CSR funding and the next best option

Andrew Sprung at Xpostfactoid has pushed back on my argument that Democrats have the advantage of inertia on the CSR funding argument. He raises an excellent point.

1. Cost. CBO projects premium increases of 20% right off the bat in 2018 and 25% by 2020. Higher premiums mean more people qualify for subsidies, and those subsidies are bigger. CBO projects a 10-year cost of $194 billion — to increase coverage by 1 million. In 2026, that 1-million coverage boost would cost a cool $37 billion. CBO’s 2016 projection for spending on marketplace subsidies in 2026 is $106 billion. Imagine the effects of increasing that spending by 36% in more rational ways. Compare, for example, the comprehensive set of subsidy sweeteners proposed by Urban’s Blumberg and John Holahan in 2015 — which included raising the AV of benchmark plans to 80%, reducing the percentage of income paid at every level and capping premiums for all buyers at 8.5% of income. The authors estimated the ten-year price tag at $221 billion over ten years.

This is a very good point. The coverage gains bought by loading all CSR costs onto Silver only are an extraordinarily inefficient way to expand coverage and improve the law. I am not disputing that at all. There are better ways to spend the money to increase coverage. The same coverage increase can be bought far more cheaply by tweaking Medicaid matching rates or encouraging some creative 1332 waivers.

In Health Affairs, Steven Chen has a good blog post on how states could use the CSR windfall to improve coverage via a 1332 waiver. He uses California and 2016 numbers for his example:

Using California as an example, Covered California showed that the termination of CSR payments by the Federal government would cause insurance premiums for silver plans in the individual market to increase by 16.6 percent in 2018. The study also showed an inverse relationship between CSR and APTC: The Federal government paid $750 million in CSR payments in 2016, but if it were to defund CSR payments, not only would it not receive any savings, it would incur an additional $976 million in APTC spending. Using these figures as illustration, if the Federal government had terminated CSR payments in 2016 and if California had provided CSR payments through a 1332 Waiver, under this scenario California would have to pay $750 million in CSR payments, but it would receive $976 million from the Federal government in lost APTC payments—payments California would have otherwise received without waiver—ending up with a total net profit of $226 million!

The ACA needs a technical corrections bill. It needs a “it’s been live for four to eight years and some things work and some things didn’t, let’s push the things that work and fix or drop the duds… bill”

Not funding CSR sets a plausible outcome absent of an agreement. It is a boundary condition. Deals get made when all sides of a deal believe that they have an outcome that is better through an agreement than the outcome which would occur without an agreement. I can easily and readily see deals.

There could be a trade where CSR is funded and $75 billion dollars are allocated to reinsurance and $30 billion dollars are allocated to increasing subsidies for people who make between 200% and 500% FPL. There could be a trade where $100 billion dollars are spent to up CSR 73 to CSR 80 and then adding a new tier of CSR for people making between 250% and 325% FPL so their Silver is now has an actuarial value of 75%. There could be a trade where Medicaid 1115 waivers can be integrated with 1331 (Basic Health Plan) and 1332 (State Innovation) ACA waivers as well as additional funding for safety net hospitals and community health centers. There could be dozens of deals that spend less money, cover more people and fix known problems. But all of these deals are premised that the outcome due to no agreement is a significant albeit inefficient Democratic policy victory.

The limits of subsidies

NBER has a fascinating new working paper out by Amy Finkelstein, Nathaniel Hendron and Mark Shepherd. They look at the discontinuities in the Massachusetts Health Connector data to estimate willingness to pay for health insurance on the individual market.

As subsidies decline, insurance take-up falls rapidly, dropping about 25% for each $40 increase in monthly enrollee premiums. Marginal enrollees tend to be lower-cost, consistent with adverse selection into insurance. But across the entire distribution we can observe – approximately the bottom 70% of the willingness to pay distribution – enrollee willingness to pay is always less than half of own expected costs. As a result, we estimate that take-up will be highly incomplete even with generous subsidies: if enrollee premiums were 25% of insurers’ average costs, at most half of potential enrollees would buy insurance; even premiums subsidized to 10% of average costs would still leave at least 20% uninsured.

