Assumption Governance and Setting Assumptions in Illustration Actuary Testing

By Mark Rowley

Small Talk, March 2021


Editor’s note: This is the second in a series of three articles published in Small Talk. The first article was published in September 2020 and was entitled “Doing Illustration Actuary Testing over the Life of a Policy.” The third article will discuss model governance and handling challenging situations faced by illustration actuaries.

Assumption Governance

Assumption Governance applies to many aspects of actuarial work. It may not often be mentioned in an article about Illustration Actuary Testing, but it very much applies.

Assumption Governance can be thought of as a checklist. If the checklist is strictly followed, it is likely that assumptions are well-governed:

  • Ideally, assumptions are set based on experience studies that are performed on a regular basis.
  • Experience studies for sensitive assumptions are done more often than those for less sensitive/critical assumptions.
  • It is important to have excellent controls around the experience studies so that all calculations are accurate.
  • The first step in setting each assumption is for the actuary to make a proposal as to how to set the assumption after reviewing the results of the applicable experience studies.
  • The second step is for an independent review of the proposal to be done which can lead to approval of the assumption. The reviewer also ensures that the documentation of the experience study and the analysis supporting the proposal is excellent.
  • The last step on the checklist is to have a centralized location to store all documentation related to assumption setting. This includes documentation of the expected impact of any assumption change.

As will be mentioned below, sometimes there is a lack of credible experience in illustration actuary work, and other approaches must be used. This may involve using other sources of data such as industry data. Many times, the illustration actuary will need to use professional judgment. Since illustration actuary testing is done over the entire life of each policy, it is a good goal to develop enough experience so that at some point, there is credible experience.

Even in situations where there is not credible data, many of the principles above are useful, such as having excellent documentation of how the assumption was set, having one person recommend an approach and another review it, and having a centralized location for the documentation.

Assumption Setting Specific to Illustration Actuary Testing

There is a lot of guidance in Actuarial Standard of Practice 24 – Compliance with the NAIC Life Insurance Illustration Model Regulation related to setting assumptions. Much of what is below is summarized, quoted or paraphrased from ASOP 24, Section 3.4 Developing the Disciplined Current Scale. Let’s go through various assumptions.

Investment Return

It should be based on recent actual investment experience, net of default costs, of the assets supporting the policy block.

Allocating investment income should be done reasonably, which can be portfolio, segmentation, investment generation, or any other method. The method needs to mirror what the company does when it allocates investment income in practice.

Investment expenses can be netted from the investment income or handled as a separate expense. Account for it exactly once—don’t miss it or double count!


Ideally, the insurer’s mortality experience is used, adjusted for risk class.

Consideration should be given to setting mortality rates by age, gender, duration, marketing method, plan, size of policy, policy provisions, risk class, and other items (or a combination thereof) consistent with the insurer’s structure of mortality experience factor classes.

However, a different approach should be used if the insurer’s experience isn’t credible. In this case, use other credible industry mortality experience, appropriately modified to reflect the insurer’s underwriting practices. If no credible experience is available, use professional judgment.

Pricing mortality can be used if it meets the criteria above. This often makes sense right after a policy is sold, but it will need to be monitored and updated over time as experience emerges.


The considerations are similar to mortality.

The actuary should consider credible variations by age, gender, duration, marketing method, plan, size of policy, policy provisions, risk class, and other items (or a combination thereof) consistent with the insurer’s structure of persistency experience factor classes.

Pricing persistency can be used if it meets the criteria above. This often makes sense right after a policy is sold, but it will need to be monitored and updated over time as experience emerges.

Direct Sales Expenses

This is defined as agent commissions, overrides and other direct compensation incurred as a consequence of sales.

This may not require an experience study, but rather just a reflection of what is happening.

All Other Expenses

The first principle is to count everything exactly once, so don’t double count by including anything that was in direct sales expenses.

There are three options, some of which only apply to illustration actuary testing. The choices are fully allocated, marginally allocated or Generally Recognized Expense Table (GRET):

Fully allocated

  • Expenses are based on a recent reasonable expense study.
  • Unit expenses are applied to both in force and newly issued policies and cover all expenses that are not direct sales expenses.
  • There are direct and indirect expenses. Direct expenses are allocated to the policy form. Indirect expenses are allocated using reasonable principles of expense allocation.
  • Non-recurring costs, such as system development costs, may be spread over a reasonable number of years.

Marginally allocated

This is the same as fully allocated except indirect expenses are excluded.


GRET is obtained from an industry expense study based on fully allocated expenses representing a significant portion of insurance companies and approved for use by the NAIC. It is updated annually. Currently the SOA’s Committee on Life Insurance Company Expenses (CLICE) makes a recommendation annually which has historically been approved by the NAIC. It is a good idea to review the new GRET each year, whether or not you use GRET. It is easy to find the annual recommendations by going to and searching on “20XX GRET.” If you are interested in the 2021 GRET, simply search for “2021 GRET” on

Deciding which of the three options to Use

There are rules that provide guidance:

  • If there is no GRET approved and available, fully allocated expenses must be used.
  • If GRET is approved and available, all three methods can be used.
  • Marginally allocated expenses can only be used if they generate aggregate expenses at least as large as those generated by GRET.

It is advisable to test all your options before making a decision, and think about what would be best from a short- and long-term perspective.

It seems unlikely that marginally allocated expenses would be an attractive option very often.


This includes Federal Income Taxes (FIT) and all other taxes.

All cash flows arising from applicable taxes are reflected. Income taxes should be recognized in accordance with their impact by duration.

Investment taxes may be treated as a deduction from the investment return or may be treated separately. Other categories of taxes, such as premium taxes or employment taxes, may be handled separately or included in the category of all other expenses.

Statements of fact and opinions expressed herein are those of the individual authors and are not necessarily those of the Society of Actuaries, the editors, or the respective authors’ employers.

Mark Rowley, FSA, MAAA, is VP, Managing Actuary at EMC National Life Company. He can be reached at