Slow and Steady or Herky-Jerky

The 2020 ACA premiums were released this morning. The press release headline is as follows:

Today, the Centers for Medicare & Medicaid Services (CMS) announced that the average premium for the second lowest cost silver plan on for a 27 year-old will drop by 4 percent for the 2020 coverage year. Additionally, 20 more issuers will participate in states that use the Federal Health Insurance Exchange platform in 2020 bringing the total to 175 issuers compared to 132 in 2018, delivering more choice and competition for consumers. As a result of the Trump Administration’s actions to stabilize the market, Americans will experience lower premiums along with greater choice for the second consecutive year.

The cynical take on this paragraph is that 2018 was a local minima due to massive policy, political and legal uncertainty. 2019 and 2020 are reactions to the re-stabilization of expectations and rule sets which were not present in 2018.

We know that 2018 was massively overpriced. We know that as insurers are paying out very large Medical Loss Ratio (MLR) rebates for the 2016-2018 plan years and are expected to pay out even larger rebates for the 2017-2019 period. If we assume that the final 2020 rates are reasonably rates that “should” be close to actuarially fair plus “normal” profits and admin costs, there are two pathways that we could have arrived at this price level with a starting point of 2017 being priced at a reasonable level high enough to cover claims and normal profits. The first is the path we took; massive rate hike in 2018 and then slight declines in 2019 and now 2020. The other is a slow and steady route.

slow and steady vs herky-jerky 2018

2 ways to get to 2020 rate levels

Both paths get to the same exact endpoint. One path produces the headline of multiple years of premium decreases. The other one produces a headline of steady rate increases.

ACO proliferation and the Exchanges

This morning, several of my colleagues have updated their most recent estimates of the spread and scale of Accountable Care Organizations (ACO) in the Health Affairs blog.

This year has been eventful for value-based care in the United States. While there has been a total reduction in ACO participants, the number of lives covered under these models has continued to grow. The new ACO program Pathways to Success has been implemented, and dropouts were significantly less than feared, while new organizations moved into the program, even with the enhanced risk requirements. There has also been an increase in the number of physician-led ACOs, and they have moved to downside risk at a slightly higher rate than ACOs with hospitals. Beyond the ACO models, there have also been significant new programs announced which will increase the opportunities for providers to move toward managing the care of populations.

Last week, CMS Adminstrator Seema Verma gave a talk at Duke University on the movement towards alternative payment models such as the ACO program. One of the things that struck me was a statement that CMS was trying to get more ACO like principles into the Exchange program in order to hold down premiums. This was intriguing for two reasons. First, it is a good indicator that the CMS priority is unsubsidized folks and not subsidized folks as lower premiums can often hurt subsidized enrollees. That is not new.

The second, and more important thing that struck me was how to operationalize an ACO in a high churn environment. Medicare and employer sponsored insurance ACOs face a far smaller churn problem. Some categories of eligibility in Medicaid face low churn while other categories of eligibility in Medicaid see high churn. The ACA individual market is a high churn segment

We know that about half the ACA enrollment pool that is signed up on January 1, 2019 will not be signed up for insurance through the Exchange on January 1, 2020. We strongly suspect that there is a differential effect by health status on re-enrollment and continual enrollment. Of the proportion of people who have at least thirteen months of continual enrollment, a decent number of them will be switching insurers. Intra-insurer churn was suppressed in 2018-2019 as monopolies were increasingly common, but more insurers entering and expanding footprints for 2020 will increase intra-insurer churn. Again, we know that this switching behavior is non-random on health factors and demographics. The folks who are continually enrolled in the same insurer for thirteen or more months are highly likely to be a much sicker and older group of beneficiaries than the pool of people who were assigned to an ACO in the first month.

I would not be surprised if prospective assignment led to an primary care physician participating in an ACA ACO model seeing eighty percent of their panel zero or one times during the year. Of the folks that they see once, I would be surprised if they see that person again in the following year. There are some folks who will require intensive case management and coordinated care who are on an ACA plan where the incentives and time horizons for ACOs to pay-off readily line up, but this is a fairly small population.

I am confused at how to operationalize ACOs in a very high churn environment.

Medicare and Medigap vs Medicare Advantage

P.A. asked a good question yesterday:

I’m 60 and not yet investigating, so here’s my Q: does MedAd ‘require’ MediGap insurance the way trad Medicare does? That would be another layer of complication.

