Earlier this week, AHIP had a tweet stating that more growth on the ACA marketplaces was related to more choices. I’m not sure.
The story that more insurers and more choice leads to more enrollment likely goes something like this:
People have varying preferences. They will buy a plan if there is an option on the option surface that is pretty close to their optimal point. If there is not an option near their optimal point of price, quality, brand, cost-sharing, network etc then they won’t buy. More insurers increases the number of points occupied on any choice surface which decreases the average distance from any person’s preference point to their first option. This reduced distance between each individual’s optimal point and the first available option leads to more people buying plans than if there were fewer insurers offering fewer options.
And that make sense.
Except that adding more insurers to a county changes the prices subsidized buyers see. We need to remember that subsidized buyers are premium spread sensitive and not premium level sensitive. More insurers in a given market tends to compress the premium spread. Compressed premium spreads means that all plans that are priced below the benchmark second cheapest silver plan are relatively more expensive.
Below is a dot plot of premium spreads by county on Healthcare.gov from 2014-2021 for the cheapest silver relative to the benchmark silver plan for a single 40 year old non-smoker stratified by the number of insurers in the county.
Generally, as more insurers are in a county, the discounts for below benchmark plans for subsidized buyers shrink!
This is not a hard and fast rule but it is a trend. I would want to add state and year effects to a model, but in general, subsidized prices go up as competition increases. That is weird.
And if we think that marginal enrollees are mostly buying on price, more competition likely leads to one of the most salient features (premium) to become less attractive.
So this is a good empirical question — does adding insurers lead to more enrollment? Or more simply, does pricing dominate choice space density for subsidized and price sensitive buyers?
Fake Irishman
Interesting point. We’d of course have to sort out the differences induced by more generous overall subsidies under the IRA and ARPA, and I suspect that would more clearly reveal the features you discuss above.
David Anderson
@Fake Irishman: I’ve been banging this drum since at least summer 2015… subsidized buyers are spread sensitive, not level sensitive.
Lobo
I think more choices lead to worse overall outcomes. Confusion and overload lead to analysis paralysis or bad choices. My preference would be to have three insurance companies offer three plans across a state or region. Insurance companies would compete to be one of three. The state would mandate certain conditions and oversee the plans. 9 similar plans would be easier to digest and compare. It is a variation on single payor, “few and similar payors”. Hopefully, the three could use their market pressure to lower prices. It is a hybrid model.
Barker
Insurers seeing growth and growth potential will also be more likely to participate – so this may be a “correlation not causation” issue in some respects.
More insurers is also likely correlated with more private marketing investments, increased attention on the exchanges, increased Broker commissions due to more competition, etc. – Given this I wouldn’t be surprised if more insurers is correlated with more growth
That being said I suspect more insurers may also effectively reduce provider choice at affordable price points due to network narrowing, incentives to price-out the competition or become competitively priced. My hypothesis would be that increased competition results in consumers (especially CSR94/87) seeing a decrease in choice of healthcare providers whom they can access at any given arbitrary price point.
Increased competition may also lead to more market turmoil as the lowest-price insurers are more likely to be underpriced. These insurers are probably more likely to exit or file large rate increases in the following years, resulting in consumers having to shop more frequently and change insurers more frequently in order to chase low premiums.
I don’t think this is a guaranteed issue though, one way to address it would be more active-purchaser like engagement from state regulators. I think California would be a good case example.
Roger Moore
I would also assume there can be an issue with analysis paralysis as the number of choices gets larger. It’s a lot easier to choose between a handful of plans, each of which may be really different from the others, than it is to sort through a bunch of plans from the same provider, each of which differs only in details. You could test for that by looking at cases where the subsidy is very small but there’s a variable number of plans to choose from.
Roger Moore
@Lobo:
I think California’s “active buyer” approach is helpful here. The state filters the available plans to make sure the plans from each provider are genuinely different from each other. The net result is that there are a reasonable number of plans in each metal band without setting an arbitrary limit on the number of providers. For example, here in LA County, there are about 10 plans in each metal band even though this is by far the largest county in the country. If I had to buy on the marketplace, I would be able to pick a plan without too much difficulty.
dnfree
I don’t think “Or, more simply” means what you think it means! The whole thing sounds complex, but worth asking.
Fake Irishman
@Roger Moore:
Compare with Harris County (Houston), which has something like 30 plans in a metal band, or Miami, which, sheesh….
Keaton Miller
Looking at cross sections to answer questions about entry is a little tricky since entry decisions are endogenous to local market conditions.
In this particular case, one could tell a story that more insurers in the market is associated with (a) denser urbanized areas (more expensive) and (b) fiercer price competition, necessarily narrowing the spread.
That said, if you can prove out a story where more entry *reduces* the size of the market, you’ve got a home-run Top 5 economics publication there. Well, maybe not. But that would be a very notable finding.
Nit pick: does that plot have the resolution necessary to resolve the density of the distribution? You’ve got roughly 10K points there, if it’s 2000 Heathcare.gov counties * 8 years…