Last night, the Center for Medicare and Medicaid Services (CMS) released their 2020 Risk Adjustment Summary report for the ACA regulated small group, catastrophic and individual health insurance markets. Risk adjustment for these markets works by looking at all of the diagnoses submitted on a claim, assigning a point value to some of the diagnoses to determine average costs for those diagnoses and then sending each individual through a model that modifies the score for known demographics. These scores are then summed for each insurer and a few adjustment factors are added in to create a standard comparable across each state. Within each state and market segment, insurers that have higher than average risk scores get money from insurers that have lower than average risk scores.
There are a lot of nuggets of interest to me in this report. One of the first things was what did the 2020 scores look like compared to 2019?
Risk adjustment state transfers as a percent of premiums remained relatively steady compared to the 2019 benefit year.
In the 2019 benefit year, the absolute value of risk adjustment state transfers as a percent of premiums averaged 9.7 percent of premiums in the individual non-catastrophic risk pool, and 4.1 percent of premiums in the small group risk pool. In the 2020 benefit year, the absolute value of risk adjustment state transfers increased slightly to 9.9 percent of premiums in the individual non-catastrophic risk pool and reduced slightly to 4.0 percent of premiums in the small group risk pool.
2020, at a 30,000 foot level, looked a lot like 2019 on a risk adjustment basis.
Now let’s get down to details….
Risk scores decreased between 2019 and 2020. In the 2020 benefit year, risk scores decreased nationally by approximately 11.9 percent in the individual non-catastrophic risk pool and decreased by approximately 11.1 percent in the small group risk pool when compared to the 2019 benefit year risk scores…
We estimate that the risk adjustment model updates between 2019 and 2020 resulted in a decrease in calculated risk scores by approximately 9.7 percent, and that most of this decrease is due to changes in the recalibration data – that is, if the underlying population was the same in both years, we estimate that 2020 benefit year risk scores would be 9.7 percent lower than 2019 benefit year risk scores as a result of the risk adjustment modeling methodology changes.
So, the early estimate is that the 2020 risk pool is not unhealthier than the 2019 risk pool and plausibly a smidge healthier.
We need to figure out how much of this was the influx of new enrollees during the Special Enrollment Periods and people flowing out of job based coverage and to the Exchanges compared to the same level of unobserved risk and disease burden but differential coding because utilization fell off the cliff for several months? That is an important and tough question.
On first glance, it looks like we don’t have to worry about 2020 adverse selection. My intuition plus what I’ve heard from people with better data is that 2021 adverse selection due to the Re-Open Enrollment Period (get signed up by August 15th) is that it is minimal at this time.