Reinsurance waivers and risk adjustment

Today, a colleague pointed me back to a December 1, 2019 comment letter to Pennsylvania on their Section 1332 Waiver that I co-authored with Coleman Drake.  We argued two things.  First reinsurance makes insurers far more sensitive to the composition of their risk pools unless there is a substantial change to risk adjustment.  Risk adjustment is intended to make insurers risk agnostic but reinsurance defeats that.  Secondly, the critical premium for enrollment is the minimum cost of coverage and not the price paid for the benchmark silver plan.  This is what Dr. Drake, myself and Dr. Golberstein found in Georgia’s reinsurance program. I want to re-highlight a chunk of the risk adjustment argument as I think there is something good here:

Pennsylvania is proposing a “caliper” approach to reinsurance with an attachment point below which claims are not eligible for reinsurance, as well as a reinsurance cap, above which claims are not eligible for further reinsurance. Pennsylvania will pay a proportion of the claims cost that lie between the attachment point and the reinsurance cap. A significant proportion of the individuals whose claims costs will quality their insurers for reinsurance payments will also have higher than average risk scores.

Three different insurers could have the same average risk score under the current risk adjustment system administered by CMS. The first insurer could cover a population with many low-cost chronic conditions, none of which will ever qualify an individual for reinsurance. The second insurer could cover enrollees that are mostly healthy with no chronic conditions and a modest number of enrollees with conditions that score moderately highly for risk adjustment. The latter group will all qualify for at least some reinsurance, and all of their conditions will not reach the reinsurance cap. The final insurer could cover a population that is overwhelmingly healthy but has a miniscule number of enrollees with extremely expensive conditions, such as history of core organ transplants or hemophilia. A significant proportion of these claims will be above the reinsurance cap.

These three insurers all have the same average risk score. However the three insurers will have very different net of risk adjustment and net of reinsurance claims expenses. The first insurer with many beneficiaries with modest risk scores will bear the entire cost of care with no reinsurance assistance. The second insurer with enrollees who have moderate claim expenses will receive some reinsurance funds. The third insurer which has a small number of enrollees with very high expenses will maximize their reinsurance receipts for a given risk score. The insurers are no longer agnostic to the health status of their enrollees. If the Commonwealth uses the federal risk adjustment system, insurers will have an incentive to attempt to market their plans to certain populations while minimizing plan attractiveness to other populations.

The Commonwealth should consider a new risk adjustment system that carves out reinsurance payments so that the risk adjustment system, in conjunction with reinsurance, transfers sufficient funds to pay for care to make insurers risk agnostic instead of risk aware.

These incentives were identified in a pre-ARPA subsidy environment. These incentives to aggressively become risk aware because risk adjustment is not changing to reflect the costs incurred by insurers net of reinsurance are universal and exploitable. This may be the result of a deliberate policy decision to make enrollees with high levels of risk adjustment the most profitable enrollees net of risk adjustment. That is a defensible position but it should be a deliberate decision.