ACA Risk Adjustment in 2021

I’ve been talking to reporters over the past few weeks of what the ACA will look like for the remaining portion of 2021 and all of 2022 now that the Biden Administration is in charge and has the American Rescue Plan passed with a new subsidy table.  This is a great universe to have questions in.  Several reporters have raised a question on what does this do to risk adjustment and so far I have not seen any of that part in print, but given multiple reporters on a variety of beats have asked that question, I think it is worthwhile to get informed speculation out there.

First, be kind to any actuary in your life.  These are rough times and their world is being rocked at the moment.  2021 premiums were priced on 2019 data and a tiny bit of 2020 data.  That is mostly standard practice as premiums are usually priced on the full year two policy years prior and a decent chunk of the prior policy year.  COVID messed this up as March-May data was inherently funky with a once in a lifetime shock event.  During the pricing process, actuaries need to figure out what the projected state wide covered enrollment pool looks like, who they will be selling to and what the average state wide premium will be.  From here, they estimate whether their pool of enrollees are going to be coded as more or less sick than the state wide average.  If they code more sick, the percentage difference is multiplied by state average premium and enrollment duration to get a risk adjustment inflow.  If they code as healthier, the opposite happens to calculate the risk adjustment outflow.

Messing up risk adjustment directions and magnitudes is a common-enough event as the actuaries are projecting plausibly knowable things (company enrollment, company  skill at maximizing risk scores, company plan offerings) and interacting those knowable things with a lot of lightly knowable to minimally knowable things. This is more true in highly competitive states with lots of sub-state regional insurers.  This is mostly irrelevant in states with a single insurer acting as a monopolist for the entire state.

Actuaries made the assumption that Open Enrollment for Healthcare.gov would freeze on December 15 for the 2021 plan year.  That is effectively not true as a second large enrollment period with plentiful publicity and supportive elite earned messaging as well as paid messaging is ongoing right now.

Actuaries made the assumption that the benchmark premium will cost someone earning 150% FPL (<~$19,000/year) will be about $60 per month.  That is no longer true.  The benchmark premium is now $0 for that individual.

These are major changes.

We now know that enrollment is running several times above normal in the “off-season”.

The menu of plans with their relative gross premiums have not changed.  This is a constant.

From here we need to make some assumptions.

I am assuming that the current wave of marginal enrollees, those people who are enrolling over and above normal “off-season” levels, will look a lot like marginal, price sensitive enrollees who signed up after December 10, 2020 for January 1, 2021 coverage.  On average, these enrollees tend to be more price sensitive and healthier than the rest of the cohort.  They are buying on price and they are likely to be on net risk adjustment outflow buyers. We know that these folks are very marginal as they had the opportunity to buy in December and the buy/no buy decision just came up no buy while increased political and public attention has increased the salience of the new opportunity to buy or not buy towards buying in February before the subsidies changed.  The change in pricing schedule for April and May enrollment will attract a slightly different type of marginal buyer (I think).

We should also assume that some people who would have disenrolled for non-payment of non-zero premium will not disenroll or have their coverage terminated due to non-payment of premium as it is now far more difficult to terminate coverage for non-payment when there is no payment that needs to be collected.

So what does this mean?

On the second point first, we should expect a different profile of partial year enrollment factors being applied to groups that were not anticipated to need partial year factors.  There are going to be a lot of people with coverage start dates of March 1, April 1, May 1 and June 1 that will remain enrolled through December 31 who will score slightly higher than actual expenses if they had enrolled on January 1.  Conversely, there will be people who will have full year enrollment who absent the subsidy table change would have disenrolled over the summer or in the fall.  I am not sure of the magnitude or direction of impact.

Furthermore, given the new subsidy tables will be visible in two weeks, I would not be surprised if people buy up a bit.  The buy-up could be minor from a bare bones zero dollar bronze plan to an expanded Bronze plan with an HSA that is also a zero premium plan.  The buy-up could be significant in some counties as zero premium gold plans are available over 200% FPL to 40 year old single non-smokers so people could be switching from bronze to gold.  The buy-up could be modest as people earning under 150% FPL could move to the benchmark plan instead of the cheapest silver plan as both cost them nothing.  In these scenarios, state wide average premium may creep up a little.

Insurers who had a pricing and risk adjustment strategy to attract low risk, price sensitive buyers by offering the cheapest plans possible in bronze and silver for their counties will likely get most of the additional post December 15th membership.  I think that membership is likely to be low to very low utilizers and score few risk adjustable diagnoses.

Overall the entire ACA individual market risk pool risk score will decrease compared to what it would have been with no February-May OEP nor ARP subsidy boost but the relative changes in risk score is unlikely to be uniform between insurers.  I am speculating wildly now, but I think it is likely that insurers that expected to pay out large risk adjustment payments for their 2021 premium years will be hit with a double whammy.  I think that they will be seeing comparatively larger declines in their covered population average risk score compared to insurers whose strategies were based on covering proportionally more sick people at higher premiums with risk adjustment inflows.

What does this mean?

Insurers that expected to pay into risk adjustment will have larger outflows on both the basis of relative health as they will have an even healthier population than anticipated relative to competitors and each point of differential coded health be worth more as state average premium will be slightly higher than projected.