must be
included with the NAIC filing for the second quarter of that year.

The commissioner, however, may reject such statement of opinion prior to Sept. 1 and require the company to
follow the requirements of VM-20 for Ordinary Life policies.

Q 1.12: If a company elects the Life PBR Exemption, how are reserves computed?

A: If a company is granted the Life PBR Exemption, the minimum reserve requirements for Ordinary Life
policies are those pursuant to VM-A and VM-C.

Q 1.13: If a company uses the Life PBR Exemption, is it allowable to use the new mortality tables as they
are released?

A: Yes. The Life PBR Exemption provides that minimum reserve requirements for Ordinary Life policies
follow VM-A and VM-C using the mortality as defined in VM-20 Section 3.C.1 and VM-M Section 1.H (e.g.,
2017 CSO).

American Academy of Actuaries 5 www.actuary.org

Q 1.14: What is a non-material secondary guarantee universal life policy?

A: The term is defined in VM-01: Definitions for Terms in Requirements. There are three conditions that need
to be met at issue of the policy.
• The policy can have only one secondary guarantee provision, and it must be of the specified annual or
cumulative premium type.
• The duration of the secondary guarantee can be no longer than 20 years from issue for issue ages
through 60, grading down by 2/3 year for each higher issue age to 82. For issue ages 83 and older, the
period can be at most five years.
• The present value of the required premiums for the secondary guarantee must be at least as great as
the present value of net premiums over the maximum secondary guarantee period allowable under the
contract (in aggregate and subject to the above duration limit). Such present values use minimum
allowable Valuation Basic Table (VBT) rates and maximum valuation interest rate.

Q 1.15: Does a company have to apply for the Life PBR Exemption the first year that VM-20 is effective or
can it use the three-year phase-in period first before applying for the exemption?

A: Some actuaries would conclude that the three-year phase-in period can be applied before applying for the
Life PBR Exemption. An actuary may wish to check with the domiciliary state to be sure it agrees with that
interpretation.

2. Available Information on Common Practice

Q 2.1: Which actuarial standards of practice (ASOPs) would apply to the actuary when performing the
tasks in conjunction with determining reserves under VM-20?

A: According to ASOP No. 1, Section 4.3, “Actuaries are responsible for determining which ASOPs apply to the
task at hand.” The following ASOPs, as of the date of this practice note, are among those the actuary may wish
to consider:

ASOP No. 1 Introductory Actuarial Standard of Practice

ASOP No. 2 Nonguaranteed Charges or Benefits for Life Insurance Policies and Annuity Contracts
ASOP No. 7 Analysis of Life, Health, or Property/Casualty Insurer Cash Flows
ASOP No. 11 Treatment of Reinsurance Transactions Involving Life or Health Insurance 1
ASOP No. 12 Risk Classification (for All Practice Areas)
ASOP No. 15 Dividends for Individual Participating Life Insurance, Annuities, and Disability
Insurance
ASOP No. 22 Statements of Opinion Based on Asset Adequacy Analysis by Actuaries for Life and
Health Insurers
ASOP No. 23 Data Quality
ASOP No. 38 Using Models Outside the Actuary’s Area of Expertise (Property and Casualty)
ASOP No. 41 Actuarial Communications
ASOP No. 52 Principle-Based Reserves for Life Products under the NAIC Valuation Manual

American Academy of Actuaries 6 www.actuary.org

The Actuarial Standards Board is likely to issue a fourth exposure draft of a new proposed ASOP, Modeling
and a second exposure draft of a new proposed ASOP, Setting Assumptions. (ASOP No. 1, Section 3.1.7
explains: “Exposure drafts of the ASOP … form part of the literature of the actuarial profession; actuaries
may look to them at their discretion for advisory guidance. An ASOP is not binding until the effective date of
the ASOP.”)

Q 2.2: Are there other practice notes that cover topics relevant to principle-based reserve calculations as
described in the Valuation Manual?

A: The Asset Adequacy Analysis practice note, the Credibility practice note, and the Model Governance
practice note may include relevant information for actuaries performing PBR reserve calculations. These
practice notes can be found at the American Academy of Actuaries website at
http://www.actuary.org/category/site-section/public-policy/practice-notes.

Q 2.3: VM-20 often uses the term “Qualified Actuary.” What does VM-20 mean by Qualified Actuary and
how does this differ from “Appointed Actuary”?

A: The term “Appointed Actuary” means a Qualified Actuary who is appointed or retained in accordance with
the Valuation Manual to prepare the actuarial opinion required in Section 3.A and/or 3.B of VM-05. The term
“Qualified Actuary” means an individual who meets the requirements specified in the Valuation Manual and
is thereby qualified to sign the applicable statement of actuarial opinion. The requirements an actuary must
satisfy to be considered qualified to sign the applicable statement of actuarial opinion can be found in the
Qualification Standards for Actuaries Issuing Statements of Actuarial Opinion in the United States (2008)
promulgated by the American Academy of Actuaries.

Q 2.4: Are there practices in other countries that an actuary can review for reference?

A: Published papers on principle-based reserve calculations in other countries may provide useful information
to U.S. actuaries. It should be noted, however, that acceptable practice in other countries is not a safe harbor
for principle-based calculations in the United States. Annotation 3-1 of the Code of Professional Conduct
states that “[i]t is the professional responsibility of an Actuary to observe applicable standards of practice …
for the jurisdictions in which the Actuary renders Actuarial Services.” U.S. actuaries subject to the Code are
obliged to follow the standards of practice promulgated by the Actuarial Standards Board. Annotation 3-2 of
the Code of Professional Conduct establishes, however, that “[w]here a question arises with regard to the
applicability of a standard of practice or where no applicable standard of practice exists, an Actuary shall
utilize professional judgment, taking into account generally accepted actuarial principles and practices.”

The Canadian Institute of Actuaries has Valuation Technique Papers (VTP) and educational notes that explain
how Canadian actuaries calculate reserves. There are some similarities between U.S. principle-based reserves
and Canadian valuation techniques. For Canadian documentation, please see the Canadian Institute of
Actuaries website at www.actuaries.ca.

3. VM-20 Calculation

American Academy of Actuaries 7 www.actuary.org

Q 3.1: VM-20 describes three components of the minimum reserve: the Net Premium Reserve (NPR), the
Deterministic Reserve (DR), and the Stochastic Reserve (SR). Is the company required to calculate all three
components for all policies?

A: The Net Premium Reserve is required to be calculated for all policies subject to VM-20. The company may
elect to perform exclusion tests that, if passed, exempt some groups of policies from the Deterministic
Reserve and/or the Stochastic Reserve. These exclusion tests are optional, and a company can decide to
calculate all three components for all policies. For certain product types, namely term and ULSG with a
material secondary guarantee provision (i.e., one that does not meet the definition of “non-material secondary
guarantee”), Deterministic Reserves must be calculated so the Deterministic Exclusion Test is not applicable.

Later sections of this practice note provide detail for each of the three components that go into the calculation
of the minimum reserve: the Net Premium Reserve, the Deterministic Reserve, and the Stochastic Reserve.

Q 3.2: What is the minimum reserve for all policies as required by VM-20?

A: Section 2.A of VM-20 defines the minimum reserve in terms of product groups. The total minimum
reserve equals the sum of the minimum reserve as determined for each product group.

It is important to note that VM-20 Section 3 defines the requirements for the policy Net Premium Reserve;
Section 3.E defines the policy minimum Net Premium Reserve as the policy Net Premium Reserve less a
credit for reinsurance ceded.

In general, the minimum reserve for a product group is determined as the sum of policy minimum Net
Premium Reserves plus the excess, if any, of the greater of the Deterministic Reserve for all policies in the
product group and the Stochastic Reserve for all policies in the product group over the quantity (A-B), where
A equals the sum of policy minimum Net Premium Reserves for all policies in the product group and B equals
any due and deferred premium asset held on account of those policies. When determining the minimum
reserve on a net-of-reinsurance basis, all three reserve components are net of any credit for reinsurance ceded
for the policies in the product group.

In mathematical terms, this could be represented as:

Minimum Reserve = AggNPR + Max(0, (Max(DR, SR) – (AggNPR – DDPA)))

Where

AggNPR = Sum of Policy Minimum Net Premium Reserves

DR = Deterministic Reserve
SR = Stochastic Reserve
DDPA = Due and Deferred Premium Asset

Section 2 defines three product groups: Term Policies, Universal Life with Secondary Guarantee policies, and
life insurance policies subject to Section 3.A.2. The sum of the minimum reserve for each product group
equals the total minimum reserve.

All policies are assumed to be subject to the calculation of all three reserve components of VM-20 (NPR, DR,
SR) unless the company has either elected to exclude the group of policies from the SR calculation or both the
SR and DR calculations and has applied and passed the applicable exclusion tests and documented the results.
American Academy of Actuaries 8 www.actuary.org
However, VM-20 does not allow the Deterministic Exclusion Test for either the Term product group or the
ULSG product group. Therefore, for these two product groups, all three reserve components of VM-20 (NPR,
DR, SR) will need to be calculated unless the company elects to exclude the Term and/or ULSG product
group of policies from the SR calculation, applied and passed the Stochastic Exclusion Test, and documented
the results.

If the company makes use of the exclusion tests, then for a group of policies that pass both the Stochastic
Exclusion Test and the Deterministic Exclusion Test, the minimum reserve for that group of policies is the
sum of the policy minimum Net Premium Reserves.

For a group of policies that pass the Stochastic Exclusion Test but not the Deterministic Exclusion Test, the
minimum reserve equals: the sum of the policy minimum Net Premium Reserves plus the excess, if any, of
the Deterministic Reserve over the sum of the policy minimum Net Premium Reserves adjusted for any due
and deferred premium asset held on account of those policies.

For a group of policies that fail the Stochastic Exclusion Test and for policies not subjected to any exclusion
tests, the minimum reserve equals: the sum of the policy minimum Net Premium Reserves plus the excess, if
any, of the greater of the Deterministic Reserve and the Stochastic Reserve over the sum of the policy
minimum Net Premium Reserve adjusted for any due and deferred premium asset held on account of those
policies.

Q 3.3: Why is the Net Premium Reserve adjusted for any due and deferred premium asset in determining
the excess, if any, of the greater of the Deterministic Reserve and Stochastic Reserve over the Net Premium
Reserve?

A: Because two of the components of the comparison (the Deterministic Reserve and Stochastic Reserve) are
calculated as of the reporting date, and require that due premiums be included in the expected future cash
flows when calculating the DR and SR, whereas the Net Premium Reserve is as of the policy anniversary
date, so per statutory accounting rules an adjustment to the Net Premium Reserve is required to put all of the
values on the same basis.

Q 3.4: How would actuaries approach the calculation of the minimum reserve requirement under VM-20?

A: One approach for completing the calculation is outlined below.

• Determine policies in scope of the VM-20 requirements. Determine in which of the three product
groups of Section 2 these policies belong.

• Within each product group, determine the model segments for all policies in scope of the
requirements. Per the definition of Model Segment in VM-01 (definition #32), this determination will
generally align with the company’s asset segmentation plan, investment strategies, or approach used
to allocate investment income for statutory purposes. It should be noted that a model segment could be
an entire block of business.

• Select the amount of starting assets for each model segment and allocate existing assets to each model
segment.

American Academy of Actuaries 9 www.actuary.org

• Build asset and liability populations in a cash flow model. This cash flow model may represent each
in-scope policy in force on the date of valuation or represent policies by grouping such policies into
representative cells of model plans as described in Section 7.B.2.

• Determine anticipated experience assumptions for all risk factors.

• Determine investment expense assumptions and asset default assumptions for each model segment.

• Determine prudent estimate assumptions for all risk factors that are not prescribed or stochastically
modeled by applying margins to the anticipated experience assumptions.

• Perform Stochastic Exclusion Test (if the company elects to do so). This may be performed for any
block of policies for which this test is deemed appropriate. If the block of policies passes the test, the
company can skip the calculation of the Stochastic Reserve for those policies.

• Determine the Stochastic Reserve as described in Section 5 of VM-20 for policies where the
Stochastic Reserve is required or deemed appropriate.

• Perform Deterministic Exclusion Test (if the company elects to do so). This may be performed only
for life insurance policies subject to Section 3.A.2 (per Section 2.A.3). If the block of policies passes
the test, the company can skip the calculation of the Deterministic Reserve for those policies.

• Calculate the Deterministic Reserve as described in Section 4 of VM-20 for policies where the
Deterministic Reserve is required.

• Calculate the Net Premium Reserve for all policies subject to VM-20 as described in Section 3 of
VM-20.

• Calculate the minimum reserve for all policies within each Section 2 product group subject to VM-20
as the sum of the following amounts:

o For the group of policies that pass both the Stochastic Exclusion and the Deterministic
Exclusion Tests, the minimum reserve equals the sum of the policy Net Premium Reserves.

o For the group of policies that pass the Stochastic Exclusion Test but fail the Deterministic
Exclusion Test, the minimum reserve equals the sum of the policy Net Premium Reserves plus
the excess, if any, of the Deterministic Reserve for those policies over the quantity (A - B),
where A = the sum of the policy Net Premium Reserves and B = any due and deferred
premium asset held on account of those policies.

o For the group of policies that fail the Stochastic Exclusion Test, and for the group of policies
not subject to the exclusion tests, the minimum reserve equals the sum of policy Net Premium
Reserves plus the excess, if any, of the greater of the Deterministic Reserve and the Stochastic
Reserve over the quantity (A - B), where A = the sum of the policy Net Premium Reserves
and B = any due and deferred premium asset held on account of those policies.

• Calculate the total minimum reserve as the sum of the minimum reserve for each of the three product
groups.

American Academy of Actuaries 10 www.actuary.org

Some actuaries would undertake approaches that are different than what is summarized above depending on
the specific circumstances of their company.

Q 3.5: In determining the minimum reserve under Section 2, how should separate accounts be handled
when comparing the NPR, DR, and SR for variable products?

A: Section II of the Valuation Manual specifies that minimum reserve requirements for variable and non-
variable individual life contracts are provided by VM-20, except for pre-need life contracts, industrial life
contracts, and credit life contracts. The NPR for variable products is determined according to Section 3.A. If
the variable contract is a term policy or ULSG, then Section 3.A.1 applies. Otherwise, Section 3.A.2 applies.
Section 3.A.2 would point to the current Commissioners Reserve Valuation Method (CRVM) requirements,
which include a provision for Separate Accounts in the reserve. So the comparison in Section 2 for variable
products does reflect separate accounts in each of the three components.

Q 3.6: When allocating the total reserve between the general account (GA) and the separate account (SA),
VM-20 states that the amount allocated to the GA must not be less than zero, and the amount allocated to
the SA must not be less than the sum of the cash surrender values and not be greater than the sum of the
account values attributable to the separate account portion of all such contracts. If the company books a
negative amount into the general account due to the CRVM expense allowance, couldn’t this result in an
increase in the total reserve (GA + SA) if the negative amount cannot be recognized?

A: Because the SA reserve has a floor of the variable cash surrender value and a ceiling of the variable
account value, the first constraint of GA reserve not less than zero also means there is an implicit floor of the
SA equal to the minimum reserve. So the sum of the SA and GA will not ever exceed the minimum reserve.

Q 3.7: Why would an actuary calculate the Stochastic Exclusion Test?

A: Some actuaries would calculate the Stochastic Exclusion Test because the block of policies does not have
material market risk and the Stochastic Reserve will not contribute to the minimum reserve calculation. The
benefit in that instance is that the time and expense to determine the Stochastic Reserve is not required. Some
actuaries may decide to perform the stochastic calculation even if the group of policies would pass the
Stochastic Exclusion Test because there may be some diversification or risk offsets the company would then
recognize in the minimum reserve calculation under VM-20 or for other reasons.

Q 3.8: Why would an actuary calculate the Deterministic Exclusion Test?

A: Some actuaries with policies that pass the Stochastic Exclusion Test would also perform the Deterministic
Exclusion Test to avoid the burden of performing the deterministic calculation. Again, some actuaries would
still calculate the Deterministic Reserve even if the policies pass the Deterministic Exclusion Test as it can be
used in the VM-20 calculation even if the Deterministic Exclusion Test is passed.

Q 3.9: How does an actuary define a model segment and determine the policies to include in each model
segment?

A: Section 7.A of VM-20 addresses the cash flow model requirements for the Stochastic and Deterministic
Reserves. This section requires the model segments to be consistent with the company’s asset segmentation
plan, investment strategies, or approach used to allocate investment income for statutory purposes. Each
American Academy of Actuaries 11 www.actuary.org
policy can be included in only one segment. Some actuaries might also consider how non-guaranteed
elements are set in determining model segments. It should be noted that a model segment can be an entire
block of business or an entire product group, where product groups are defined by Section 2.A.

Q 3.10: What is the difference between the grouping of policies described in Section 7.B.2 and the product
groups described in Section 2.A?

A: Section 7.B.2 addresses the level of granularity when constructing the cash flow model. VM-20 allows
policies to be grouped into modeling cells for both the Stochastic Reserve and Deterministic Reserve
calculation, rather than requiring a seriatim, policy-by-policy reserve calculation. VM-20 requires that the
grouping of policies must be done in a manner consistent with Section 2.G. VM-20 requires a seriatim
calculation (i.e., with no grouping) for the Net Premium Reserve.

Section 2.A refers to the minimum reserve in terms of three product groups, where these groups consist of
Term policies, ULSG policies, and other life insurance policies subject to Section 3.A.2. Within each of these
product groups, the concept of aggregation refers to the combining of cash flows when calculating the
Stochastic Reserve for the purpose of recognizing the amount of risk diversification among the policies.
Aggregating policies into a common model segment allows the cash flows arising from the policies for a
given stochastic scenario to be netted against each other (i.e., allows risk offsets between policies to be
recognized). Full aggregation would find the cash flows for all policies are combined together in one large
group. Currently, full aggregation is not allowed due to the product groups defined by Section 2.A and the
allocation provision of Section 5.G, which would come into play if a company includes policies from two or
more product groups in the calculation of a Stochastic Reserve.

Q 3.11: What considerations should be taken into account when deciding how to group policies when
defining modeling cells for the cash flow model under Section 7.B.2?

A: First, the actuary will want to decide which of the Section 2.A product groups each policy belongs in.
Then, within each product group, the actuary may wish to consider the similarities between policies and their
respective assumptions when grouping policies together. Some actuaries may use model office projections for
a subset of scenarios to determine the impact various groupings may have on the resulting reserve amount to
ensure that the policy groupings do not have a material impact. Some actuaries may rely on seriatim
projection output data across a sampling of scenarios to gather data that suggests how policies should be
grouped together. The actuary may wish to consider Section 3.3 of ASOP No. 52, Principle-Based Reserves
for Life Products under the NAIC Valuation Manual, which discusses considerations in modeling for
principle-based calculations.

Q 3.12: What is the required modeling time step / frequency of projection?

A: While there is no required model time step in the VM-20 requirements, actuaries commonly use monthly,
quarterly, or annual time steps in cash flow projections. In choosing a time step, actuaries may wish to
consider factors such as product characteristics, the frequency and method of setting credited interest rates or
other non-guaranteed elements, the sensitivity of the projection to the time step, and practical limitations.
Some actuaries may have a quarterly time step for a specific model segment while using a monthly time step
for other model segments. Some actuaries might consider longer (annual) time steps for very stable model
segments with little interest rate sensitivity.

Q 3.13: What is the required length of the projection period?

American Academy of Actuaries 12 www.actuary.org

A: Section 7.A.1.d mandates that the model project “cash flows for a period that extends far enough into the
future so that no obligations remain.” However, Section 2.G allows for approximations when these
approximations do not cause a material understatement of the reserve. Some actuaries might interpret this to
mean that shorter projection periods are appropriate when no material liabilities remain after some period of
time, or when the actuary can demonstrate that longer projection periods would not result in a materially
greater reserve. Some actuaries would instead assume a 100 percent termination rate (either through death or
surrender) to ensure no obligations remain at the end of the projection period where the actuary believes that
this assumption would not materially understate the reserve.

Q 3.14: How would the actuary determine whether using a longer projection period would result in a
“materially greater reserve”?

A: Some actuaries may be able to determine that the amount of business in force after a certain period is
immaterial and could not lead to a materially greater reserve.

It is also possible that some actuaries would use current or historical results to determine that the greatest
present value of accumulated deficiencies is achieved within the projection period for every scenario in the
Stochastic Reserve calculation. This analysis could include actually performing the projection with a longer
projection period (potentially with a more compressed model) and determining that there is no material
increase in the Stochastic Reserve calculation. Other analysis could be used including an analysis of when the
greatest present value of accumulated deficiency (GPVAD) occurs in the calculation (i.e., if the GPVAD
occurs at a point within the projections for all of the scenarios where it is not possible for future deficiencies
to become the greatest).

Other actuaries may rely on historical reserve calculations. For example, assume a longer projection period
was run historically and showed that in all cases the GPVAD occurred prior to a certain projection year.
Assuming no changes in the policy mix or assumptions have been made that would affect this outcome, a
projection period that includes the GPVAD year but does not go all of the way out to the end of the policy
period for all policies may be shown not to materially understate the reserve.

Q 3.15: How is an individual policy reserve defined under VM-20?

A: The seriatim Net Premium Reserve is the floor for the reserve for any specific policy. Section 2.C specifies
that the reserve for each product group, as defined in Section 2.A, is allocated to each policy within that
product group in the same proportion as the minimum Net Premium Reserve for that policy to the minimum
Net Premium Reserve for the product group.

Each policy’s share of the excess reserve is determined by multiplying the Net Premium Reserve for that
policy by the ratio of the reserve excess divided by the sum of policy Net Premium Reserves for the
applicable product group. In this context, the Net Premium Reserve is net of any credit for reinsurance ceded.
For example, consider policy (x):

NPR(x) = 100
DR Excess of Product Group A (containing (x)) = 80
Sum of policy minimum NPRs of Product Group A = 1,000

Min Reserve(x) = 100 + [100 x (80 / 1000)] = 108

American Academy of Actuaries 13 www.actuary.org

Q 3.16: How might an actuary determine that simplifications and approximations do not cause a material
understatement of the reserve, as required by Section 2.G?

A: Some actuaries may test the impact of using simplifications / approximations by calculating the minimum
reserve without the simplifications / approximations on a small block of policies that are a good proxy for the
entire group of policies, and then comparing the result to the minimum reserve on the same small block of
policies with the simplifications / approximations. Another approach used by some actuaries may be to
calculate the minimum reserve on all policies both with and without the simplifications / approximations
every three to five years to see whether there are material understatements. This comparison could occur at a
time other than the valuation date.

4. VM-20 Calculation Overview—Part A. Net Premium Reserve (NPR)

Q 4.1: How does an actuary determine which of the Net Premium Reserve calculations in Section 3 of VM-
20 apply?

A: Section 3.A.1 of VM-20 provides that term insurance policies and universal life policies with a secondary
guarantee should follow the calculations in Section 3. Section 3.A.2 provides that all other policies subject to
VM-20, but for which 3.A.1 does not apply, are subject to the requirements in VM-A, VM-C, and VM-M for
basic reserves. VM-A, VM-C, and VM-M reproduce the CRVM methodology and assumptions in existence
prior to VM-20.

VM-01 includes a definition for the term “secondary guarantee.” It states:

A “secondary guarantee” is a conditional guarantee that a policy will remain in force for either:

• more than five years (the secondary guarantee period), or

• five years or less (the secondary guarantee period) if the specified premium for the secondary
guarantee period is less than the net level reserve premium for the secondary guarantee period based
on the CSO valuation tables defined in VM-20 Section 3.C and VM-M and the valuation interest rates
defined in this Section, or if the initial surrender charge is less than 100% of the first year annualized
specified premium for the secondary guaranteed period, even if its fund value is exhausted.

Once the policy type has been determined, the NPR methodology follows according to policy type as
referenced below.

Policy Type Applicable NPR Methodology

Term Insurance Section 3.B.4.
ULSG—during SG period Max(Section 3.B.5; 1 Section 3.B.6)
ULSG—after expiration of the SG period Section 3.B.5

1
Where 3.B.5 is calculated assuming the policy has no secondary guarantee(s).

American Academy of Actuaries 14 www.actuary.org

Q 4.2: What are the steps for determining the NPR for term insurance policies?

A: Per Section 3.B.4.b, determine the adjusted gross premiums for the policy. These will be equal to the
annual mode guaranteed gross premiums for the policy multiplied by the factors below.

Policy Year 1: 0%
Policy Years 2-5: 90%
Policy Years 6+: 100%

Then, determine the uniform percentage of the present value of adjusted gross premiums equivalent to the
present value of benefits (PVB) at issue plus $2.50 per $1,000 of insurance. The product of the uniform
percentage and the adjusted gross premiums is the vector of valuation net premiums (unless adjustments as
described next are required).

Adjustment to the valuation net premium may be required for policies subject to the shock lapse provisions
(please see Section 3.C.3.b.v of VM-20) if, for periods following the shock lapse, the present value of
valuation net premiums (PVP) exceeds the PVB by more than 35%. In this situation, the valuation net
premium following the shock lapse must be reduced uniformly to produce a PVP/PVB ratio of 135%. If the
application of the 135% limitation results in an adjustment to the net valuation premiums following the shock
lapse, increase the valuation net premiums for policy years prior to the shock lapse by a uniform percent. At
issue and after adjustments, the present value of adjusted gross premiums equals the PVB at issue plus $2.50
per $1,000 of insurance. This situation results in two uniform percentages, one for the policy years prior to the
shock lapse and one for the policy years following the shock lapse.

The Net Premium Reserve equals the present value of future benefits less the present value of future valuation
net premiums but not less than the greater of the policy’s cash surrender value and the cost of insurance to the
date to which the policy is paid. The cash surrender value used should be consistent (from the standpoint of
determining the value on other-than-anniversary dates) with that used to determine the Net Premium Reserve
on other-than-anniversary dates.

For valuation dates other than on policy anniversary, the Net Premium Reserve is intended to assume an
annual premium mode for the policy and the actual valuation date relative to the policy issue date. A deferred
premium asset may be required under accounting rules using the method in Section 3. The approach of
calculating the reserve for a given policy taking into account the exact issue date of the policy may be similar
to approaches undertaken prior to the Valuation Manual adoption. Background on these approaches can be
found in SSAP 51.
Q 4.3: What are the steps for determining the Net Premium Reserve for a universal life policy without a
secondary guarantee, and for a universal life policy with a secondary guarantee?

A: Per VM-20 Section 3.A.2, if a universal life policy form has no secondary guarantee as defined in VM-20,
then the Net Premium Reserve must be determined according to the requirements in VM-A, VM-C, and VM-
M. These are the same requirements applicable before the operative date of the Valuation Manual.

Per VM-20 Section 3.A.1, if a universal life policy form has a secondary guarantee, then the Net Premium
Reserve must be determined according to the requirements of Sections 3.B.5 and 3.B.6.

American Academy of Actuaries 15 www.actuary.org

Q 4.4: VM-20 Section 3.B.5 addresses requiring the calculation of a reserve determined without
consideration for the secondary guarantee provision. What purpose does this serve, and why is it necessary
for a ULSG NPR?

A: The process for calculating the NPR for a ULSG policy involves the comparison of two calculated reserve
amounts. The first reserve amount is defined in Section 3.B.5. This amount prevails when it is larger than the
second reserve amount, which is defined in Section 3.B.6. The Section 3.B.5 reserve amount also prevails
when the secondary guarantee period has expired. The steps to get a Section 3.B.5 reserve amount are
described below.

• First determine the level gross premium at issue such that if this premium is paid each year for which
premiums are permitted, the policy would remain in force for the entire coverage period. This
determination is made using the policy’s guarantees of interest, mortality, and expense. *Note that
unlike Universal Life Model Regulation methodology, “maturity” (i.e., endowment) of the policy is
not required, only inforce status.

• The premium derived in the step above is used in defining the expense allowance components for the
policy at issue as shown below.

o Policy Year 1: 100% of the premium and $2.50 per $1,000

of insurance
o Policy Years 2-5: 10% of the premium
o Policy Years 6+: 0% of the premium

• Determine the valuation net premium ratio from Section 3.B.5.c. The ratio is derived by taking the
PVB at issue divided by the present value at issue of future gross premiums from the first step. These
actuarial present values are calculated using the interest, mortality, and lapse assumptions that are
appropriate for the policy from Section 3.C.

• The valuation net premium is the gross premium from the first step times the valuation net premium
ratio.

• At any valuation date t, the Net Premium Reserve will equal the product of mx+t times rx+t. The
quantity mx+t is the present value of future benefits less the present value of future valuation net
premiums and less the unamortized expense allowance for the policy. Determination of the
unamortized expense allowance is provided in Section 3.B.5.b.

• The quantity rx+t is a ratio of two fund values, ex+t and fx+t. The amount ex+t equals the actual policy
fund value on the valuation date floored at zero. The amount fx+t is determined as that amount which,
together with the payment of future level gross premiums from the first step above, will keep the
policy in force for the entire period that coverage is provided, using the policy guarantees for
mortality, interest, and expense. The quantity rx+t equals 1 if fx+t is less than or equal to zero, otherwise
rx+t equals the quantity ex+t divided by fx+t capped at 1.

Q 4.5: What are the steps for determining the Net Premium Reserve for universal life policies with a
secondary guarantee?

American Academy of Actuaries 16 www.actuary.org

A: The net premium approach for universal life policies with a secondary guarantee provision that meets the
definition of secondary guarantee as provided by VM-01 is explained below. If, however, the secondary
guarantee period has expired, then the methods described in Question 4.4 apply. If a policy has more than one
secondary guarantee period, the approach assumes the longest guarantee period for which the policy can
remain in force.

• During the secondary guarantee period, the Net Premium Reserve is the greater of the reserve
calculated assuming no secondary guarantee and the reserve assuming the secondary guarantee. After
the end of the secondary guarantee period, the reserve is calculated according to the requirements for
universal life policies without secondary guarantees. (Section 3.B.6.a.)

• As of the policy issue date, find the level gross premium that will maintain the policy in force for the
length of the secondary guarantee period based on the secondary guarantee provisions for mortality,
interest, and expenses. The valuation net premium is the uniform percentage of this gross premium
such that at issue and over the secondary guarantee period, the present value of future valuation net
premiums equals the present value of future benefits. The uniform percentage is the Valuation Net
Premium Ratio (VNPR) and will not change for the policy. (Section 3.B.6.c.iii.)