What does this mean?

Most people aren’t willing to pay the average price of care. This makes sense to me on first glance. Most people won’t run up the average claim dollar value. Health care is a concentrated. The bottom 50% of the spending uses 3% of the resources. The top 5% of the population uses half of the medical spend.

Healthy people or people who believe that they will be healthy in the next contract year won’t vlaue health insurance much because the odds of them needing tens of thousands of dollars worth of care are very low. So healthy buyers leave if the post-subsidy price is significantly higher than their expected medical costs plus a small insurance risk aversion function. This makes sense.

So what are the policy implications?

Cutting back subsidies will lead to a sicker risk pool as more marginally healthy people leave the pool. However we can’t say that increasing subsidies to the point where the typical reasonably healthy/marginal buyer is paying nothing in post-subsidy premiums will bring everyone into the pool. We saw strong evidence with the Medicaid woodworker phenomena.

In 2014, people who were Medicaid eligible but not signed up for Medicaid went on the Exchanges for the first time. They went on because they were in an environment where health insurance sign-ups were high salience and there was a lot of help to push these people onto the Exchanges. And instead of being sent to go shop for subsidized plans, the Exchanges identified that the individual was eligible for Legacy Medicaid. Legacy Medicaid has no premiums and nominal cost sharing. Yet it took a massive external force and messaging effort to get people to sign up for free to them health insurance.

In order to get people to sign up for insurance, we need very low monthly premiums, aggressive messaging and most likely a fairly significant and broadly perceived to be frequently enforced and legitimate penalty for not signing up. Or we can go for an auto-enrollment system with an opt-out.

One of my major efforts here at Balloon Juice has been promoting Silver Gap strategies. These are an explicit hack on the subsidy formula. I am advocating for creative strategies that effectively increase the subsidy benchmark which means for healthy families and individuals, their post-subsidy premiums would go down. In some counties in 2017, they could make $38,000 and pay nothing for a Silver plan.

However, the Finkelstein et al paper show that this is a marginal strategy. It helps people get insurance, it helps improve the risk pool. It does not hurt, but it does not get everyone covered.

The CBO and Super Silver Gapping

Kevin Drum asked for a blog post instead of a tweet storm regarding my argument that over the long run not paying the Cost Sharing Reduction (CSR) subsidies is a massive liberal policy win. So here it goes.

Under current law, there are two sets of subsidies. The first is Premium Tax Credits. These apply to eligible people making between 100% and 400% Federal Poverty Line (FPL). They are based on filling the gap between an individual’s personal contribution and the premium of the second least expensive Silver plan. The Premium Tax Credit is a fixed amount. If a person buys a plan with a lower premium than the benchmark Silver plan, the person pays less in premiums. If a person buys a more expensive plan than the benchmark Silver, they pay the incremental difference.

The second subsidy type is the CSR subsidies. These apply to people earning between 100% and 250% FPL. CSR is a value bump built on top of Silver plans to reduce the out of pocket maximum and increase actuarial value. Actuarial value is shorthand for the percentage of allowed costs the insurance company pays. These bumps are 94%, 87% and 73% for people earning up to 150% FPL, 200% FPL, and 250% FPL respectively.

The House sued the Obama Administration on whether or not the CSR payments were properly appropriated instead of just authorized but not funded. Currently, CSR is adjudicated to have not been properly appropriated as a mandatory entitlement. This issue is slowly working it way to the DC Circuit. The Trump Administration could drop the appeal and stop payments immediately. Insurers have been working to protect themselves from this possibility in 2018 by raising rates dramatically. This is the baseline of the story.

Right now Silver plans are targeted to be 70% actuarial value which means the insurance company pays roughly 70% of all allowed amount from claims with premium dollars and individuals pay the remaining 30% of allowed amount for the entire group out of their own pocket. Pragmatically, the benchmark Silver plan will tend to have an actuarial value under 70% due to allowed pricing variations. If we make some assumptions that everything is relatively equal, most subsidized buyers will see Bronze plans (60% AV) are less expensive than the Benchmark Silver and one additional Silver plan is less expensive than the Benchmark Silver. Gold (80% AV) and Platinum (90% AV) are, for the same network at the same insurer in the same plan type, more expensive than the Benchmark Silver.

