Advantages of state based exchanges

Virginia’s governor is considering running a state based exchange for the Affordable Care Act. The Roanoke Times reports:

The group also likes the idea of creating a state-operated insurance exchange, which the General Assembly initially supported and then abandoned in 2011 at the recommendation of Gov. Bob McDonnell…

Several other states such as New Mexico and Nevada are considering leaving and moving to their own State Based Marketplaces (SBMS) in the next two years.

There are several reasons why this might be attractive. The biggest one risk pool composition.

Right now charges 3.5% of premium as the price to run With that money, the feds are supposed to upgrade the system, buy server space, run phone centers, perform outreach and keep the entire process humming. Since January 20, 2017, the federal government has dramatically reduced advertising and outreach expenditures while still collecting the entire fee. The code base of is fairly stable at this point and the demand for server space is also stable as enrollment is slightly declining for three years in a row now. should have the money to spend on outreach. And they are not spending money on outreach.

Shafer, Fowler, Baum and Gollust have shown that engagement drives enrollment activities. Shafer and I have a forthcoming paper that shows immediate enrollment activity changes on due to changes in messaging regimes. We know that engagement works to drive marginal buyers to and away from whenever the political powers that be are in favor of one of those results. The SBMS, like California, New York, Massachusetts etc. have been spending money on engagement. They are seeing flat or increasing enrollment despite a negative federal messaging and support environment.

We can make a strong assumption that the people who are the most marginal of buyers and who will either buy or not buy an ACA plan conditional on outreach and messaging tend to be healthier and lower cost than average. More low cost people in a pool lowers average premiums and lowers federal advanced premium tax credits per insured individual. States that engage in aggressive outreach should, all else being equal, see a healthier risk pool and thus a lower benchmark premium. If Virginia was to buy an Exchange out of the box and run their own system, the exchange fee could either be held constant with far more local outreach being funded or it could be lowered while holding current outreach constant. That is an actuarial and political question as to what would buy on net lower premiums but that is the biggest play for a state based exchange.

The other major reason that a state may want to have their own exchange is that it is far easier to customize that portal to play nicely with other state portals and policies. Smooth transitions between Medicaid Expansion and Exchange plans may be a desired policy outcome. It is easier to achieve when both platforms are owned by the same entity. Updating verified income across social service programs and tax records may also be easier. Building a complex 1332 waiver with significant alterations to the current program design are probably easier on a state owned platform compared to

These are some of the reasons why Virginia and other states may want to think about building out an exchange on their own or in collaboration with neighboring states. It gives them control and it offers a fairly low cost way to hold premium increases down.

Messy is not bad…

Some smaht guy who writes here makes a good point on healthcare:

I am in full agreement with Tom. My priority on a personal and political level is universal coverage without too much concern for the actual mechanics of that coverage as long as those mechanics produce an outcome that creates affordable care at both the point of service and for society, some justice in the distribution of care and health, and continues to produce reasonably strong innovation incentives.

And all of those goals will require dealing with current stakeholders, rent-seekers, political roadblocks and non-optimal (locally or universally) equilibrium. It is going to be messy as an overtired and caffeinated toddler at a birthparty which has a build your own sundae as the main sweet.

To achieve that goal, a political coalition will need to cater to the marginal votes. To achieve that goal, a political coalition will need to make hard trade-offs. To achieve that goal, a political coalition will need to smooth off pain points at the cost of optimality (however that may be defined).

It will be messy politics as it is a core political question of who holds and how does power over life and the future function.

Mandates vs frictionless receptivity

A new NBER paper just came out that looks at school choice. The researchers want to know if differential enrollment patterns among SES and behavioral needs between default public schools and either selective public schools or charter schools is a matter of preference or a matter of schools engaging in differential friction setting to screen the potential applicant pool out.

I am not an education researcher. However I am familiar with insurance and risk adjustment. This research sounds amazingly like insurance mandates and risk adjustment problems.

The Affordable Care Act mandates insurers pay for certain services with no cost sharing. Vaccines and contraception are some of the better known recommended services. Most of these services will not have a positive return on payment for the insurer in six to eighteen months. People leave their individual market insurer fairly quickly. Most of the gain for these preventive care services accrue to either the patient, society in general or future insurers.

Insurers have to pay for the claim once it is submitted for the service rendered. They don’t have to be enthusiastic about it. They don’t have to send a reminder e-mail and a follow-up call to everyone who is missing a recommended HPV vaccine dose. They don’t have to remind every eligible patient to get a colonoscopy or a FIT screen . Some insurers will because they believe it is the right thing to do. Some insurers will because outside accrediting entities prioritize and reward preventive care services being performed on schedule for a large proportion of the population. Some insurers will because they think that they will capture significant downstream benefits because they are a dominant local monopoly. Other insurers won’t.

Little frictions matter. Adding lubrication to a bumpy process matters too.