• Base the valuation expense allowance components on the level premium amount determined in the
prior step. (Section 3.B.6.c.ii.) The expense allowance components are:

o Policy Year 1: 100% of the premium and $2.50 per $1,000 of insurance
o Policy Years 2-5: 10% of the premium
o Policy Years 6+: 0% of the premium

Section 3.B.6.c.ii shows that the expense allowance is further adjusted by the VNPR.

• After issue and on each future valuation date, t, the Net Premium Reserve is determined as described
below. The terms in this step are based on the definitions found in Section 3.B.

𝐴𝐴𝐴𝐴𝐴𝐴𝑥𝑥+𝑡𝑡
𝑀𝑀𝑀𝑀𝑀𝑀 � , 1� ∗ 𝑁𝑁𝑁𝑁𝑁𝑁𝑥𝑥+𝑡𝑡 − 𝐸𝐸𝑥𝑥+𝑡𝑡
𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝑥𝑥+𝑡𝑡
Where

𝐴𝐴𝐴𝐴𝐴𝐴𝑥𝑥+𝑡𝑡 = the amount of the policy’s actual secondary guarantee on the valuation date

𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝑥𝑥+𝑡𝑡 = the amount necessary to fully fund the policy’s secondary guarantee on the valuation date

𝑁𝑁𝑁𝑁𝑁𝑁𝑥𝑥+𝑡𝑡 = the net single premium on the valuation date for the coverage provided by the secondary
guarantee for the remainder of the secondary guarantee period, using the interest lapse and mortality
assumptions found in Section 3.C; the 𝑁𝑁𝑁𝑁𝑁𝑁𝑥𝑥+𝑡𝑡 includes consideration for death benefits only

𝐸𝐸𝑥𝑥+𝑡𝑡 = the policy’s unamortized expense allowance at the valuation date calculated in a manner
consistent with that for the policy without secondary guarantee, but using secondary guarantee
parameters and a different valuation interest rate

The amount determined in this step is to be compared with the amount determined for the policy,
absent the secondary guarantee, and the greater amount used as the Net Premium Reserve for the

American Academy of Actuaries 17 www.actuary.org

 policy.

Q 4.6: How is the Net Premium Reserve calculated for policies with non-level death benefits or death
benefits where the level of benefits is not guaranteed?

A: Similar to the pre-principle-based reserve (pre-PBR) Standard Valuation Law, the NPR section of VM-20
was drafted in the context of a level annual premium, level benefit contract. Adjustments to handle non-level
benefits would generally follow the similar adjustments made pre-PBR. The only item in the NPR calculation
that specifically relates to benefits is the first-year expense allowance of $2.50 per $1,000. No explicit
instructions are included in VM-20 about whether a non-level death benefit would affect the first-year value
to which the $2.50 is applied, and therefore some actuaries would use the benefits payable in the first year.

Q 4.7: How is the Net Premium Reserve calculation affected if structural changes are made to the policy
(i.e., changes separate from the automatic workings of the policy and initiated by the policyholder)?

A: In the NPR calculation, some actuaries would follow the principles and practices that would apply under
pre-PBR CRVM if there were structural changes to the policy.

Q 4.8: How are non-annual modes reflected in the Net Premium Reserve calculation?

A: The Net Premium Reserve requirements were written to provide a standard for terminal reserves under an
assumption of an annual mode premium similar to the pre-PBR formulaic requirements. Standard actuarial
adjustments such as inclusion of a deferred premium asset or unearned premium liability can be made to
accommodate actual premium modes that are not annual. For flexible premium products, some actuaries
would not calculate a deferred premium asset or unearned premium liability, which is consistent with actuarial
practice pre-PBR.

Q 4.9: How are policy fees considered?

A: For universal life products, the Net Premium Reserve approach uses a “solved-for” premium and, as a
result, any policy fees required by the product would not directly enter into the calculation. For term insurance
products, the Net Premium Reserve approach requires the use of the maximum guaranteed gross premium for
the policy year as the gross premium used in setting the valuation net premium. Section 3 does not
specifically discuss how policy fees enter into this calculation; however, some actuaries would include the
policy fee for the policy in determining the uniform percentage of Section 3.B.4.a.

Q 4.10: Do return of premium (ROP) products introduce any special considerations for the Net Premium
Reserve?

A: Section 3.B.7 requires the actuarial present value of future benefits to include death benefits, endowment
benefits including endowments intermediate to the term of coverage, and cash surrender benefits. Some
actuaries would conclude that “endowments” includes ROP benefits. All these benefits are determined before
consideration for reinsurance and before consideration for policy loans in force. In the situation of a return of
premium benefit feature, all interim return amounts must be included in the Net Premium Reserve calculation.
In this situation, Section 3.C.3.b.iii specifies a 6 percent lapse rate for the first half of the initial level premium
period, and 0 percent for the remainder of the initial level premium period.

Q 4.11: Is the Net Premium Reserve calculation similar to Universal Life Model Regulation Reserves
(ULMR)?
American Academy of Actuaries 18 www.actuary.org
A: In terms of Section 3.B.5 (Net Premium Reserve for universal life without considering any secondary
guarantee provisions), there are some similarities. However, while there are concepts that are similar, there
are some differences. Both methods employ premium-solves at issue and at valuation dates. However, the
ULMR is solving for a maturity premium, while the VM-20 method is solving for a premium that keeps the
policy in force and not necessarily matures the policy. The “r” ratio is similar in that it is a ratio of the actual
policy account value to a solved-for account value, but in the context of ULMR, the solved-for account value
considers maturity of the policy, while in the context of VM-20, the solved-for account value considers only
maintaining the policy in force.

Q 4.12: For fixed premium policies, if an interpolated terminal plus an unearned premium liability CRVM
reserve is calculated instead of a mean reserve, would the actuary still need to calculate the Deferred
Premium Asset?

A: Terminal Net Premium Reserves are computed as of the end of a policy year and not the reporting date, so
the terminal reserve as of policy anniversaries immediately prior and subsequent to the reporting date are
adjusted to reflect that portion of the net premium that is unearned at the reporting date. This is generally
accomplished using either the mean reserve method or the mid-terminal method. Other appropriate methods,
including an exact reserve valuation, are also commonly used (see SSAP 51). Whatever valuation reserve
method is used, the actuary may wish to confirm that all three components (Deterministic Reserve, Stochastic
Reserve, and Net Premium Reserve as adjusted for the reporting date) are internally consistent with respect to
the assumption of premiums in the reserve. Consistency may be achieved by using the mean reserve method,
the mid-terminal method, or the exact reserve method with appropriate adjustments to the terminal Net
Premium Reserve amount as summarized below.

• Mean Reserve Method: Net Premium Reserve (using annual premium assumption) less deferred
premium asset.
• Mid-Terminal Reserve Method: Net Premium Reserve (using annual premium assumption) plus
unearned premium reserve.
• Exact Reserve Method: Net Premium Reserve (using actual premium mode for the policy) with no
adjustment needed.

Q 4.13: For flexible premium policies, would the actuary still need to calculate the Deferred Premium
Asset?

A: For products with flexible premiums, some actuaries would not calculate a deferred premium asset or
unearned premium liability because the methodology results in a reserve that is consistent with the valuation
date.

Q 4.14: Is the deferred premium asset on a gross or valuation net premium basis?

A: With respect to adjusting entries such as the deferred premium asset, the Net Premium Reserve
methodology is intended to behave similar to the pre-PBR formulaic Standard Valuation Law (SVL)
methodologies. Some actuaries, therefore, would determine the DPA on a net basis and make any other
adjustments consistent with pre-PBR adjustments.

Q 4.15: How would the actuary calculate the deferred premium asset for fixed premium policies in the first
year?

American Academy of Actuaries 19 www.actuary.org

A: For term insurance-based policies, the adjusted gross premium in the first year is zero, so the valuation net
premium and the deferred premium asset in the first year would also be zero. For fixed premium universal life
policies, the valuation net premium is a level percentage of the solved-for gross premium, so some policies
may have a non-zero deferred premium asset.

Q 4.16: Are there any requirements with respect to the timing of benefits (i.e., curtate vs. continuous) in the
Net Premium Reserve?

A: The requirements are written in terms of an annual premium and some actuaries would interpret this as
beginning-of-year timing. With respect to the benefits (death benefits, surrender benefits, etc.), Section 3 of
VM-20 does not explicitly state requirements for the timing of such benefits. However, some actuaries would
assume immediate payment of claims in the VM-20 reserve calculation as that is a requirement of Actuarial
Guideline XXXII.

Q 4.17: How does the actuary apply the floors that are defined in Section 3.D?

A: For a term insurance policy, some actuaries would compare the terminal Net Premium Reserve amounts to
terminal cash value amounts and take the larger of the two. Then they would use these values to interpolate
between policy anniversaries and compare the result to the modal cost of insurance as described in Section
3.D.1.

Other actuaries may calculate the interpolated Net Premium Reserve amount prior to comparing to the
consistently interpolated cash value amount and modal cost of insurance.

Other methods may develop as practice matures.

Q 4.18: Are lapse rates allowed in the NPR calculation?

A: Section 3.C.3 specifies the lapse rates required to be used in the NPR calculation that vary by product type
and by the number of guarantee years.

Q 4.19: Are there specific lapse assumptions required in the NPR calculation for level premium term
products?

A: Section 3.C.3.b of VM-20 specifies the required lapse assumptions that must be used in NPR amounts
calculated according to Section 3.B.4. The rates vary based on the level premium period. If a level premium
period is less than five years, a 10 percent lapse rate is required to be used during this level premium period. If
a level premium paying period is five or more years, then a 6 percent lapse rate is used during this level
premium period.

There are two exceptions to this requirement: the first relates to return of premium term products (i.e., policies
subject to Actuarial Guideline XLV in VM-C). The lapse rates for this situation are 6 percent for the first half
of the initial level premium period and 0 percent for the balance of the initial level premium period. The
second exception applies when the initial level premium period is less than five years. In this case the lapse
rate is 10 percent during any subsequent premium paying period.

American Academy of Actuaries 20 www.actuary.org

The lapse rate for the final year of the level period, applied after any benefits assumed payable in the final
year, and prior to the payment of the increased premium rate (also known as the shock lapse rate) shall be
adjusted from that described above. The adjusted lapse rate varies by the length of the premium period prior
to the increase, the length of the premium period after the increase, and the percent increase in the gross
premium per $1,000 (as specified in the table in Section 3.C.3.b.v). The table is reproduced below.

Lapse for the Final

Length of Premium Percent Increase in
Length of Premium Year of the Level
Period Prior to Gross Premium Per
Period after Increase Premium Period
Increase $1,000
(Shock Lapse)

1<PP≤5 ART ANY 50%

1<PP≤5 1<PP ANY 25%
5<PP≤10 ART < 400% 70%
5<PP≤10 ART Over 400% 80%
5<PP≤10 1<PP≤5 ANY 50%
5<PP≤10 5<PP ANY 25%
10<PP ART < 400% 70%
10<PP ART Over 400% 80%
10<PP 1<PP≤5 ANY 70%
10<PP 5<PP≤10 ANY 50%
10<PP 10<PP ANY 50%

Q 4.20: Are there specific lapse assumptions required in the NPR calculation for universal life policies
with secondary guarantees?

A: Section 3.C.3.c specifies the formula for determining the lapse rates at each duration for policies with
secondary guarantees.

Q 4.21: In the NPR calculation, should assumed lapses occur annually, monthly, or on premium due
dates?

A: VM-20 does not provide any requirement as to the timing of assumed lapses. Some actuaries will assume
lapses occur on a basis consistent with how their cash flow testing is performed. It is common for a monthly
assumption as to lapses to be made in future projections for those actuarial systems that process monthly cash
flows. Other actuaries may assume average lapse rates during the year and process lapses on an annual basis.

Q 4.22: If the actuary expects policies to be limited pay, but they contractually allow for future premiums,
what premium period should be used for the reserve calculation?

A: For universal life policies, Sections 3.B.5 and 3.B.6 require the calculation of an annual premium for the
period over which the premiums are permitted to be paid that would keep the policy in force over the entire
period coverage is to be provided or under the provisions of the longest secondary guarantee.

American Academy of Actuaries 21 www.actuary.org

For term insurance policies, Section 3.B.4 requires the valuation net premium to be based on the maximum
guaranteed gross premium. Two uniform percentages may be required for policies subject to the shock lapse
provisions in Section 3.C.3.b.v.

Q 4.23: How do the Net Premium Reserve interest rates compare to current SVL interest rates for term and
universal life policies?

A: Per Section 3.C.2.a, the valuation interest rate for policies with NPR amounts calculated according to
Section 3.B.5 is the same as the current SVL. For policies with NPR amounts calculated according to Section
3.B.4 or 3.B.6, the valuation interest rate is the same as the current SVL increased by 1.5 percent, but limited
to 125 percent of the current SVL rate. The final rate is rounded to the nearest one-quarter of one percent (1/4
of 1 percent).

Q 4.24: How is the NPR calculated on an IUL policy?

A: The NPR requirements in VM-20 Section 3 do not distinguish between indexed and non-indexed UL
policies. Like UL, the NPR requirements differ depending on whether or not the IUL policy has a secondary
guarantee as defined in VM-01.

If the IUL policy has a secondary guarantee as defined in VM-01, then the policy meets the definition of a
ULSG policy that is subject to VM-20 sections 3.B.5 and 3.B.6, which together define the NPR methodology
for ULSG policies whether indexed or not. When calculating the NPR on IUL policies, practice varies as to
the guaranteed minimum credited rate used in the Section 3.B.5 part of the calculation. For example, some
actuaries may use the fixed account minimum credited rate, some may use the floor for indexed account
crediting, others may use methods consistent with CRVM such as the implied guarantee rate method from
AG36.

If the IUL policy has a conditional guarantee that does not meet the definition of a secondary guarantee as
defined in VM-01 or has no secondary guarantee at all, then it is subject to the requirements of VM-A and
VM-C and per VM-C must follow AG36.

Q 4.25: What equity return rates should be used in calculating Net Premium Reserve interest for separate
account products?

A: VM-20 does not give clear direction on the assumption for separate account returns for NPR. Some
actuaries may assume that the equity return assumptions under the NPR method would be developed in a
manner similar to how it is developed under the CRVM approach for variable life policies.

Q 4.26: Are the assumptions in the NPR calculation (specifically interest and mortality) locked in by year
of issue? Does a company’s use of the three-year transition period matter in this regard?

A: With respect to valuation mortality, a Guidance Note in Section 3.C.1 (at the end of this section) suggests
that adoption of new valuation tables (e.g. new “CSO” tables) after the operative date of the Valuation
Manual would apply to all business issued since the adoption of the Valuation Manual. This Guidance Note
states “The details of how to implement any unlocking of mortality tables will need to be addressed in the
future.”

American Academy of Actuaries 22 www.actuary.org

With respect to valuation interest, Section 3.C.2 includes a Guidance Note at the beginning of the section that
makes clear the valuation interest rate assigned to a policy at issue date is not unlocked at future valuation
dates.

Q 4.27: If a product has anticipated mortality higher than the CSO valuation table, should the higher
mortality be used for the NPR calculation? Additionally, if the Deterministic Reserve is calculated for this
same product, and presumably the higher anticipated mortality serves as the basis for the prudent estimate
assumption, and if the NPR does not include the extra mortality expectation, how should the company
adjust the NPR so that it is consistent with the DR?

A: The actuary might wish to consider any differences in product design in assessing an approach for this
situation. In particular, the practice that develops for term insurance may be different than that which develops
for ULSG. Requirements of Section 3.C.1 provide for use of the 2017 CSO mortality table in its various
forms (smoker-distinct, preferred structure, etc.). This section does not include any specific requirements or
adjustments for substandard risks. Some actuaries may choose to value the reserve for the additional mortality
using a method such as one-half the substandard extra gross premium amount for the policy, and consider this
a part of the aggregate NPR reserve, while also recognizing the additional mortality in the Deterministic
Reserve component. Other actuaries may choose to assign no value to the additional mortality within the
aggregate NPR, but only to recognize the increased mortality in the Deterministic Reserve component.

Q 4.28: VM-20 Section 3.B.6.d talks about the funding ratio in terms of the components ASGx+t and
FFSGx+t. How are these components defined in the context of a shadow account product or in the context
of a specified premium product?

A: Refer back to the definitions found in Sections 3.B.1 and 3.B.2 in evaluating how to calculate ASGx+t and
FFSGx+t components of the ULSG NPR 3.B.6 reserve for these different secondary guarantee designs.

5: VM-20 Calculation Overview—Part B. Deterministic Reserve

Q 5.1: How would actuaries approach the calculation of the Deterministic Reserve (DR) for each model
segment as required by VM-20?

A: Unlike the NPR, which is a seriatim reserve based on specific formulas using prescribed assumptions, the
DR is an aggregate reserve based on general cash flow projections using company-specific assumptions. As
such, instead of prescribing the exact steps and formulas to calculate the DR, VM-20 Section 4 on the DR
generally describes two methods actuaries may use to calculate the DR using cashflows projected from the
company’s integrated asset-liability models, i.e. the Gross Premium Valuation method and the Direct Iteration
Method (see Q 5.7). Both methods have iterative elements to them. Unlike the NPR, the specific steps taken
to calculate the DR are largely left to actuarial judgement. Some actuaries would begin either method with
similar initial steps, followed by different steps depending on the method chosen, for example:

Initial steps:
a. Determine margins and prudent estimate assumptions for all risk factors as described in Section 9
of VM-20.
b. Determine the makeup of starting assets as described in Section 7.D of VM-20.

American Academy of Actuaries 23 www.actuary.org

 c. Select the level of starting assets. For the first iteration, some actuaries may use a simple estimate
based on the current period’s NPR or an approximate roll forward of last period’s DR.
d. Run the integrated asset-liability cash flow model to project asset and liability cash flows under
the prescribed economic scenario.

Gross Premium Valuation method:

e. Determine the path of net asset earned rates as described in Section 7.H of VM-20.
f. Calculate a gross premium reserve by discounting projected benefits and expenses less premiums
using the path of net asset earned rates, as described in Section 4.A of VM-20.
g. Calculate DR as the gross premium reserve, less the positive or negative pretax interest
maintenance reserve (PIMR) balance at the valuation date allocated to the model segment as
described in Section 7.D.6 of VM-20, plus the balance of separate account assets on the valuation
date, and plus the policy loan balance at the valuation date if policy loans are modeled explicitly
under Section 7.F.3.b of VM-20.
h. Iterate starting with step c. until the level of starting assets for all model segments combined is
within the asset collar described in Section 7.D.3 of VM-20, i.e. is greater than 98% of the
modeled reserve and less than the greater of the NPR and 102% of the modeled reserve, where
modeled reserve is defined in VM-20 Section 1.B.8 as the DR on policies determined under
sections 2.A.1.a, 2.A.2.a, and 2.A.3.b plus the greater of the DR and the SR on policies
determined under sections 2.A.1.b, 2.A.2.b, and 2.A.3.c. Although VM-20 only requires the asset
collar be satisfied once for all model segments combined, to ease implementation some actuaries
may choose to satisfy the asset collar for each model segment individually subject to verifying that
this more restrictive implementation of the asset collar doesn’t have a material impact on
aggregate results. Alternatively, VM-20 allows a company to stop iterating before the asset collar
is satisfied as long as the company documents and provides reasonable assurance in the PBR
Actuarial Report that the modeled reserve is not materially understated as a result of having
starting assets outside the collar.

Direct Iteration Method:

i. Iterate starting with step c. until the level of starting assets is sufficient to fully liquidate all policy
obligations, as described in Section 4.B of VM-20.
j. Calculate DR as the level of starting assets less the positive or negative PIMR balance at the
valuation date allocated to the model segment as described in Section 7.D.6 of VM-20.
k. If starting assets for all model segments combined fall outside the asset collar described in Section
7.D.3 of VM-20, the actuary will need to document and provide reasonable assurance in the PBR
Actuarial Report that the modeled reserve is not materially understated as a result of having
starting assets outside the collar. Some actuaries may include documentation based on step i.
above showing the level of starting assets is sufficient to fully liquidate all policy obligations.

Some actuaries may use modeling systems that automate the iterative nature of either method.

Some actuaries may find they need to justify and perform additional steps if asset and liability cash flow
models are not integrated.

Q 5.2: What economic scenario assumptions are used to determine the Deterministic Reserve?

A: Per Section 7.G.1, the Deterministic Reserve assumes the U.S. Treasury interest rate curves for Scenario
12 from the set of prescribed scenarios used in the stochastic exclusion ratio test (defined in Section 6.A.2 of
American Academy of Actuaries 24 www.actuary.org
VM-20). The total investment return paths for general account equity assets and separate account fund
performance also use those investment returns for corresponding investment categories contained in Scenario
12 from the set of prescribed scenarios used in the stochastic exclusion ratio test defined in Section 6.A.2.

Q 5.3: How does the actuary calculate the path of net asset earned rates?

A: Companies that use the Gross Premium Valuation method to calculate the Deterministic Reserve must
discount projected cash flows using the path of Net Asset Earned Rates (NAER) as described in VM-20
Section 7.H:

Determine a path of projected portfolio NAER for each projection interval used in the model (monthly,
quarterly, annual, etc.). The NAER for each projection interval is calculated in a manner that is consistent
with the timing of cash flows and length of the projection interval of the related cash flow model.

The NAER for each projection interval will equal Net Investment Earnings divided by Invested Assets.

Net Investment Earnings include:

• Gross investment income (including any amortization of premium or accrual of discount) less
prescribed default costs and less investment expenses
• Capital gains and losses
• Income from derivative asset programs

Invested Assets:
• Are determined in a manner that is consistent with the timing of cash flows within the cash
flow model
• Include the annual statement value of derivative instruments or a reasonable approximation
thereof

Thus, the NAER will depend on:

• Projected Net Investment Earnings from the portfolio of starting assets
• Projected Net Investment Earnings from reinvestment assets
• The pattern of projected net liability cash flows (premiums less benefits and expenses)
• The pattern of projected asset cash flows from starting assets and reinvestment assets

However, per Section 7.H.2.a, separate account income and assets and interest (investment) income on policy
loans and the policy loan asset are not included in the NAER calculation.

VM-20 permits the use of approximations and simplified approaches to determine appropriate NAER if the
approach used is consistent with the requirements of Section 2.G. Considerations include differences in the
way invested assets are defined for NAER, the timing of investment income within each projection interval,
and the treatment of non-cash items. See Q 5.7 for additional information regarding NAER within each of the
two methods for calculating the DR.

Q 5.4: How will actuaries model assets in the deterministic and stochastic calculations?

A: See Section 11 of this practice note for questions and answers about modeling of assets and the starting
asset amount.

American Academy of Actuaries 25 www.actuary.org

Q 5.5: How are separate account cash flows and balances reflected in the Deterministic Reserve?

A: VM-20 Section 4.A on the Gross Premium Valuation method says to include the balance of separate
account assets on the valuation date when calculating the DR using the GPV method; that is:
o Actuarial present value of benefits, expenses, and related amounts
o Less actuarial present value of premiums and related amounts
o Less the applicable positive or negative PIMR balance at the valuation date
o Plus the balance of separate account assets on the valuation date
o Plus the applicable policy loan balance at the valuation date (if policy loans are explicitly
modeled)
o Equals Deterministic Reserve (using GPV method)

VM-20 Section 4.A.4.b addresses separate account cash flows, stating that the present value of future net cash
flows from the separate account to the general account is to be included in the actuarial present value of
premiums and related amounts when calculating the DR using the GPV method.

VM-20 Section 4.B.3 addresses separate account balances and changes in those balances when calculating the
DR using the Direct Iteration Method, stating that the balance of separate account assets on the valuation date
is modeled each period in compliance with the applicable changes to those asset balances as defined in VM-
20 Section 7.

Separate account cash flows include, but are not limited to, deposits of policyholder premiums to the separate
account, transfers between fixed and variable investment options, transfers of separate account values to pay
death or withdrawal benefits, and amounts charged to separate account values for cost of insurance, expense,
or other similar amounts.

Q 5.6: How is reinsurance incorporated into the Deterministic Reserve?

A: Section 8.C, titled “Reflection of Reinsurance Cash Flows in the Deterministic Reserve or Stochastic
Reserve,” outlines all the considerations for including reinsurance cash flows in the modeled reserves.

Q 5.7: VM-20 allows two different methods for the calculation of the DR. What are the two methods, and
how are they different?

A: The two methods allowed for the calculation of the DR are known as the Gross Premium Valuation (GPV)
method described in Section 4.A, and the Direct Iteration Method (DIM) described in Section 4.B. In theory,
both methods produce exactly the same DR amount. Both methods serve to assure adequacy by identifying
theoretically equivalent levels of starting assets that support and mature policy obligations under prudent
estimate assumptions and a moderately adverse economic scenario. The two methods can be summarized as
follows:

• The GPV method identifies the level of starting assets by projecting liability cash flows, then
discounting them at net asset earned rates (NAER) derived from an integrated asset-liability model,
iterating as necessary until the discounted liability cash flows are within 2 percent of the starting
assets.
• The DIM identifies the level of starting assets by projecting asset and liability cash flows to the end of
the projection period, iterating as necessary until the starting assets are sufficient to fully liquidate all
policy obligations, leaving zero assets at the end of the projection period.

American Academy of Actuaries 26 www.actuary.org

The primary difference between the two methods is that the GPV method has the separate NAER derivation
and discounting step, whereas the DIM makes direct use of asset cash flows already being produced by the
asset-liability model to arrive at the same answer.

Calculating NAER (net investment earnings divided by invested assets) subject to VM-20 requirements can
be complicated and may require modifications and/or approximations. Considerations include differences in
the way invested assets are defined for NAER, the timing of investment income within each projection
interval, and the treatment of non-cash items. Although theoretically equivalent, consideration should be
given to the cumulative impact of approximations over long periods, which can lead to differences between
the two methods in practice.

Q 5.8: Is the decision to select the method to calculate the DR completely at the company’s discretion, or do
certain conditions have to be met?

A: No conditions need to be met when deciding which method to use to calculate the DR. The choice is made
at the discretion of the company.

Q 5.9: Can the company move back and forth between the two methods in subsequent valuation dates?

A: There is no provision prohibiting the movement back and forth between methods in subsequent valuation
dates. However, switching methods would likely result in additional model governance, auditing, and
documentation considerations.

Q 5.10: What are the primary advantages of the Direct Iteration Method (DIM) compared to the Gross
Premium Valuation (GPV) method?

A: The primary advantage of the DIM is that eliminates the need for a separate NAER derivation and
discounting step. Calculating NAER can be complicated and may require modifications and/or
approximations related to the timing of asset cash flows, policy loan cash flows, and the treatment of non-cash
accrual items. Consideration should be given to the cumulative impact of approximations over long time
periods.

Another advantage of the DIM is that it automatically provides proof of reserve adequacy, because starting
assets by definition are adequate to fund all future cash flow obligations as evidenced by the ending asset
value. Also, although solve routines can be programmed to automate the iterative aspects of both methods,
many asset-liability models will already have routines to solve for the starting assets to fully liquidate all
policy obligations, leaving zero assets at the end of the projection period. Note that both methods may require
modifications to the asset-liability model related to negative assets in order to converge.

Q 5.11: Are there any disadvantages of the Direct Iteration Method (DIM) compared to the Gross Premium
Valuation (GPV) method?

A: The primary disadvantage of the DIM is that most actuaries are very comfortable with the concept and
practice of discounting future cash flows. GPV calculations have been done for years for a variety of
applications, including U.S. statutory regulations. Some actuaries may be less familiar with the DIM of
solving for starting assets that accumulate to zero over time. However, the DIM is not entirely new and has
been used by actuaries in certain applications and regulatory frameworks (e.g., the Canadian Asset Liability
Method (CALM)).
American Academy of Actuaries 27 www.actuary.org
Some actuaries may find the GPV method better suited for certain situations where the actuary is limited to a
liability-only model and can justify a method for obtaining NAER from a separate asset only model (e.g.,
from the investments area).

6: VM-20 Calculation Overview—Part C. Stochastic Reserve

Q 6.1: How is the Stochastic Reserve determined as required by VM-20?

A: Section 5 of VM-20 contains the requirements for determining the Stochastic Reserve (all section
references below are to VM-20).

• Determine the number of product groups for which the Stochastic Reserve is to be calculated. The
definition of product groups is found in Section 2 and includes: Term, ULSG, and policies subject to
Section 3.A.2. It is important to note that Section 5.A specifies that the company may determine the
number and composition of policy subgroups for aggregation purposes in a manner consistent with
how the company manages risk across the different product types. However, if an aggregation
subgroup includes policies from more than one product group, then the Stochastic Reserve for
policies from each product group must be calculated on a stand-alone basis by following the steps in
Section 5.A through Section 5.F. With that in mind, for each product group, the following steps apply.
• Determine margins and Prudent Estimate Assumptions for all Risk Factors as described in Section 9.
• Generate the stochastic economic scenarios as described in Section 7.G.2.
• Determine the number of stochastic economic scenarios to use in the calculation as described in
Section 7.G.2.c, d, and e.
• For each product group, calculate the Stochastic Reserve as follows:
o Project cash flows for each model segment under each economic scenario as described in
Section 5.
o Determine the Scenario Reserve for each scenario as described in Section 5.B.3.
o Calculate the Stochastic Reserve by ranking the Scenario Reserves from lowest to highest and
taking the average of the highest 30 percent (CTE 70) of these Scenario Reserves.
• If necessary, add amounts to the Stochastic Reserve to capture any material risk included in the scope
of VM-20 but not already reflected in the Stochastic Reserve, per Section 5.E.

Q 6.2: How is the Scenario Reserve calculated?

A: Section 5.B describes how the Scenario Reserve is calculated. The Scenario Reserve may include more
than one model segment. A model segment is defined in Section 1.C.7 of VM-20 and discussed in Section
7.A.1.b. It is a group of policies and associated assets that are modeled together. Model segments should be
consistent with the company’s asset segmentation plan, investment strategies, or approaches used to allocate
investment income for statutory purposes. If a company is managing the risks of two or more different
product types as part of an integrated risk management process, then the products may be combined into the
same subgroup (an aggregation subgroup). The Guidance Note in Section 5.A states that aggregating policies
into a common subgroup allows the cash flows arising from the policies for a given stochastic scenario to be
netted against each other. Section 5.A also requires that if policies from more than one product group are
included in an aggregation subgroup, the Stochastic Reserve for each product group shall also be determined,

American Academy of Actuaries 28 www.actuary.org

as described in Section 5.G. Section 5.G. requires that the Stochastic Reserve be calculated on a stand-alone
basis for each product group.

