Deductibles and Actuarial Value

Late last week, I saw the following tweet:

I don’t think the deductible per se is the problem. It is the overall level of cost-sharing that is the problem with a secondary distributional challenge. Let’s think this through a bit with a two period model.

In period 1, there there is a 1,000 person pool with an average annual claims expense of $10,000 per person. This means that there is $10,000,000 in total claims. Let us also assume that this pool has 85% actuarial value insurance which is fairly standard US employer sponsored insurance. This means the insurer will pay $8.5 million in claims and the individuals in the pool will for $1.5 million in claims through the combination of deductibles, co-pays and co-insurance. We should also assume that this pool has fairly typical distribution of claims. Most people in the year barely touch the healthcare system. A dozen folks will have a six figure claim. In period 1, the deductible is low, the co-pays are fairly high and co-insurance is applied to most services up to an out of limit maximum.

The question on cost-sharing is who pays how much to cover the $1.5 million that is not covered by the insurer. On first order, the cost-sharing question is primarily a distributional question.

People who incur no claims (probably a fifth of the pool) won’t care as they have no claims to cost share on.

People who have a $100,000 claim year will max out whatever their cost sharing limit.

People who have some light contact with the healthcare system will care a lot about the distribution of cost-sharing elements. An individual who has a sick PCP appointment, a generic prescription and an urgent care visit is highly motivated by the benefit design. Deductible only designs will make light users pay for their entire medical utilization. Co-pay and co-insurance designs have the insurer pay for a significant proportion of the routine costs of “maintenance” healthcare. This means that the individuals with five and six figure claims have to pay more cost sharing to make sure that the pool puts up their $1.5 million dollars in cost sharing.

Now let’s go to Period 2. The employer group has shifted to a high deductible benefit design with an 80% actuarial value. The insurer says that there is a 10% trend rate growth in claims cost, so the total claims are $11,000,000. The individuals in the employer group are now responsible to cover $2.2 million dollars in the allowed claims expense through cost sharing. The distribution of claims is the same as it was in Period 1. A few of the super high cost claims are incurred by new individuals having a bad year who replace folks who had a bad year in the previous period and are having a good year in the current period.

The shift to a deductible heavy design means that the light users are paying almost everything out of pocket. The people with huge claims are shouldering a smaller percentage of the group’s total cost sharing obligations.

The pain point with the proliferation of deductibles is primarily having the light utilizers paying more while concurrently seeing actuarial value of coverage drop even as the cost of claims increase at a rate higher than wage growth, at a rate higher than GDP growth, at a rate higher than inflation. That is the pain.

Switching insurance designs from deductible heavy plans to coinsurance heavy plans will shift who bears the pain on first effect (let’s not consider immediate incentive effects). Switching to a co-insurance and co-pay heavy designs will make the expensively ill pay more and the relatively healthy/cheap pay less at the point of service but the group’s cost sharing obligation will still be there, and that is an increasing obligation. That is the problem, in my mind, more than the specific cost sharing design choices.

Medicaid expansion also helps the middle class

In Morning Consult, I argue that Virginia’s Medicaid expansion will obviously help the folks who directly qualify for Medicaid. It will also help the folks who earn too much money to qualify for ACA subsidies in two ways:

There are two ways that Medicaid expansion in Virginia will drive down individual market rates that non-subsidized, middle-class families pay…. Evidence shows that individuals who earn between 100 percent and 138 percent of the federal poverty level (single individuals earning between $25,100 and $34,638) and currently buy ACA insurance have more chronic conditions and more expensive health care needs than individuals who earn over 138 percent FPL….

Virginia — like many other states — had its insurers load the cost of providing CSR into the premiums for silver plans, which are the plans that set the local benchmark from which all premium subsidies are calculated. This led to a significant spike in silver benchmark premiums. Other plans saw significant but far lower premium increases. On average, CSR workarounds led to an extra $960 to $1,040 in premiums for silver plans.

The data shows us that people who were previously eligible for an ACA health plan and will now be eligible for Medicaid under the expansion were the highest per-capita recipients of the cost-sharing reduction subsidies. As these individuals move to Medicaid expansion, the cost of funding CSR through silver premiums will decline…

I am expecting the initial rate requests for Virginia and Maine to decrease as the combination of a morbidity adjustment and backing out a significant part of the CSR bump will need to happen as Medicaid expands for the 2019 plan year.

The corollary to this good news is that the people who currently benefit from Silver Loading, who primarily earn between 200-400% FPL are going to be a bit worse off than they would have been with no Medicaid expansion. The reduced Silver Load will make the least expensive Gold plans slightly more expensive relative to the Silver benchmark than they had been in the past.

Everything is a trade-off in health policy, but this is, I think, a very worthwhile trade-off as there is still a decent size Silver Load which will reduce post-subsidy Gold and Bronze premiums compared to the universe where James Comey actually followed FBI procedure.

Value does not always mean lower prices

Value based pricing means we should be willing to pay for things that work and not pay (much) for things that don’t work. It does not necessarily mean that all prices nor net expenditures will decrease.