One of the big advantages Medicare Advantage has over traditional Medicare is that it does provide a catastrophic cap on financial exposure.

Traditional fee for service Medicare Part has no out of pocket limit.  Beneficiaries can buy that protection with a MediGap supplemental policy.  There are roughly a dozen standardized supplemental designs at various premium points.  Each plan design has a different coverage matrix for the deductible, co-pays and coinsurance as well as a maximum allowed out pocket limit.

Medicare Advantage plans may also wrap in Medicare Part D drug benefits into the same plan.  There are some nifty incentive and substitution effects at the plan management level for health economists to geek out about but the primary point of interest is that a comprehensive Medicare Advantage plan reduces the number of decisions that need to be made and pieces of paper that need to be tracked.  People who stay within traditional Medicare need to manage hospital and professional insurance with their own unique cost sharing (Part A and B), potentially buy a Medicare Part D drug plan and look into a MediGap supplement to get some catastrophic protection.  People who buy A&B only Medicare Advantage can avoid thinking about supplements.

Medicare is complicated.  It is a weird benefit design that was cutting edge Blue Cross/Blue Shield thinking from 1964 with several kludges added to it since then to solve real problems.  Medicare Advantage plans have the potential and ability to be a bit less kludgy.  And sometimes that non-kludginess is quite valuable to people in the right circumstances.



Oh plans and jam

Michael McWilliams and others in 2011 looked at the impact of cognitive capacity and the number of Medicare Advantage plans available to Medicare beneficiaries to see if those factors mattered in plan choice in a Health Affairs paper.

The number of plans mattered a lot:

People got overloaded with choice and chose the default, traditional Medicare, when there were too many plans to sort out.  There is a great study on the overload of choice in choosing jelly where a big spread of choices could attract attention but not purchases while a curated set of choices led to more conversions of interest into purchases.  This applies to at least Medicare Advantage as well.

I have two wonders from this study:

  • Do insurers intentionally overload a market to risk select in a strategic manner?
  • Does this generalize to the ACA population

I am not sure where to get started with the first question.

The second question requires some data that I can not freely access (if you are a potential collaborator who knows health surveys with linked IDs to or state based marketplaces, let’s talk). However, we can freely identify the variance in plan offerings. I am using 2019 Public Use Files to count the distinct number of plans on by county for all on-exchange plans, including catastrophic.  Individuals over the age of 30 and those who do not file for an exemption will see slightly fewer plans.

The range is a single county (Holmes County, Ohio) with 1 plan being offered to a single county (Seminole County, Florida) with 110 plans being offered. 2019 is a little less spread out than previous years:


Plans offered by County on 
YearMinimum plans offeredMaximum plans offeredMean Plans offered


I think a priori that a single plan county is giving up potential  enrollment. Lost enrollment will happen for two reasons in a single plan county; first, that plan may not have desired attributes and secondly, far more mechanically the single plan can not create anything cheaper than the benchmark plan so the pricing won’t be attractive to low risk, subsidy eligible, individuals.

I also think, a priori, that counties with 50 or more plans in them are probably losing some counter-factual enrollment as well. Again, this is partially a pricing problem, but to a lesser degree than in the single plan only county as there are numerous bronze priced below benchmark. But I also think it is a search cost problem. Figuring out the minuscule differences between the plans is mentally exhausting. Individuals who know that they have a strong need for insurance will pay that cognitive management tax. Individuals who are flipping a coin between insurance and not-insurance have a non-zero probability of rage quitting.

I don’t know where the optimal range of plans being offered lies if we are optimizing on enrollment. I am very confident that the number is greater than one and significantly less than 110.  Beyond that, I am uncertain.  9 to 15 plans does not sound crazy to me. 15 to 30 plans also sound facially plausible to me.  I’m guessing at this point.

I am curious.

Open enrollment thoughts

The Medicare open enrollment period has started. There are some issues with the plan assistance tool, Medicare Plan Finder. I think it is worthwhile to go over some general thoughts and decision rules about open enrollment. We’ll talk generally and then specifically for Medicare.

Picking an insurance plan, be it Medicare, be it group insurance through work or individual insurance on the Exchange is tough. It is a cognitively demanding task that has multiple threads of uncertainty pulling your decision processes in conflicting directions.
Read more