A Scenario Reserve is the negative of the present value of accumulated net cash flows at the beginning or end
of the year when the present value of the accumulated deficiency is the most negative during the projection
period. This method is often called the Greatest Present Value of Accumulated Deficiency (GPVAD). In this
method, future sufficiencies beyond the minimum point are not taken into account in the calculation.

The reserve for each scenario is determined as follows:

• At the valuation date, and at the end of each projection year, calculate the negative of the projected
statement value of assets (may be positive or negative) for all model segments. The negative of the
projected statement value of assets is called the negative accumulated deficiency.
• Discount the negative accumulated deficiencies at the end of each projection year to the valuation
date.
• Discount rate = 105% of 1-year Treasury rate.
• Sum the discounted negative accumulated deficiencies across all model segments.
• Scenario Reserve = the largest of discounted values plus the starting asset amount.

Below is a simplified hypothetical numerical example showing the calculation of the Scenario Reserve across
all model segments:

Hypothetical Example for Scenario Reserve Calculation

Scenario #: 123 of 10,000

Product: 5 Year Term Insurance
Projection Period: 5 Years (assumption of 100% lapse at end of year 5)
Starting Assets: $1,000

Projection Year (t) 0 1 2 3 4 5

[1] Statement Value of Assets $ 1,000 $ 500 $ (100) $ (50) $ (105) $ 500
[2] Negative of the Statement Value of Assets = -1 * [1] $ (1,000) $ (500) $ 100 $ 50 $ 105 $ (500)
[3] One-Year Treasury Rate (beginning of year) 0.200% 1.000% 3.000% 3.000% 2.000%
[4] 105% of One-Year Treasury Rate = 1.05 x (3) 0.210% 1.050% 3.150% 3.150% 2.100%
[5] Discount Factor (t) = Discount Factor (t-1) / (1+[4]@t-1) 1.000 0.998 0.988 0.957 0.928 0.909
[6] Discounted Negative Accumulated Deficiencies = [2] x [5] $ (1,000) $ (499) $ 99 $ 48 $ 97 $ (455)
[7] Greatest Present Value of Accumulated Deficiency (GPVAD) = Max([6]) $ 99
[8] Scenario Reserve = Starting Assets + [7] $ 1,099

Q 6.3: Can Scenario Reserves be negative, and should such negatives be reflected in a CTE 70 calculation?

A: Section 5.B of VM-20 states that a scenario reserve equals starting assets plus the greatest present value of
accumulated deficiencies (GPVAD) for the scenario. Because a scenario’s GPVAD can be negative, the
scenario reserve can also be negative if the level of starting assets is sufficiently small. Negative scenario
reserves will only be reflected in a CTE 70 calculation to the extent more than 70 percent of the scenario
reserves are negative.

Q 6.4: What is the appropriate level of aggregation when determining the Stochastic Reserve?

American Academy of Actuaries 29 www.actuary.org

A: Section 5.G requires that, for a company that includes policies from two or more product groups (as
defined in VM-20 Section 2) in an aggregation subgroup, the Stochastic Reserve for each such product group
must be determined on a stand-alone basis. Therefore, the highest level of aggregation allowed is at this
Section 2 product group level.

Aggregation in Section 5.A refers to the number of aggregation subgroups of policies used to combine cash
flows when calculating the Stochastic Reserve for the purpose of limiting / allowing the amount of risk
diversification between policies within a VM-20 Section 2 product group. Aggregating policies into an
aggregation subgroup allows the cash flows arising from the policies for a given stochastic scenario to be
netted against each other (i.e., allows risk offsets between policies to be recognized).

An actuary may decide to group policies into more than one aggregation subgroup. VM-20 requires that the
level of aggregation must be consistent with how the company manages risks across the different product
types and must reflect the likelihood of any change in risk offsets that could arise from shifts between product
types. For example, if a company is managing the risks of two or more different term insurance periods as
part of an integrated risk management process, then these forms may be combined into the same aggregation
subgroup. The level of aggregation within a Section 2 product group is up to actuarial judgment, consistent
with the risk practices of the company.

Q 6.5: Will term insurance products with different level premium periods be projected in an aggregated
basis or separately in the Stochastic Reserve?

A: Some actuaries may model all of the level term products in one model in aggregate because the products
are managed together and are believed to have similar risks. Other actuaries may combine some level
premium periods but model others separately (e.g., 30-year level premium period) because they believe that
the product risks are different. Also, some actuaries may model each different level premium period
separately because the risk for each level premium period is managed separately or they believe separate
models are more appropriate.

Q 6.6: Are the prudent estimate assumptions the same for the Stochastic Reserve as for the Deterministic
Reserve?

A: Per Section 9.A.5, prudent estimate assumptions for the Stochastic and Deterministic Reserves must be
consistent, with an exception for assumptions that are dependent (i.e., dynamically linked) on the economic
scenario, such as lapse rates that move up or down with changes in interest rates.

Q 6.7: Does the calculation of the Stochastic Reserve include the use of a “working reserve” similar to the
Actuarial Guideline XLIII?

A: VM-20 does not mention “working reserves.” This is equivalent to a working reserve equal to zero.

Q 6.8: How are the discount rates determined?

A: Per Section 7.H.4, the discount rates for the Stochastic Reserve are set equal to the path of one-year
Treasury rates at the beginning of each projection year for the scenario, multiplied by 1.05. The Guidance
Note in this section has additional information on why this discount rate was chosen and why it is different
from the one used for the Deterministic Reserve.
American Academy of Actuaries 30 www.actuary.org
Q 6.9: How does the actuary determine the amount (if any) to add to the Stochastic Reserve to capture any
material risk not already reflected in the Stochastic Reserve per Section 5.E of VM-20?

A: VM-20 is not specific on how this adjustment will be determined. Some actuaries may make an adjustment
for material risks where it is known that their Stochastic Reserve models do not cover specific policy risks.
For example, if there is a certain rider that would pay an additional benefit if some contingent event happened
and that event is not being modeled, then the actuary may include an adjustment for the payment of this
benefit.

7. Stochastic Exclusion Test

Q 7.1: What is the Stochastic Exclusion Test?

A: As described in Section 6.A of VM-20, the Stochastic Exclusion Test can be used to identify groups of
policies that do not have material interest rate or asset return volatility risk. Such groups of policies would
likely have minimal variation in financial outcomes over varying economic conditions. Companies may elect
to use this test to exclude groups of policies from the calculation of the Stochastic Reserve.

Q 7.2: What methods are available to pass the Stochastic Exclusion Test?

A: The actuary may use any of three approaches to pass the Stochastic Exclusion Test. One approach, the
Stochastic Exclusion Ratio Test (SERT), as described in Section 6.A.2, is to run 16 specified scenarios that
are used in a ratio to demonstrate minimal variation by economic scenario in the present value of cash flows.
Alternatively, as described in Section 6.A.3, the actuary can demonstrate that the Stochastic Reserve would
not increase the minimum reserve required for the group of policies. The third approach, which can be used
only for policies that are not variable life or not universal life with secondary guarantees, is for a qualified
actuary to certify and report that the policies are not subject to material interest rate risk or tail risk, or asset
risk.

Q 7.3: What products might be good candidates for the Stochastic Exclusion Test? What products may not
be excluded? What products are only allowed to pass the test under specific testing methods?

A: Life insurance products where the main risks are not highly dependent on interest rate movements or
equity performance, or where these types of risks can be shared or passed on to the policyholder, are strong
candidates for the Stochastic Exclusion Test. Some actuaries may use the Stochastic Exclusion Test for
products such as term life insurance, which focus primarily on mortality risk, or for some variations of
traditional permanent life insurance and accumulation-oriented universal life insurance where non-guaranteed
elements help transfer the asset performance risk to the policyholder. Variable life or universal life insurance
products that contain secondary guarantees may not use the certification method to pass the Stochastic
Exclusion Test. These types of policies may only be excluded through the use of the Stochastic Exclusion
Ratio Test under Section 6.A.2 or the Stochastic Exclusion Demonstration Test under Section 6.A.3.

Q 7.4: Can a group of policies with a clearly defined hedging program be excluded from calculating the
Stochastic Reserve?

American Academy of Actuaries 31 www.actuary.org

A: No. Section 6.A.1.b states that a company may not exclude a group of policies for which there is one or
more clearly defined hedging strategies from the Stochastic Reserve requirements.

Q 7.5: Is it necessary to perform the Stochastic Exclusion Test on all blocks of life insurance?

A: No. Section 2.A states if a company elects to calculate Stochastic Reserves for one or more groups of
policies, the Stochastic Exclusion Test is optional for those groups of policies. In other words, within a given
product group, the test may be performed for all groups of policies, for selected groups of policies, or for no
groups of policies.

Q 7.6: How does a company determine whether a group of policies passes the Stochastic Exclusion Ratio
Test?

A: The Stochastic Exclusion Ratio Test is one method to pass the Stochastic Exclusion Test, thereby
excluding a group of policies from calculation of Stochastic Reserves. The company calculates a ratio that
evaluates the sensitivity of the reserve to changes in the economic scenario.

The numerator of the ratio is equal to the difference between two items: the Adjusted Deterministic Reserve
calculated under the baseline economic scenario, and the largest Adjusted Deterministic Reserve calculated
under any of the 15 other economic scenarios. The method for creating the economic scenarios is found in
Appendix 1 of VM-20, and these can be produced using the prescribed Economic Scenario Generator. The
Adjusted Deterministic Reserve is calculated following one of two possible methods defined in Section
6.A.2.b.

The first method is to calculate the Adjusted Deterministic Reserve similarly to the Deterministic Reserve in
Section 4, but with adjustments specified in Section 6.A.2.b.

The second method is to calculate a gross premium reserve from the cash flows from the company’s Asset
Adequacy Analysis models. This gross premium reserve would use the experience assumptions of the
company’s cash flow analysis models except that:

• the interest rate and equity return assumptions used are specific to each of the 16 exclusion test
scenarios; and

• the rates used to discount the cash flows follow the methodology defined in Section 7.H but use the
company’s cash flow testing assumptions for default costs and reinvestment earnings.

The denominator of the Stochastic Exclusion Ratio is the present value of benefits for the group of policies, as
determined in the baseline economic scenario, with an adjustment to reflect reinsurance by subtracting out
ceded benefits.

If the ratio of the numerator to the denominator is less than 6.0 percent, it implies that the reserve calculation
is relatively insensitive to variation in economic scenarios, and the group of policies passes the Stochastic
Exclusion Ratio Test.

Q 7.7: What are the differences between the Deterministic Reserve and the reserve in the numerator of the
Stochastic Exclusion Ratio Test?

American Academy of Actuaries 32 www.actuary.org

A: The reserve in the numerator of the Stochastic Exclusion Ratio is called the “Adjusted Deterministic
Reserve.” The calculation of the Adjusted Deterministic Reserve uses assumptions that do not include
margins.

Net asset earned rates used across the 16 scenarios in calculating the Adjusted Deterministic Reserves are
specific to each scenario in order to discount cash flows, whereas the discounting assumption in the
Deterministic Reserve would be based on one specified scenario (the Deterministic Scenario).

As in the calculation of the Deterministic Reserve, the Adjusted Deterministic Reserve should still use
dynamic adjustments for experience assumptions that depend on the economic scenario.

Q 7.8: What is the difference between the two methods of calculating an Adjusted Deterministic Reserve?

The Adjusted Deterministic Reserve method that follows Section 4.A uses liability assumptions equal to the
anticipated experience assumptions defined in VM-20. These assumptions do not include margins, so the
company does not need to add a provision for margins as required for the prudent estimate assumptions.
However, the other requirements of VM-20 regarding the use of prescribed assumptions and development of
anticipated experience assumptions must be met. For example, the prescribed asset spreads and defaults must
be used.

On the other hand, the gross premium reserve approach uses assumptions that are used by the company in its
asset adequacy analysis. The development of these assumptions does not need to follow the VM-20
requirements, but instead uses the same assumptions determined by the appointed actuary used for asset
adequacy analysis. For example, the asset spread and default assumptions from the asset adequacy analysis
may be used instead of the asset spreads and defaults prescribed by VM-20.

Another possible difference between the two approaches is that if the appointed actuary has added margins in
the asset adequacy analysis assumptions, the gross premium reserve used in the SERT calculation includes a
margin for conservatism. Whether a margin is included depends on the asset adequacy assumptions.

Q 7.9: Can the actuary include mortality improvement beyond the valuation date when performing the
Stochastic Exclusion Ratio Test?

A: No—per Section 6.A.2.b.v, mortality improvement past the projection start date is not allowed in the
anticipated experience assumptions used in the Stochastic Exclusion Ratio Test.

Q 7.10: What amounts should be included and excluded from the present value of benefits in the
denominator of the Stochastic Exclusion Ratio?

A: The present value of the benefits used in the denominator of the Stochastic Exclusion Ratio specifically
should include benefits such as death benefits, surrenders, withdrawal benefits, and policyholder dividends.
These benefits should be net of reinsurance benefits when the Stochastic Exclusion Ratio Test (SERT) is
calculated on a net-of-reinsurance basis. Section 6.A.2.a.iii states that premium, ceded premium, expenses,
reinsurance expense allowances, modified coinsurance reserve adjustments, and reinsurance experience
refund cash flows may not be included in the calculation of the benefit amount.

Q 7.11: Is the SERT calculation net of reinsurance?

American Academy of Actuaries 33 www.actuary.org

A: Yes, when applying the SERT for the purpose of calculating the Stochastic Reserve component of the
minimum reserve. No, when applying the SERT for the purpose of calculating the Stochastic Reserve
component for the pre-reinsurance reserve required by Section 8.D.2.

Q 7.12: Does the company need to pass the Stochastic Exclusion Ratio Test on a pre-reinsurance basis in
order to exclude policies from the pre-reinsurance Stochastic Reserve calculation, even if the policies have
passed the ratio test on a post-reinsurance basis?

A: Yes. Section 8.D.2 states that the ratio test must be passed on a pre-reinsurance basis in order to exclude
the policies in the pre-reinsurance Stochastic Reserve calculation, even if the policies pass the ratio test on the
post-reinsurance basis.

Q 7.13: What happens when the policies pass the ratio test on a pre-reinsurance basis, but not the post-
reinsurance basis (or vice versa)?

A: The ratio test must be passed separately for both the pre-reinsurance reserve calculation and the post-
reinsurance calculation in order for the policies to be excluded from the respective Stochastic Reserve
calculations.

Q 7.14: If a block of business fails the SERT post-YRT reinsurance, but passes it pre-YRT reinsurance, is
there a demonstration that will satisfy the test and allow the use of the exclusion?

A: Section 6.A.2.c allows for a group of policies to still pass the SERT if the company can demonstrate that
the sensitivity of the adjusted Deterministic Reserve to economic scenarios is comparable between pre-YRT
and post-YRT reinsurance. One such example of a demonstration as given in Section 6.A.2.c.i is to calculate
the Largest Percent Increase in Reserve (LPIR), both pre- and post-YRT reinsurance.

Given the SERT is the ratio (b-a)/c as defined in Section 6.A.2.a, the LPIR is defined as (b-a)/a. Letting the
subscript “gy” denote “gross of YRT” and “ny” denote “net of YRT,” if SERTgy × LPIRny/LPIRgy < 0.06, then
the block of policies passes the Stochastic Exclusion Ratio Test.

Q 7.15: How is YRT reinsurance addressed in the Stochastic Exclusion Ratio Test?

A: For the pre-reinsurance Stochastic Reserve calculation (which is required by Section 8.D.2), the impact of
YRT reinsurance is ignored when performing the SERT. For the minimum reserve calculation (i.e., the post-
reinsurance reserve), the total YRT reinsurance cash flows should be included in the numerator, and the
denominator should reflect the YRT reinsurance benefit cash flows. The latter includes claim settlements and
dividends, but not reinsurance expense allowances, mod-co reserve adjustments, and experience refunds (per
Section 6.A.2.a.iii).

Q 7.16: Is coinsurance treated differently from YRT reinsurance when performing the SERT?

A: No. The applicable reinsurance cash flows for each reinsurance agreement are to be reflected in both the
numerator and denominator of the SERT regardless of the type of reinsurance agreement (YRT, coinsurance,
stop-loss, etc.).

Q 7.17: How does the company create the 16 economic scenarios used for the Stochastic Exclusion Ratio
Test?

American Academy of Actuaries 34 www.actuary.org

A: The methodology for creating the 16 economic scenarios is defined in Appendix 1 of VM-20.

Q 7.18: Are there other requirements the actuary should understand when determining the minimum
reserve requirement for a group of policies passing the Stochastic Exclusion Test?

A: If a group of policies passes the Stochastic Exclusion Test, then Stochastic Reserve calculations are not
required in determining the minimum reserve.

If the Stochastic Exclusion Test is relied upon to avoid calculating the Stochastic Reserve, then, according to
Section 4.A.5, future transactions associated with non-hedging derivative programs may not be included in
calculating the Deterministic Reserve. If the Stochastic Exclusion Ratio Test is used to pass the Stochastic
Exclusion Test, the demonstration must be done annually and within 12 months before the valuation date.

Some actuaries may still calculate the Stochastic Reserve and use it in the determination of the minimum
reserve even if the block of business passes the Stochastic Exclusion Test.

Q 7.19: What happens if a group of policies fails the Stochastic Exclusion Test?

A: If a group of policies fails the test, both the Stochastic Reserve and Deterministic Reserve must be
calculated for that group of policies. Section 2.A specifies the definition of the minimum reserve in this
situation.

Q 7.20: How does a company demonstrate that a group of policies passes the Stochastic Exclusion Test
because it is not subject to material risks? What methods are acceptable? What information is needed?
What demonstrations and reports need to be prepared?

A: Section 6.A.3 outlines several methods that may be used to demonstrate that the group of policies is not
subject to material risk.

• The company may opt to show that the greater of A and B is greater than the Stochastic Reserve,
where:

o A equals the Deterministic Reserve, and

o B equals the Net Premium Reserve less any associated due and deferred premium asset.

• The company may show that the greater of A and B is greater than a Scenario Reserve in each of a
sufficient number of adverse deterministic scenarios.

• The company may show that the greater of A and B is greater than the Stochastic Reserve calculated
on a stand-alone basis, but using a smaller representative subset of the policies in the stochastic
calculation, rather than the entire group of policies.

• The company may show that the main risk characteristics that would drive the Stochastic Reserve to
exceed the larger of A and B have been substantially mitigated through various company actions.
These actions may include hedging, risk reducing investment strategies, reinsurance, or transferring of
the risk to the policyholder through contractual provisions.

The company may also use another method that is acceptable to the commissioner.

American Academy of Actuaries 35 www.actuary.org

If any of these methods are used, the company must provide a demonstration in the PBR Actuarial Report.
The demonstration must provide reasonable assurance that if the Stochastic Reserve were calculated, the
minimum reserve would not increase.

The report must be prepared in the first year that the demonstration is performed, and at least once every three
years thereafter.

Q 7.21: How would a qualified actuary certify that a group of policies may be excluded from the Stochastic
Reserve calculation?

A: Per the Guidance Note in Section 6.A.1.a.iii, “the qualified actuary should develop documentation to
support the actuarial certification that presents the analysis clearly and in detail sufficient for another actuary
to understand the analysis and reasons for the actuary’s conclusion that the group of policies is not subject to
material interest rate risk or asset return volatility risk. Examples of methods a qualified actuary could use to
support the actuarial certification include, but are not limited to:

• A demonstration that reserves for the group of policies calculated according to Section 5 — Section 9
of VM-05, VM-A, and VM-C are at least as great as the assets required to support the group of
policies using the company’s cash-flow testing model under each of the 16 scenarios identified in
Section 6 or alternatively each of the New York seven scenarios.

• A demonstration that the group of policies passed the stochastic exclusion ratio test within 36 months
prior to the valuation date and the company has not had a material change in its interest rate risk.

• A qualitative risk assessment of the group of policies that concludes that the group of policies does not
have material interest rate risk or asset return volatility. Such assessment would include an analysis of
product guarantees, the company’s NGE policy, assets backing the group of policies and the
company’s investment strategy.”

Q 7.22: What analysis would be done on a regular basis for groups of policies that satisfy the Stochastic
Exclusion Test?

A: Some actuaries would review the characteristics of each group of policies on a regular basis to make sure
that the group of policies still passes the test, particularly if the block only passed the test by a small margin.
Other actuaries may identify certain economic parameters or other characteristics of the block of policies that
drive variability to economic conditions and would monitor those items to see whether the calculations need
to be recomputed. An example of this may be a block of universal life policies without secondary guarantees,
with cash values that are a very low percentage of the death benefit. It may be that this block is not currently
sensitive to economic conditions, but the sensitivity of that block could increase to the point where it no
longer passes the test if the cash values increase.

Other actuaries may only perform the test when it is required under VM-20 if they believe the block has not
changed enough to make retesting worthwhile.

Q 7.23: If a group of policies becomes subject to material risks after having previously been certified, what
transition does the company need to make in light of the Stochastic Exclusion Test?

American Academy of Actuaries 36 www.actuary.org

A: In this situation, the company would need to re-evaluate whether the group of policies requires a
calculation of the Stochastic Reserve. It may be that the group of policies still satisfies one of the other
exclusion tests; if not, the exclusion needs to be discontinued. The company would then be required to
calculate Stochastic Reserves in the determination of the minimum reserve.

Q 7.24: Does the Stochastic Exclusion Test need to be performed with year-end valuation data?

A: According to Section 6.A.1.a.i, groups of policies pass the Stochastic Exclusion Test if annually and within
12 months before the valuation date the company demonstrates that the groups of policies pass the Stochastic
Exclusion Ratio Test defined in Section 6.A.2. Therefore, year-end data does not need to be used.

Q 7.25: If a base policy passes the Stochastic Exclusion Test, does an associated rider with a separate
charge need to be tested separately for the Stochastic Exclusion Test?

A: According to VM-20 Section 2.H, riders and supplemental benefits are treated as described in Valuation
Manual, Section II. According to Section II, if the base policy is subject to VM-20 and the rider has a separate
premium or charge, then the rider reserve is calculated as if it were a stand-alone policy of a type most similar
in benefits to the rider. An exception is made for riders that enhance or modify the terms of the base contract,
e.g., a return of premium rider, cash value enhancement rider, or secondary guarantee rider. In these cases, the
rider must be valued as part of the base contract even if it has a separately identified premium or charge. If the
rider does not have a separate premium or charge, then all cash flows associated with the rider must be
included in the reserve calculation of the base policy. Supplemental benefits may be calculated on a stand-
alone basis.

Some actuaries might interpret Section 2.H as being implemented from the point of the exclusion tests, and
therefore riders would be tested separately. Other actuaries, however, might group the rider and base policy
together for purposes of the exclusion test, assuming this grouping is consistent with VM-20 requirements
regarding the exclusion test. Some actuaries would group return of premium riders with the base policy for
purposes of the exclusion test.

8: Deterministic Exclusion Test

Q 8.1: What is the Deterministic Exclusion Test?

A: The Deterministic Exclusion Test is a test that a company may elect to perform in order to simplify its
valuation process. For a group of policies considered as term or ULSG that does not meet the definition of
“non-material secondary guarantee” (definition within VM-01), the Deterministic Exclusion Test is not
available. Otherwise, if a group of policies passes both the Stochastic Exclusion Test and the Deterministic
Exclusion Test, then Section 2.A.3 specifies that the minimum reserve would be the aggregate Net Premium
Reserves for those policies. In other words, the Deterministic Reserve does not need to be calculated for those
policies. However, if the group of policies fails the Stochastic Reserve Test, then both the Deterministic
Reserve and the Stochastic Reserve must be calculated.

Q 8.2: How would a group of policies pass the Deterministic Exclusion Test?

American Academy of Actuaries 37 www.actuary.org

A: As described in Section 6.B, a group of policies passes the Deterministic Exclusion Test when the
company demonstrates that the sum of valuation net premiums for all future years is less than the sum of the
corresponding guaranteed gross premiums.

A company may elect to perform this test for a group of policies only if the group has passed the Stochastic
Exclusion Test, to determine whether it is required to calculate the Deterministic Reserve for that group of
policies.

Q 8.3: Is it necessary to perform the Deterministic Exclusion Test on all blocks of life insurance?

A: No. Section 2.A states that the Deterministic Exclusion Test is optional. The test may be applied to all
groups within the entire inforce life insurance business with the exceptions noted in Section 6.B.1.

Q 8.4: Are there any types of products that cannot be excluded from calculating the Deterministic Reserve
through the Deterministic Exclusion Test?

A: Section 6.B.1 notes, three types of policy groups are automatically deemed to not pass the Deterministic
Exclusion Test.

• Universal life policies with a secondary guarantee that do not meet the definition of a non-material
secondary guarantee (Section 6.B.1.a);
• Groups of policies that are not excluded from the Stochastic Reserve calculation (Section 6.B.1.a);
and
• Term insurance policies subject to Section 3.A.1 (Section 6.B.1.b).

Deterministic Reserves must be calculated for these policies.

Q 8.5: What limitations exist for grouping policies when using the Deterministic Exclusion Test?

A: Section 6.B.3 states that different contract types may not be grouped together for a combined test if the
contract types present significantly different risk profiles. Section 2.A also requires distinct product groups for
the calculation of the minimum reserve, which consist of “Term Policies” (Section 2.A.1), “ULSG Policies”
(Section 2.A.2), and “Life Insurance Policies Subject to Section 3.A.2” (Section 2.A.3).

Some actuaries would use the same grouping for the Stochastic Exclusion Test and the Deterministic
Exclusion Test due to the similar requirements of Section 6.A.2.b.iv and 6.B.3. For example, a block of
policies with high guaranteed gross premiums relative to their valuation net premiums may have less
mortality risk than other blocks within the company. Some actuaries would, therefore, not combine this block
with other contract types that have high mortality risk in order for the combined group collectively to pass the
Deterministic Exclusion Test. As noted in Q 8.4, some products are deemed to not pass the Deterministic
Exclusion Test and would therefore not be grouped together with other products when performing this test.

Actuaries who expect to calculate both the Stochastic Exclusion Test and the Deterministic Exclusion Test
may wish to consider the grouping requirements for both tests when determining the appropriate grouping for
modeling the policies.

Q 8.6: What valuation net premiums are used in the Deterministic Exclusion Test?

American Academy of Actuaries 38 www.actuary.org

A: The valuation net premiums used in the Deterministic Exclusion Test are defined in Section 6.B.5 as
follows:

For policies where the Net Premium Reserve is the minimum reserve according to the minimum reserve
requirements of Section 2.A of the Standard Valuation Law that applied to policies issued prior to the
Valuation Manual becoming effective (see VM-A, VM-C, and VM-M), then the valuation net premiums are
determined according to those minimum reserve requirements.

The Deterministic Exclusion Test is not available for term life insurance.

For any group of policies for which the Deterministic Exclusion Test is being performed, if anticipated
mortality is expected to be higher than the valuation mortality for the group of policies, then anticipated
mortality must be used in the determination of the valuation net premium.

Q 8.7: What gross premiums are used in the Deterministic Exclusion Test?

A: Per Section 6.B.5.e, guaranteed gross premiums used in the Deterministic Exclusion Test will generally be
the guaranteed gross premiums specified in the contract. For some universal life policies, there may not be a
schedule of guaranteed gross premiums. If no guaranteed gross premiums are specified, VM-20 defines the
gross premium to use as the level annual gross premium at issue that will keep the life insurance policy in
force for the entire coverage period based on the policy guarantees.

Q 8.8: Is the comparison of valuation net premiums and guaranteed gross premiums done before (direct)
or after reinsurance?

A: Per Section 6.B.2, the test is performed on either a direct or assumed basis. This means it is performed
before reinsurance ceded and including all assumed business.

Q 8.9: What products might be good candidates for the Deterministic Exclusion Test?

A: Traditional products that have guaranteed gross premium levels in excess of valuation net premiums are
good candidates for the Deterministic Exclusion Test. These product types generally have gross premiums
that provide sufficient revenues to cover the development of adverse mortality, expense, or investment income
experience. However, the Deterministic Exclusion Test is not mandatory, and some actuaries may calculate
the Deterministic Reserve even for policies that likely would pass the test.

Q 8.10: What are the implications of a group of policies passing the Deterministic Exclusion Test?

A: If a group of policies passes the Deterministic Exclusion Test, then Deterministic Reserve calculations are
not required in determining the minimum reserve. Some actuaries may still calculate the Deterministic
Reserve, however, and use it in the determination of the minimum reserve.

Q 8.11: How frequently is the Deterministic Exclusion Test performed?

A: Section 6.B.4 states that for a group of policies that is no longer adding new issues and the test has been
passed for three consecutive years, the test is assumed to be passed each year as long as the test is computed
once each five years going forward. Some actuaries see this requirement as implying that the test is required
to be passed annually unless the group of policies meets the requirements of Section 6.B.4.

American Academy of Actuaries 39 www.actuary.org

There is no specific requirement in VM-20 on how often the test must be performed. Some actuaries would
perform the test annually as of the valuation date. Other actuaries would perform the Deterministic Exclusion
Test in conjunction with the Stochastic Exclusion Test if the group of policies is eligible for passing the
Deterministic Exclusion Test.

Q 8.12: What changes may occur that could cause a group of policies where no new business is being
issued in that group to fail the Deterministic Exclusion Test after passing for several consecutive years?

A: Significant changes in the risk profile of a group of policies could cause a group of policies that had been
passing to start failing the test. For example, if poor experience emerges in the group over time that has
sufficient credibility, then the actuary may change the future anticipated mortality assumption such that it may
begin to approach or exceed the valuation mortality. In this case, the new, higher anticipated mortality will be
used to determine valuation net premiums, and the resulting net premiums may be higher than the guaranteed
gross premiums for the group.

9. Difference From Cash Flow Testing—Scenario Reserve Calculation

Q 9.1: May the actuary use an asset adequacy testing cash flow model to calculate the scenario reserve in
the Stochastic Reserve calculation for VM-20?

A: Some actuaries will make use of their cash flow testing (CFT) models in calculating the scenario reserve.
However, the actuary may need to adjust the CFT model to reflect the different purposes of the calculations.
Some of the potential differences between CFT and VM-20 are listed below. Note that this list is not intended
to be a complete listing of the differences.

