Adverse selection is an important consideration when pricing in health care that may need to be analyzed and quantified. Adverse selection occurs when potential policyholders have an opportunity to choose from different health care benefits and can choose the benefit that most closely matches the policyholder’s needs. Among scenarios are policyholders who generally use more medical coverage choosing higher benefit plans than healthier potential policyholders. Another situation is when one benefit plan covers a particular service that is not covered, or is covered at a substantially lower level, by other plans. Still another scenario is if one benefit plan allows services at different providers than other plans. Not all situations involve adverse selection but adverse selection should be analyzed to help ensure rate sufficiency. Quantifying adverse selection can be difficult and there may be subjectivity in the impact quantification process in certain situations.
One of the simplest adverse selection scenarios is a situation where potential policyholders have a choice between two or more benefits with significantly different benefit levels. Various tables quantifying adverse selection exist. Attached is an Excel file containing a sample table with worksheets for two-benefit and three-benefit plan scenarios. Two key assumptions in these types of tables are the minimum selection factor if a very small portion enrolls in the lowest benefit and the highest selection factor if only a small portion enroll in the highest benefit. The assumption made in the attached tables is that if only a few members enroll in the lowest benefit level then the result would be a selection factor of 0.700, meaning their experience is expected to be 30% lower than average for the group in total. Likewise, if only a few individuals enroll in the highest benefit plan then the selection factor is assumed to be 1.500, meaning that these individuals may have costs 50% higher than the average costs of the entire enrollment group.
As an example of the use of the tables, assume an employer group with 800 potential policyholders gives its employees a choice of three plans. The lowest level of benefits is free, is a high deductible health plan through Insurer A with a $5,000 but has no cost-sharing after the deductible is met. The middle benefit plan is through Insurer B, requires the employee to contribute $100 a month towards the premium, and only has a $2,500 deductible with no cost-sharing after the deductible. The third benefit plan through Insurer C is a $250 deductible plan with 80% up to $1,000 out of pocket, which costs the employees $300 per month. Under the assumption that the average enrollment in the low, middle, and high benefit plans will be 500, 200, and 100 employees, respectively, then the corresponding selection factors are 0.923, 0.995, and 1.398.
The knowledge gained in the example above is important for many reasons:
From a pricing perspective, Insurer C needs to know that its benefit offering to this group is the highest benefit offering and that it potentially (based on its projected enrollment) could get enrollees that have costs 40% higher than the group, on average. If this is not built into the rates for this group, either by use of experience which has this same 40% adverse selection or by applying a 1.398 selection factor to claims from a population where little or no adverse selection exists, then Insurer C will likely take substantial losses on individuals enrolled in this plan.
Insurer A and Insurer B would also want to ensure that the anticipated selection factor for this group is consistent with the experience from which the rates were built or that an appropriate adjustment would be made. There is often conservatism built into rates due to uncertainties surrounding the actual enrollment that would be obtained. For example, Insurer B may decide to make no adjustment for adverse selection since it is an estimate and the actual adverse selection could easily by higher than the anticipated 0.995. Insurer A may also decide to not adjust its rates lower even if the 0.923 selection factor is lower than the selection factor for the population from which the claims experience was taken.
The employer may also wish to use adverse selection, in addition to benefit differential, as part of its contribution strategy.
Provider groups with certain forms of payment arrangements, such as capitation, also need to be concerned about adverse selection. If Insurer C physicians are capitated then the physicians who receive this capitation may receive insufficient funds to cover the costs of the individuals capitated, even if the capitation is age-graded.
Use of adverse selection factors in underwriting (when allowed), pricing, or setting contributions by plan can be difficult since the rates and contributions are based on assumed enrollment by plan, which may vary substantially from the enrollment actually obtained. Insurers sometimes retain the right to recalculate rates if enrollment differs by more than 10% from the enrollment assumed in the rates. Part of the reason for this is the possibility that this membership differential may substantially affect the adverse selection inherent in the claims experience.
Often determination of the existence of adverse selection is very difficult, such as a situation where Plan A has higher drug benefits than Plan B but a higher medical deductible. It may also be the case that there are, direct or indirect, limits on certain types of services in Plan A or Plan B, or one of the plans may include dental benefits or gym club membership discounts. These situations may lead to a subjective estimate of the impact of adverse selection.
Adverse Selection Spiral
The concept of an adverse selection spiral exists in many aspects of our health care system. In the example above, Insurer C would need to add roughly 40% to its rates in order to ensure that it does not lose money on that particular group, given its anticipated enrollment level. However, the additional 40% added to the rate might cause fewer individuals to enroll in the product, thus increasing the adverse selection factor. In the rates for the following year, Insurer C would need to add an additional adjustment, along with normal trend, to the rates to account for this unforeseen adverse selection. This larger increase may result in an additional number of persons switching out of the plan, feeling they are not getting their money’s worth out of the benefits and contributions. This would result in further adverse selection to be added into the rates the following year and the cycle would continue. Insurer C may never get its rates to be sufficient as it may always be trying to catch up with actual adverse selection. This is known as an adverse selection spiral. Some insurers do not want to have the highest benefits in an offering due to the possibility of an adverse selection spiral.
Some blocks of businesses for certain insurers are currently in an adverse selection spiral, of varying magnitudes, as a result of outdated benefit levels. For employer plans, some employers set employee contributions in an attempt to avoid an adverse selection spiral, although other employers are willing to have the rates for higher benefits escalate each year.
Offsets to Adverse Selection
Complicating some adverse selection analyses are some state and federal programs that provide additional funding to insurers based on health status of members enrolled in its plans. These plans include risk adjuster and reinsurance programs, most notably in Medicare and health care reform enacted through PPACA. This means that even though the factor for Insurer C in the example above shows a selection factor of 1.398, Insurer C might be able to reduce, or even eliminate, the adverse selection if it were entitled to risk adjuster and/or reinsurance payments for these individuals. There may even be situations where insurers seek out certain individuals otherwise thought of as selection risks as the insurer may believe that the additional risk adjuster and reinsurance payments are more than enough to offset the actual adverse selection of the individual.
Adverse selection is one of the many nuances in health care for which there is no correct answer. Every situation is different and the individual doing the analysis must be attentive to the many different aspects of the situation being reviewed. This narrative is only intended to give the reader an idea of the components of adverse selection so that a proper analysis process can be designed.