Last week, I had two papers go through meaningful milestones. The first paper on dominated plan choices in California Petra Rasmussen and I sent back to the editors after we approved the copy-edits. It is almost ready for public consumption and I can’t wait to share it with everyone! The second paper was a joint collaboration with Petra and Coleman Drake, another very frequent co-author. We submitted the manuscript last Thursday. It went under review over the weekend. That manuscript integrates three streams of research that the three of us have been working on for several years. Petra and I have been working on dominated choices; Coleman and I have been working together on zero premium plans while Coleman has been working with other researchers on choice inertia. We got to play hard with those concepts. During the process of creating that manuscript, I realized how much my thinking has changed in the past five year on ACA marketplaces.
In 2014-2017, I was almost exclusively a price and relative price above all else in how I thought about plan choice. I assumed that lower net premiums were the cause and solution to most problems. I geeked out on silver gapping, I gawked at large county level discontinuities in pricing. I thought that advertising and navigation mattered, at least a little bit but not a ton. However, my thinking at this point was mostly micro-founded economic analysis where low prices are better and high prices are bad. I, of course, acknowledged that plans would have different attributes and some plan attribute sets would be worth a higher willingness to pay for some buyers than others. Some people would prefer and pay for a bigger network or a PPO plan and thus the pricing of those plans would be higher. But I was fundamentally thinking of everything in terms of pricing with the assumption that future year markets were very flexible and responsive to changes in relative pricing from current consumers. Some of that was misreading the data that 40% of plan buyers in each year were new to their insurer — these were mostly people coming into the market for the first time. Some of it was just misreading the market.
I really started to think that price mattered, but so did everything else in spring 2017. We had seen enrollment fall off in the last 10 days of the 2017 Open Enrollment Period when the federal government went from supportive to opposition of the law. There was no price change at that time. There was a series of cool papers that looked at advertising effects in Kentucky and elsewhere. By the fall of 2017, I was thinking in terms of pricing but had started to think in terms of information as well for the first time. By fall of 2018, I was starting to think that pricing was being dominated by other factors as I had observed that there were free Gold plans in Oklahoma for anyone subsidy eligible but not a mad dash of folks into those free gold plans from dominated bronze and silver plans. I had also seen in Tennessee a massive affordability shock in some counties that had gone from no Silver gap between the cheapest silver plan and the benchmark offered by a monopolistic insurer to the largest Silver gap in the country as a new monopolist with a different strategy took over the “bare” market with massive discounts. If pricing mattered, then zero premium CSR-94 silver plans should have juiced enrollment in these counties. It did not.
This was weird. Crespin and Deleire had shown that insurer exit led to large enrollment losses. Coleman and I had shown that something meaningful and big was happening with zero premium plans on the level of county-year enrollment. We were also seeing huge relative discounts in subsidized pricing showing up in monopolistic counties in non-expansion states while Trachtman found adverse selection from the political lean of a rating area. This was shaping my thinking as well as everything that Herd and Moynihan have written on administrative burden. And then between Coleman, Petra and myself, we had a very productive research year in 2020 with stuff to come out in 2021 and 2022 on inertia, frictions and other non-cash considerations for enrollment.
Right now, my head is at the following place:
- Good pricing is better than bad pricing
- Pricing is not everything
- Information costs are massive
- Inertia is massive
- Lubrication is critical for decent (but not optimal) choice
I’m not sure where my head will be in the next couple of years, but I felt like I owed it to you all to explain how my thinking has evolved over the past several years to a very heavy behavioral econ point of view.