• CFT models may have some approximations or model simplifications that are not appropriate for
VM-20.
• VM-20 uses prudent estimate assumptions; CFT assumptions may include margins for adverse
experience, but these may not be the margins required by VM-20.
• Some assumptions in VM-20 are prescribed and must be used in the calculation. CFT has no
prescribed assumptions.
• There may be some difference in projection periods / time step / model segments.
• Some business / model segments for which the scenario reserve needs to be determined may not be
cash flow tested.
• The starting assets used in the model and reinvestment assumptions may be different.
• Cash flow testing models may include issue years that are not within the scope of VM-20.
• Treatment of IMR and AVR may be different.

Q 9.2: May the actuary use the same assumptions for VM-20 as those used for cash flow testing?

A: VM-20 Section 9.A.1 requires that prudent estimate assumptions must be used. VM-01 defines “prudent
estimate assumption” as “a risk factor assumption developed by applying a margin to the anticipated
experience assumptions for that factor.” Some modifications to cash flow testing assumptions may be
required if the cash flow testing assumptions are not prudent estimate assumptions. Also, some assumptions
are fully or partially prescribed for the Stochastic Reserve calculation. Examples include lapse rates for

American Academy of Actuaries 40 www.actuary.org

certain universal life policies per Section 9.D.5 and asset reinvestment and default cost assumptions as
described in Section 9.F.

Q 9.3: In calculating the Stochastic Reserve, may the actuary use the same interest rate and equity
scenarios as used for the CFT projections?

A: Section 7.G.2.a states that a prescribed economic scenario generator with prescribed parameters is required
to be used to generate the interest rate and equity scenarios used in the determination of the Stochastic
Reserve. Appendix 1 of VM-20 states that the prescribed economic scenario generator is available on the
SOA website. Scenario reduction techniques are allowed to reduce the number of scenarios, but only if they
do not materially reduce the reserve. In order to use a smaller set of scenarios for VM-20 calculations, the
scenario picker tool within the economic scenario generator must be used to select that subset. CFT analysis
does not require a prescribed generator and parameters.

Q 9.4: Can the CFT projection period be used without modification for VM-20 calculations?

A: Section 7.A.1.d of VM-20 states that “The Company shall use a cash flow model that projects cash flows
for a period that extends far enough in the future so that no obligations remain.” There is also a Guidance
Note in this section that states “For example, it may be reasonable to assume 100% deaths or 100% surrenders
after some appropriate period of time.”

Actuaries wishing to use the same projection period in the VM-20 calculation as is used for CFT will have to
make sure these conditions are satisfied.

Q 9.5: Would the actuary generally use the same starting assets in both the VM-20 and CFT calculations?

A: Section 7.D requires the assignment of a set of starting assets satisfying certain criteria to each model
segment. Some actuaries will assign the same assets for both calculations, but it is not a requirement. Because
VM-20 applies to new issues only, some actuaries may select a limited subset of assets that were purchased
during the time period when the sales occurred and that meet the criteria. Other actuaries may use a pro-rata
portion of the CFT assets for VM-20 if these assets meet the criteria.

Q 9.6: Would the actuary generally use the same model aggregation in VM-20 as is used in the CFT
calculations?

A: The minimum reserve requirements specified by Section 2 of VM-20 clearly state that policies subject to
VM-20 requirements must be included in a specific product group, where the product groups are (1) term
policies; (2) universal life with secondary guarantee policies; and (3) life insurance policies subject to Section
3.A.2. Further, there is a requirement in Section 5.G that the Stochastic Reserve for any product group be
calculated on a stand-alone basis. This approach may or may not be consistent with the CFT calculations of
the company. Some actuaries might wish to review carefully the CFT model structure before using that
structure for VM-20 minimum reserve calculations.

Q 9.7: How are IMR and AVR treated in Stochastic and Deterministic Reserve calculations?

A: Because VM-20 is on a pre-tax basis, the IMR in the Stochastic and Deterministic Reserve calculation is
adjusted to a pre-tax basis. The VM‐20 reserve methodology (for both DR and SR) recognizes that the pretax
IMR (PIMR) is redundant under a PBR regime because the PIMR only impacts the timing of recognizing

American Academy of Actuaries 41 www.actuary.org

investment income over the lifetime of the investment. Hence, the approach to reflect the PIMR in the DR
calculations is to forgo adjusting the NAER calculation by including the amortization of the PIMR and,
instead, simply make a direct post‐calculation reduction/increase to the modeled reserve in the amount of the
positive/negative PIMR existing on the valuation date. With respect to the SR, the approach is to forgo
adjusting the modeled statutory asset values at all projection durations for unamortized PIMR (arising from
PIMR amounts existing at the start of a projection and amounts arising as a result of capital gains and losses
occurring during a projection) and, again, simply make a direct post‐calculation reduction/increase to the final
CTE 70 reserve in the amount of the existing allocated positive/negative PIMR at the valuation date.

The AVR does not enter into the calculation of either the Stochastic Reserve or the Deterministic Reserve.

Q 9.8: How are federal income taxes treated in VM-20 versus CFT?

A: VM-20 requires that federal income taxes be excluded from cash flows. Typically, federal income taxes
would be reflected in CFT.

Q 9.9: Are taxes other than federal income taxes ignored for the purpose of calculating reserves under
VM-20? For example, should the actuary ignore premium taxes?

A: Section 7.B.1.e states that taxes (excluding federal income taxes and expenses paid to provide fraternal
benefits in lieu of federal income taxes) should be included in the cash flows. Section 9.E contains more
information on taxes and expenses, including that foreign income taxes are excluded (Section 9.E.1.g).

10. Considerations When Performing Reserve Calculations on Other Than the Valuation Date

Q 10.1: What valuation date is required for the Stochastic and Deterministic Reserves?

A: Per Section 2.E of VM-20, the calculation of the Stochastic and Deterministic Reserves can be performed
as of a date up to three months prior to the valuation date as long as an appropriate method is used to adjust
the reserves to the valuation date.

Although the calculation may use prior data, some actuaries would take into account relevant experience up to
the valuation date in their determination of Anticipated Experience and Prudent Estimate Assumptions used in
the calculations.

Q 10.2: Does the Net Premium Reserve have to be calculated using data as of the valuation date?

A: The Net Premium Reserve is a seriatim reserve that is calculated for each policy in force on the valuation
date. It is expected that actuaries would perform the Net Premium Reserve calculation as of the valuation date
using current inforce data, if possible.

Q 10.3: How might an actuary “use an appropriate method to adjust” the Deterministic and Stochastic
Reserves to the valuation date when it is calculated as of an earlier date?

A: VM-20 does not prescribe a method for this adjustment. Some actuaries may use a model office as of the
earlier date and make several adjustments to take into account changes to the valuation date. The adjustments

American Academy of Actuaries 42 www.actuary.org

could include taking into account lapses and new policies issued since the projection start date but prior to the
valuation date through a pro-rata adjustment to the relevant policy information (e.g., face amount, account
value) so the overall characteristics for the group of policies approximate those characteristics on the
valuation date. Additionally, the actuary could review all prudent estimate assumptions in the model and
make any adjustments necessary for information / experience from the projection start date to the valuation
date. Adjustments to the assets may or may not be made depending on how materially the asset portfolio has
changed. In particular, some actuaries would use the economic scenarios as of the valuation date if the
economic conditions have changed materially.

Some actuaries may apply a high-level adjustment to the overall reserve amount or to each of several portions
of the reserve. The adjustment may be based on analysis of past trends in the reserve, including “roll-forward”
analysis of the change in the reserve balance. Such analysis would typically isolate the main drivers of the
reserve change and would enable the actuary to estimate short-term changes in the reserve after the projection
start date based on correlation with those drivers. This could include using some characteristic of the group of
policies in the calculation (e.g., face amount for a group of term insurance policies) and a ratio method to
modify the Scenario Reserve for the stochastic calculation or the Deterministic Reserve for the group of
policies actually in force on the valuation date. Other actuaries may calculate the Stochastic and Deterministic
Reserves on a small, representative sample of policies from the group of policies on the valuation date and
compare this amount to the Stochastic and Deterministic Reserves for those policies as of the earlier
projection start date. This information could be used in a number of ways to modify the calculations made
using the earlier projection start date to the valuation date. Of course, the actuary would need to be prudent in
determining the number of representative samples that are appropriate and which policies to use in this
calculation.

It would be prudent for the actuary to confirm that the approach used is appropriate for the group of policies
being valued. If the policy characteristics or economic environment have changed materially, a Projection
Start Date prior to the valuation date may not be appropriate.

Q 10.4: Should changes in economic conditions from the projection start date also be incorporated in the
adjustment of the reserves?

A: Some actuaries performing the calculations would modify the reserve calculations to take into account the
economic conditions as of the valuation date if the valuation date is different from the projection start date.
This adjustment would take into consideration both the impact on the liability cash flows and the net asset
earned rate.

For products whose liability cash flows are dependent on changes in the economic environment, such as
current assumption universal life products, some actuaries would reflect the economic conditions on the
valuation date (if there have been material changes) and adjust the liability cash flows accordingly. This may
require a re-computation of the reserve amounts or could include high-level adjustments reflecting expected
changes to the liability cash flows. Some actuaries would apply adjustments that were based on sample
calculations or prior sensitivity testing.

For products whose future expected liability cash flows are not sensitive to changes in market conditions, it is
expected that some actuaries would not make modifications to the liability calculation. It is likely that these
actuaries would disclose their rationale in the documentation of the calculation. An example would be no or
very low Cash Surrender Value term policies with fixed premiums. The liability cash flows would not be
dependent on a change in the economic environment, so no adjustments would generally be made. Some
actuaries might make adjustments in order to reflect the impact of changes in interest rates on discount rates.
American Academy of Actuaries 43 www.actuary.org
Some actuaries believe that actuarial judgment should be used to determine whether adjustments to the
reinvestment assumptions and the projected net asset earned rate used in the calculation are needed. Those
actuaries would modify the asset projection based on updated economic conditions as of the valuation date if
the changes would lead to materially different reserves.

It is anticipated that some actuaries would include a specific description and discussion of how the asset and
liability cash flows were adjusted in the documentation supporting the calculation per the requirements in
VM-31 Section 3.D.11.g.

Q 10.5: Will actuaries adjust the calculation for new business and lapses if company inforce data from
prior to the valuation date is used?

A: If policy inforce data from a date prior to the valuation date is used, it is expected that some actuaries
would make no adjustments to the models to recognize changes in the inforce data between the projection
start date and the valuation date. If large volumes of new business have been issued, some actuaries would
estimate the additional reserves from the new issues. It is also expected that some actuaries would have a
lapse and new business assumption to reflect changes in the inforce data between the projection start date and
the valuation date.

Some actuaries would analyze whether including policies no longer in force as of the valuation date in the
Stochastic Reserve modeling would materially affect the reserves. If there were a material difference in actual
to expected terminations, some actuaries would make modifications to the models to take into account the
changes. Some actuaries would include a specific description and discussion of how the inforce policies were
modified to take into account changes up to the valuation date in the documentation supporting the calculation
and why it was determined to be material or immaterial.

Q 10.6: Does the calculation need to be adjusted for policyholder actions such as additional premium
payments, taking or repayment of a policy loan, partial withdrawals, face amount adjustments, and other
policy changes that have occurred since the projection start date but prior to the valuation date?

A: It is expected that some actuaries would not make adjustments to the reserve for policyholder actions
between the projection start date and the valuation date. However, if there have been policyholder actions that
would materially affect the minimum reserve, some actuaries would make modifications.

Q 10.7: Are there other types of changes occurring between the projection start date and the valuation date
that could lead to adjustments in the calculations?

A: In addition to those items addressed above, some actuaries would consider the impact of changes to the
business’s risk profile, such as:
• Significant changes in asset allocation or mix of assets (by quality, duration, or other characteristics);
• New, terminated, or recaptured reinsurance;
• Changes in counterparty risk;
• Changes in non-guaranteed elements; and
• Merger and acquisition activity.

American Academy of Actuaries 44 www.actuary.org

11. Details on Starting Assets and Asset Modeling

Q 11.1: How are due premiums reflected in the starting assets?

A: The due premium asset associated with expected future due premium cash flows should not be included in
the level of starting assets used when calculating the DR and SR. This is consistent with the following two
VM-20 requirements. First, VM-20 requires that due premiums be included in the expected future cash flows
when calculating the DR and SR. Second, the definition of the minimum reserve includes an adjustment to the
aggregate Net Premium Reserve for any due and deferred premium held on account of those policies. See Q
3.3 for more detail.

Q 11.2: If the company passes the Stochastic Exclusion Test and Deterministic Exclusion Test, does the
actuary still have to model assets?

A: Assets do not need to be modeled to determine the Net Premium Reserve. However, assets may still need
to be modeled to perform the Stochastic Exclusion Test. Some actuaries performing the Stochastic Exclusion
Test would conclude that they need to model assets, while others may find alternative methods per Section
6.A.1.a.iii to certify that they would pass this test if it were performed.

Q 11.3: What is the level of starting assets required in the calculation of the Deterministic and Stochastic
Reserve?

A: Section 7.D.1 states that for each model segment, the level of starting assets is required to equal the
estimated value of the modeled reserve plus the PIMR balance on the projection start date.

The final aggregate modeled reserve depends on the level of starting assets, making it difficult to estimate the
final aggregate modeled reserve with 100% accuracy. Accordingly, Section 7.D.3 requires that if, for all
model segments combined, the value of starting assets—using asset valuation methods consistent with their
annual statement values—is less than 98% of the final reserve or greater than the larger of zero, the NPR less
the due and deferred premium asset, or 102% of the final reserve (where the final reserve is the final
aggregate modeled reserve [whether stochastic or deterministic]), the company must supply documentation in
the PBR Actuarial Report that provides reasonable assurance that the aggregate modeled reserve is not
materially understated as a result of the estimate of starting assets.

Some actuaries will set the level of starting assets equal to their best estimate of what the modeled reserve will
end up being on the valuation date based on historical reserve patterns and trends, and the impact of such
things as new sales, new products, and new reinsurance agreements since the prior valuation date, or any other
items that impact the expected reserve level.

Some actuaries may target starting assets to be within 2% of the final aggregate modeled reserve but set
starting assets to NPR if larger than the final aggregate modeled reserve. Some actuaries using the Direct
Iteration Method for the Deterministic Reserve may do an additional run with starting assets bumped up to the
NPR, if larger than the final aggregate modeled reserve, as part of their documentation showing the
aggregated modeled reserve hasn’t been materially understated as a result of using starting assets outside the
collar.

Q 11.4: How would the actuary meet the 98 percent to 102 percent corridor required in Section 7.D.23?

American Academy of Actuaries 45 www.actuary.org

A: If the starting assets used in the initial reserve calculation fall outside the prescribed corridor, some
actuaries would perform iterative calculations for each model segment until the starting assets used meets the
asset corridor requirements to ensure that additional recalculations are not required. Other actuaries might
evaluate the asset corridor across all model segments and make adjustments as necessary. However, VM-20
does not require either method to be used. Therefore, some actuaries may choose other methods to
demonstrate that the reserve is adequate if the starting asset level falls outside of the required corridor. These
could be based on additional calculations that have been performed historically or showing that the reserve
change is immaterial if the starting assets were changed. The latter may be used in particular if the reserve
amount is quite small.

Q 11.5: How should the actuary choose which assets to include in the calculation?

A: Model segments are needed to determine the group of policies and associated assets that are modeled
together to determine the path of net asset earned rates. Section 7.A.1.b states that model segments should be
selected that are consistent with the company’s asset segmentation plan, or the approach used to allocate
investment income for statutory purposes. Further, Section 7.D.1.a and Section 7.D.4 outline the required
selection of assets for each model segment at the Projection Start Date should include:

• All Separate Account assets supporting the policies;

• Policy loans, if they are explicitly modeled per Section 7.F.3.b.
• Due, accrued, or unearned investment income;
• All derivative investments held at the projection start date that are part of a derivative program and
can appropriately be allocated to the model segment; and
• General Account assets that result in the sum of all starting assets equal to the estimated value of the
modeled reserves plus the PIMR balance on the projection start date.

Regarding the last bullet point above, some actuaries would choose General Account assets that, for
management purposes, are generally associated with policies modeled. For example, where asset
segmentation is used, the starting assets would be selected from the asset segment that supports the block of
policies subject to PBR (i.e., the asset segment that is used to track investment strategies and allocation of
investment income to the policies they support).

In other cases, the assets may not be clearly defined to support policies subject to VM-20 Sections 4 and 5,
versus those that are not, and other approaches will be used. Section 7.D.2 requires that for an asset portfolio
that supports policies both subject and not subject to VM-20, an equitable method to apportion the total
amount of starting assets shall be determined. Additionally, per Section 7.D.4.e., the assets shall generally be
selected on a consistent basis from one reserve valuation to the next.

Q 11.6: If a company doesn’t have enough real assets in a portfolio, and the Stochastic or Deterministic
Reserve calculation requires more assets to get within the 2 percent collar, what should it do? Can you add
more assets to the projection if they do not actually exist?

A: Some actuaries would proportionally increase the assets in the portfolio to equal the starting asset amount
under the assumption that assets would be transferred into the portfolio with the same average characteristics
of the existing portfolio if more assets are deemed to be required.

American Academy of Actuaries 46 www.actuary.org

Other actuaries would identify specific assets in the surplus portfolio or in other portfolios (e.g., corporate) so
there are enough assets for the projection.

Q 11.7: To the extent that starting assets have unstable market values, what impact would this have on the
projected cash flows?

A: Per Section 7.D.6, the projected value of starting assets shall be determined in a manner consistent with
their values at the start of the projection. Assets are selected at the projection start date such that the aggregate
annual statement value of the assets equals the estimated value of the modeled reserve plus the PIMR balance.
Section 7.F.1.c also requires modeling the proceeds arising from modeled asset sales, and determining the
portion representing realized capital gains and losses.

During any projection interval, Section 7.F.1.d requires the actuary to reflect any uncertainty in the timing and
amounts of asset cash flows related to the paths of interest rates, equity returns or other economic values.

In considering this requirement, actuaries would not make any modification to prescribed asset assumptions
or the modeled interest rates or equity returns because they are generated stochastically using a prescribed
generator.

Some actuaries may take into account market instability by making modifications to nonprescribed
assumptions, where appropriate, using actuarial judgment.

Q 11.8: Is it possible for the General Account portion of starting assets to be negative?

A: Yes. The Guidance Note following Section 7.D.5 indicates that the General Account starting assets
described in Section 7.D.4.e may be negative. The Guidance Note also states that for the General Account
starting assets described in Section 7.D.4.e, negative assets or short-term borrowing can be used, resulting in a
projected interest expense.

An example of this situation is when a substantial portion of policyholder funds are allocated to Separate
Accounts and the modeled reserve is less than the Separate Account Value.

Q 11.9: How will the actuary choose assets to include in the model if there are more General Account
assets associated with the modeled policies than are needed as starting assets?

A: Some actuaries would choose a pro-rata portion of a segmented portfolio or other grouping of assets. Some
actuaries would follow how assets are assigned for determining credited rates. Some actuaries might choose
specific assets based on modeling constraints. However, the same asset would not be included in the starting
assets of different model segments where the total amount of such asset is more than what the company owns.

Q 11.10: What types of grouping of assets might be used in the projection?

A: Section 7.F.1.a states that grouping of assets is allowed if the company can demonstrate that grouping does
not materially understate the minimum reserve that would have been obtained using a seriatim approach.
Some assets may be similar enough in nature that some actuaries might reasonably expect their cash flows
and valuations to react in similar fashions as economic conditions change and therefore would group them
together for modeling purposes. For grouping of fixed-income investments, some actuaries might ensure that
parameters such as credit quality, time to maturity, duration, and interest and principal patterns are similar
American Academy of Actuaries 47 www.actuary.org
before considering grouping of individual securities. Simplification of cash flow modeling, including asset
modeling, is discussed in Section 7.A.2, which references Section 2.G. Other actuaries would still model
assets on a seriatim basis.

Q 11.11: How are investment cash flows (returns and principal repayments) for general account fixed-
income assets modeled?

A: VM-20 states that fixed-income assets include public bonds, convertible bonds, preferred stocks, private
placements, asset-backed securities, commercial mortgage loans, residential mortgage loans, mortgage-
backed securities, and collateralized mortgage obligations. (See Guidance Note after Section 7.F.1.c.)

Investment cash flows on fixed-income assets are modeled for each projection interval per Section 7.F.1 as
follows:

• Project the gross investment income and principal repayments for each asset (or grouping of assets).
• Subtract prescribed default costs from gross investment income.
• Subtract investment expenses, using prudent estimate assumptions, from gross investment income.
• Project the proceeds from asset sales, and determine the portion representing any realized capital gains
and losses.
• Reflect any uncertainty in the timing and amounts of the asset cash flows (calls, puts, prepayments,
extensions, etc.).
• Model the impact of the derivative asset programs (if any) associated with these assets.

For starting assets, gross investment income shall be modeled in accordance with the contractual provisions of
each asset and in a manner that is consistent with each economic scenario. For reinvestment assets, gross
investment income is determined by adding a prescribed spread (see Tables F through I on the Related
Documents tab of the LATF page of the NAIC website [https://naic.org/cmte_a_latf.htm], updated on a
monthly basis for current spreads, and quarterly basis for long-term spreads) to the prescribed Treasury rate in
each projection interval (where the prescribed Treasury rate comes from the economic scenario).

Q 11.12: How are investment cash flows for general account total return equity investments modeled, such
as money market funds, diversified bond funds, real estate, U.S. or international equity, private equity,
etc.?

A: VM-20 states that equity assets are non-fixed-income investments having substantial volatility of return,
such as common stocks and real estate investments.

Investment cash flows on equity assets are modeled per Section 7.F.2 as follows:

• Determine the grouping for equity asset categories (called “proxy funds” in VM-20) that reflect the
types of equity assets of the company (e.g., large-cap stocks, international stocks, owned real estate,
etc.) as described in Section 7.I of VM-20. Each proxy fund is typically expressed as a linear
combination of recognized market indices.
• Allocate specific equity assets to each asset category (i.e., each proxy fund).
• Project the future gross investment return, including realized and unrealized capital gains, for each
proxy fund in a manner that is consistent with the prescribed general account equity returns from the
economic scenario as described in Section 7.G.

American Academy of Actuaries 48 www.actuary.org

 • Model the timing of asset sales in a manner that is consistent with the investment policy of the
company for that type of asset.
• Reflect investment expenses through a deduction to the gross investment return using prudent estimate
assumptions.

This process involves actuarial judgment to determine the appropriate proxy funds and methodology to “map”
the return from proxy funds to the prescribed general account equity returns from the prescribed economic
scenario. See Q12.1 for details on the different returns produced by the prescribed generator.

Actuaries might wish to effect modeling of proxy funds in different ways. For example, some actuaries might
perform a regression of past fund performance against past performance of market indices consistent with the
scenario generator classifications, and develop a scenario set for each proxy fund that is consistent with an
appropriate weight of such scenario returns derived from the regression. Alternatively, funds might be
decomposed using regression techniques into funds corresponding to each of the indices, or some other way
could be used. Whatever method is used, the actuary must be able to document that the projected returns for
the proxy funds are not overly optimistic.

VM-20 provides a Guidance Note (after Section 7.G.2.b) to assist actuaries in performing this mapping
process.

Q 11:13: How are investment returns on Separate Account assets modeled?

A: Per Section 7.F.5, investment cash flows on Separate Account assets are modeled as follows:

• Determine the number and type of Separate Account funds (called “proxy funds” in VM-20) that
reflect the types of variable subaccounts in the Separate Account assets (e.g., large-cap stocks,
international stocks, owned real estate, etc.) as described in Section 7.J. Each proxy fund is typically
expressed as a linear combination of recognized market indices.
• Allocate specific variable subaccounts to each asset category (i.e., each proxy fund).
• Project the future total investment return, including realized and unrealized capital gains, for each
proxy fund in a manner that is consistent with the prescribed return from the economic scenario as
described in Section 7.G.

Thus, investment returns on Separate Account assets follow the same modeling process used to determine the
investment returns on General Account equity investments described above. Similar to the process for
General Account equity investments, actuarial judgment is used to determine the appropriate proxy fund, and
determine the methodology to “map” the return from proxy funds to the prescribed returns from the
prescribed economic scenario. A Guidance Note after Section 7.G.2.b may assist actuaries in performing this
mapping process.

Note that Separate Accounts may also include fixed-income funds, but the same mapping process is used.

Q 11.14: How are policy loans modeled?

A: Section 7.F.3 defines two different approaches to modeling policy loans, explicitly or by proxy:

• Explicit modeling. A direct approach is to treat policy loan activity as an aspect of policyholder
behavior as described in Section 7.F.3.b. This approach models policy loan activity explicitly by

American Academy of Actuaries 49 www.actuary.org

 modeling all future policy loan activity in each projection period (i.e., interest, principal repayments,
new principal amounts, etc.) as liability cash flows in the Deterministic Reserve and Stochastic
Reserve calculations. Thus, policy loan interest and loan principal repayments are excluded from the
Net Asset Earned Rate (NAER) calculation. Policy loans at the policy level can then be handled based
on each policy’s situation based on policyholder behavior assumptions subject to the requirements of
Section 9.D additional requirements for explicit modeling of policy loans, covered in Section 7.F of
VM-20, are below:

o Section 7.F.3.b.i requires that policy loan activity is treated as an aspect of policyholder behavior.
o Section 7.F.3.b.ii requires loan balances are assigned to either exactly match each policy’s loan
utilization or reflect average utilization over a model segment or sub-segments.
o Section 7.F.3.b.iii requires that policy loan interest rates be modeled in a manner that is consistent
with policy provisions and with the economic scenario and interest paid in cash is treated as a
policyholder cash flow (not as investment income cash inflow), but interest due that is added to
the loan balance is ignored. This is because the increased balance will require increased repayment
in future periods.
o Section 7.F.3.b.iv requires policy loan principal repayments be modeled as a cash inflow,
including those which occur automatically upon death or surrender.
o Section 7.F.3.b.v requires that additional policy loan principal be modeled as a negative policy
loan cash flow.
o Section 7.F.3.b.vi requires any investment expenses allocated to policy loans be modeled with
policy loan cash flows or insurance expense cash flows (not as an investment expense).

• Proxy. This approach substitutes assets that are a proxy for policy loans (e.g., a bond) in the starting
asset amount, and then includes the investment income from the proxy assets in the NAER (used in
the Deterministic Reserve calculation), and as an investment cash flow in the Stochastic Reserve
calculation. If this approach is followed, Section 7.F.3.a.i requires the company to demonstrate that
the resulting reserve is not less than what would be produced by modeling policy loan activity
explicitly under the first approach.

Regardless of which approach is used, the policy loan activity must reflect the company’s expected utilization
of policy loans by policyholders and comply with the policyholder behavior requirements stated in Section
9.D.

Some actuaries will choose to follow the explicit modeling approach to avoid the demonstration that is
required under the proxy approach and to more directly satisfy the policyholder behavior requirement that
may be difficult to replicate without modeling policy loan activity explicitly.

If the explicit modeling approach is selected, Section 4 and Section 7.F describe the approach to reflect policy
loan activity in the modeled reserve calculation. The treatment of the various policy loan items can be
summarized as follows:

1. If the Deterministic Reserve is calculated as the present value of cash flows (Section 4.A):
a. The initial policy loan balance is included in the Deterministic Reserve. Section 4.A defines
the Deterministic Reserve as the actuarial present value of benefits and expenses, less the
actuarial present value of premiums, less the PIMR balance, plus separate account assets, plus
the policy loan balance at the valuation date, with appropriate reflection of any relevant due,
accrued, or unearned policy loan interest present value of benefits. In effect, this is equivalent

American Academy of Actuaries 50 www.actuary.org

 to adding the initial policy loan balance to the reserve that is determined by taking the present
value of cash flows.
b. Section 4.A.4.c requires that future net policy loan cash flows are included the actuarial
present value of premiums in the Deterministic Reserve calculation if policy loans are
explicitly modeled.
c. The Guidance Note after 4.A.4.c states that future net policy loan cash flows include
policy loan interest paid in cash, plus repayments of policy loan principal, including
repayments occurring at death or surrender (note that the future benefits in Section
4.A.3.a. are before consideration of policy loans), less additional policy loan principal.
These following items are modeled as future cash flows, either increasing or decreasing the
amount of invested reinvestment assets and present value of future cash flows:
i. New additional loan principal amounts are treated as negative cash flows in future
projection periods (Section 7.F.3.b.v).
ii. Loan principal repayments are treated as positive cash flows in future projection
periods. Principal repayments from death claims and surrenders are included (Section
7.F.3.b.iv).
iii. Future policy loan interest (if paid in cash) is treated as a positive cash flow in future
projection periods (Section 7.F.3.b.iii).
iv. Future policy loan interest that is assumed to be added to the loan balance is not
treated as a cash flow (Section 7.F.3.b.iii).
d. Any investment expenses allocated to policy loans shall be modeled, either with loan cash
flows or insurance expense cash flows (Section 7.F.3.b.vi).

2. If the Deterministic Reserve is calculated under the Direct Iteration Method, where the reserve equals
the amount of starting assets that produces an ending accumulated asset amount of zero (Section
4.B.):
a. The initial policy loan balance on the valuation date is projected forward using the projected
additional loan principal amounts, loan principal repayments, and future policy loan interest
added to the loan balance.
b. The future policy loan cash flows are reflected in the projection of the amount of non-policy
loan reinvestment assets.

3. Per Section 5.B.1, the projected assets used to calculate the Stochastic Reserve follow the same
approach to project assets described in item 2 for calculating the Deterministic Reserve under the
Direct Iteration Method above.

Under explicit modeling, the net impact of policy loan activity on the Deterministic Reserve will vary
depending on the relationship between the policy loan interest rate and the emerging NAERs.

Q 11.15: How are derivative programs modeled?

A: Section 7.K.1 requires that the cash flow model reflect the appropriate costs and benefits of derivatives
currently held for the Deterministic Reserve and the Stochastic Reserve.

The requirements for the modeling of future derivative program transactions, however, vary by the type of
derivative program.

American Academy of Actuaries 51 www.actuary.org

If the company has a clearly defined hedging strategy (CDHS), the model must reflect appropriate costs
and benefits of future derivative transactions associated with its execution. Additionally, Section 6.A.1.b
excludes any group of policies for which there is a CDHS from the stochastic exclusion tests, so these
groups of policies must calculate a Stochastic Reserve.