Insurers that face mandates that could negatively impact their desired performance metrics will find ways to increase friction. Anything and anyone who makes it through the obstacle course for an undesired treatment will see the claim paid, but the frictions will deter some utilization.

Conversely, mandates that positively impact desired performance metrics will see the insurers spray oil on squeaky gears and smooth out the process.

The second insurance isomorph in this paper is the a risk adjustment problem for kids with disabilities.

A key question is whether these results differ between traditional public schools in areas with school choice and
charter schools. Charter schools represent the fastest growing form of school choice in the country.
To explore this question, our sample includes charter schools matched to nearby traditional public schools of choice with the same entry grade level. We find that, overall, traditional public schools’ response rates are similar to the response rates from charter schools across treatment messages. However, there is a different response rate to messages that signal a child has a significant special need.
Traditional public schools exhibit no differential response rate to these messages, but charter schools are 7 percentage points less likely to respond to them than to the baseline message. This result is important because students with disabilities are twice as expensive to educate than the typical student without a disability (Moore et al., 1988; Chambers, 1998; Collins and Zirkel, 1992), and students with the severe disabilities can cost 8-to-14 times to educate compared to the typical non-disabled student (Griffith, 2008)

If we can assume that charter schools are paid a capitated amount for each kid in a non-individualized risk cell and there is non-zero variance in costs to educate all members of any given risk cell, we can say that some kids in a given risk cell are lower cost than the capitated payment and some kids are higher cost than the capitated payment.

This is similar to insurance risk adjustment. Some diseases have tight spreads with little variance while other diseases like hemophilia have both very high average costs and incredible right hand skew in the cost distribution. In 2018, a hemophilia diagnosis had a risk adjusted transfer value of roughly $450,000 on the ACA exchanges. An individual with hemophilia that is well controlled could have a yearly claims cost of $200,000-300,000. They are a very profitable patient for the insurer due to the large risk adjustment revenue inflow. At the same time, an individual with hemophilia who has either a severe bleed or inhibitor resistance to the common treatments could easily have a $1,000,000 month. They are extremely unprofitable.

Any single insurer that operates in a risk adjusted environment would love to be able to perfectly detect and cover the cohort of individuals with hemophilia who won’t have any unusual months or inhibitor resistance. However if there is any imperfection in the detection/recruitment methods and a single individual who was thought to be a $300,000 case with a $450,000 risk adjustment payment has a $5,000,000 claim year, the entire model blows up. In a competitive insurer market, insurers will look at the variance and then look at their ability to buy reinsurnace or eat the risk in a half million person risk pool before they decide if they want to throw up frictions such as not covering the three doctors in the region who treat 90% of the individuals with hemophilia.

If kids with disabilities are in risk adjusted cells and there is high variance within this particular cell, unless the school can perfectly detect which kids are still profitable to enroll, adding friction can make business sense. As policy makers, we need to recognize that tables can be tilted and frictions can be created when private entities are ordered to do things that they would prefer not to do (or at least not do at the current offered reimbursement).

Open Enrollment wraps up

The Center for Medicare and Medicaid Services (CMS) released the final initial count of enrollments for’s open enrollment period last night. Platform SnapshotWeek 7: Dec 9 – Dec 15Cumulative:  Nov 1-Dec 15
Plan Selections4,322,4508,454,882
New Consumers918,6482,025,611
Consumers Renewing Coverage3,403,8026,429,271
Consumers on Applications Submitted4,291,903



Total plans purchased went down by 368,000 before the data gets cleaned up. Assuming this year looks like most other years, I expect to lose a net 50,000-100,000 more enrollments after files are reconciled. Charles Gaba has a very helpful chart that lays out the details year over year:

There are two things to note that I think tell the stories of enrollment. I am speculating wildly with little to no direct evidence but I think I have reasonably well grounded intuition.

The first story is that the renewals went up by a minuscule amount. We know that more people are being exposed to Silver Loaded and Silver Gapped deals due to Cost Sharing Reduction (CSR) subsidies having been terminated and baked into the premium structures. This is not uniform. It was a plus in North Carolina and Iowa but a big minus in Philadelphia and Phoenix.

Pricing for subsidized folks was an enrollment driver:

More states are silver-loading. Colorado and Delaware shifted from a broad-load of CSRs, meaning that CSR costs were incorporated into the premiums for all plans, to a silver load of CSR costs. North Dakota and Vermont are incorporating CSR costs into premiums for the first time. These steps will increase the value proposition for subsidized buyers.

The other story is to look at the new enrollees. This year saw a 15% decline in new enrollees.

I think this is a sign that the repeal of the individual mandate as well as increased focus on expanding eligibility for underwritten plans is working as expected.

I am speculating wildly right now.