Now let’s assume that insurers and their state regulators either think that CSR won’t be paid or know that it won’t be paid. Insurers are still obligated to increase the actuarial value of the Silver plans bought by people with incomes between 100%-250% FPL. They won’t give that benefit away for free. Instead they will increase premium rates. They can either increase the rates on all plans or only increase the rates on Silver plans. Most will increase the rates only on Silver plans.

So what happens if CSR costs are incorporated into the premiums of Silver plans. Silver plans are now priced as a function of the proportion of people who make between 150% and 200% FPL and thus qualify for an 87% AV plan and the proportion of people who make between 100% and 150% FPL and qualify for a 94% AV plan. Almost no one who makes between 200% and 250% FPL will buy a Silver plan. This means that the typical Silver plan will be priced as if it about 90% AV. This is effectively a platinum plan serving as the benchmark.

Yet it is an artificially inflated Silver plan. The benchmark is now at priced at 90% AV. What happens to buying decisions?

If we assume that the benchmark Silver has cousins at Platinum, Gold and Bronze where all of these plans have the same networks and same basic benefit design, the relative, post subsidy price order will look like the following: Platinum and Silver are roughly equal, Silver is more expensive than Gold by 15% and Silver is more expensive than Bronze by 40% or more. This is a significant inversion of the current scenario where Benchmark Silver has a price advantage over Bronze by a little bit, and is significantly under-priced compared to Gold and Platinum.

Most of the people who earn under 200% FPL will stay in their CSR enhanced Silver plans. They still get good to very good deals. Their behavior won’t change much. Some people, especially those who are healthy and earning between 150% and 200% FPL may shift their purchases from Silver plans to the less expensive but higher deductible Gold plans that still offer some decent protection.

The big changes in behavior are for people earning between 200% and 400% FPL. The Congressional Budget Office in their analysis of the effects of terminating CSR payments has a wonderful illustration of this enhanced Silver Gapping:

Gold becomes cheaper than Silver despite offering, all else being equal, a better value with lower out of pocket costs and lower deductibles. Subsidized Silver buyers who make more than 200% FPL will migrate overwhelmingly to Gold plans. This increases their average AV from roughly 70% to roughly 80% even as post-subsidy premiums decrease.

More importantly, Bronze plans are highly likely to be very low cost or free. This will increase purchases of these plans by people who are currently uninsured. The CBO projects an average of 300,000 additional covered lives per year because of this effect.

I don’t think this is cost effective way to cover more people. Expanding Medicaid and then expanding Basic Health Plans would be much more efficient.

What this does is anchor expectations of what “reasonable” publicly subsidized insurance looks like. Right now the benchmark plan for people making over 200% FPL is a plan that they are paying $140 or more for a plan with a $3,000 or more deductible. This is better than nothing but it is still under-insurance and not particularly helpful for people. In a Silver loaded CSR universe, the typical Gold plan will have lower premiums after the subsidy and a $1,000 deductible with a $1750 out of pocket maximum. That is truly useful insurance for far more people. And people will get used to this being the standard and we can work to improve from there.

Pragmatism and pre-negotiations

In comments to the post on pragmatic evolution of US health policy on Monday, The Question raised a point that I want to respond to:

on health care why do we always have to pre-negotiate with ourselves and have ourselves primed to accept half a loaf? I am so tired of being sensible when there is no gorram reward. If loudly shouting the most extreme thing we want gets us even half what the republicans have gotten out of it why the hell not??

I want to raise an empirical point and then a broader political/policy point that explains my thought process.

First, empirically, what has “shouting the most extreme thing” gotten Republicans?

It has gotten them power.

What have they done with it so far? In 2009, Democrats at this point had a smaller functional majority in the Senate and a slightly larger majority in the House than the Republicans have today. Democrats had passed and signed into law the stimulus, CHIP re-authorization, Lily Ledbetter, and the Dodd-Frank CARD ACT by now.  They were grinding their way through what would become the ACA.

