Response to March 2021 Practice Note on ASOP 6

By Wes Edwards

Health Watch, September 2022

It is no surprise that some aspects of actuarial practice draw more attention from the public and the profession at large than others. The testimony of the chief actuary of Social Security, for example, rightfully gets more attention than the consulting actuary’s statement of opinion on a private employer’s financial statement disclosure liability. With the decline in prevalence of defined benefit post-retirement medical plans in the non-union private sector, the attention given to such plans has declined as well, but open or closed, they remain a material liability for many U.S. companies and public sector entities. As is so often the case, apparently minor changes to actuarial methods and assumptions can potentially have a large impact on the present value of future benefits and therefore have great importance to plan sponsors, plan members and the perceived value of their owners/taxpayers.

ASOP 6 was adopted in 2001. This addressed for actuaries the accounting standard requirement to value the “assumed per capita cost (by age)” and defines this further as the cost “for one year at each age from the youngest age to the oldest age.” There were articles and letters to the editor for the antecedent of this publication, but generally it was agreed that per capita cost by age was not to be dismissed lightly. When ASOP 6 was revised in 2012, the most extensive comments in Appendix 2 related to sections 3.7.7–3.7.8, addressing this precise issue. The only mention of a Medicare Advantage-Prescription Drug (MA-PD) program in the ASOP occurs on page 57 (Figure 1), and the drafters appear to leave the audience hanging by agreeing not to address the issue in response to one comment about what “could be” appropriate.

Figure 1
Comment from ASOP 6 Appendix 2 (2012)

I agree that addressing that issue now, some years later, is worthwhile. It warrants broad discussion of how and whether Medicare Advantage and Part D plan reimbursement increase in proportion to overall morbidity by age.

I also appreciate that a group of Academy members have stepped up with a practice note to address this topic. Issued in March 2021, it appears to reference what would be an important new exception for valuing liabilities for MA plans, exempting them from the requirements in accounting standards that actuarial valuations of per capita cost reflect morbidity differences by age, when it states:

c. Medicare Advantage (MA)¹² plans exist in the form of either individual or group-issued contracts. Each type receives a risk-adjusted federal subsidy that is intended to eliminate any subscriber cost differences due to age, gender, or health status, resulting in an average medical cost for all subscribers.¹³ It is also true that Medicare prescription drug plans (whether offered as stand-alone or bundled with MA as an MA-PD) have a relatively flat age and gender curve after federal payments. In our view, the age independent cost curve created by federal subsidies supports the practice of not age-rating the premiums for these individual or group MA, MA-PD, or stand-alone Medicare Part D drug plans. [Italics are mine.] Note that this exception would likely not apply to individual Medicare supplement plans or to traditional employer-sponsored Medicare-integrated retiree health insurance plans, as those generally would have underlying costs that vary by age.

¹² Medicare Advantage plans must offer benefits that are at least actuarially equivalent to the original fee-for-service Medicare program (Part A Hospital Insurance and Part B Supplemental Medical Insurance), while the benefits for any Medicare prescription drug plan must be at least actuarially equivalent to the standard Medicare Part D design.

¹³ This intent was described to us by CMS actuaries and is supported by the following quote from page four of the Advance Notice of Methodological Changes for 2022 (https://www.cms.gov/files/document/2022-advance-notice-part-i.pdf): “… risk adjustment models … are used to calculate risk scores that adjust capitated payments made for aged and disabled beneficiaries enrolled in MA plans … A risk score represents a beneficiary’s expected medical cost relative to the average expected medical cost of beneficiaries entitled to Part A and enrolled in Part B …”

Some MA plans may have a zero premium, and for those plans, the issue of an accounting liability may be a moot point. There is no liability for a plan for which there are no costs if that situation is expected to continue for the life of participants in a sponsored group retiree benefit plan.

For MA plans that currently have a premium (generally due to plan costs associated with enhanced benefits or variance in care management/administrative costs) or are expected to develop one in the future due to an assumption that Centers for Medicare and Medicaid Services (CMS) funding will not keep pace relative to general trends, age—as measured by aging factors—is still appropriate as the best proxy for the changing morbidity risk of the participants and therefore the measure of plan liability.

If, for a given plan population, it is true that (1) CMS payments to health plans are equal to plan costs at any one plan benefit level and CMS payment (subsidy) level, and (2) both costs and CMS payment are proportionally increased with change in risk as captured by risk scores, then mathematically, it cannot be simultaneously true at a higher plan benefit level if the plan benefits costs increase but CMS revenue is not increased for the higher benefit level. I believe the intent by CMS is for revenue paid to the MA insurer to be adequate but not excessive at the traditional Medicare benefit plan level, but many employers offer plans at a higher level, resulting in non-zero MA premiums. These are the plans the accounting standards require plan sponsors to value. At those higher benefit levels, the net premium (gross cost less CMS revenue differences) attributable to each individual in the group plan will bear the same relationship to the plan’s average net premium as they do to the average gross plan cost. Thus, aging is still necessary, and assumptions recognizing morbidity differences in the valuation are appropriate.

A morbidity by age assumption is the standard proxy in valuation for risk mix. It is easily argued that there are better measures of risk than age. Actuaries agree that risk scoring underlying CMS revenue determination is a better measure of risk, but it is not practical to obtain with census data for use in actuarial valuations, nor have tables been developed for future projected changes in risk scores. Furthermore, MA chief actuaries will advise that “quantifying our ability to outperform other MA providers is easier when we have all of the current data … as part of our job,” so such tables would need to be insurer-specific.

It is the group-specific claims/illness burden relative to the group-specific CMS revenue expected that drives rates. However, that isn’t meant to imply older/younger individuals are higher/lower net cost on average. The measure of claims/illness burden is measured using diagnosis information in claim experience. The ability to capture that experience completely and accurately is based on the time period available for that population and the insurer’s ability to capture those data from claims. The first can vary for reasons beyond the insurer’s control (new business from Medicare Supplement or individual products), and the second can vary by insurer technical skill in data capture. These are additional reasons that standard tables for using measures other than age may yet be some way off for plan liability valuation actuaries.

From a practicing valuation actuary’s standpoint, I am concerned that we should not be quite so quick to eliminate valuation of morbidity by age for MA plans.

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.

Wes Edwards, FSA, MAAA, is a principal with Mercer (US) Inc. Wes can be reached at wes.edwards@mercer.com.