I think that once someone shows up on, the odds of converting an account creation or even a windowshopping moment into a plan purchase is as high if not higher this year than in past years. The pricing due to CSR termination creates a lot of high value proposition deals. However the odds of someone who is potentially in the individual market for health insurance actually going to is lower this year than in prior years.

That is speculation. I think it is reasonable speculation but it is not evidence based (yet).

Overall, given the environment, this is a surprisingly strong end of the open enrollment period enrollment report for

Changes, law and time re the Warren bill

Senator Warren (D-MA) introduced a bill that would authorize the federal government to directly manufacture or contract for some high cost generic drugs including insulin. I don’t have a well formed opinion about this bill yet as I am still trying to think through the dynamic effects. But there is one very good point that Alexander Gaffney is making



The bill has an aggressive schedule for success:

MANUFACTURING LEVELS.—Not later than 1 year after the date of enactment of this section, the Office shall manufacture, or enter into contracts with entities for the manufacture, of not less than 15 applicable drugs. Not later than 3 years after such date of enactment, the Office shall manufacture, or enter into contracts with entities for the manufacture, of not less than 25 applicable drugs.

This is a new program. The federal government does not do new programs quickly. For success to be achieved, especially first year success, the agency would need to find at least an interim director, recruit at least a core staff including several exceedingly hard to find and hire scientific, manufacturing and legal advisors, lease office space, find the coffee pot, get complex contracts negotiated and reviewed and then get things started. This is a complex undertaking even if we assume there are no significant legal challenges from any of the entities that are probably going to be losing money from this policy.

As we think about healthcare bills and changes, we need to think about implementation timelines. If there is anything other than shifting money flows, the federal government needs time especially if program success counts on private sector actors active and enthusiastic participation. They need to know what the rules are, they need to prepare their bids, they need to rework their internal processes.

This applies for any of the pharmacy bills. It will apply to ACA 3.1 or Medicare for All or Medicare Buy-in. I was recently having a beer with a fellow health policy nerd, and we were stuck debating whether or not the earliest major implementation of anything that re-opens Title 1 of the ACA (the coverage requirement section for guaranteed issue and community rating) would be a twenty four or a thirty six month slog. I’m leaning towards thirty six months. My friend thought that depending on the definition of “major”, twenty four months would be a plausible time frame. Medicare for X packages would need several years to ripen after a Presidential signature before implementation could proceed well for large parts of the population.

Implementation timelines are not sexy. They are not fun. The project managers who have to think about these things scare me. But these timelines partially define what is plausibly promised.

Could have taken a W

Yesterday, John embedded the President screaming into the void that he can fix healthcare again:

He could have taken a W on healthcare after last year.

I made this argument in the New York Times in October 2017:

President Trump once promised a health care plan that would have much lower premiums and deductibles while at the same time taking care of pre-existing conditions.”

That plan may be Obamacare. Mr. Trump’s decision to end cost-sharing-reduction subsidies, known as C.S.R.s, and perhaps to derail a bipartisan bill by Senators Lamar Alexander, Republican of Tennessee, and Patty Murray, Democrat of Washington, that would restore C.S.R. funding through 2019 may actually lead to better coverage for more people paying lower monthly premiums….

By pulling the plug on C.S.R. subsidies, President Trump can claim that he will provide a better deal on health insurance for more Americans. He and his fellow Republicans should declare victory, with or without a version of the Alexander-Murray bill — and check health care reform off their legislative list.


North Carolina is a good example of the facts on the ground regarding prices paid by subsidized individuals.

People earning between 100% and 400% FPL have lower premium plans with lower deductibles now under Trump than they did under Obama.  In Wilmington, North Carolina, a family of four can get a $0 Gold plan with a family income of at least $65,000.  That same family can earn up to $98,000 to get a $0 premium Bronze plan.  These types of pricing spreads are not uncommon.  They are the direct result of Trump Administration policies.  The biggest driver is the termination of Cost Sharing Reduction subsidies that led to the overinflation of Silver benchmark premiums.

The problem with this fundamental argument of people actually being able to find insurance with lower premiums and higher actuarial value is that this is not the vision of health insurance that animates conservative policy making. It would not be considered a victory. Instead, the conservative vision is low actuarial value plans that are tied to health savings accounts.  There is a two fold theory of change:

theory of change with the use of HSA in both a single year and over a lifetime. The single year theory of change is that high first dollar expenses will lead to lower utilization with minimal real health consequences as people become expert shoppers and evaluaters of health care need and value. The lifetime theory of change is that an HSA can be built up while an individual is young and healthy and spent when an individual is old and sick. It prefunds some of the expected health cost obligations on an individual level.

The reality on the ground would allow for the President to claim a victory as he at least partially delivered on his rhetoric but it conflicts with the policy class of the GOP’s policy vision.

Do not panic, carry your towel and enroll

Do not panic. Carry your towel. Enroll on

the exchanges are still open. You can get insurance. Ignore the Texas judge. His ruling will be overturned shortly.