The linked study establishing the new standard of care is from a 2012 article in the New England Journal of Medicine** and the differences in complications are significant:

The primary outcome was post-ERCP pancreatitis, which was defined as new upper abdominal pain, an elevation in pancreatic enzymes to at least three times the upper limit of the normal range 24 hours after the procedure, and hospitalization for at least 2 nights.

A total of 602 patients were enrolled and completed follow-up. The majority of patients (82%) had a clinical suspicion of sphincter of Oddi dysfunction. Post-ERCP pancreatitis developed in 27 of 295 patients (9.2%) in the indomethacin group and in 52 of 307 patients (16.9%) in the placebo group (P=0.005). Moderate-to-severe pancreatitis developed in 13 patients (4.4%) in the indomethacin group and in 27 patients (8.8%) in the placebo group (P=0.03).

There was a 45% decrease in the primary outcome of pancreatitis. The study defined the outcome as something that required at least 2 nights in the hospital because of the undesired complication. Hospital days are expensive. An extra hospital day can lead to an average of an extra $2,200 in expenses. Avoiding at least 2 hospital days avoids at least $4,400 in expenses.

And the study shows that the indomethacin performed a very valuable function that saved society $100,000+ in claims expense while leading to higher quality of life for the patients who were not hospitalized under the new protocol and who would have been hospitalized under the placebo.

A sole source supplier of the drug has new information showing that the drug has a lot more value than they previously thought. Our economic system gives an extraordinarily strong incentive for the supplier to raise prices as they see that the willingness to pay is a lot higher than previously thought.

Value based purchasing means we should pay for things that work well. Avoiding hospital days is valuable so we should expect to pay a good amount for things that avoid numerous hospitalization days. It is not a panacea for price cuts, it is a system of avoiding doing dumb things by lowering the willingness to pay for ineffective treatments while increasing the incentives for price increases for effective treatments.

Now if we are concerned about the significant increase in effective sales price of a generic drug there are a few policy options available. The first is a cap on prices. The downside is that the sole source supplier has a strong incentive to pull the drug from the market until patients and doctors start crying on TV for the drug at the previous price plus 20%. The second option is to make it easier for competitors to offer more generic versions of the drug. This could mean importing the drug from countries with similar regulatory regimes or streamlining FDA approval processes or a number of other things.

Finally, this could be a good case example for the new non-profit generic drug manufacturer that a number of large hospitals have funded. I anchor on the Friedman and Weiner brief on 5 stories of drug costs as my primary analytical framework. Story #3 is relevant as I discussed last February:

Story 3: Cheap generics get expensive fast — the Martin Shkreli story and the Epi-Pen story. Here the exploit is a lack generic manufacturers that can quickly shift to produce near substitutes. The time and cost of other manufacturers to set up a production line to make a cheap competitor won’t ever return a profit as the original manufacturer/distributor will drop prices to or below marginal cost as soon as they see a threat.

The recent agreement by a number of large hospitals to set up a non-profit generic manufacturer is a response to this story. The new entity would be willing to lose money to set up a production line for a generic drug that just saw its charged price increase by 1,000%. I think the entity’s leadership would be totally happy to certify the capability to get a few simple drugs and one complex generic approved as a demonstration of capability and then just use their capability as a looming threat to tamp down on these pump and dump schemes. That would be a stunning success even if the entity never ships a single pill for anything other than demonstration purposes.

In this story, the ability of a new manufacturer to produce the generic at a cost somewhere between the current incumbent’s marginal cost and the current sales price would see a significant price snap back and then a transfer of more surplus value to the patient or the risk bearing entity instead of the generic drug manufacturer.

The key thing to remember though is that paying for value sometimes means paying more when new information shows more value unless there is a robust and well functioning market that can produce low marginal cost inputs.

** April 12, 2012 N Engl J Med 2012; 366:1414-1422 DOI: 10.1056/NEJMoa1111103

Go comment

A friend and collaborator makes a very good point in this tweet:

When you see something that you think is wrong in federal rule-making, go comment.

Comments must be read and at a broad level, they must be responded to by the regulating agency. The response might not be the on one that you want, as it is usually along the lines of “Yes, 33 comments expressed concern on Issue X impacting Issue Y, we will monitor that situation….” but it requires a response.

This is a form of effective feedback to our government so use it.

Half a loaf with Indiana’s HIP 2.0

Health Affairs has a recent evaluation of Indiana’s HIP 2.0 Medicaid Expansion program **. It is a middling result. More people are covered in Indiana than in states where there is no Medicaid expansion and fewer people are covered compared to states where there is a full fledged, no strings attached Medicaid expansions:

We estimated the impact of this expansion on coverage rates and compared the effects to results from other states that expanded Medicaid after 2014. We found that Indiana’s coverage gains (relative to pre-ACA uninsurance rates) were smaller than gains in neighboring expansion states, but larger than those in other states. These results imply that while one potential reason for Indiana’s lower gains relative to neighboring states was its cost-sharing requirements, expansion led to unquestionable coverage gains in the state.

