This morning, Petra Rasmussen, Coleman Drake and I have published new research in JAMA Health Forums (open access license) where we attempted to first estimate the number of people who would be automatically re-enrolled in strictly dominated plans under both original ACA and the new American Rescue Plan enhanced subsidies. We then propose and analyze the impact of a “smart default” that seeks to respect peoples’ revealed preferences while moving as many people as possible out of strictly dominated plans.
So what did we do?
First, we used Covered California 2018 enrollment data and simulated re-enrolling these folks back into these plans for the 2021 plan year. Ideally we would have had 2020 enrollment data but that is still an outstanding records request for us at the moment for future projects. We estimate everyone’s premiums and premium subsidies under both the ACA regime and the ARP regime to get the net premium of both their chosen plan and every other plan in the same vertical. We define “strictly dominated” in the same way that Petra and I defined it for our Milbank Quarterly article: The presence of another plan other than the default offered by the same insurer on the same network with the same plan type that has the same or lower net premium and better cost-sharing for the consumer.
So what did we find?
Under the ACA subsidies, 3.2% of enrollees would be placed into strictly dominated plans. The income distribution is overwhelming folks (98.3%) earning under 200% Federal Poverty Level (FPL). The most common scenario is likely folks who bought Gold plans in 2018 being automatically re-enrolled in Gold plans that are now more expensive than the corresponding silver plan with medium Cost-Sharing Reduction (CSR) benefits. The Silver CSR has the same or lower premium and lower deductible and out of pocket expenses.
Under the current policy of enhanced ARP subsidies, 5.8% of enrollees would be placed into dominated plans. This is still mostly a <200% FPL with 89% of dominated defaults occurring for this group. The story is different though. Under ACA subsidies people who bought bronze plans made up 8% of the dominated defaults. With the ARP subsidies, bronze buyers are over 40% of the dominated defaults. These folks had bought plans under ACA subsidies where bronze was cheaper than the corresponding silver plans with CSR benefits. The cost-sharing was higher but premium was cheaper — this is a plausibly rationale expression of preferences. However, under the ARP enhanced subsidies, both the bronze plan and a higher value plan both have a $1 premium, so the higher value plan from the same insurer/network/plan type dominates on cost-sharing.
So what do we propose to do?
We proposed a smart default for automatic re-enrollment where people would be placed into a plan with the same or lower premium and better cost-sharing if it is offered by the same insurer/network/plan type triad.
Dominated defaults are expensive. We estimated that the dominated defaults would cost single buyers an average of an extra $1,200 in premiums and single individuals are seeing over $1900 more in deductibles. Given that the overwhelming majority of individuals effected by dominated defaults earn under 200% FPL (~$25,000), $1200 in premium is 4.8% of household income. That is a lot of money even before we think about the care deterring effects of large deductibles and the stress of likely medical debt. We also find co-pays go down a lot and maximum out of pocket decreases by almost 20% of income for someone at 200% FPL.
We also estimate that there are 327,000 individuals on Healthcare.gov (4.4% of subsidized enrollees) who may also be exposed to dominated defaults.
Choosing insurance is tough. It is real easy to make errors. Good choice architecture and smart defaults can help this challenging choice problem by helping people avoid objectively hideous and inferior choices.
I think that we’re a bit too late into the 2022 OEP prep season for many if any exchanges to adapt our proposed smart defaults, but this is something that insurance exchanges should adapt for 2023.