• CDHS requirements are similar to Actuarial Guideline XLIII & C3 Phase II. Some actuaries would
consider examples of CDHS associated with VM-20 products to include:

o Hedging for equity-indexed products

o Hedging variable life long-term guarantees
o Rebalancing portfolio using derivatives to meet ALM targets

The Guidance Note after Section 7.K.1 clarifies that the modeled reserve requirements prohibit projection
future hedging transactions other than those associated with a clearly defined hedging strategy CDHS.
This prohibition was added to address concerns that reserves could be unduly reduced by including
programs that may or may not actually be executed.

Section 7.K.1 also requires that the model reflect appropriate costs and benefits of future derivative
transactions associated with non-hedging derivative programs (e.g. replication, income generation) if they
are part of the company’s risk assessment and evaluation processes. The company may choose whether or
not to reflect non-hedging programs that are not part of the regular risk assessment process.

For each derivative program that is modeled, the company shall reflect any established investment policies
and procedures for that program, project expected program performance along each scenario, and recognize
all benefits, residual risks, and associated frictional costs:

• Residual risks include basis, gap, price, parameter estimation, and variation in actuarial assumptions.
• Frictional costs include transaction costs, opportunity cost of margin requirements, and administrative
costs.
• Also, an addition to Stochastic Reserve is required to reflect material derivative risks not fully
captured within the “CTE 70” cash flow model. Emerging actuarial practice should produce sound
approaches.

Reflection of a qualified CDHS in the model may either increase or decrease the reserve. Not reflecting a
hedging program similar to a qualified CDHS can also increase or decrease the reserve relative to what would
be held if the program qualified as a CDHS. The direction of such impact can vary with economic conditions.
The Guidance Note in VM-20 encourages future consideration of a graded approach for recognizing the
impact of non-qualifying hedging strategies.

Each derivative program must be classified as associated with assets or with liabilities for purposes of
calculating reserves, especially the Deterministic Reserve:

• “Asset-associated derivatives” or “derivative asset programs” are included in the net asset earned
rates.
• “Liability-associated derivatives” or “derivative liability programs” are included in the liability cash
flows.

American Academy of Actuaries 52 www.actuary.org

Some actuaries will conclude that if derivative cost and payoff generate relatively smooth increments to the
net asset earned rate (more “bond-like”), it is a good candidate for asset treatment. Otherwise, liability
treatment may be more appropriate.

Q 11.16: What investment strategies for modeling reinvestment assets are permitted?

A: Section 7.E.1.a requires that the modeled investment strategy for reinvestment assets:

• Must be a representation of the company’s actual investment policy.

• Is permitted to be complex and incorporate assets for which spreads are not prescribed, or simple and
be expressed as a function only of assets for which spreads are prescribed (or zero for Treasuries) for
ease of demonstrating compliance with the minimum floor requirement.

Some actuaries will use a mix of asset types and final maturities that produce a desired pattern of gross
investment income and principal repayments. Other actuaries will assume fixed or floating rate coupons, or
zero coupons. Other actuaries will prefer to utilize a complex model representation for reinvestment assets,
which may include, for example, illiquid or callable assets. Other actuaries will prefer a simple model
representation, which may involve mapping of more complex assets to combinations of asset types, such as
combinations of public non-callable corporate bonds, U.S. Treasuries, and cash.

Whatever the approach, Section 7.E.1.a requires that the actuarial report includes documentation supporting
the appropriateness of the assumptions relative to actual investment policy.

Furthermore, the requirement in Section 7.E.1.g, which is covered in more detail in the following question,
may also affect the modeled investment strategy. The modeled investment strategy may be altered to achieve
compliance with the floor or reduce the need to calculate the modeled reserve twice to demonstrate
compliance.

Q 11.17: How are gross spreads determined for reinvestment assets?

A: Section 7.E.1.d states that the gross spreads on public non-callable corporate bonds purchased in the model
subsequent to the valuation date are determined by a prescribed methodology. The methodology is based on
current and historical spread levels relative to U.S. Treasuries and is consistent with the prescribed approach
to determine default costs on existing assets. For these bonds:

• Gross spreads are determined based on actual current and historical market data, using the same tables
used in the calculation of default costs.
• The prescribed gross spreads start at current average market spreads in effect at the valuation date
(published by the NAIC from a market source on a monthly basis) and grade linearly in yearly steps to
the long-term benchmark spreads (derived and published by the NAIC based on actual historical data
from the same market source on a quarterly basis) by the start of projection year four.
• Prescribed default costs are then deducted explicitly for purchased assets, using the same approach to
calculate default costs as for existing assets (but ignoring the maximum net spread adjustment factor).

For investments other than public, non-callable corporate bonds, gross spreads are not prescribed, but are to
be consistent with and in reasonable relationship to the prescribed spreads for public, non-callable corporate
bonds.

American Academy of Actuaries 53 www.actuary.org

Per Section 7.E.1.g., the methodology also incorporates a floor on the modeled reserve.

• The modeled reserve cannot be less than would be obtained by substituting an alternative investment
strategy in which all fixed income reinvestment assets are public non-callable corporate bonds with a
credit quality blend of 50% “A2/A” and 50% “Aa2/AA”.
• The blend of 50 percent A and 50 percent AA is intended to represent an approximate equivalent of
the industry average asset allocation. This is based in part on data incorporated in a NAIC Rating
Agency Work Group report.

The VM-20 Actuarial Report requires that the actuary provide documentation demonstrating compliance with
the minimum floor requirement. Some actuaries may decide to run the reserve calculation twice, the first
using spreads following the company’s investment strategy, and a second calculation assuming a strategy
following the 50/50 mix of A2/A and Aa2/AA bonds. However, if the model investment strategy does not
involve callable assets, some actuaries will conclude that demonstration of compliance will not require
running the reserve calculation twice. For example, an analysis of the weighted average net reinvestment
spread on new purchases by projection year (gross spread minus prescribed default costs minus investment
expenses) of the model investment strategy compared to the weighted average net reinvestment spreads by
projection year of the alternative strategy may suffice.
.
Some actuaries will conclude that they need to adjust the assumed mix of asset types, asset credit quality, or
the levels of nonprescribed spreads for other fixed-income investments to achieve compliance.

Q 11.18: How are asset market values modeled when assets are sold?

Per Section 7.E.2, gross spread assumptions are needed to compute market values when modeling the sale of
starting assets as well as the purchase and sale of reinvestment assets. Spreads used in calculating the market
value when assets are sold are to be consistent with, but not necessarily the same as, the spreads used for
purchases, recognizing that specific starting assets may have different characteristics than the modeled
reinvestment assets.

Q 11.19: What is the Pretax IMR (PIMR) and why is it used instead of the IMR?

A: The PIMR is the statutory interest maintenance reserve (IMR) adjusted to a pretax basis for each model
segment. It is adjusted to a pretax basis because the minimum reserve calculation in VM-20 ignores the
impact of income taxes. The PIMR is recognized as a reduction to the present value of cash flows in the
Deterministic Reserve determination and as a reduction to the CTE 70 amount in the Stochastic Reserve
determination.

Q 11.20: Should the assets backing the PIMR be included in the starting assets?

A: The PIMR liability is described as a negative asset that equals the portion of the total company PIMR
allocable to each model segment. The PIMR may be positive or negative. The allocated PIMR is not included
in the starting assets. Rather, Section 7.D.1 notes that the value of the starting assets on the projection start
date equals the estimated value of the modeled reserve plus the PIMR balance on the projection start date.
Note that the deterministic and stochastic reserves already deduct the PIMR balance allocated to the group of
policies at the valuation date allocated to the group of policies, as defined in Section 4.A and Section 5.F.

American Academy of Actuaries 54 www.actuary.org

Conditions for allocating PIMR balances to the policies being modeled are provided in Section 7.D.7. Some
actuaries will use the same approach to allocate the PIMR that is used for cash flow testing. Simplified
approaches are explicitly allowed if the impact of the PIMR on the minimum reserve is minimal.

Q 11.21: How will the actuary model negative assets when they arise in the projections?

A: There are several ways actuaries will model negative assets or “borrowing.” Some actuaries may assume
that the borrowing takes place using their assumed reinvestment strategy (particularly if a model segment is
going to be aggregated with other model segments). Other actuaries may assume that the negative assets will
come from the surplus portfolio or that the company will borrow and consider the company’s ability to secure
external or internal financing during situations that are representative of stressful economic “tail” scenarios.

Q 11.22: What approaches are acceptable to model asset disinvestments?

A: Section 7.E.2 of VM-20 requires that negative cash flows be modeled in a manner that is consistent with
the company’s investment policy, and that reflects the company’s borrowing cost, if applicable.

Spreads used in calculating the market value of assets sold are to be consistent with but not necessarily the
same as the spreads used for purchases, recognizing that specific starting assets may have different
characteristics than the modeled reinvestment assets. However, buying negative assets or other techniques for
modeling disinvestment are not prohibited.

Some actuaries will model negative cash flows by selling assets. Other actuaries will model buying negative
assets (i.e., borrow money) that reflects the company’s borrowing cost.

Q 11.23: What approach is used to model asset defaults?

A: As described in section 9.F.1, a three-step process is required to determine default costs for fixed-income
assets with an NAIC designation or NAIC commercial mortgage designation (i.e., corporate bonds, preferred
stocks, residential mortgage-backed securities (RMBS), commercial mortgage-backed securities (CMBS), and
commercial mortgages). Total annual default cost factors are calculated by projection year for each asset, and
are expressed as a percentage of the statement value each year. The default cost factors are the sum of three
components:

• Baseline annual default cost factor. Cost factors are based on historical corporate default and
recovery experience and include a margin for conservatism. Factor is a table “look-up” based on a
“PBR credit rating” and weighted average life (WAL) of the asset.

• Spread related factor. Based on the corporate bond spread environment as of the valuation date.
Adjustment can be positive or negative, and grades off over three years (subject to a floor and a cap).

• Maximum net spread adjustment factor. Portfolio-wide upward adjustments, graded off over three
years, if the net spread of the portfolio exceeds the net spread of a “regulatory threshold” index bond.

A different prescribed methodology applies to fixed-income assets that lack an NAIC designation. This may
include but is not limited to, some commercial mortgage loans, CMBS, RMBS, and residential mortgages
(whole loans). Section 9.F.5 provides the requirements for default cost factors for such assets without an
NAIC designation.

American Academy of Actuaries 55 www.actuary.org

Q 11:24: What company inputs are needed to determine the prescribed asset default rates for assets with an
NAIC designation?

A: Per Section 9.F.2, there are three items that need to be provided from company records:

• Investment expense assumption for each asset type.

o This is the company’s anticipated experience assumptions (with no margin).
o Expressed as an annual percentage of statement value.

• Option adjusted spread (OAS) for each asset.

o Average spread over zero coupon Treasury bonds that equates a bond’s market price as of the
valuation date with its modeled cash flows across an arbitrage free set of stochastic interest rate
scenarios.
o For floating rate bonds, the OAS equals the equivalent spread over Treasuries if the bonds were
swapped to a fixed rate.
o VM-20 allows for market conventions and other approximations.

• Weighted Average Life (WAL).

o WAL equals the weighted average number of years until 100 percent of the outstanding principal
is expected to be repaid (rounding to the nearest whole number, but not less than 1).
o For bonds or preferred stocks that are perpetual or mature after 30 years, the WAL shall be 30.
o VM-20 allows for market conventions and other approximations.
o WAL as commonly understood by investment professionals is the weighted average of the times
of the principal repayments, weighted by the amount of principal repayment; it’s the weighted
average time until a dollar of principal is repaid.

Q 11.25: How is the PBR credit rating used to determine asset default rates?

A: Section 9.F.3 summarizes the determination of PBR credit rating to calculate asset default rates. VM-20
uses a numeric rating system from 1 to 21 to determine the credit quality of each asset. Table K, referenced in
Appendix 2 in VM-20 and found on the Related Documents tab of the LATF page of the NAIC website
(https://naic.org/cmte_a_latf.htm), converts the ratings of NAIC Approved Ratings Organizations (AROs),
NAIC designations, and NAIC commercial mortgage designations to this numeric rating system. A rating of
21 applies for any ratings of lower quality than those shown in the table.

The 1-21 PBR credit rating system attempts to provide a more granular assessment of credit risk than has been
used for establishing NAIC designations for risk-based capital (RBC) and asset valuation reserve (AVR)
purposes. Unlike for RBC and AVR, the VM-20 reserve cash flow models start with the gross yield of each
asset and make deductions for asset default costs. The portion of the yield represented by the purchase spread
over Treasuries is often commensurate with the more granular rating assigned, such as A+ or A-. Thus, use of
the PBR credit rating system may provide a better match of risk and return for an overall portfolio in the
calculation of VM-20 reserves. However, for assets that have an NAIC designation that does not rely directly
on ARO ratings, a more granular assessment consistent with the designation approach is not currently
available.

American Academy of Actuaries 56 www.actuary.org

For an asset with an NAIC designation that is derived solely by reference to underlying ARO ratings without
adjustment, the PBR credit rating is determined by taking the average of the numeric ratings corresponding to
each available ARO rating, rounded to the nearest whole number. Corporate bonds are in this category.

For example, for a bond with an NAIC designation of 1 that is rated A+ by S&P and A3 by Moody’s, the
PBR rating is 6, determined as follows:

S&P rating of A+ → PBR numeric rating of 5 (from Table K)

Moody’s rating of A3 → PBR numeric rating of 7 (from Table K)

PBR rating = average of 5 and 7 = 6

For a bond with an NAIC designation of 5 that is rated CCC by S&P and Caa3 by Moody’s, the PBR rating is
19, determined as follows:

S&P rating of CCC → PBR numeric rating of 18 (from Table K)

Moody’s rating of Caa3 → PBR numeric rating of 19 (from Table K)

PBR rating = average of 18 and 19 = 18.5 = 19 (rounded)

For an asset with an NAIC designation that is not derived solely by reference to underlying ARO ratings
without adjustment, the PBR credit rating is determined by taking the second least favorable numeric rating
associated with the NAIC designation. Non-agency RMBS and traditional private placements fall into this
category. For example, for a private placement designated NAIC 1, there are seven potential PBR ratings
shown in Table K, 1 through 7. So the PBR rating is the second least favorable, that is, a PBR rating of 6.

For commercial mortgage assets with an NAIC commercial mortgage designation, Table K is used to look up
the corresponding PBR credit rating.

Once the PBR credit rating is determined for each asset, the baseline default cost is a table “look-up” based on
the PBR credit rating and the WAL of the asset.

Q 11.26: What is the purpose of the spread related factor, and how is it used to determine asset default
rates?

A: The purpose is to incorporate in the default cost the impact of current market spreads (i.e., spreads over
Treasuries) on the valuation date, based on the premise that a portion of the difference between the current
spread and historical spreads represents near-term expected default experience (could be a positive or a
negative adjustment).

A floor and cap have been provided to assure that this component cannot reduce the total default cost factor in
year one by more than the baseline default cost factor and cannot increase the total default cost factor in year
one by more than two times the baseline default cost factor. The cap is intended to help limit reserve
volatility, which remains a concern for both the spread related factor and the maximum net spread adjustment
factor. The spread related factor grades linearly from the prescribed amount in year one to zero in years four
and after.

American Academy of Actuaries 57 www.actuary.org

Per Section 9.F.1.b, the prescribed amount in year one may be positive or negative and is calculated as
follows:

• Multiply 25% by the result of (a) minus (b), where:

(a) equals the current market benchmark spread published by the NAIC consistent with the
PBR credit rating and WAL of the asset on the valuation date (see Tables F and G on the
Related Documents tab of the LATF page of the NAIC website
(https://naic.org/cmte_a_latf.htmwww.naic.org), updated by NAIC on a monthly basis).

(b) equals the most current available historical mean benchmark spread published by the NAIC
consistent with the PBR credit rating and WAL of the asset on the valuation date (see Tables
H and I on the Related Documents tab of the LATF page of the NAIC website
(https://naic.org/cmte_a_latf.htmwww.naic.org), updated by NAIC on a quarterly basis)

• The resulting amount shall not be less than the negative of the baseline annual default cost in year
one and shall not be greater than 2 times the baseline annual default cost in year one.

Q 11.27: What is the maximum net spread adjustment and how it is used to determine asset default rates?

A: The maximum net spread adjustment increases the required assumed default costs in the VM-20
calculations if the average credit quality of the company’s asset portfolio falls below the targeted threshold.
The targeted threshold is a Baa2/BBB Bond (NAIC 2, PBR credit rating of 9).

Per Section 9.F.1.c, for each model segment, a comparison is to be made of two spread amounts, both being
net of the default costs calculated thus far and net of investment expenses. In each case, the gross option
adjusted spread is based on current market prices at the valuation date.

• The first spread amount in the comparison represents the weighted average net spread of all the assets
in the model segment, as if all the assets were purchased at their current market spreads.
• The second spread amount in the comparison represents the net spread for a portfolio of index
Baa2/BBB bonds (NAIC 2, PBR credit rating of 9), as if the index Baa2/BBB portfolio were
purchased at the current average market spread.

If the first result is higher than the second, additional default costs must be added to each asset until the two
results are equal for the first projection year. This additional amount of default cost on each asset then grades
off linearly in the model until it reaches zero in year four and after. This process is repeated each actual
valuation date.

A company that invests in an asset mix earning an average gross spread greater than Baa2/BBB bonds
initially, or an asset mix whose average market spread could widen significantly relative to market spreads for
Baa2/BBB bonds, are examples of situations likely to trigger additional assumed default costs either initially
or in the future.

The calculation is as follows:

The prescribed amount in year one shall be the excess, if any, of (a) over (b):

American Academy of Actuaries 58 www.actuary.org

 (a) Weighted average net spread for the asset portfolio, calculated as follows:

1. For each asset, calculate a preliminary year one net spread equal to the option adjusted spread
of the asset on the valuation date less the sum of the amounts from Sections 9.F.1.a and
9.F.1.b (baseline and spread related factor) and less the investment expense for the asset.

2. Calculate a weighted average preliminary year one net spread for the total asset portfolio (i.e.,
assets subject to this section) using weights equal to each asset’s statement value on the
valuation date multiplied by the lesser of three years and the asset’s WAL on the valuation
date.

(b) The net spread for a “hypothetical asset,” which is called the “regulatory threshold asset.”

The regulatory threshold asset is determined as follows:

• Calculate the preliminary year one net spread for a hypothetical asset with the following assumed
characteristics (the regulatory threshold asset):

o A PBR credit rating of 9 (equivalent to Baa2/BBB).

o A WAL equal to the average WAL on the valuation date for the assets in the portfolio.
o An option adjusted spread equal to the current market benchmark spread published by the
NAIC for the assumed PBR credit rating and WAL (see Table F on the Related
Documents tab of the LATF page of the NAIC website
(www.naic.orghttps://naic.org/cmte_a_latf.htm), updated by NAIC on a monthly basis).
o Investment expense of 0.10 percent.

• The preliminary year one net spread is equal to the option adjusted spread of the hypothetical asset
on the valuation date less the sum of the baseline and spread related factor for the hypothetical
asset, and less the investment expense for the hypothetical asset.

Q 11.28: What are the default rates required in the VM-20 calculation for commercial mortgages, CMBS
and RMBS investments, and residential whole loans?

A: If the asset does not have an NAIC designation or an NAIC commercial mortgage designation, per Section
9.F.5, a prescribed default assumption is established such that the net yield shall be capped at 104 percent of
the applicable historical U.S. Treasury yield rate most closely coinciding with the purchase date and maturity
structure, plus 25 basis points. If the assets have an NAIC designation or an NAIC commercial mortgage
designation, their default assumptions are determined according to Sections 9.F.3 and 9.F.4.

Q 11.29: If a company is using nontraditional assets (e.g., letters of credit, contingent notes, etc.) as part of
a reserve financing arrangement, how are these nontraditional assets handled in the VM-20 calculation?

A: Some actuaries will include the nontraditional assets in starting assets if such assets are treated as admitted
assets on the balance sheet of the company. Because the nontraditional asset is backing a portion of the
statutory reserve, some actuaries would conclude that it is appropriate to include the nontraditional asset in
starting assets, even though these assets are not income-producing assets. In addition, some actuaries would
also include the fee paid to the nontraditional asset provider as an expense in the future cash flows when

American Academy of Actuaries 59 www.actuary.org

performing the minimum reserve calculation, because it is an expense that is directly related to the underlying
policy.

Actuarial Guideline XLVIII has limited the use of such nontraditional assets in some ways. Refer
to requirements found in Actuarial Guideline XLVIII for more information.

12. Details on Scenarios / Scenario Generators / Economic Assumptions

Q 12.1: What economic assumptions are stochastically generated?

A: The prescribed economic scenario generator produces the following economic parameters:

• Treasury rates at maturities of 0.25, 0.50, 1, 2, 3, 5, 7, 10, 20, and 30 years. (Filename: UST)
• The following equity return indices, with file names shown in parenthesis
o Aggressive or specialized equity (AGGR)
o Diversified international equity (INT)
o Diversified large-cap U.S. equity (US)
o Intermediate-risk equity (SMALL)
o Money market / Short term (MONEY)
o Intermediate-term U.S. government bond (INTGOV)
o U.S. long-term corporate bonds (LTCORP)
o Diversified fixed income (FIXED)
o Diversified balanced (BALANCED)

Treasury rates may be generated either as bond equivalent rates or annual effective rates.

Equity return indices are total returns, with one row for each scenario and one column for each time period.

Actuaries may or may not choose to generate other measures stochastically, including inflation rates, or
currency exchange rates, or dividend and volatility rates for the equity return indices. Consideration should be
given to the internal consistency of other stochastic quantities with the scenario data.

Q 12.2: How would actuaries generate the scenarios for the VM-20 calculations?

A: Section 7.G.2 requires the actuary to use (i) U.S. Treasury interest rate curves following the prescribed
economic scenario generator with prescribed parameters, as described in Appendix 1; and (ii) total investment
return paths for general account equity assets and separate account fund performance generated from a
prescribed economic generator with prescribed parameters, as described in Appendix 1. In Appendix 1, the
link to the prescribed economic scenario generator website is provided: https://www.soa.org/tables-calcs-
tools/research-scenario/.

The generator is updated periodically, and users would be prudent to verify that they have the current version
of the generator prior to use. In addition to the prescribed scenario generator, the website also contains
relevant documentation on how to use the scenario generator, including options for the main Scenario
Generator tab, and how to maintain the Historical Curves tab by entering the last daily yield curve for each

American Academy of Actuaries 60 www.actuary.org

month as time passes using data available from the U.S. Treasury, which can be found at:
https://www.treasury.gov/resource-center/data-chart-center/interest-rates/Pages/TextView.aspx?data=yield.

Similarly, Section 7.G.1 requires using the same tool to generate the scenario used to determine the
Deterministic Reserve, and Section 6.A.2 requires using this tool to generate the Stochastic Exclusion Ratio
Test scenarios.

Q 12.3: How many scenarios should be run to determine the Stochastic Reserve?

A: The current version of VM-20 does not define the appropriate number of scenarios to run. The prescribed
scenario generator allows the user to choose the “full set of 10,000 scenarios,” as well as representative
scenario subsets of 1,000, 500, 200, and 50 scenarios. From this, some actuaries would run all 10,000
scenarios to meet the regulatory requirements under VM-20, while some actuaries would use fewer than the
10,000 scenarios. See Q 12.4 for additional details.

Q 12.4: How would the actuary determine the appropriate number of scenarios if less than the full 10,000
scenario set is used?

A: A range of methods might be appropriate to pick such a subset of the full 10,000 scenarios, including, but
not limited to, the following:

• Using the scenario picker embedded in the prescribed economic scenario generator, or
• Selecting a random subset of the scenarios, or
• Using an alternative scenario picking or reduction tool.

The actuary may wish to use caution in using the scenario picker, as it relies solely on the 20-year Treasury
rate for scenario selection. Thus, for products that are sensitive to equity performance or other maturities, the
scenario picker may not be an effective scenario reduction technique.

Section 7.G.2.c indicates that the use of fewer scenarios rather than a higher number of scenarios is
permissible as a model efficiency technique provided scenario reduction techniques may be considered
acceptable as long as:

• The smaller set of scenarios is generated using the prescribed scenario; and
• The use of the technique is consistent with Section 2.G.

Section 2.G states that a company may use simplifications, approximations and modeling efficiency
techniques provided that:
• The company can demonstrate that the use of such techniques does not understate the reserve by a
material amount, and
• The expected value of the reserve calculated using simplifications, approximations, and modeling
efficiency techniques is not less than the expected value of the reserve calculated that does not use
them.

VM-20 does not specify how to demonstrate compliance with Section 2.G. Some actuaries might follow a
process similar to the following:

1. First, tentatively select a subset of the scenarios.

American Academy of Actuaries 61 www.actuary.org

 2. Next, calculate reserves across these scenarios using the tentatively selected subset of the liabilities
and also using the entire inforce business population. If these results are similar, then the actuary
might reasonably conclude that the subset of the liabilities (i.e., the smaller or representative model) is
appropriate to use to measure the appropriateness of a scenario subset.
3. Finally, calculate reserves for the subset of the liabilities across both the reduced set of scenarios and
the full set of scenarios. If the reserves calculated for the subset of liabilities using the scenario subset
are not materially smaller than the reserves calculated for the subset of liabilities using all the
scenarios, then the actuary might reasonably conclude that the subset of scenarios is appropriate to
use.

In addition, some actuaries might conclude that if the same technique is used to pick a representative subset of
the liabilities in successive periods, then it is not necessary to revalidate the appropriateness of that liability
subset for use in validating scenario subsets. Some actuaries might conclude that it is necessary to repeat this
process only when the characteristics of the liabilities, the assets backing them, or the economic environment
have materially changed.

For randomly selected subsets of scenarios, one commonly accepted method was published by Manistre and
Hancock. 2 However, it is important to understand that this test has limitations and is appropriate only when
using randomly selected subsets of a pool of scenarios.

Q 12.5: What if there are a lower number of scenarios that results in a higher reserve than a greater
number of scenarios?

A: VM-20 does not prevent the actuary from selecting a scenario subset that produces a reserve that is greater
than what would result if the full set is used. Some actuaries would hold the lower reserve based on the larger
number of scenarios to take into account the additional credibility of the larger scenario set. Other actuaries
would hold the higher reserve based on the smaller scenario set (if in the actuary’s judgment the number of
scenarios in the smaller scenario set is reasonable for the purpose of calculating the Stochastic Reserve
requirement).

Q 12.6: Can the actuary use scenarios other than prescribed scenarios for calculating the Stochastic
Reserve?

A: Section 7.G.2 specifically requires use of the “prescribed economic scenario generator with prescribed
parameters, as described in Appendix 1.”

Q 12.7: How are separate account funds mapped into the specific equity return indices of the prescribed
economic scenario generator?

A: A Guidance Note in Section 7.G.2.b states that:

Mapping of the returns on the proxy funds to the prescribed fund’s returns is left to the judgment of
the qualified actuary to whom responsibility for this group of policies is assigned, but the returns so
generated must be consistent with the prescribed returns. This does not imply a strict functional
relationship between the model parameters for various markets/funds, but it would generally be
inappropriate to assume that a market or fund consistently “outperforms” (lower risk, higher expected
return relative to the efficient frontier) over the long term.
2
Manistre and Hancock; “Variance of the CTE Estimator”; NAAJ; Vol 9, No.2 (April 2005); pages 129-156.

American Academy of Actuaries 62 www.actuary.org

Actuaries might wish to effect modeling of funds in different ways. For example, some actuaries might
perform a regression of past fund performance against past performance of market indices consistent with the
scenario generator classifications and develop a scenario set for each such fund that is consistent with an
appropriate weight of such scenarios derived from the regression.

Alternatively, funds might be decomposed using regression techniques into funds corresponding to each of
the indices.

Q 12.8: How can the actuary obtain the scenarios used for the Stochastic Exclusion Ratio Test?

A: VM-20 Section 6.A.2 requires that the same prescribed economic scenario generator used for stochastic
scenario generation be used to generate the 16 scenarios used for the Stochastic Exclusion Ratio Test. This is
done by selecting the “Stochastic Exclusion Test Scenarios” on the Scenario Generator tab of the workbook.

Q 12.9: How can the actuary obtain the scenario used for the Deterministic Reserve?

A: VM-20 Section 4.A.1 specifies that the Deterministic Reserve is to be calculated using scenario 12 from
the set of prescribed scenarios used in the Stochastic Exclusion Ratio Test.

Q 12.10: What changes should be made to the prescribed economic stochastic scenario generator
parameters for non-U.S. economies?

A: Section 7.G.2 requires the actuary to use the prescribed economic scenario generator provided by the
American Academy of Actuaries, with prescribed parameters. That said, some actuaries might conclude that it
is unreasonable to use these parameters for non-U.S. interest rates. In a situation where non-U.S. interest rates
are required, some actuaries would deem it appropriate to:

• Use historical interest rates in that market, in the same formula used in the United States, to determine
the mean reversion target appropriate for that market; and
• Update the starting yield curve to match government bond yields in that market.

In addition, some actuaries might wish to update volatility, correlation, and other parameters to be consistent
with the market in question.

For equity returns in non-U.S. markets, the actuary might want to use techniques similar to what is used for
U.S. funds to map each fund into one or more of the funds generated by the prescribed economic scenario
generator.

The Guidance Note in Section 7.G.2.b states that it would generally be inappropriate to assume that a market
or fund consistently outperforms (lower risk, higher expected return relative to the efficient frontier) over the
long term.

Q 12.11: How should the actuary generate interest rates for maturities not generated by the economic
scenario generator?

American Academy of Actuaries 63 www.actuary.org

A: The prescribed economic scenario generator only generates interest rates for selected terms to maturity.
For some purposes, the actuary might need a full set of interest rates at other maturities. A range of options
are available for this. VM-20 gives no guidance.

Techniques some actuaries might find reasonable could include, but would not be limited to, the following:

• Linear interpolation;
• Bootstrapping the curve to find forward rates that increase linearly between the given maturity points
and are consistent with the given rates;
• Using the Nelson-Siegel parameters and formulas documented in the scenario generator release notes;
or
• Cubic spline interpolation.

13. Setting Prudent Estimate and Anticipated Experience Assumptions

Q 13.1: How are prudent estimate assumptions determined?

