Georgia, reinsurance and risk adjustment

Earlier this month, Georgia released the draft for a two part Section 1332 ACA individual market waiver.  The first part is a big reinsurance program.  The second part is an attempt to implement  the Cruz Amendment for the individual market while injecting additional state funds.  Both of these policies are legitimate attempts to solve real problems of affordability.  I have been thinking about the waiver as I am trying to figure out if and how I want to write a public comment over the Thanksgiving break.

 

Risk adjustment is intended to make insurers risk agnostic.  Insurers are supposed to be paid for the average level of expected medical costs so that they can compete on customer service, networks, ability to negotiate better prices for goods and services, and care management instead of optimizing the competition on the ability to avoid sick people.  The ACA risk adjustment system is not perfect but it is pretty good.  The Georgia reinsurance proposal has significant risk adjustment concerns due to its program design.

• Georgia proposes three levels of pay-out from the reinsurance program
• Georgia proposes a fairly low start point ($20,000 in claims)
• Georgia proposes a high end point ($500,000 in claims)

Current ACA risk adjustment is based on a regression model that attempts to estimate the incremental cost of a disease category after controlling for a bunch of factors. That coefficieinct is then multiplied by the state wide average premium and adjusted for actuarial value and geographic pricing differences to estimate the value of a given diagnosis.  Some conditions, like asthma, are fairly common but also have fairly low risk adjustment values.  Other conditions like hemophilia are rare and very high value.

 

Georgia’s reinsurance proposal will make insurers far more risk aware once implemented then they are today.  This risk awareness will vary by region and it will be highest where the reinsurance pool pays out at the highest rate.  i will want to illustrate why this is the case by taking the adult silver coefficiencts from 2019 for some relevant disease categories and simulate what different insurers will end up paying.  I equalized all diseases to be the equivalent of the expected risk adjustment value of one individual with hemophilia at national average premium.

 

Insurers that are heavy on very expensive and rare conditions will receive more reinsurance funds than insurers that have the same risk score but whose enrollees have far more common and lower cost conditions.  This is doubly true in regions where the reinsurance pool is paying out 80% of the qualified claims.  Between risk adjustment and reinsurance, an individual with hemophilia or metastatic cancer or a history of organ transplants could become an extraordinarily profitable individual for an insurer operating in the 45% or 80% reinsurance zones.  Insurers in these regions will have very strong business case incentives to aggressively enroll everyone who has a qualifying condition or find ways to transfer individuals who have high risk adjustment value and high claims costs from one side of a county line to another if the difference between a 45% and an 80% pay-out is only a few hundred feet.

[UPDATE 1:  A fellow insurance nerd pinged me to point out that hemophilia is a bad risk adjustment and a great reinsurance example due to high in-group variability of claims ranging from $100,000/year to $12,000,000/year.  They are right.  Metastatic cancer that is treated with on-patent drugs that run in the six figures would be a better example ]

 

What should Georgia, or any other state that is enacting big reinsurance programs, do?

 

They should run their own risk adjustment programs where the coefficiencts are modified to account for reinsurance or they can put a back-end fix to correct for risk selection.  Running state based risk adjustment is not a simple task as shown by the fact that while each state is allowed to run their own risk adjustment, every state elects to use the federal system.  However, state based risk adjustment would restore risk agnosticism to the system instead of this massive invitation to game the system and rent really nice apartment blocks 10 feet over selected county lines for very high cost individuals.