What have the Republicans accomplished as of today? Read more

CSR and the policy of inertia

Andrew Sprung at Xpostfactoid asks a common question about Cost Sharing Reduction (CSR) subsidies.  

the primary agenda for Democrats is obvious: appropriate funding for Cost Sharing Reduction payments and for some kind of reinsurance program to replace the program that expired in 2017….
To have any real hope of getting these measures passed in a Republican Congress, however, Democrats are going to have to face up to the question: What pound of flesh will they let Republicans extract as payment for these essential, common-sense fixes?

He envisions a negotiating environment where Democrats must concede significant waiver flexibility in return for Cost Sharing Reduction subsidies.  This model of leverage is wrong.  

Democrats have no reason to trade CSR funding for policies that they don’t prefer.  Inaction gives them an incredible policy victory.  Conservatives are the ones who need to make concessions to fully fund CSR.  

The Congressional Budget Office projects that most states will allow insurers over the long run to load the cost of their obligated but not reimbursed CSR obligations to only their Silver plans.  This will have an incredible change in the dynamics of the market.  

Total federal subsidies for health insurance in the nongroup market—in particular, the sum of the premium tax credits and the CSR payments—would increase for two reasons: The average amount of subsidy per person would be greater, and more people would receive subsidies in most years….


By CBO and JCT’s estimates, the number of people receiving subsidies for nongroup health insurance would increase under the policy in most years. In particular, because tax credits would increase and gross premiums for plans other than silver plans in the marketplaces would not change substantially, many people with income between 200 percent and 400 percent of the FPL would, compared with outcomes under the baseline, be able to pay lower net premiums for insurance that pays for the same share (or an even greater share) of covered benefits….reducing the number of uninsured people, on net, in most years.

Insurers are increasingly explicit that they are loading the full cost of the uncertainty onto only the Silver plans.  This is in states as ranging in size from  Idaho to California.  Bronze, Gold and Platinum plans will be priced on the basis of changes in medical costs, changes in enrollment and other normal insurance industry factors.  Silver plans will be priced on those basis and then there will be a significant second price increase on top of the baseline increase.  

The Center for Medicare and Medicaid Services (CMS) issued a guidance letter on August 10th to states contemplating loading CSR costs onto Silver plans only.  This letter states that for risk adjustment purposes, Silver load only plans will be treated as if they are Platinum plans.

For the risk adjustment transfer formula, we intend to propose considering the 87 percent and 94 percent silver plan variants (as well as the limited cost-sharing and zero cost-sharing variants) to have plan metal level actuarial values of 0.9 in order to account for the higher relative actuarial risk associated with these plans

From a mechanical point of view, this is good guidance and a reasonable solution to the problem of running risk adjustment where Silver plans cost more than Gold plans.  It is an incredible admission of a massive policy change regarding the sufficiency of the subsidy.  

Subsidies in the ACA are calculated by taking the difference from the second lowest premium Silver plan and an individual’s expected contribution.  The individual expected contribution is a function of their income.  Any premium that is not covered by the expected individual contribution for the benchmark Silver is covered by the federal premium subsidy.  

The ACA designated Silver plans as having 70% actuarial value (AV) with allowed minimal variation.  In 2018, this means Silver plans will range from 66% AV to 72% AV. In most competitive markets, the benchmark Silver plan will be close to 66% AV.  However in states that load the cost of CSR onto only Silver plans , the Silver plans will have an AV of 90% and this is what the subsidies will be calculated from.  

The ACA exchanges have had difficulty in signing people up who make more than 200% FPL because the cost of the post-subsidy premiums rise too quickly in comparison to perceived value.  Silver, if it is priced at 90% AV, will lead to incredibly lower prices for individuals making between 200% and 400% FPL.   Most people making just under 400% FPL will be able to buy Bronze plans for no out of pocket premius.  Gold plans with $1,500 deductibles will be significantly more affordable in this scenario than Silver plans with $3,000 deductibles are today to individuals and families earning more than 200% FPL.