A: VM-01 defines margins and prudent estimate assumptions, where prudent estimate assumptions equal
anticipated experience assumptions plus a margin. The margin may increase or decrease the assumption as
appropriate to cover adverse deviations and estimation error. The margin should result in a larger reserve than
would otherwise result without it. There is no requirement in VM-20 to add a margin to assumptions that are
stochastically modeled because the applicable CTE measure provides a margin.

VM-20 requires a margin for each material risk factor. If sensitivity testing shows that the risk factor does not
have a material impact on the reserve, some actuaries will establish the prudent estimate assumption for this
risk factor without separately identifying a margin.

Section 4.3 of ASOP No. 1, Introductory Actuarial Standard of Practice, states: “Actuaries are responsible
for determining which ASOPs apply to the task at hand.” Relevant actuarial standards of practice may
include, but are not limited to, the following:

ASOP No. 2 Nonguaranteed Charges or Benefits for Life Insurance Policies and Annuity Contracts
ASOP No. 7 Analysis of Life, Health, or Property/Casualty Insurer Cash Flows
ASOP No. 11 Financial Statement Treatment of Reinsurance Transactions Involving Life or Health
Insurance
ASOP No. 15 Dividends for Individual Participating Life Insurance, Annuities, and Disability Insurance
ASOP No. 22 Statements of Opinion Based on Asset Adequacy Analysis by Actuaries for Life and Health
Insurers
ASOP No. 23 Data Quality
ASOP No. 25 Credibility Procedures
ASOP No. 52 Principle-Based Reserves for Life Products under the NAIC Valuation Manual

Q 13.2: Are the same prudent estimate assumptions used for deterministic and stochastic calculations?

American Academy of Actuaries 64 www.actuary.org

A: Section 9.A.5 of VM-20 requires prudent estimate assumptions to be consistent for the two calculations.
Some actuaries would use the same assumptions except where the risk factor is scenario-dependent.

Q 13.3: Which risk factors do not require a prudent estimate assumption?

A: According to Sections 9.A.1 and 9.A.4, the company is not required to develop prudent estimates for
assumptions that are prescribed, or stochastically modeled. Per Section 9.A.3, risk factors that must be
stochastically modeled, and therefore do not require prudent estimate assumptions, include interest rate
movements (i.e., Treasury interest rate curves) and equity performance.

Some actuaries might not develop a prudent estimate assumption for risk factors that do not materially affect
modeled reserves as permitted by Section 9.B.4. In this case, anticipated experience assumptions should be
used pursuant to Section 9.A.6.

Q 13.4: How will actuaries set anticipated experience assumptions?

A: Per Section 9.A.6, the actuary shall use company experience, if relevant and credible, to establish the
anticipated experience assumption for any risk factor. To the extent the company experience is not available
or credible, the actuary may use industry experience or other data to establish the anticipated experience
assumption, making modifications as needed to reflect the circumstances of the company.

Some actuaries may develop anticipated experience assumptions based on the assumptions used for pricing
and development of the product. Those actuaries may wish to consider whether it is necessary to work closely
with the product development actuary to understand the basis for assumptions and review the studies
performed prior to using them in the reserve calculations to assure they meet the criteria set in Section 9.A.6.

Q 13.5: Can actuaries use company experience if it predates the effective date of VM-20?

A: Some actuaries will use company experience that predates the effective date of VM-20 if it is relevant and
credible for the assumption or margin that is being determined. There is no requirement in VM-20 that the
experience data used to determine anticipated experience assumptions and/or margins for prudent estimate
assumptions has to occur after the effective date of VM-20.

Q 13.6: When will the actuary update the anticipated experience assumptions?

A: Section 9.A.6.d requires the actuary to annually review relevant emerging experience, and Section 9.A.2
requires that the assumptions be periodically reviewed and updated. Therefore, it is expected that actuaries
will update assumptions if there is new experience data or other information that changes the actuary’s
expectation of future experience. Actuaries will use materiality on testing results, company practice for
updating assumptions, and the credibility of the new data and other information in determining whether to
update the experience assumptions. Experience assumptions are not locked in at issue.

Q 13.7: What data sources may be used to set anticipated experience assumptions?

A: Per Section 9.A.6, the company shall use its own experience, if relevant and credible, to set anticipated
experience assumptions. In instances where company-specific data is not available or credible, actuaries may
wish to consider looking to alternative sources as discussed in Section 9.A.6.a, Section 9.A.6.b, and Section
9.A.6.c.

American Academy of Actuaries 65 www.actuary.org

A non-exhaustive list of possible data sources to consider for developing anticipated experience assumptions
is listed below:

• Company data on the same or similar products

• Industry or reinsurer data on the same or similar products
• General population data
• Predictive models or algorithms
• Sound actuarial judgment, if risk factors have limited or no experience, or other applicable data
available

For assumptions that are established based on limited data, Section 9.A.6.d requires the actuary to sensitivity
test the assumption chosen to ensure the assumption is set at the conservative end of the plausible range.

Q 13.8: What other issues would the actuary take into account when setting anticipated experience
assumptions?

A: The actuary may wish to consider whether anticipated experience assumptions developed from the data
should be modified as needed to reflect the circumstances of the company. For example, if the data from
alternative sources includes mortality experience based on three underwriting classes, and the company has
just two underwriting classes, then adjustments to the assumptions may be needed to make them appropriate
for the company’s underwriting classes.

The actuary may wish to consider adjusting assumptions that are based on historical experience to consider
contractual guarantees that are present in the contracts being valued that were not present in the contracts that
contributed to the experience base. The actuary may also wish to consider the possibility of antiselection
affecting assumptions. For example, antiselection may involve a combination of lapses, persistency, mortality,
and the level of guarantees.

The actuary may also wish to consider reviewing contractual guarantees to determine whether and to what
degree these contractual guarantees will impact future cash flows of the model.

Q 13.9: When might it be appropriate to represent extreme or catastrophic behavior?

A: Section 9.D.2.c states that the company is not required to model extreme or “catastrophic” forms of
behavior in the absence of evidence to the contrary.

Q 13.10: How should trends in data affect the anticipated experience assumption?

A: Some actuaries might use the smallest breakdown of data that is available and credible over time for
comparison to experience of other exposure periods. However, judgment is needed here, as the actuary has to
balance precision against credibility, considering the practical ability to do the necessary studies. An example
would be in analyzing lapse data for a certain product over calendar years 2000–2004 to the same grouping
over calendar years 2005–2009.

Some actuaries might project trends that in their judgment are likely to be sustained in the future in the
anticipated experience assumption, subject to any applicable restrictions. Some actuaries may make a
judgment on the uncertainty surrounding the projection of the trend and increase the margin as uncertainty
increases, such as in later durations.

American Academy of Actuaries 66 www.actuary.org

In Section 9.A.6.d, the last paragraph states that “the qualified actuary, to whom responsibility for this group
of policies is assigned, shall annually review relevant emerging experience for the purpose of assessing the
appropriateness of the anticipated experience assumption. If the results of statistical or other testing indicate
that previously anticipated experience for a given factor is inadequate, then the qualified actuary shall set a
new, adequate, anticipated experience assumption for the factor.”

Q 13.11: How does the actuary determine whether company experience data is credible?

A: A discussion of credibility in the context of the mortality assumption is found in Section 15 of this practice
note. ASOP No. 25, Credibility Procedures, is available on the Actuarial Standards Board’s website. To make
a determination on the level of credibility that a set of company experience data may have, some actuaries
might use concepts from classical credibility theory. In this sense, the actuary may determine that full
credibility (i.e., the subject experience will have full predictive value) will be established when enough
observations of an event occur so that the actual result of its frequency will be within a defined percentage of
the expected results with a specified probability. Additional adjustments to the definition of full credibility
may be used in cases where the actual observed events can range in their magnitude of severity. The actuary
may wish to review the Credibility practice note. 3

If the number of observations is fewer than the amount needed for full credibility, partial credibility of the
data may be established. This can be done by using information such as the expected number of observations
and the number of observations needed for full credibility.

Some actuaries with data that is only partially credible might use actuarial judgment to combine partially
credible data with other industry experience to determine their anticipated experience assumptions.

Q 13.12: Should the actuary perform sensitivity testing to set anticipated experience assumptions or
prudent estimate assumptions?

A: Section 9.A.7 requires sensitivity testing to be performed to understand the materiality of prudent estimate
assumptions on the minimum reserve. The sensitivity testing shall be updated periodically and may be
performed less frequently when the tests show less sensitivity of the minimum reserve to changes in the
assumptions being tested or the experience is not changing rapidly. Some actuaries might perform the
sensitivity testing using anticipated experience assumptions. Other actuaries might instead perform the
sensitivity testing using prudent estimate assumptions. If the sensitivity testing shows that the assumption is
material to the minimum reserve level, Section 9.B.3 says that greater analysis and more detailed justification
are needed to determine the level of uncertainty when establishing margins for risk factors that produce
greater sensitivity on the modeled reserve. Therefore, it may be prudent for the actuary to review the
credibility of the data used to set the assumption and whether the assumption is set at the appropriate level.

14. Setting Margins

Q 14.1: What are the VM-20 requirements when determining appropriate margins?

3
Available on the Academy’s website at
http://www.actuary.org/files/publications/Practice_note_on_applying_credibility_theory_july2008.pdf.

American Academy of Actuaries 67 www.actuary.org

A: Section 9.B states that assumptions that are neither stochastically determined nor prescribed should
incorporate appropriate margins in the prudent estimate assumption for adverse deviation and estimation
error.

The actuary is required to set an explicit margin for each material assumption independently but, if applicable,
the actuary may apply an adjustment to take in to account the fact that risk factors are not normally 100%
correlated. A margin is not required for assumptions that do not have a material impact on the minimum
reserve.

Section 9.B.2 states that higher margins should be used when:

• The experience data has less relevance or lower credibility;
• The experience data are of lower quality, such as incomplete, internally inconsistent, or not current;
• There is doubt about the reliability of the anticipated experience assumption, such as, but not limited
to, recent changes in circumstances or changes in company policies; or
• There are constraints in the modeling that limit an effective reflection of the risk factor.

Greater analysis and more detailed justification are required when establishing margins for risk factors that
produce greater sensitivity on the modeled reserve.

Section 9.B.5 states that in setting the margin for a risk factor, the actuary must consider the magnitude of
fluctuations in the historical experience of the company for that risk factor, as appropriate.

Section 9.B.6 states that the company shall apply the method used to determine the margin consistently on
each valuation date, but is permitted to change the method from the prior year if the rationale for the change
and the impact on modeled reserve is disclosed.

Sections 9.C.5 and 9.D.3 provide specific requirements for setting margins for mortality and policyholder
behavior, respectively.

Q 14.2: What other references are available when determining appropriate margins?

A: Some actuaries may look to the following:

• A Society of Actuaries research paper titled Analysis of Methods for Determining Margins for
Uncertainty Under a Principle-Based Framework for Life Insurance and Annuity Products (March
31, 2009).
• ASOP No. 52, Principle-Based Reserves for Life Products under the NAIC Valuation Manual,
adopted September 2017.
• The C3 Phase III Report and associated practice note.
• The AG-43 / C3 Phase II practice note.
• Internal documentation of margins used for other financial reporting methods, for example, U.S.
GAAP FAS 60 margins.
• Various Canadian valuation technique papers and educational notes
• ASOP No. 25, Credibility Procedures

Q 14.3: What general method should be used in setting margins?

American Academy of Actuaries 68 www.actuary.org

A: Some actuaries will focus on the key assumptions for the underlying product first and then analyze how
those margins affect other assumptions. For example, for an annually renewable term product, the actuary
may initially focus on setting the anticipated mortality experience and margin assumptions (because it is a
principal risk) and then set other assumptions such as lapse rates and expenses such that these assumptions are
internally consistent with the mortality assumption. One advantage of concentrating on the key assumptions is
avoiding the situation where the application of a margin to one assumption affects the direction of the
appropriate margin to apply to another assumption (e.g., a high mortality margin may cause lower rather than
higher lapses to be conservative).

Some actuaries would develop an approach to setting margins in a manner consistent with the underlying
principle of reflecting the underlying risk characteristics of the product.

Q 14.4: When can and how does the actuary apply an adjustment in setting the margins to account for risk
factors not being 100% correlated?

A: Per Section 9.B.1, the initial level of a particular margin may be adjusted to take into account the fact that
risk factors are not normally 100 percent correlated. The initially determined margin may only be reduced to
the extent the company can demonstrate that the method used to justify such a reduction is reasonable
considering the range of scenarios contributing to the CTE calculation, recognizing that risk factors may
become more correlated in adverse circumstances, or considering the scenario used to calculate the
Deterministic Reserve as applicable or considering appropriate adverse circumstances for risk factors not
stochastically modeled.

Industry practice determining covariance adjustments is expected to evolve over time.

Q 14.5: Does the sensitivity of an assumption’s impact on the overall reserve level affect how the margin is
set?

A: In setting the margins in the calculation, Section 9.B.5 states that the actuary must consider the magnitude
of fluctuation in the historical experience of the company for the risk factor. One approach to determining the
uncertainty of a risk factor is to measure the standard deviation around the mean or other standard statistical
measure (if meaningful historical experience data is available for the risk factor).

Section 9.B.2 lists some considerations that would lead to a larger margin. They are when:

• The experience data has less relevance or lower credibility.

• The experience data is of lower quality, such as incomplete, internally inconsistent, or not current.
• There is doubt about the reliability of the anticipated experience assumption, such as, but not limited
to, recent changes in circumstances or changes in company policies.
• There are constraints in the modeling that limit an effective reflection of the risk factor.

Some actuaries will not include a larger margin for risk factors having a material impact on the reserves but
little uncertainty. For example, if an expense assumption has a large impact on the reserve but is known with
certainty (has been outsourced for a fixed contractual cost), then a smaller margin could be appropriate.

Q 14.6: Are there instances where it may be appropriate to use no margin?

American Academy of Actuaries 69 www.actuary.org

A: Section 9.B.4 indicates that a margin is not required for an assumption when variations in the
assumption do not have a material impact on the modeled reserve.

In addition, there may be instances where the anticipated experience assumption (before margin)
produces the most conservative result. In these instances, where either increasing or decreasing
the anticipated experience assumption (before margin) would produce a less conservative result,
some actuaries may find it appropriate to use no margin. An example of this situation might be
the anticipated assumption for ULSG paid premium timing and/or amount.

Q 14.7: Are there additional considerations for setting margins for specific assumptions?

A: The specific considerations for mortality, policyholder behavior, and expense margins are discussed in the
respective sections of this practice note covering those items.

Q 14.8: Can the actuary use Canadian prescribed margins?

A: Some actuaries may use Canadian prescribed margins, but VM-20 does not list them as a “safe harbor.”
The actuary would need to make a determination of their appropriateness for use in the reserve calculation as
set forth in VM-20.

Q 14.9: To what extent should the size of the company affect the size of the margins used?

A: Some actuaries would not modify the size of the margins based solely on the size of the company. For
example, smaller margins should generally coincide with a reduction in uncertainty that might be due to a
large number of observations, but is not necessarily dependent on company size because a large number of
observations could be generated over time from a smaller company.

Q 14.10: What are the considerations for setting margins across blocks of business?

A: Some actuaries would assess whether the margins were appropriate by product as well as adequate over a
broad set of policies. This assessment would be particularly useful for products with uncertain risks or
particularly unique characteristics. Some actuaries would review the margins in aggregate, particularly where
policy characteristics are similar or have homogeneous risks.

Q 14.11: How should margins be set when the impact of assumption movements changes over the duration
of the business?

A: For some products, the impact on the overall reserve of increasing or decreasing an assumption may vary
according to the duration of the business. Because a margin is intended to add conservatism, in the case where
the impact of the margin varies by duration, some actuaries would set a margin that changes based on policy
duration. However, some actuaries would set a specific margin on an assumption that complies with Section
9.B requirements overall. In other words, the margin increases the overall reserve even if the assumption may
not be conservative for every policy duration.

Where there is a clear change in the policy, such as a dramatic change in premiums or surrender charges,
some actuaries will develop a margin that differs before and after this point.

Q 14.12: How are margins determined for dynamic assumptions?

American Academy of Actuaries 70 www.actuary.org

A: As implied by Sections 9.A.3 and 9.A.4, where an assumption is dependent on interest rates or equity
returns and a dynamic formula is included in the modeling, some actuaries may not add an additional margin
to the calculation, on the basis that conservatism is provided by the conservatism inherent in the tail measure
(i.e., CTE) and in their judgment this implicit margin would satisfy the requirements of Section 9.B.

However, other actuaries may add additional conservatism, as they might feel that the use of the tail measure
will only inject conservatism regarding the interest rate or equity risk, but not necessarily the dynamically
related risk, which they may see as a distinct risk. Some of these actuaries might add conservatism by making
the dynamic formula slightly more or less dynamic (depending on what would be more conservative) than
anticipated.

Some actuaries might consider margins on the base underlying assumption to be applicable to the resulting
assumption including dynamic components, so would not add an additional margin.

Q 14.13: How often should margins be updated?

A: Some actuaries would use consistent margins or the same margins from one reporting date to the next
unless there is a particular reason the actuary believes the credibility, quality, or reliability of the assumption
has changed. Other actuaries may review and update margins, where necessary, on a regular basis every time
they perform the calculation.

15. Setting Mortality Assumptions

Note to the reader: The 2019 and 2020 versions of the Valuation Manual are expected to include
changes to the VM-20 requirements for setting mortality assumptions. For valuations after Dec. 31,
2018, please consult the most recent version of the Valuation Manual, which may include
requirements not contemplated in this practice note.

Q 15.1: How are the applicable industry basic tables chosen?

A: A company may apply the underwriting criteria scoring procedure described in Section 9.C.3.d to choose
the industry basic tables. However, in determining the applicable industry basic table, a company must take
into account factors not recognized in the underwriting scoring algorithm that may be applicable. To reflect
factors not recognized in the underwriting scoring algorithm, the company may, to the extent it can justify,
adjust industry basic tables up or down two tables from that determined by the underwriting scoring
algorithm, as noted in Section 9.C.3.d.iii. As an alternative, a company may use other actuarially sound
methods to determine the applicable basic tables.

Q 15.2. In what circumstances is a modified industry basic table permitted?

The actuary may wish to consider whether the industry basic table developed pursuant to Section 9.C.3.a
should be modified as needed to reflect the circumstances of the company. This is only allowed in a limited
number of situations as discussed in Section 9.C.3.b:

American Academy of Actuaries 71 www.actuary.org

 A modified industry basic table is permitted in a limited number of situations where an industry basic
table does not appropriately reflect the expected mortality experience, such as joint life mortality,
simplified underwriting, or substandard or rated lives. In cases other than modification of the table to
reflect joint life mortality, the modification must not result in mortality rates lower than those in the
industry table without approval by the commissioner.

Q 15.3. For what purposes may the underwriting criteria scoring procedure be used?

A: The underwriting criteria scoring procedure may be used to determine the industry basic table that can
serve as industry experience rates. These can be used when company experience data is limited or not
available as well as for grading from company experience to industry experience mortality.

Q 15.4. How are risk classes scored under the underwriting criteria scoring procedure?

A: The underwriting criteria scoring procedure is the algorithm embedded in the Underwriting Criteria Score
Calculator adopted by the Life Actuarial (A) Task Force and maintained on the Society of Actuaries website. 4
The tool found at this site is called the “Relative Risk Tool (RR Tool).” Risk classes are scored based on the
specific underwriting criteria used by a company.

Q 15.5. How should the actuary take into consideration factors not recognized in the underwriting criteria
scoring algorithm?

A: Such factors can be taken into account to the extent these can be justified by adjusting the industry tables
up or down two tables from that determined by the application of the underwriting criteria scoring procedures.
Further adjustments to reflect risk characteristics not captured by the Underwriting Criteria Scoring Tool may
be allowed upon approval by the commissioner.

Q 15.6. Can the underwriting criteria scoring procedure be used to determine the applicable industry basic
tables for policies with simplified or no underwriting?

A: The Underwriting Criteria Scoring Tool is applicable only to underwriting systems with a preferred class
structure, and not to underwriting systems utilizing simplified issue or guaranteed issue methodologies.
Alternative methods are allowed. Section 9.3.C.e of VM-20 states that the company shall document the
analysis performed to demonstrate the applicability of the chosen method and resulting choice in tables
and reasons why the results using the Underwriting Criteria Scoring Tool may not be suitable. (For these
programs, the company would develop an alternative method for determining a prudent estimate for mortality
and might consult with the commissioner prior to implementation.)

Q 15.7. How should the applicable industry basic table be determined for policies that were rated
substandard but issued as standard?

A: Some actuaries might use the results of the underwriting criteria scoring tool for the rated substandard
class to determine the applicable industry basic table. Other actuaries might use the results of the underwriting
criteria scoring tool and adjust upward by up to two tables to the extent this can be justified.

4
https://www.soa.org/research/topics/indiv-val-exp-study-list/.

American Academy of Actuaries 72 www.actuary.org

Q 15.8. What procedure should be used if the actuary believes it is appropriate to make a greater
adjustment?

A: If adjustments greater than two tables up or down are felt to be appropriate, Section 9.C.3.d.iii indicates the
actuary must obtain the approval of the commissioner.

Q 15.9. What alternatives may be used to the Underwriting Criteria Scoring Tool and what is required to
use such alternatives?

A: Some companies may use other actuarially sound methods to determine the applicable basic tables. The
company must document the analysis performed to demonstrate the applicability of the chosen method and
the resulting choice in tables and reasons why the results using the Underwriting Criteria Scoring Tool may
not be suitable.

Q 15.10. What should be done if no industry basic table appropriately reflects the risk characteristics of the
mortality segment?

A: Per Section 9.C.3.f, if there is no appropriate basic industry table, a company may use any well-established
industry table that is based on the experience of policies having the appropriate risk characteristics for the
mortality segment. This might be the case when the policies involve group-type mortality; there are possibly
other situations.

Q 15.11: When will 2017 CSO become the prevailing mortality table for tax reserves under IRC 807(c)?

A mortality table becomes prevailing for tax reserves when it is the most recent table permitted to be used in
computing reserves by at least 26 states as of the date of issue of the contract. Because the Standard Valuation
Law, which includes the Valuation Manual, has been approved by at least 26 states as of Jan. 1, 2017, the
mortality tables included in VM-M will be prevailing as of Jan. 1, 2017, for use in tax reserves.

This means that the 2017 CSO is the prevailing mortality table for tax reserves as of Jan. 1, 2017.

However, the tax law provides for a transition period. If the prevailing commissioners’ standard mortality
table as of the beginning of any calendar year (e.g., year of change) is different from the prevailing
commissioners’ standard mortality table as of the beginning of the preceding calendar year, the insurer may
use the former table as of Jan. 1 of the “year of change” for any contract issued after the change and for a
three-year period beginning on Jan. 1 of the “year of change.” The “year of change” is the calendar year in
which the prevailing commissioners’ standard mortality table, as of the beginning of the year, is different
from the prevailing commissioners’ standard table as of the beginning of the preceding year.

Thus, because the 2017 CSO table became prevailing as of Jan. 1, 2017, for policies issued as of Jan. 1, 2017,
and later, a company may elect to use 2001 CSO for policies issued from Jan. 1, 2017, through Dec. 31, 2019
(three-year period beginning on Jan. 1 of the “year of change”). It would then need to use 2017 CSO for tax
purposes beginning with policies issued Jan. 1, 2020.

This three-year transition period for tax reserves is consistent with the transition period provided in VM-20
for the 2017 CSO for statutory reserving.

It should be noted that the Valuation Manual contains the following Guidance Note with respect to tables
adopted after the Valuation Manual is effective:
American Academy of Actuaries 73 www.actuary.org
 The Valuation Manual can be updated by the NAIC to define a new valuation table. Because of the
various implications to systems, form filings, and related issues (such as product tax issues), lead time
is needed to implement new requirements without market disruption. It is recommended that this
transition be for a period of about 4.5 years—that is, that the table be adopted by July 1 of a given
year, that it be permitted to be used starting Jan. 1 of the second following calendar year, that it be
optional until Jan. 1 of the fifth following calendar year, thereafter mandatory.

Q 15.12: For universal life contracts with secondary guarantees, can different tables be used for the Net
Premium Reserve main guarantee, Net Premium Reserve secondary guarantee, and cost of insurance
(COI) floor?

The mortality standards for the main guarantee, secondary guarantee, and COI floor are provided in Section
3.C.1. It is expected that the same mortality standard will apply to these calculations, except possibly during
the short period of transition to new tables. As an example, during the transition from 2001 CSO to 2017
CSO, a company may have flexibility in applying different tables until it must use 2017 CSO beginning Jan.
1, 2020.

Q 15.13: What are the steps to determining mortality assumptions for the Deterministic Reserve and/or
Stochastic Reserve?

A: For these reserves, mortality assumptions are discussed in Section 9.C of VM-20. (For the Net Premium
Reserve, mortality assumptions are provided in Section 3.C.1.) The following approach is defined for
determining the prudent estimate mortality assumption:

1. Determine mortality segments as provided in Section 9.C.1.a.

2. Determine company experience rates as provided in Section 9.C.2, or, if company experience data is
limited or not available, use an applicable industry basic table or other applicable experience in lieu of
company experience as provided in Section 9.C.3.
3. If the company determines company experience mortality rates as provided in Section 9.C.2, then use
the procedure described in Section 9.C.3 to determine the applicable industry table for each mortality
segment to grade company experience to the industry table. If the company is using an applicable
industry basic table in lieu of company experience as provided in Section 9.C.3, skip to step 5.
4. Determine the level of credibility of the underlying company experience as provided in Section 9.C.4.
Per Section 9.C.4.b, credibility may be determined at either the mortality segment level or at a more
aggregate level if the mortality for the sub-classes was determined using an aggregate level of
mortality experience.
5. Determine the prescribed mortality margins as provided in Section 9.C.5. Separate mortality margins
are determined for company experience mortality rates and the applicable industry basic tables.
Prescribed margin percentages shall be increased, as appropriate, to reflect the level of uncertainty
related to situations, including, but not limited to, those described in Section 9.C.5.d.
6. Use the procedure described in Section 9.C.6 to determine the prudent estimate mortality assumption.

Q 15.14: Is the company required to determine its own company experience mortality rates, or can it just
use an industry mortality table?

A: Section 9.A.6 states that the company shall use its own experience, if relevant and credible, to establish an
anticipated experience assumption for any risk factor. Then, Section 9.C.1.b.i states that the company can use
American Academy of Actuaries 74 www.actuary.org
an industry mortality table in lieu of company experience if the company experience data is limited or not
available. Section 9.C.6.b.iii states that if the credibility of company’s data is less than 20%, the company
cannot use its own experience; 100% of the applicable industry tables should be used instead.

Q 15.15: What is the Applicable Industry Table?

A: The Applicable Industry Table is the most recent valuation basic table listed in VM-M Section 2, including
the Primary, Limited Underwriting, and RR Table form, if available.

Q 15.16: Should the mortality rates be based on experience for each cell, for each mortality segment, or for
the company?

A: Section 9.C.2.b states that company experience data shall be based on experience from the following
sources:

• Actual company experience for books of business within the mortality segment.
• Experience from other books of business within the company with similar underwriting.
• Experience data from other sources, if available and appropriate, such as actual experience data of one
or more mortality pools in which the policies participate under the term of a reinsurance agreement.
Data from other sources is appropriate if the source has underwriting and expected mortality
experience characteristics that are similar to policies in the mortality segment.

VM-20 states that company experience mortality rates shall be determined for each mortality segment
(Section 9.C.2.a), but the rates can be based on more aggregate experience (Section 9.C.2.d).

If experience is aggregated, Section 9.C.2.d states that the company should use other techniques to further
subdivide the aggregate class into various sub-classes or mortality segments. In doing so, the company must
ensure that when the mortality segments are weighted together, the total number of expected claims is not less
than the company experience data for the aggregate class.

Q 15.17: How should the mortality rates that were established based on the aggregate company experience
be subdivided into sub-classes (mortality segments)?

A: VM-20 does not specify a method for subdividing aggregate company experience into sub-classes or
mortality segments. However, Section 9.C.2.d does state that the company must ensure that when the
mortality segments are weighted together, the total number of expected claims is not less than the company
experience data for the aggregate class.

Some actuaries would use the Conservation of Deaths and / or the Preservation of Total Deaths methods to
subdivide aggregate mortality rates into sub-classes. Other approaches may be acceptable.

Please see the following source documents for more information on each method.

Conservation of Deaths
Society of Actuaries Study Note ILA-D107—07; Experience Assumptions for Individual Life
Insurance and Annuities; Richard F. Lambert; page 25.

Preservation of Total Deaths

American Academy of Actuaries 75 www.actuary.org

 Life Insurance Products and Finance; Section 3.2.4.3, “Effect of Selective Lapses”; David B.
Atkinson and James W. Dallas; pages 149-151.

Q 15.18: How long of an exposure period should be used in the experience study used to determine the
mortality assumptions?

A: Section 9.C.2.e.ii states that the company experience should be based on the most recent experience study
and the exposure period should be at least three exposure years, but not more than 10 exposure years.

Q 15.19: How often do experience studies need to be updated?

A: Section 9.C.2.e indicates that the company shall review the mortality experience (in a mortality segment)
at least once every three years and update as needed. Some companies may review their experience every
year, while others may establish a rotating basis for annual reviews of experience across mortality segments
so that each segment is reviewed at least every three years. Some companies may leverage experience studies
that are being performed for other purposes within the company, or the experience reporting for VM-51, and
perform reviews at that time.

Section 9.C.2.e indicates that the company shall reflect changes that are implied by any updated experience
data to the extent that such changes are significant and are expected to continue into the future. Such changes
should be reflected in the current year.

Q 15.20: Can mortality improvement be used?

A: Section 9.C.2.g indicates that the company may reflect mortality improvement from the central point of the
underlying company experience data to the valuation date. The mortality rates from the applicable industry
basic table may be improved from the year of the table to the valuation date, as indicated in Section 9.C.3.g.
The company is prohibited by VM-20 from reflecting mortality improvement beyond the valuation date.

Some actuaries would use published mortality improvement rates to incorporate mortality improvement into
the company experience mortality rates and update the improvement factors every year. Other actuaries may
only update the projected improvement factors in conjunction with the review of underlying experience and
updates of the experience mortality. Actuaries may use improvement factors for the applicable industry table,
applied to the industry table that will be determined and published by the Society of Actuaries.