This has immediate policy relevance because this is the desired form of Medicaid that the CEnter for Medicare and Medicaid Services (CMS) is supporting. CMS will quickly approve waivers for increased cost sharing, lock-out periods and premiums. CMS will quickly approve waivers for work requirements.

The key question from Medicaid advocates is if the Indiana style Medicaid is what is on the table is it better compared to what?

This evaluation shows that HIP 2.0 is better than not expanding Medicaid. If a state has not expanded Medicaid already, than we have a strong revealed preference of at least one veto-position holding entity that straight Medicaid expansion via a state plan amendment is not going to easily happen unless the possession of veto control points is switched by electoral change. Therefore, the realistic choice is either continued no expansion or a convoluted kludge that is HIP 2.0, I’m on board with a convoluted kludge as it can be improved upon in the future. That is what is happening in Virginia. That is what may happen in Utah.

On the other hand, Medicaid expansions via state plan amendments are superior to “private market facsimiles” like HIP 2.0 on the metric of enrollment and usefulness. States that have already expanded without a waiver are going backwards. Those are the states where there should be a fight as it is hard to argue that a waiver would allow for any improvement in outcomes compared to the status quo as the current evidence base shows that the status quo of a state plan amendment is superior to a pseudo-HIP 2.0 waiver plan.

** doi: 10.1377/hlthaff.2017.1596 HEALTH AFFAIRS 37, NO. 6 (2018): 936–943

Farm Bureau plans and yet another out

The North Carolina Farm Bureau wants to sell underwritten health benefit plans next year. The Republican supermajority in the North Carolina state legislature is considering the idea of authorizing “non-profit health benefit plans” that aren’t insurance despite quacking and walking like insurance.

My analysis of the similar plan in Iowa stands:

And if an individual can pass underwriting and does not qualify for significant ACA subsidies, these plans are a good deal as that group of people are mainly looking for hit by a meteor protection. Well subsidized individuals will be indifferent to these plans as the subsidies buys affordable comprehensive insurance on the Exchange. The people in trouble from this scheme will be individuals who either can never pass underwriting or will pass with so many upcharges or restrictions that passing is still pointless. Their options are to either move out of the state or to drop their incomes so that they qualify for ACA subsidies….

But at some point this is fundamentally irrelevant. The upcoming short term limited duration (364 day) plan rule and the association health plan rule will do the same thing. Those two rules will create very large outs in the ACA individual market for people who can pass underwriting. The Farm Bureau wants to underwrite and it won’t be much different except that it will be called a “non-profit health benefits plan” instead of “health insurance”.

Allowing the Farm Bureau to sell these unregulated plans won’t be much more than allowing slightly worsening of morbidity within the ACA QHP risk pool as the damage would have been done already.

What’s going on in Texas

I am not a lawyer, so please don’t take any of the following as informed legal analysis.

The Trump Administration has decided to argue that the $0 penalty individual mandate is unconstitutional and therefore community rated guaranteed issue requirements are unconstitutional as they are non-severable from the mandate. That would destroy the individual market reforms of the ACA.

Take Care has the legal argument and demolition:

In the government’s brief, the Trump DOJ makes two arguments. (A) The individual mandate, which the Supreme Court upheld in NFIB v. Sebelius, is unconstitutional; and (B) because the mandate is unconstitutional, the most important provisions of the Affordable Care Act should also be struck down, on the ground that they are not severable from the now-unconstitutional mandate.

The first of these arguments is excruciatingly stupid, but has the complementary virtue of being irrelevant on its own…

To this we say: Whatever. We’re law professors, and not even we can get worked up about the difference between “do it, or pay the price, which is zero” and “do what you want…”
there are actually instances where severability doctrine is capable of generating clear answers to obvious questions. The easiest case—hard to imagine, but stay with us—would be if Congress actually passed a piece of legislation that eliminated, as a formal or practical matter, one provision of a law, but left the rest of that law in place. In that case, the reconstruction of Congressional intent would be straightforward: What Congress wanted was a law without that provision.

But that is exactly what Congress did in the tax bill, with respect to the Affordable Care Act. And that is what makes the Justice Department’s argument so transparently dishonest.

Yeah, this is dumb.

As a reminder, Jonathan Adler was one of the prime proponents of the “Moops” argument that tried to get subsidies tossed for citizens of states using He thinks this is a bullshit argument.

It is a bullshit set of arguments that are being made in bad faith. And usually that would be enough. But since this is related to the ACA and the district judge is very conservative, we’re may be riding this out for a while as it gets appealed up and down the chain multiple times.

On a pragmatic basis, this introduces uncertainty into the market. Insurers hate uncertainty. They respond to it in one of two ways. They can either run like hell from the market or they can jack up their premiums to incorporate the risk that mid-contract they have to operate under a different regulatory regime.

On a political level, this highlights the popular parts of the ACA (community rating and no denial of coverage for pre-exisiting conditions) just as the election season heats up.