Liberals will have achieved an incredible policy victory in the states that force insurers to load the cost of CSR onto only Silver plans.  In these states, the benchmark plans will be sufficient to buy 90% actuarial value coverage.  That is better than Medicare. That is an incredible improvement over the ACA as plans will become more affordable to many more people as premiums and deductibles will decrease and the risk pool will get healthier as the value proposition gets better.  

Since it is an incredible policy victory that will be cemented into place by inaction, giving it up for short term funding of a secondary set of subsidies would be counterproductive.  


And then there were two

Now let’s hope that they Silver Gap instead of Silver Spam themselves so that more people can get good deals.

There are two counties, one each in Wisconsin and Ohio, where there are no announced insurers for the Exchange. I am betting with no inside knowledge that someone will take these orphans on. They should be able to monopoly price with almost no pushback from the state insurance commissioner and choose their own risk pool by product offering choices.

This is what happens when state regulators give a damn. Good work Centene. Good work Gov. Sandoval.

Medicaid partial expansion’s consequences

Adrianna Macintryre, Allen Joseph and Nicholas Bagley had a recent perspective in the New England Journal of Medicine regarding the possibility of a partial Medicaid expansion model. I want to explore some of the distributional and macro-economic implications of this possibility. (Disclaimer, I helped Adrianna with some of the data). But first, let’s see the argument and the potential policy:

another decision of arguably greater long-term significance has been overlooked: whether to allow “partial expansions” pursuant to a state Medicaid waiver. Arkansas has already submitted a waiver request for a partial expansion, and other states may well follow its lead… waivers became more consequential in 2012, when the U.S. Supreme Court gave states a choice about whether to expand their Medicaid programs to cover everyone with an income of up to 138% of the federal poverty level.

In general, Obama-era expansion waivers permitted adoption of rules congenial to Republican policymakers….
These waivers, however, did not grant red states everything they requested. The Obama administration refused to approve waivers that would have conditioned Medicaid eligibility for some beneficiaries on their ability to find work. It denied waivers that would have terminated coverage for beneficiaries with incomes below the poverty level if they failed to make out-of-pocket payments for medical care. And it declined states’ requests to partially expand their Medicaid programs to enroll beneficiaries with incomes up to 100% of the poverty level, but not those between 100% and 138%.

Why were states interested in these partial expansions? Starting in 2020, states are responsible for covering 10% of the costs associated with the Medicaid expansion. Because of a drafting mistake, however, the ACA says that the 100-to-138 population can receive subsidies to purchase a private health plan on the exchanges — but only if they are ineligible for Medicaid. For those people, the federal government bears the entire cost of subsidizing private coverage, with no contribution from the states. As a result, the states save money for every beneficiary whom they can move from Medicaid into their exchanges.

The waivers that have been approved for non-standard expansions to 138% FPL have had premiums and cost sharing. Premiums have been limited to 2% of income and cost sharing is limited to 5% of income. That roughly translates to $20 per month in premiums and a $600 deductible. This is the level of a 94% Actuarial Value Silver plan with Cost Sharing Reduction (CSR) subsidies.

The distributional consequences are important. For people who earn between 100-138% FPL in states that have not expanded Medicaid, nothing will change for them. They are no worse off. People who live on less than 100% FPL in these states will be dramatically better off as they will have Medicaid for their coverage. People who earn between 100-138% FPL in waiver states won’t be significantly worse off. A few benefits (vision and non-emergency transportation for instance) may no longer be available but these are marginal changes. People earning under 100% in waiver states won’t see a change.

The major area of change would be individuals who earn between 100-138% FPL and live in states that made a straight forward Medicaid expansion. These individuals will have higher premiums and higher cost sharing. They would be worse off if Pennsylvania or Louisiana or Oregon or anywhere else with a simple expansion reduced the expansion eligibility to only 100% FPL and shifted them onto the Exchanges.

If this is primarily a tool used by states that have not expanded Medicaid or have current cost sharing and premium requirements through Obama era waivers, then the distributional consequences have minimal harms and significant improvements.

I want to geek out a bit below the fold on secondary points.
Read more