Q 15.21: Would the actuary need to use a specific credibility method to determine the credibility of the
mortality experience used to develop the mortality assumption?

A: Section 9.C.4 requires the use of the Limited Fluctuation Method or the Bühlmann Empirical Bayesian
Method to be used to determine credibility of mortality experience for valuations in which the 2015 VBT is
the industry basic mortality table. Please see Q 15.23 for additional details on choosing a method and/or
changing the method.

The following are additional resources on credibility methods:

• American Academy of Actuaries, Credibility practice note, at
http://www.actuary.org/files/publications/Practice_note_on_applying_credibility_theory_july2008.pdf
;

American Academy of Actuaries 76 www.actuary.org

 • The Society of Actuaries Research Department, Credibility Theory Practices, at
https://www.soa.org/Files/Research/Projects/research-cred-theory-pract.pdf; and
• The Canadian Institute of Actuaries, Expected Mortality Educational Note, at http://www.cia-
ica.ca/docs/default-source/2002/202037e.pdf?sfvrsn=2.

Q 15.22: How would the actuary determine whether company mortality experience data is credible?

A: Section 9.C.4.a provides the requirements for the credibility calculation:

• For valuations in which the industry basic table is the 2008 VBT, an aggregate level of credibility
over the entire exposure period must be determined using a methodology that follows common
actuarial practice as published in actuarial literature (for example, but not limited to, the Limited
Fluctuation Method or Bühlmann Empirical Bayesian method).

• For valuations in which the industry basic table is the 2015 VBT, an aggregate level of credibility
over the entire exposure period must be determined using either the Limited Fluctuation Method by
amount, such that the minimum probability is at least 95% with an error margin of no more than 5%,
or the Bühlmann Empirical Bayesian Method by amount. Once chosen, the credibility method must be
applied to all business subject to VM-20 and requiring credibility percentages. A company seeking to
change credibility methods must request and subsequently receive the approval of the commissioner.
The request must include the justification for the change and a demonstration of the rationale
supporting the change.

To make a determination on the level of credibility that a set of company experience may have, actuaries will
look to the credibility procedure selected.

Section 9.C.4.a provides the formula to determine the credibility level by amount under the Limited
Fluctuation Method.

Limited Fluctuation Z = min{1, rm/zσ}

where,

r = error margin ≤ 5%
z = normal distribution quantile ≥ 95%
m = mortality ratio; i.e., A/E ratio by amount
σ = standard deviation of the mortality ratio

Section 9.C.4.a also provides a formula the actuary can use in conjunction with the 2015 VBT industry basic
table to compute an approximation to the Bühlmann Empirical Bayesian Method.

A
Bühlmann Z =
A+
(109% * B) − (120.4% * C )
(0.019604 * A)

The numerical result under either of these methods should be rounded to the nearest whole integer.

American Academy of Actuaries 77 www.actuary.org

Q 15.23: Can company mortality experience be aggregated to establish credibility?

A: VM-20 states the following with respect to establishing the level of credibility of the company experience:

Section 9.C.4.b: Credibility may be determined at either the mortality segment level or at a more
aggregate level if the mortality for the sub-classes (mortality segments) was determined using an
aggregate level of mortality experience.

Section 9.C.4.b: A single level of credibility shall be determined over the entire exposure period,
rather than for each duration within the exposure period.

Q 15.24: There are five prescribed mortality margin tables in VM-20. Which tables apply to the experience
mortality rates and which to the industry table?

A: The table in Section 9.C.5.b.i applies to the margin for company experience mortality rates for valuations
in which the industry mortality table is the 2008 VBT limited underwriting table (the 2015 VBT does not have
a limited underwriting table). The table in Section 9.C.5.b.ii applies for valuations in which the industry
mortality table is the 2015 VBT and where the credibility is determined using the Bühlmann Empirical
Bayesian Method by amount. The table in Section 9.C.5.b.iii applies for valuations in which the industry
mortality table is the 2015 VBT and where the credibility is determined using the Limited Fluctuation by
amount method. The margin could vary by credibility level, credibility method, valuation basis, and attained
age.

The table in Section 9.C.5.c.i applies to the applicable industry table for valuations in which the industry
mortality table is the 2008 VBT limited underwriting table. The table in Section 9.C.5.c.ii applies to the
applicable industry table for valuations in which the industry mortality table is the 2015 VBT. The margin
could vary by attained age and valuation basis.

The margins need to be added to company experience mortality and to the applicable industry table before the
grading process takes place. The margin is in the form of a percentage increase applied to each mortality rate.

Q 15.25: How are the prescribed mortality margin tables used in VM-20?

A: Section 9.C.5.a indicates that there are two separate sets of prescribed margins. One set of prescribed
margins is for company experience mortality rates. The tables in Section 9.C.5.b show the margin to be used
for company experience mortality rates. Another set of prescribed margins is for the applicable industry basic
tables. The tables in Section 9.C.5.c apply to the margin for the applicable industry table.

The prescribed margin percentages for the company experience mortality rates vary by the valuation’s
applicable industry basic tables, credibility method, level of credibility, and attained age. The prescribed
margin percentages for the applicable industry basic tables vary by attained age.

Pursuant to Section 9.C.5.a of VM-20, the mortality margin shall be in the form of a prescribed percentage
increase applied to each mortality rate.

Section 9.C.5.d states that the prescribed margin percentages shall be increased, as appropriate. The section
provides a possible list of situations to consider in making adjustments to the prescribed margin percentages.

American Academy of Actuaries 78 www.actuary.org

Q 15.26: Should the actuary apply a margin in addition to the prescribed margins?

A: Some actuaries may apply a margin in addition to the prescribed margins to reflect the level of uncertainty
in the company mortality experience. Section 9.C.5.d requires that the prescribed margins be increased in
situations that include, but are not limited to, the following:

• The reliability of the company’s experience studies is low due to imprecise methodology, length of
time since the data was updated, or other reasons.
• The length of time since the experience data was updated.
• The underwriting or risk selection risk criteria associated with the mortality segment have changed
since the experience on which the company experience mortality rates are based was collected.
• The data underlying the company experience mortality rates lacks homogeneity.
• Unfavorable environmental or health developments are unfolding and are expected to have material
and sustained impact on the insured population.
• Changes to the company’s marketing or administrative practices or market forces expose the policies
to the risk of antiselection.
• Underwriting is less effective than expected.

Q 15.27: What other considerations should be taken into account in setting margins for the mortality
assumption?

A: Some actuaries would include additional margins to reflect uncertainty in antiselection under the following
circumstances:

• Term plans with low initial premiums followed by substantially higher premiums;
• Extended term insurance as the nonforfeiture option;
• Conversion at the expiry or very close to expiry of the term coverage period;
• Policies that have elected to exercise a guaranteed purchase option; and
• Products that offer composite rates (i.e., same rate for male/female, various issue/attained ages,
underwriting classes, etc.).

Q 15.28: How does the actuary grade the experience mortality rates into industry rates?

A: Section 9.C.6.b states that if the company determines company experience mortality rates, the prudent
estimate assumptions will be determined as follows:

• For each mortality segment, use the company experience mortality rates for policy durations in which
there exists sufficient company experience data plus the prescribed margin as provided in Section
9.C.5.b. and any additional margin as provided in Section 9.C.5.d.

• The company shall determine the sufficient data period by identifying the last policy duration at which
sufficient company experience data exists (using all the sources defined in Section 9.C.2.b). This
period ends at the last policy duration that has 50 or more claims (i.e., no duration beyond this point
has 50 claims or more) subject to the limits provided in Section 9.C.6.B.iii.b. The sufficient data
period may be determined at a more aggregate level than the mortality segment if the company based

American Academy of Actuaries 79 www.actuary.org

 its mortality on aggregate experience and then used a methodology to subdivide the aggregate class
into various sub-classes or mortality segments.

• Beginning in the first policy duration after the sufficient data period, use the guidelines in the table
referenced in Section 9.C.6.b.iii to linearly grade from the company experience mortality rates with
margins to 100% of the applicable industry table with margins. (The determination of the applicable
industry table is described in Section 9.C.3). Grading must begin and end no later than the policy
durations shown in the table referenced in Section 9.C.6.b.iii, based on the level of credibility of the
data as provided in Section 9.C.4. For valuations on or after Jan. 1, 2015, if the credibility level is less
than 20%, the company is not allowed to use their company experience and must use 100% of the
applicable industry table.

Q 15.29: How is “sufficient data” defined for the purposes of grading into the industry table in Section
9.C.6.b? Can the actuary aggregate data across policy durations and mortality segments to determine the
sufficiency of data?

A: Section 9.C.6.b.ii indicates that the “sufficient data period” is the last policy duration that has 50 or more
claims, subject to limits provided in Section 9.C.6.b.iii.b. This precludes a company from aggregating data
across policy durations to determine the sufficient data period. However, Section 9.C.6.b.ii allows a company
to determine the “sufficient data period” by aggregating mortality segments if the company based its mortality
on aggregate experience and then used a methodology to subdivide the aggregate class into various sub-
classes or mortality segments. Based on this VM-20 guidance, some actuaries may consider aggregating data
across issue ages to determine the sufficient data period.

Q 15.30: Over how many years should the actuary completely grade the experience mortality rates into the
industry table?

A: Sections 9.C.6.b.iii and 9.C.6.b.iv describe the limits and requirements for the grading period used for
grading experience mortality rates to the industry table. These sections are quite expansive and will not be
repeated in this practice note.

Q 15.31: Can the mortality rates produced by following the VM-20 procedure be adjusted?

A: Section 9.C.6.d indicates that the company may adjust the resulting mortality rates within each mortality
segment to ensure the resulting prudent estimate rates reflect reasonable relationships with assumptions in
other mortality segments. The adjustment to the mortality rates must be done in a manner that does not result
in a material change in the total expected claims for all mortality segments in the aggregate. Section 9.C.6.c
indicates that smoothing may be utilized within each mortality segment to ensure a reasonable relationship
exists by attained age within each mortality segment.

Q 15.32: How are mortality assumptions modified for impaired lives?

A: Section 9.C.6.e requires adjustments to the mortality assumption for impaired lives if the assumption does
not appropriately reflect the expected mortality experience. If the block has an immaterial amount of impaired
lives, some actuaries would develop a combined mortality assumption for the mortality segment. Some
actuaries may exclude the impaired lives from the mortality segment and later include an adjustment for them.
This method is described in Section 9.C.2.f. Some actuaries will make this adjustment in the form a
percentage or flat addition to standard mortality experience. Where there is sufficient credibility of

American Academy of Actuaries 80 www.actuary.org

experience, some actuaries might make a more complex adjustment that changes the shape of the mortality
curve.

Where there is a significant amount of substandard business, some actuaries would determine a mortality
segment and assumption for substandard business separately (i.e., develop a prudent estimate assumption for
mortality specifically for substandard policies).

Section 9.B.2 requires a larger margin where there is greater uncertainty in the anticipated experience
assumption. Therefore, some actuaries may also increase the margin on the mortality assumption that applies
to impaired lives (in addition to the higher base assumption) because there is more uncertainty regarding the
assumption.

Q 15.33: Should policyholder behavior be taken into account in setting the mortality assumption?

A: Actuaries may wish to consider making adjustments for policyholder behavior when the product design
leads to potential antiselection from policyholder behavior. This is required by Sections 9.C.2.f and 9.C.6.e if
the actuary expects the results to vary due to the behavior. An example of this would be a term product with
dramatic increases in tail premiums and the assumption of less than 100 % lapse after the premium increase.

Other areas where an actuary may consider making adjustments due to policyholder behavior include where
there is a guaranteed purchase option, guaranteed renewability, or similar features where the mortality
expectation of those electing the option would be different than the base assumption.

Q 15.34: Is the use of stochastic mortality allowed under VM-20?

A: VM-20 does not specifically allow the use of stochastic mortality. However, Sections 9.A.1 and 9.A.4
imply that assumptions can be stochastically modeled, and if mortality is stochastically modeled, the
requirements of Section 9 relating to prudent estimate assumptions are not applicable. Therefore, some
actuaries may choose to model mortality stochastically. If the election is made to model mortality
stochastically, there is no guidance in VM-20 on how to do this and very little established actuarial practice
around stochastic modeling of mortality experience.

16. Setting Premium Assumptions

Q 16.1: What are some of the general considerations that actuaries may take into account in setting
premium assumptions for flexible premium products under VM-20?

A: According to Section 9.A.6.b, for risk factors that do not lend themselves to the use of statistical credibility
theory (such as a flexible premium pattern), a company shall establish an anticipated premium assumption in
a manner that is consistent with accepted actuarial practice and that reflects any available relevant company
experience, any available relevant industry experience, or any other experience data that is available and
relevant. Some actuaries may choose to vary the premium assumption by one or more of the following
factors: product type (e.g., cash-value accumulation products vs. protection-only products), year of issue,
premium funding level, underwriting risk class, and gender.

American Academy of Actuaries 81 www.actuary.org

Note that some actuaries may choose to model cash flows in a manner that reflects the actual modal premium
distribution using average factors in the aggregate.

Section 9.A.6.d requires that the actuary use sensitivity testing and disclose such analysis to assure that
assumptions set using these approaches are set at the conservative end of the plausible range.

Q 16.2: Are there any other areas where sensitivity testing may be done?

A: According to Section 9.D.4, a company shall, at a minimum, examine the sensitivity of the minimum
reserve to changes in premium payment patterns, premium persistency, surrenders, partial withdrawals,
allocations between available investment and crediting options, benefit utilization, and other option elections
if relevant to the risks in the product. Section 9.D.4 requires the following sensitivity tests to be performed, at
a minimum, for policies that offer policyholders flexibility in the timing and amount of premium payments:

i. Minimum premium scenario;

ii. No further premium payment scenario;
iii. Prepayment of premiums—Single premium scenario; and
iv. Prepayment of premiums—Level premium scenario

For flexible premium products with both primary and secondary guarantees, some actuaries may choose to
define the above premium payment patterns in light of the primary guarantee, whereas some others would
consider them in light of the secondary guarantee. Some actuaries may choose to define the premium to
encompass both the primary guarantee as well as the secondary guarantee. Some actuaries would use the
above scenarios to examine the sensitivity of the reserve to the premium payment pattern assumption. If there
is uncertainty about the level of future premiums and the reserve is sensitive to the premium payment pattern,
then some actuaries would use the sensitivity test results in the determination of the premium prudent estimate
assumption. At a minimum, VM-31 requires disclosure of these results.

Q 16.3: How might the actuary model the four required premium payment pattern sensitivities required by
Section 9.D.4?

A: Some actuaries would interpret these premium sensitivity tests as illustrations of minimum premium
patterns that would keep the policies in force until the end of the insurance period (with the exception of ii).
Some actuaries would model the minimum premium on a cell-by-cell basis but might combine cells or use a
subset of the policies to complete these sensitivities in a timely manner.

Q 16.4: For dividend-paying business, how would the modeling account for dividends used to reduce
premium?

A: It is expected that some actuaries will model based on how policies are actually utilizing dividends to pay
premiums by including both the dividend and the implicit premium account. However, it is also likely that
some actuaries would model based on a net basis (fixed premium minus dividend used to pay premium)
because the cash flow impact is similar. In that circumstance, those actuaries may make other adjustments
where necessary so that other projected items that are based on premium (e.g., premium taxes) or dividends
are captured correctly. Actuaries may also wish to consider how dividends that are to reduce premiums would
impact lapses and other policyholder behavior assumptions. It would be prudent for the actuary to make sure
the modeling of dividends used to reduce premium is well documented if the net basis is used.

American Academy of Actuaries 82 www.actuary.org

Q 16.5: Flexible premium products often have minimum required premium payments, excess premium
payments, cessation of premium payments, and irregular premium payments. How does one capture this
flexibility in the cash flow model if the underlying experience is not fully credible?

A: Section 9.A.6.b discusses risk factors that do not lend themselves to credibility theory, such as flexible
premium patterns. Some actuaries would determine the premium assumption on flexible premium policies
taking into account how the product was marketed and sold, if such information is available. However, as
there is flexibility in payment of premiums, historical payment patterns likely would also be taken into
account. Some actuaries would review policy features that would impact premium payments in the future. For
example, if a policy has a guarantee that would expire if a specific premium is not paid in a period, then some
actuaries would look at historical or expected experience for similar policies with that feature in setting the
expected future premium payments.

Additionally, some actuaries model flexible premium products based on how the products are expected to be
used by policyholders. For example, one could have a separate assumption for policies expected to be used for
accumulation compared to those expected to be used for protection.

17. Setting Policyholder Behavior Assumptions Other Than Premiums

Q 17.1: What policyholder behaviors other than premiums might be considered in the calculation of
reserves?

A: Section 9.D.4.a requires sensitivity tests of at least the following policyholder behavior assumptions:

Premium payment patterns, premium persistency, surrenders, partial withdrawals, allocations between
available investment and crediting options, benefit utilization, and other option elections if relevant to
the risks in the product.

Because sensitivity testing of these behaviors is required, these behaviors will need to be captured in the
model or will need to be sensitivity-tested in some other manner.

Some actuaries would also consider the election of dividend, conversion, guaranteed purchase, and
nonforfeiture options to the extent these options are relevant to the risks in the product.

Section 7.F.3.b provides guidance on how to model policy loan behavior when policy loans are modeled
explicitly, but explicit modeling of policy loans is not mandatory. However, the requirements in Section
7.F.3.a must be met if policy loans are not explicitly modeled. In this case, the reserves can’t be less than
those produced when modeling loans explicitly and policyholder behavior must comply with Section 9.D.

Q 17.2: When should dynamic policyholder behavior assumptions be used?

A: Section 9.D.2.a requires that the company shall use a dynamic model or other scenario-dependent
formulation to determine anticipated policyholder behavior unless the behavior can be appropriately
represented by static assumptions. No guidance is given as to how to determine when an assumption can be
represented “appropriately” by static assumptions. It would be prudent for an actuary to document the

American Academy of Actuaries 83 www.actuary.org

rationale for using static assumptions if static assumptions are used in place of dynamic policyholder
behavior.

Some actuaries would use dynamic policyholder behavior assumptions in instances where an external
environment or actions of the company, which can be reflected in the model, affect policyholder behavior.

Some examples the actuary may wish to consider include, but are not limited to, the following:

• If a company raises non-guaranteed premiums or cost of insurance rates, more policyholders could
surrender / lapse / convert (especially the healthy lives), which could worsen the overall mortality of
the remaining lives.
• Under decreasing- / low-interest scenarios, the guarantees may become attractive, possibly leading to
lower lapses or additional premium payments or exercise of guaranteed settlement options.
• Reductions to interest crediting rates or dividend scales may lead to additional lapses or premiums for
certain specific policy forms, such as those with premium expectations based on higher illustrated
interest rate or dividend levels.
• For interest-sensitive products, when interest rate change (in particular, under increasing-interest
scenarios), the assumption of how credited rates are set (i.e., how fast they increase) might be
assumed to affect the lapse assumption, which may affect other assumptions such as mortality and
premiums.
• For policies with secondary guarantees, lapses might be modeled to significantly decrease when the
guarantee is in the money. Section 9.D.5 of VM-20 gives specific guidance on this topic.

Q 17.3: Are there specific considerations in setting margins on assumptions for policyholder behavior?

A: Section 9.D.3 states that to the extent that there is an absence of relevant and fully credible data, the
company shall determine the margin such that the policyholder behavior assumption is shifted toward the
conservative end of the plausible range of behavior, which is the end of the range that serves to increase the
reserve. Margins should reflect an increase in policyholder efficiency over time, unless the company has
relevant and credible experience or clear evidence to the contrary. Also, the margin for surrender or partial
withdrawal should be higher in the case where the company’s marketing or administrative practices encourage
antiselection. Note, the company must reflect the data uncertainty when using assumptions from similar but
not identical blocks of business.

Given the level of uncertainty in the estimation of policyholder behavior, especially when there are valuable
contract options, some actuaries would increase margins for these risk factors as compared to risk factors
where fewer policyholder options exist. Some actuaries would look to the credibility of experience data in
setting the margins around this assumption using the considerations summarized above.

Q 17.4: How would the actuary reflect nonforfeiture options in the cash flow projections?

A: Some actuaries would assume that all nonforfeiture benefits are cash surrenders. If various nonforfeiture
options are not equivalent, then some actuaries might use the greatest of the values of the nonforfeiture
options similar to the taking the highest value under the Commissioner’s Annuity Reserve Valuation
Methodology. Other actuaries would model a nonforfeiture option cost that is consistent with an expected
level of nonforfeiture option utilizations.

American Academy of Actuaries 84 www.actuary.org

Another approach may be to project the cash flows under the various nonforfeiture paths with expected
election rates that vary by path.

Q 17.5: How are dividend options considered in performing the VM-20 projections?

A: Some actuaries may model dividend options explicitly in the cash flow projections, especially if the
election of particular dividend options is integral to how a policy was marketed and sold. This would include
dividends used to pay premiums and the purchase of paid-up additions or one-year term. Alternatively, when
the minimum reserve is not materially affected by the choice of dividend options, some actuaries may adopt a
more simplified approach, such as modeling all dividends as paid in cash.

Section 7.C.6 discusses the liability for dividends declared but not yet paid as of the valuation date. It states
that this liability may or may not be included in the Cash Flow model at the company’s discretion. According
to statutory accounting principles, dividends are reported separately from the statutory reserve. Therefore, if
the policyholder dividends that give rise to the dividend liability are not included in the cash flow model, then
no adjustment is needed to the resulting aggregate modeled (whether deterministic or stochastic) reserve. If
the policyholder dividends that give rise to the dividend liability are included in the cash flow model, then the
resulting aggregate modeled (whether stochastic or deterministic) reserve should be reduced by the amount of
the dividend liability.

Q 17.6: How might policyholder behavior assumptions differ between the various exclusion tests and
projections required under VM-20?

A: Some actuaries would conclude that the Deterministic Reserve and Stochastic Reserve calculations as well
as the Stochastic Exclusion Ratio Test should use the same assumptions, except that margins are not included
in the Stochastic Exclusion Ratio Test per Section 6.A.2.b.i.a.1, and prudent estimate assumptions will vary
dynamically for the 16 scenarios to reflect scenario-dependent risks.

If the qualified actuary chooses to pass the Stochastic Exclusion Test by certifying that a group of policies are
not subject to material interest rate or asset return volatility risk, then different assumptions may be used, for
example results of the New York seven scenarios from cashflow testing, as opposed to the 16 scenarios from
Appendix 1.

Per section 6.B.5, for the Deterministic Exclusion Test, if the net premium reserve follows Section 3.A.1, then
the lapse rates are 0% when calculating the valuation premiums for this comparison. Per Section 6.B.5.d, if
the anticipated mortality exceeds the valuation mortality, then the anticipated mortality will be used to
calculate the net valuation premium.

Q 17.7: How should policyholder options such as term conversion options and guaranteed purchase
options be treated in determining reserves?

A: Section 9.D.1.e states that the actuary must reflect the likelihood that policyholder behavior will be
affected by any significant increase in the value of a product option, such as term conversion privileges or
policy loans.

Some actuaries will take into account the potential exercise of policyholder options in their policyholder
behavior assumptions for those options that have value. However, some actuaries would use a simplified
model for companies where the impact of term conversions and exercise of the guaranteed purchase options is
not material.
American Academy of Actuaries 85 www.actuary.org
Section 2.G indicates that:

A company may use simplifications, approximations and modeling efficiency techniques to calculate
the Net Premium Reserve, the Deterministic Reserve and/or the Stochastic Reserve required by this
section if the company can demonstrate that the use of such techniques does not understate the reserve
by a material amount and the expected value of the reserve calculated using simplifications,
approximations and modeling efficiency techniques is not less than the expected value of the reserve
calculated that does not use them.

For companies for which term conversions are significant, some actuaries would include additional
decrements in the term line to take into account policies converting to whole life policies. The impact (i.e.,
additional cash flows from the conversion or purchase) would be included in the cash flow projection to take
into account the impact of the policies converting. Some actuaries would reflect this impact using a “cost of
conversion or purchase” charge or credit from studies on the cost of term conversions.

The cost / credit of conversion or purchase could be determined by projecting some typical post-conversion or
-issue policies using appropriate prudent estimate assumptions and including this cost / credit at the point of
conversion or purchase. Some actuaries may consider reflecting the potential of mortality antiselection and
other option-specific behavior.

Some actuaries may consider the impact from more than one possible post-conversion product type. Term
policies converting to newer products such as a universal life with secondary guarantees could theoretically
generate a significantly different cost of conversion or purchase than those converting to a traditional whole
life product.

Some actuaries might include projected new policies arising from term conversions with their own respective
cash flow projections in the reserve calculations. However, this may involve separate modeling of the new
policies that convert from term to whole life because these new policies must be modeled separately in the
“Other” product group independent from the “Term” product group. Other actuaries might include the present
value of the cost / credit of conversion in conjunction with the election rate in the term projection to affect the
cash flows at the time of the expected election of an option.

Q 17.8: Could the existence of a guaranteed purchase option lead to a reduction in the VM-20 reserves?

A: Yes. A guaranteed purchase option is simply a charge being made for a potential future exercise of the
purchase of additional insurance. Prior to the exercise of any remaining future options, the impact to reserves
would be negative if the actuary projects (using Prudent Estimate Assumptions, including any applicable
antiselection and margins) that the present value of expected profits from future exercised policies plus the
present value of guaranteed purchase option premiums less guaranteed purchase option commissions and
guaranteed purchase option expenses is greater than zero.

Q 17.9: What are some things actuaries consider for policies that have already converted?

A: Some actuaries model policies based on the status of the policy on the valuation date. However, other
actuaries might take into account expected future experience that may be based on a policy coming into its
current status from another policy type. When considering conversion options, some actuaries will reflect
expected mortality antiselection arising from converted term policies, either explicitly for these policies, or as
part of the aggregate experience for the block.
American Academy of Actuaries 86 www.actuary.org
Q 17.10: Are there any additional considerations related to policyholder behavior assumptions for term
policies with premium increases after the end of the level term period?

A: Section 9.D.6 provides guidance on the treatment of policyholder behavior assumptions in calculating the
reserves for term policies with material premium increases at the end of the level term period.

For a term life policy that guarantees level or near level premiums until a specified duration followed by a
material premium increase, or for a policy for which level or near level premiums are expected for a period,
followed by a material premium increase, for the period following the premium increase, the lapse and
mortality assumptions shall be adjusted or margins added such that the present value of cash inflows in excess
of cash outflows assumed shall be limited to reflect the relevance and credibility of the experience,
approaching zero for periods where the underlying data has low or no credibility or relevance.

Additionally, for the calculation of the Deterministic Reserve, for a term policy issued Jan. 1, 2017, and later
that guarantees level or near level premiums for more than five years until a specified duration followed by a
material premium increase, or for a policy for which level or near level premiums are expected for more than
five years followed by a material premium increase, for the period following that premium increase, the cash
inflows or outflows shall be adjusted such that the present value of cash inflows does not exceed the present
value of cash outflows. In other words, in this situation, Section 9.D.6 requires that the calculation of the
Deterministic Reserve not include any post-level term period positive net cash flows.

18. Setting Expense Assumptions

Q 18.1: What types of expenses should be included in the models for determining reserves?

A: Section 7.B.1.e provides that all types of expenses—including but not limited to overhead expenses,
commissions, fund expenses, contractual fees and charges, and taxes (excluding federal income taxes and
expenses paid to provide fraternal benefits in lieu of federal income taxes)—be reflected in the modeling. Per
Section 7.H.2.c.i, investment expenses are considered as a part of the net asset earned rate calculation. Some
actuaries may consider ASOP No. 2, Nonguaranteed Charges or Benefits for Life Insurance Policies and
Annuity Contracts, and ASOP No. 7, Analysis of Life, Health, or Property/Casualty Insurer Cash Flows,
when determining the expenses included in the projections where it is not specifically addressed in VM-20.
Specific guidance is provided in Section 7.H.2 for reflecting investment expenses and in Section 9.G for
expenses (and expense reimbursements) related to revenue-sharing agreements.

Section 9.E.1.c requires prudent estimate expense assumptions to assume that the company is a going
concern.

Q 18.2: Must acquisition expenses be included?

A: Per Section 9.E.1.m, acquisition costs associated with the business in force as of the valuation date and
significant nonrecurring expenses that are expected to be incurred after the valuation date shall be included in
the reserve calculation.

Q 18.3: How should overhead be reflected and allocated in the calculation?

American Academy of Actuaries 87 www.actuary.org

A: Per Section 9.E.1.l, an appropriate portion of indirect costs and overhead expenses should be included
along with other expenses consistent with the block of policies being modeled.

Some actuaries would use the company’s practices of expense allocation among lines of business if it includes
overhead expenses and is reasonable for the specific cohorts of business in scope for VM-20. Some actuaries
may need to develop an allocation specifically for reserve modeling that might be based on pricing or the
Illustration Actuary methodology.

Section 9.E.1.j states that allocations must be determined in a manner that is within the range of actuarial
practice and methodology and consistent with applicable actuarial standards of practice and may not be done
for the purpose of decreasing the modeled reserve.

Q 18.4: How should inflation be reflected?

A: Section 9.E.1.e requires that expense assumptions reflect the impact of inflation, and Section 9.E.1.a states
that expense assumptions for the stochastic scenarios may differ from expense assumptions for the
deterministic scenario due to the application of inflation rates. Some actuaries would thus make an assumption
for inflation that is related to the scenarios being modeled. One way to do this is to base the rate of inflation
on the interest rate scenario being projected, such as assuming a base real rate of return with some portion of
the additional amount assumed to be inflation. Some actuaries may model inflation separately.

Q 18.5: What future improvements in expenses may be included?

A: Under Section 9.E.1.f, prudent estimate expense assumptions may not assume future expense
improvements. However, according to Section 9.E.1.k, expense efficiencies that are derived and realized from
the combination of blocks of business only can be assumed when the costs of achieving these efficiencies are
also recognized.

Q 18.6: How should federal income taxes be reflected in the calculation?

A: Section 9.E.1.g, the following shall not be included in the calculation:

• federal income taxes;

• expenses paid to provide fraternal benefits in lieu of income taxes; and
• foreign income taxes.

Q 18.7: How would expense assumptions be set for new products or new lines of business where there is no
experience within the company?

