The ACA health insurance markets rely on individuals to choose a single plan from a menu of plans of varying lengths. At some points, in some counties, there was a single plan to choose from. Other counties in other years have had over 150 plans to choose from. There are some notable divisions of relevant plan characteristics. Different insurers have different reputations and different product profiles to offer. The metal bands function as a signal of one interpretation of value. But it can still be a complex choice environment with 45% of Healthcare.gov enrollees seeing 50 or more choices in 2021, and likely to see even more choices in 2022.
There can be meaningful difference between all fifty or more plans. There can also be plans that are barely differentiable besides a few benefits that no one is likely to use. The Obama Administration had very weak meaningful difference rules where plans had to be distinct from other plans on one of seven criteria. The Trump Administration got rid of those rules. I don’t think there was much of an impact either way.
I think that the choice space is challenging and that more distinctive choices would be beneficial which means more aggressive choice curation would improve consumer and social welfare. I am coming at this from a perspective advanced by Herb Simon which says that we’re messy barely evolved East African Plains Apes that get easily confused and have limited cognitive resources to figure things out. Information and search costs can be expensive. We have bounded rationality. Therefore we adopt heuristics, and schemas. We see a good enough solution but likely suboptimal solution. We satisfice.
We also have predictable and consistent known failure modes. We hyperbolically discount. We anchor on primacy and recency. We anchor on attention. We are a priori rational when we devote a whole lot of energy and effort to that, and retrospectively rational once we invent a good story to tell ourselves.
This is how I approach the world. I think that at least in the health insurance choice space, we have pretty good evidence that inertia is important and plenty of people make dominated choices which are fundamentally irrational. From here, changing the size, scope and complexity of the choice space is an empirical question as to what direction social welfare moves.
The other way to think about decision choice is a Milton Friedmanesque set of assumptions where people are either rational or act as if they are rationally utility maximizing based on currently available information. From this set of assumptions, increasing the choice space either reduces the variance between an individual’s ideal point and the closest available option or does nothing at all. In either case, increasing the choice space does not decrease welfare while it can increase social welfare.
In either set of assumptions, a proliferation of expert tools and decision supports to help people act as if they are rational at low cost to themselves are a good thing. If there are five hundred hidden possibilities and only half a dozen presented options that match to revealed preferences, then the quality of choice improves and social welfare is likely better off. But there is a big difference in analytical outcomes if there are over a hundred and fifty visible choices and we apply either a bounded rationality perspective or an as-if rational maximization perspective.
Betty
I hate this healthcare system, and I am only faced with the multitude of Medicare supplemental plans. It is fundamentally unfair.
Butch
I have to admit I’m not sure I understand this post. We live in an area where there’s no competition; the only plans we can afford are Bronze and they’re pretty much like being uninsured, so I don’t know what “choice” we really have.
The Castle
I endorse this blogpost. 50 or more choices is an overwhelming number, and without guideposts of some kind, is simply beyond most human brains’ capacity to analyze with any kind of rigor. This is a large reason why people stay in sub-optimal health plans year after year — the experience of trying to pick a new one is so confusing and onerous.
This kind of artificial complexity serves the insurance companies much more than it does consumers.
As a behavioral scientist, it has dawned on me over the years that the cognitive psychology paradigm of brain=computer is dramatically wrong and has led us astray. Brains are not computers, and as computers have grown to be super fast and thus able to easily solve problems through brute force computing, the analogy of brain as computer has become less and less apt.
Similarly, it has been obvious for a long time that the Milton Friedman idea of seeking maximal utility is a nice theory, but that brains don’t act that way. Yet the fiction lives on…this idea that people should act like corporations (and why not, if corporations are people!)
As a side note, and only somewhat anecdotally, it seems to me that company-offered health plans have decisively moved to fewer choices and smaller differences among these plans. Why the difference? I would bet that a company with 50+ health plan offerings at open enrollment would get a lot of complaints, no?
TheOtherHank
The image is broken for me on Firefox on a Mac
David Anderson
fixed
Anonymous At Work
The Friedman scenario you present has more than a few “magic asterisks” about the calculating capacity, self-awareness (both of their own health and their own rationality/intelligence ala Dunning-Kruger), and ability of users to turn on a dime (aka not getting locked into a preferred network even if the plans no longer are “rational” to use in that network.
I did a lot of analysis of Simon and bounded rationality as it applies to antitrust law way on back in law school. I preferred to use Thaler, Kahneman, and Tversky (plus, you know, Akerlof, the grand-daddy) instead. Choice overload among options not deeply-understood is a recipe for disaster and insurance options are not “deeply” understood among the general populace (I know, True Irishman Fallacy/Begging the Question), so I have doubts.
Still, miles better than U-Chicago models and their Homo Econimus assumptions.
MattF
Optimizing a single variable in a multi-dimensional space with constraints is hard. Very hard. My own experiences with computations of constrained optimization are grim— those cartoons you see of optimizers climbing Gaussian peaks and sitting happily on the top are not how optimization works in-real-life. IRL, optimizers will gleefully run to any nearby edge case and stay there. There is actually a theorem in linear optimization theory stating that this will happen in any (overly) simplified case. And, yeah, ‘utility’ is a mythical beast.
Lobo
” We have bounded rationality.”
I am a student of Kahneman and Tversky. As a behavioral economist, the pitfalls are everywhere when operating in a Friedmanesque universe. The simple experiment that shows when choices reach a certain point, people narrow their picks. Choice overload becomes the issue. Realizing this, I operate in a good enough universe where the goal is to not find the optimal unicorn but a choice close enough to optimal to work. What this means in health insurance is difficult. There should be a strong basic package of essential health coverage and optional modules to fill it in. Of course what is “basic” becomes the question.
Anonymous At Work
@Lobo: If you look into the mirror and say the word “Satisficing” 3 times, a neo-classical economist’s head explodes…
David Anderson
@Anonymous At Work: I need to try that more often then.