I want to talk about some research that my co-author, Petra Rasmussen, and I are presenting at Academy Health’s Annual Research Meeting this week.
We wanted to look at how people made choices when there was a clearly dominated plan on their choice menu.
The short version is not too well.
The long version is not too well. Choice architecture, defaults and inertia matters a lot.
A dominated choice happens when there are two choices, A and B who share basic relevant attributes. On all of the relevant attributes A either ties or beats B. In that case A is the dominant choice and B should never be chosen if we’re working from a rational choice framework. Dominated choices are fairly rare in insurance because there are seldom opportunities for all attributes to be equal. Dominated choices are unlikely to occur in the ACA markets because Silver plans are usually a distinctly different item than Gold plans.
However, silver-loading and California’s actively managed marketplace produced the possibility of strictly dominated plans being offered in 2018. We looked into this.
California has a rule that an insurer can only offer a single plan in each metal level/network/plan type. Furthermore, California only allows insurers to offer standardized benefits. All silver plans have the same deductible, co-pays and co-insurance and maximum out of pocket limits. All gold plans share the same cost-sharing structures. The only variation at a particular metal level that consumers, in a given rating area, see will be premium, insurer name, plan type (HMO, EPO or PPO) and network. From 2014-2017, this does not produce dominated plans. There is a fairly orderly trade-off within insurer/plan type/network verticals. A Bronze plan has lower premiums but more cost sharing than a standard silver which has lower premiums and more cost sharing than a gold offered by the same insurer. Controlling for insurer/plan type and network, we see a clear premium-cost sharing trade-off. People with a mix of preferences and expected future healthcare costs can rationally calculate that Gold is the best plan or Bronze is the best plan for their personal scenarios. Nerds will lay out the odds of each of these choices being retrospectively right, but there is fuzziness in what the optimal choice is.
In 2018, this changes. California, in response to the threat and the eventual termination of direct federal reimbursement to insurers for Cost Sharing Reduction (CSR) subsidies, engaged in a practice known as Silver-switching. Silver-switching is the practice of placing all of the costs of CSR subsidies onto only on-Exchange silver plans. Pragmatically, this means Silver premiums spiked relative to Gold premiums. Petra found significant plan switching in a previous paper while I, with another batch of co-authors, found massively increased affordability for the cheapest subsidized plans. ** Relative price spreads shrank massively for all plans. For some insurer/county combinations in California in 2018, the premium spreads inverted so gold was cheaper than some insurers’ silver plans in some counties.
This was our opportunity. We looked at people, earning over 200% FPL to wash out CSR messiness, in these counties who bought either a particular insurers’ dominant gold plan or dominated silver plan. A pure rational choice model would predict that no one in this universe bought silver and everyone bought gold from the same insurer/plan type/network.
Is that what happened?
Lots of people bought silver plans for 2018 that cost them more in monthly premiums than the gold plan that they had available and if there was any medical utilization during the course of the year, their cost sharing in silver was never lower and often higher on both a per service and cumulative basis than the gold plan. We calculated that this population on average spent several hundred dollars more per family in annual premiums for a bad choice and could not calculate the cost-sharing loss but can assume it is, on average, significant.
Choosing insurance is tough.
** Side note, this paper originated from a long discussion after a presentation session where soon to be Dr. Rasmussen presented her plan switching work at ARM-2019. We had similar interests in ACA choice behaviors so we stayed in touch and kept on chatting back and forth on ideas that could be interesting to poke at. And now we have a paper that is working its way through the review process.
I have a hard time seeing how these types of quasi-random interactions happen during virtual conferences. I wonder if there are economists already prepping discontinuity shocks by discipline on new co-author ties depending on if a major conference was before or after the shut-downs.