A: Section 9.E.1.n provides that the company should review expense factors used to determine anticipated
experience for an existing block of mature policies and adjust those factors for any differences in long-term
expense factors that would be expected between the new product and existing mature block of business,
taking steps to ensure that all expenses are fully allocated.

Some actuaries would look to pricing assumptions; expense experience for other similar products within the
company; or to studies done by industry groups, reinsurers, or consultants for similar products to adjust the

American Academy of Actuaries 88 www.actuary.org

anticipated experience assumption. Some actuaries would increase the margin on this assumption compared to
other expense assumptions to reflect the greater uncertainty in the anticipated experience assumption.

Q 18.8: How would the actuary set margins for an expense assumption?

A: A process that some actuaries might use would be to review the historical experience data (for example,
unit costs for the last five years) for the line of business or relevant block. Assuming the actuary is
comfortable that the level of expenses will not significantly increase or decrease in the future, the actuary
would then determine what type of modification to this assumption would increase the reserve (likely an
increase to expenses). Then, taking into account the criteria above for determining margins, including the
uncertainty, credibility, quality of experience data, and the level of the anticipated experience assumption
relative to the historical values, the actuary would develop a range of potential outcomes for future expenses.
This range could vary by duration with a tighter range expected in the next few projection years and a wider
range further out in the projection. The margin would be set so that in the actuary’s judgment, the range of the
prudent estimate assumption includes the potential deviations from the anticipated experience in a manner
consistently conservative as the CTE requirement for the entire reserve. See Section 9.B for general guidance
on assumption margins.

Q 18.9: How should capital expenditures, such as technology investment costs, be considered in the
expense assumptions?

A. Section 9.E.1.b and the subsequent Guidance Note provide that certain capital expenditures may be spread
over a reasonable number of years in accordance with statutory accounting principles as defined in the
Statements of Statutory Accounting Principles but that care should be taken with regard to the potential
interaction with other assumptions.

Q 18.10: Can expenses for items like death claim processing and loan processing be included with
maintenance expenses?

A. Section 9.E.1.d states that an actuary shall choose an appropriate expense basis that properly aligns an
expense with the assumption. For example, the Guidance Note states that the expense assumption for handling
death benefit payments should be allocated per death incurred. Section 9.E.1.d also states that costs that are
not significant can be aggregated into another base assumption. The actuary should use judgment to decide
which assumptions can be aggregated and which should be separated.

19. Setting Non-Guaranteed Element Assumptions

Q 19.1: What are non-guaranteed Elements (“NGEs”) and how should they be included in the models
under VM-20?

A: Non-guaranteed elements (“NGEs”), as defined in Section 1.B.12 of VM-20, refer to charges or credits to
a policyholder’s account value, benefit, premium, or consideration that are both established and which may be
adjusted at the discretion of an insurance company. NGEs include, for example, policyholder dividends for
participating policies, and participation rates and asset fee charges for equity-indexed universal life policies.
Section 7.C provides guidance on how NGEs should be modeled. In addition, Section 8.C.7 outlines items

American Academy of Actuaries 89 www.actuary.org

that companies should include in setting assumptions for the NGE in reinsurance cash flows. Section 8.C.8
provides that certain actions under reinsurance agreements may be thought of as comparable to NGEs.

Q 19.2: Can the actuary modify (up or down) the assumed NGEs scale or spread in response to the
experience unfolding in the scenario?

A: Section 7.C.3 and 7.C.4 indicates that projected levels of NGEs in the cash flow model must be consistent
with how actual NGEs are determined (7.C.3) and consistent with the experience assumptions used in each
scenario (7.C.4). Policyholder behavior assumptions in the model must be consistent with the NGE assumed
in the model.

Q 19.3: What are some of the factors that should be considered when deciding to include NGE in reserves
covered by VM-20?

A: According to Section 7.C.2, the projected NGE shall reflect factors that include, but are not limited to, the
following:

• The nature of the contractual guarantees;

• The company’s past NGE practices and policies;
• The timing of any change in NGE relative to the date of recognition of a change in experience; and
• The benefits and risks to the company of continuing to authorize the NGE.

Q 19.4: When determining the NGE assumptions for each scenario, what considerations might the actuary
take into account when modifying the current NGE scale or spread?

A: Examples of considerations the actuary may wish to consider include:

• Existence of contract guarantees;

• The company’s ability to modify its non-guaranteed dividend scale and/or non-guaranteed elements
on items such as credited rates, expense charges, COIs, etc.;
• Effect on policyholder behavior of maintaining the current non-guaranteed dividend scale and/or non-
guaranteed elements under the scenario;
• Effect of the NGE assumptions on the competitive position of the product under the scenario;
• The extent to which a change in experience is recognized in the non-guaranteed dividend scale or non-
guaranteed elements;
• The timing lag from when a change in experience occurs to when it is recognized in the non-
guaranteed dividend scale or non-guaranteed elements; and
• Management philosophy.

Q 19.5: What adjustments to the model can be made to take into account lags in the changing of NGEs?

A: Some actuaries may find it difficult to model lags in changing NGEs in these calculations. One technique
that some actuaries use is to assume that NGEs are determined based on last year’s values, or the last period if
quarterly or monthly time steps are used in the model. Some actuaries would consider whether the modeled
timing of changes to NGEs materially distorts the minimum reserve.

Q 19.6: Can the actuary just review the NGE impact for the tail scenarios, because those scenarios are the
only ones that go into the Stochastic Reserve calculation?

American Academy of Actuaries 90 www.actuary.org

A: Some actuaries would pay particular attention to how the model assumes NGEs are modified as the market
conditions change across all scenarios. This would include scenarios outside of the tail scenarios where NGEs
play a significant part in risk mitigation because an error or misstatement in one of those scenarios could lead
to that scenario (that was not previously in the tail) now becoming part of the tail. This is particularly true for
blocks where changes to NGEs are likely to have a significant impact on calculation of reserves.

Q 19.7: Will actuaries assume the prudent estimate assumptions or anticipated experience assumptions
actually occur when projecting the NGEs as part of VM-20?

A: Section 7.C.4 states that projected levels of NGEs in the cash flow model must be consistent with the
experience assumptions used in each scenario. Some actuaries may interpret this to imply the NGEs be
established based on the prudent estimate assumption set.

Like all other assumptions, VM-20 requires that the NGEs include a margin. Hence, some actuaries will not
assume that the company will fully reflect the projected experience in the model, because doing so could
eliminate the margin built into the prudent estimate assumption. As an example, if future policyholder
dividends are reduced to exactly offset the projected higher claims due to the margin in the mortality
assumption, the mortality margin is essentially eliminated. Some actuaries will delay the timing of when the
dividend scale changes are implemented in the model as the NGE margin. Other actuaries will not fully reflect
the needed dividend scale change in future dividends as the NGE margin.

Q 19.8: Are there situations where an NGE may be excluded in the determination of the Deterministic or
Stochastic Reserves?

A: According to Section 7.C.5, the actuary may exclude any portion of the NGE that:

• is not based on some aspect of the policy’s or contract’s experience; and

• is authorized by the board of directors and documented in the board’s minutes where the
documentation includes the amount of NGE that arises from other sources.

However, if the board has guaranteed a portion of the NGE into the future, the company must model that
amount (unless excluded by Section 7.C.6, which discusses policyholder dividends declared but not yet paid).

Some actuaries believe that, if the board has identified a distribution to policyholders like a special dividend,
(e.g., arising from a sale of a company or block of business unrelated to the specific block of business) but
that special dividend is not guaranteed, then the language in Section 7.C.6 would allow them to exclude it
from the model.

Q 19.9: Are dividends declared but not yet paid allowed to be included in the cash flow model?

A: This topic is covered in Section 7.C.6.

Some actuaries would not model payment of dividends declared but not yet paid to the extent a separate
dividend liability exists as required by statutory accounting principles. Doing so would have the dividends
captured in both the dividend liability reserve and the Deterministic or Stochastic Reserve.

American Academy of Actuaries 91 www.actuary.org

Some actuaries may include all dividend payments due to system constraints, or in recognition of second-
order impacts, and then remove the declared but unpaid dividends via a top-side adjustment, such as
subtracting the dividend liability from the Deterministic or Stochastic Reserve.

20: Treatment of Reinsurance

Q 20.1: When should the actuary include a reinsurance agreement in the calculation?

A: Section 8.A.3 states that a company shall include a reinsurance agreement in calculating the minimum
reserve if under the terms of the Accounting Practices and Procedures Manual the agreement or amendment
qualifies for credit for reinsurance.

Section 8.A.4 and the accompanying Guidance Note state that if a reinsurance agreement or amendment does
not qualify for credit for reinsurance, but treating the reinsurance agreement or amendment as if it did so
qualify would result in a reduction to the company’s surplus, then the company shall increase the minimum
reserve by the absolute value of such reductions in surplus.

Some actuaries would include a reinsurance agreement that has an effective date after the valuation date in the
future valuation cash flows if the treaty had been signed by both parties on or prior to the valuation date and it
qualifies for reinsurance credit under the terms of the Accounting Practices and Procedures Manual.

Q 20.2: Can the actuary rely on calculations performed by the other party to the reinsurance transaction?

A: The Guidance Note in Section 8.A.1 states that an actuary can rely on calculations performed by the other
party, but the reserve calculation must be rerun or appropriate adjustments must be made if different
assumptions are chosen.

The actuary may wish to consider whether the margins set by the other party satisfy the requirement that
margins are conservative with respect to their reserve amount. The margins from the other party’s perspective
may not necessarily meet that requirement.

Q 20.3: How is reinsurance reflected in the Net Premium Reserve?

A: Section 8.B.1 states the determination of a credit to the Net Premium Reserve to reflect reinsurance ceded
is done in accordance with SSAP No. 61R. Some actuaries would perform this calculation for each policy and
then aggregate based on Section 8.B.3.

Q 20.4: How is the reinsurance credit for the Net Premium Reserve calculated when a company cedes a
portion of the policy under multiple reinsurance agreements?

A: Section 8.B.2 states that if a company cedes a portion of a policy under more than one reinsurance
agreement, then the company shall calculate a credit separately for each such agreement. The credit for
reinsurance ceded for the policy shall be the sum of the credits for all such agreements.

American Academy of Actuaries 92 www.actuary.org

Q 20.5: When calculating the ratio for the Stochastic Exclusion Ratio Test (SERT), the company can get a
dramatically different result before reinsurance and after reinsurance when the reinsurance is non-
proportional (e.g., YRT mortality risk reinsurance). What approach are companies taking to properly
reflect non-proportional reinsurance in the SERT calculation?

A: The Stochastic Exclusion Ratio Test (SERT) is defined as: SERT = (b-a)/c

Where:

a = the adjusted Deterministic Reserve described in Section 6.A.2.b.i using the baseline economic
scenario described in VM-20 Appendix 1.
b = the largest adjusted Deterministic Reserve described in Section 6.A.2.b.i under any of the other 15
economic scenarios described in VM-20 Appendix 1.
c = an amount calculated from the baseline economic scenario described in Appendix 1 that represents
the present value of benefits for the policies, adjusted for reinsurance by subtracting ceded benefits.

If SERT < 0.060, SERT is “Passed.”

In many, but not all cases, quota-share coinsurance scales “b,” “a,” and “c” roughly in proportion to 1- QS%,
leaving the SERT result similar both pre- and post-quota share coinsurance.

However, under non-proportional reinsurance, “a” is likely to be of similar magnitude pre- and post-
reinsurance, as is “b,” but “c” will be reduced by the ceded mortality benefits. As a result, the SERT ratio
post-YRT reinsurance is likely to be larger than pre-YRT reinsurance—perhaps significantly so, if a large
percentage of the mortality risk is ceded out. The SERT may “Fail” simply because of the YRT cession, even
though actual reserves, and reserve sensitivity to yield rates, might be similar pre- and post-YRT reinsurance.

If a reserve calculation segment passes SERT pre-YRT reinsurance but fails SERT post-YRT reinsurance,
some actuaries may attempt to demonstrate that sensitivity of the adjusted Deterministic Reserve to economic
scenarios is comparable pre- and post-YRT and thus argue that the SERT result should be determined to be a
“Pass” both pre-YRT reinsurance and post-YRT reinsurance. One form of such a demonstration that some
actuaries might use is as follows (and is found in Section 6.A.2.c.i).

For convenience in notation, call the pre-YRT reinsurance results “gross of YRT,” with a subscript “gy,” and
the post-YRT results “net of YRT,” with subscript “ny.” If a block of business being tested is subject to one or
more YRT reinsurance cessions as well as other forms of reinsurance, such as coinsurance, take “gross of
YRT” to mean net of all non-YRT reinsurance but ignoring the YRT contract(s), and “net of YRT” to mean
net of all reinsurance contracts. That is, treat YRT reinsurance as the last reinsurance in, and compute certain
values below with and without that last component.

So, if we have

SERTgy ≤ 0.060 but SERTny > 0.060, then move to a second-stage test as follows:

Compute the Largest Percent Increase in Reserve = LPIR = (b – a)/a , both “gross of YRT” and “net of YRT.”

LPIRgy = (bgy – agy)/agy

LPIRny = (bny – any)/any

American Academy of Actuaries 93 www.actuary.org

If SERTgy × LPIRny/LPIRgy < 0.060, then declare the Stochastic Exclusion Ratio Test result to be a “Pass” on
the post-reinsurance basis.

Note that the scenario underlying bgy could be different than the scenario underlying bny.

For quota-share coinsurance and non-proportional reinsurance arrangements some actuaries might consider
another more qualitative demonstration as provided in Section 6.A.2.c.ii. This qualitative approach is to
calculate the adjusted Deterministic Reserves for the 16 scenarios of the Stochastic Exclusion Ratio Test both
gross and net. Array the results in table and chart, demonstrating a similar pattern of sensitivity by scenario.

Q20.6: Do the assumptions used by both the assuming and ceding company have to be the same for a
specific block of business subject to a reinsurance agreement?

A: Section 8.C.1 of VM-20 states that the ceding and assuming companies are not required to use the same
assumptions and margins for the reinsured policies.

Q 20.7: How might non-proportional reinsurance (e.g., YRT, stop loss, or limits on benefits receivable) be
reflected in the cash flow projections used to determine the Deterministic and Stochastic Reserves?

A: Some actuaries would not include non-proportional reinsurance ceded benefits in the cash flows if such
benefits are immaterial.

Some actuaries would include non-proportional reinsurance cash flows in the projections directly where
possible (e.g., YRT).

Other actuaries will perform separate stochastic analysis as discussed in Section 8.C.2.b to quantify the impact
where a single deterministic valuation assumption for a risk factor will not adequately capture the risk (e.g.,
mortality on a stop-loss reinsurance agreement). Some actuaries would consider stochastic analysis to be
essential for assumed non-proportional reinsurance.

Q 20.8: What does VM-20 mean by treating counterparties to a reinsurance treaty as “knowledgeable
counterparties”?

A: Section 8.C.7 addresses knowledgeable counterparties in the following way:

Assume that the counterparties to a reinsurance agreement are knowledgeable about the contingencies
involved in the agreement and thus likely to exercise the terms of the agreement to their respective advantage,
taking into account the context of the agreement in the entire economic relationship between the parties. In
setting assumptions for the non-guaranteed elements in reinsurance cash flows, the company shall include, but
not be limited to the following:

• The usual and customary practices associated with such agreements.

• Past practices by the parties concerning the changing of terms, in an economic environment similar to
that projected.
• Any limits placed upon either party’s ability to exercise contractual options in the reinsurance
agreement.
• The ability of the direct-writing company to modify the terms of its policies in response to changes in
reinsurance terms.

American Academy of Actuaries 94 www.actuary.org

 • Actions that might be taken by a party if the counterparty is in financial difficulty.

Q 20.9: Will actuaries assume 100 percent selection against the company by knowledgeable counterparties
in all instances?

A: Some actuaries will set assumptions for non-guaranteed elements in reinsurance agreements taking into
account expected experience based on their actuarial judgment informed by historical experience.

Some actuaries will assume less than 100 percent selection against the company, but the actuary may wish to
consider taking into account the financial impacts to the counterparty when setting these assumptions,
including, for example:

• The estimated level of profit being earned (or losses experienced) on the reinsurance agreement in the
year of the setting of any non-guaranteed element;
• The dollar amount of cumulative losses assumed not to be passed on through increased reinsurance
costs; and
• Expected future profitability of the agreement with a higher likelihood of changes in any non-
guaranteed elements by the counterparty if the future profitability is low or negative.

Some actuaries are expected to assume significantly higher likelihood of selection against the company if the
financial impact to the counterparty is significant.

Sections 8.C.11 through 8.C.17 provide specific considerations for setting these assumptions.

Q 20.10: How are assets that support the reserve modeled when held by another party?

A: Section 8.C.14 provides considerations for when these assets should be modeled and how it must be done.

Q 20.11: How should the actuary reflect the creditworthiness of a reinsurance counterparty?

A: Sections 8.C.15 and 8.C.16 require a margin for the risk of default to be included when the one party has
knowledge that the other party is financially impaired. However, if there is no known financial impairment,
then there is no requirement to establish a margin for the risk of default.

Industry practice regarding the definition of financial impairment is expected to develop over time.

Q 20.12: How should creditworthiness of the ceding company be modeled if on assumed business the
assuming company may terminate the reinsurance upon nonpayment by the ceding company?

A: Section 8.C.16 states that the margin for default risk otherwise established for the risk of default may be
reduced or eliminated in this circumstance.

Q 20.13: Is there a difference in the modeling or reporting of authorized versus unauthorized reinsurers?

A: Section 8.C.17 requires the actuary to take into account the ratings, risk-based capital ratio, or other
available information related to the probability of the risk of default of any counterparty. Only if there were
some difference in the probability of the risk of default would there be a difference in the modeling.

American Academy of Actuaries 95 www.actuary.org

Q 20.14: How is the reinsurance reserve credit determined?

A: Per Section 8.D, the reinsurance reserve credit is the excess of the pre-reinsurance-ceded minimum reserve
over the post-reinsurance-ceded minimum reserve.

Q 20.15: How is the pre-reinsurance-ceded reserve determined?

A: Where such a pre-reinsurance-ceded amount is required, Section 8.D.1 requires the minimum reserve be
recalculated, ignoring the impact of any reinsurance ceded.

Section 3.E states that the policy minimum Net Premium Reserve is defined to be the policy Net
Premium Reserve determined in sections 3.A through Section 3.D, less a credit for reinsurance ceded as
defined in Section 8. The Deterministic and Stochastic pre-reinsurance-ceded reserves are determined by
excluding all of the ceded reinsurance cash flows. Section 8.D.2.a requires the Stochastic Exclusion Test and
Deterministic Exclusion Test to be performed, ignoring ceded reinsurance.

Section 8.D.2.b allows the actuary to adjust assumptions to reflect company experience in the absence of
reinsurance such as assuming, for example, that the business was managed in a manner consistent with the
manner that retained business is managed. Some actuaries believe this would include assumptions regarding
asset modeling for the assets backing reserves no longer ceded unless directly addressed in VM-20. Some
actuaries would do this by grossing up the existing asset models.

Q 20.16: How should a company ceding YRT reinsurance model reinsurance rates when calculating
reserves based on prudent estimate mortality?

A: For reinsurance-ceded cash flows, the direct company must determine the appropriate assumption for the
reinsurance premium rates when projected mortality is based on a prudent estimate. The reinsurer likely did
not establish its premiums using VM-20 prudent estimate mortality. If the treaty allows for premiums or
expense allowances to be adjusted and modeled reinsurance premiums are not adjusted, the modeled
reinsurance cash flows may be less of a net cost and potentially a net benefit to the ceding company. Some
actuaries would consider the following sections of VM-20 as relevant to this issue.

• Section 8.C.10 requires the company to use assumptions that account for any actions that the
assuming company (the reinsurer) has taken or is likely to take. This requirement is followed by a
Guidance Note that puts reinsurance premiums and expense allowances in the category of
assumptions that qualify for this treatment.
• Assuming the reinsurer can contractually change premiums or expense allowances, then Section
8.C.11 requires the company to treat reinsurance premiums and expense allowances as similar to non-
guaranteed elements and subject to the requirements in Section 7.C (non-guaranteed elements
section). Within Section 7.C, one of the requirements (7.C.4) is to project NGEs in the cash flow
model consistent with the experience assumptions used in each scenario. Some actuaries may interpret
this language to imply that the modeled reinsurance premiums be consistent with the modeled
mortality assumption, potentially with a timing lag consistent with similar NGE management lags.

Section 8.C.1 requires the company to use assumptions and margins appropriate for each reinsurance
agreement. There are no specific Guidance Notes, practice notes, or established practice to draw upon
that indicate how such margins are applied to reinsurance premiums, particularly after consideration
for the NGE nature of the reinsurance premiums. If the reinsurance premiums have already been

American Academy of Actuaries 96 www.actuary.org

modified to align with the prudent estimate mortality such that the modeled reinsurance cash flows are
reflecting an ultimate net cost in the model similar to the net cost the company expects to actually
emerge, some actuaries may consider this sufficient. It is not clear in reading VM-20 whether the
regulators expect a separate margin to be added to the now-adjusted reinsurance gross premiums. If
so, then this additional margin layer would further increase the net cost of reinsurance in the model.

21. Treatment of Hedging / Derivative Programs

Q 21.1: Do the terms “derivative instrument,” “derivative program,” and “hedging program” all mean the
same thing as used in VM-20?

A: These terms do not have similar meanings. The terms “derivative instrument” and “derivative program”
are defined in VM-01.

The term “derivative instrument” includes, but is not limited to, an option, warrant, cap, floor, collar, swap,
forward, or future, or any other agreement or instrument substantially similar thereto or any series or
combination thereof. Each derivative instrument shall be viewed as part of a specific derivative program.

The term “derivative program” means a program to buy or sell one or more derivative instruments or open or
close hedging positions to achieve a specific objective, and includes both hedging programs (derivative
programs used to reduce risk) and non-hedging programs (derivative programs used for replication or income
generation objectives).

Q 21.2: Are there limitations on including the impact of derivative programs in the minimum reserve
calculation?

A: Section 7.K.1 of VM-20, “Modeling of Derivative Programs,” states that when determining the
Deterministic Reserve and the Stochastic Reserve, the company shall include in the projections (i) the
appropriate costs and benefits of derivative instruments that are currently held by the company in support of
the policies in the calculation; (ii) the appropriate costs and benefits of anticipated future derivative
instrument transactions associated with the execution of a clearly defined hedging strategy (CDHS); and (iii)
the appropriate costs and benefits of anticipated future derivative instrument transactions associated with non-
hedging derivative programs (e.g., replication, income generation) undertaken as part of the investment
strategy supporting the policies, provided they are normally modeled as part of the company’s risk assessment
and evaluation processes. However, Section 4.A.5 states that if a group of policies is excluded from the
Stochastic Reserve requirements, the company may not include future transactions associated with non‐
hedging derivative programs in determining the Deterministic Reserve for those policies.

Q 21.3: What is a CDHS?

A: Section 7.L states that a CDHS must identify:

• The specific risks being hedged (e.g., cash flow, policy interest credits, delta, rho, vega, etc.).
• The hedge objectives.
• The risks that are not hedged (e.g., variation from expected mortality, withdrawal, and other
utilization or decrement rates assumed in the hedging strategy, etc.).
• The financial instruments used to hedge the risks.

American Academy of Actuaries 97 www.actuary.org

 • The hedge trading rules including the permitted tolerances from hedging objectives.
• The metrics for measuring hedging effectiveness.
• The criteria used to measure effectiveness.
• The frequency of measuring hedging effectiveness.
• The conditions under which hedging will not take place.
• The person or persons responsible for implementing the hedging strategy.
• Areas where basis, gap, or assumption risk related to the hedging strategy have been identified.
• The circumstances under which hedging strategy will not be effective in hedging risks.

Section 7.L also states that a CDHS may be dynamic, static, or a combination of dynamic and static. A
hedging strategy involving the offsetting of the risks associated with other products outside of the scope of
these requirements is not considered a CDHS.

Q 21.4: Should costs and benefits of a hedging program that does not qualify as a CDHS be included in the
cash flow model in some way?

A: Section 7.K.1 states that the company shall include in the projections the appropriate costs and benefits of
derivative instruments that are currently held by the company in support of the policies subject to VM-20
requirements. Some actuaries would conclude this includes derivative instruments currently held regardless of
whether these are part of a CDHS. For an anticipated future derivative instrument transaction, however,
Section 7.K.1 only requires inclusion if the transaction is associated with the execution of a CDHS, or with a
non-hedging derivative program that is normally modeled as part of the company’s risk assessment and
evaluation processes.

Q 21.5: How will actuaries determine the expected future hedging program cash flows?

A: Per Section 7.K.2, the company shall reflect the company’s established investment policy and procedures
for that program, project expected program performance along each scenario, and recognize all benefits,
residual risks, and associated frictional costs.

Some actuaries may run their hedging model to project the hedging cash flows (costs and benefits) under each
scenario included as part of the VM-20 stochastic projection. These hedging cash flows would be included as
an offset (or cost) to the cost of the guaranteed benefits projected in the cash flow model that are being
hedged.

Q 21.6: If a derivative program is going to be revised or changed in the future, should this change be
included in the calculation?

A: Section 7.K.1 states that the company shall also include in the cash flow projections the appropriate costs
and benefits of anticipated future derivative instrument transactions associated with the execution of a CDHS
and the appropriate costs and benefits of anticipated future derivative instrument transactions associated with
non-hedging derivative programs (e.g., replication, income generation) undertaken as part of the investment
strategy supporting the policies, provided they are normally modeled as part of the company’s risk assessment
and evaluation processes.

If changes are going to be made to a derivative program, some actuaries may include the changes in the
model, while other actuaries may wait until the changes are implemented to the derivative program,
particularly if the changes to the program are not final or approved.

American Academy of Actuaries 98 www.actuary.org

If future changes or revisions are going to be included in the model, it would be prudent for the actuary to
document the basis for including changes to derivative programs included in the reserve model.

Q 21.7: Do the limitations of projected hedge effectiveness in other reserve and capital calculations such as
Actuarial Guideline XLIII and C3 Phase II apply to the minimum reserve calculation in VM-20?

A: There are no required limits to hedge effectiveness in VM-20. Section 7.K.2 states that for clearly defined
hedging strategies, the company may not assume that residual risks and frictional costs have a value of zero,
unless the company demonstrates in the PBR Actuarial Report that “zero” is an appropriate expectation.

Also, Section 7.K.3 states that where one or more material risk factors related to a derivative program is not
fully captured within the cash flow model used to calculate CTE 70, the company shall reflect such risk
factors by increasing the Stochastic Reserve as described in Section 5.E.

Some actuaries may not use any effectiveness measures because they are not required in VM-20. Other
actuaries may use effectiveness measures consistent with the hedge effectiveness factors used in their
Actuarial Guideline XLIII and C3 Phase II calculations if they have variable annuity business. Other actuaries
may develop hedge effectiveness factors specifically for the business being modeled for VM-20. The actuary
may wish to consider whether hedge effectiveness will vary under extremely adverse or favorable scenarios.

Q 21.8: Are there other things that the actuary should take into account when modeling a derivative
program?

A: As discussed in Section 7.K.2, some actuaries would aim to fully capture the practical realities associated
with the program and the modeling of the program—frequency of rebalancing, transaction costs, margin
costs, basis risk, gap risk, pricing risk, parameter estimation, assumption limitations, and trading limits. These
actuaries would determine these other costs and limitations in the framework of the tail scenarios that end up
determining the Stochastic Reserve.

With dynamic hedging strategies, in some cases the cash flow model supporting the CTE calculation may be
able to adequately reflect residual risks through margins in program assumptions, adjustments to costs and
benefits, etc. In other cases, reference to additional external models or analyses may be necessary where such
results cannot be readily expressed in a format directly amenable to a CTE calculation. In such cases, the
actuary may consider combining the results of such models by some method that is consistent with the
objectives of VM-20 requirements.

Q 21.9: Is there a certification needed if a derivative program is included in the calculation?

A: There is no explicit requirement in VM-20 for certification regarding any derivative program included in
the calculation. Section 3.C.3.e of VM-31 requires a brief description of the approach used to model
derivative programs in VM-20 calculations and a description of any CDHS. Section 3.D.6.f of VM-31
requires a detailed description of model risk management strategies (e.g., hedging) and other derivative
programs, including any CDHS, specific to the groups of policies in scope and not already covered in 3.C.3.e.
Section 3.D.13.a of VM-31 requires a certification from a duly authorized investment officer that the modeled
asset investment strategy is consistent with the company’s current investment strategy. Section 3.D.13.b
requires a certification by a qualified actuary regarding the modeling of clearly defined hedging strategies.

American Academy of Actuaries 99 www.actuary.org

Q 21.10: What are the Valuation Manual definitions of derivative asset program and derivative liability
program, and what VM-20 guidance distinguishes between these programs?

A: Section 7.B.1.h states that the cash flows projected by the company shall include “cash flows from
derivative liability and derivative asset programs, as described in Section 7.K.” For the Valuation Manual that
is effective on Jan. 1, 2018, these terms are not defined in VM-01, “Definitions for Terms in Requirements,”
or in Section 1 of VM-20, “Purpose and Definitions.” Some actuaries believe that an appropriate definition for
derivative asset program would be similar to a definition for “asset-associated derivative” (which is defined in
VM-01 but not used in VM-20), which is a derivative program whose derivative instrument cash flows are
combined with asset cash flows in performing the reserve calculations. Some actuaries believe that an
appropriate definition for derivative liability program would be a derivative program whose derivative
instrument cash flows are combined with liability cash flows in performing the reserve calculations.

Section 7.F.1 of VM-20 specifies that cash flows from invested assets includes cash flows from derivative
asset programs. Likewise, for the calculation of the net asset earnings rate, Section 7.H.2.c.ii specifies that net
investment earnings shall include income from derivative asset programs. Also, Section 4.A.4.e specifies that,
for the calculation of the Deterministic Reserve, the actuarial present value of premiums and related amounts
includes the present value of future derivative liability program net cash flows.

Thus, although not articulated by the Valuation Manual effective on Jan. 1, 2018, some actuaries may
conclude that VM-20 implicitly requires that each derivative program must be classified as a derivative asset
program or a derivative liability program, depending on how the program is modeled for purposes of the
Deterministic Reserve calculation and for each scenario for the Stochastic Reserve calculation.

American Academy of Actuaries 100 www.actuary.org