Friday, April 01, 2016

A major divide in the ACA individual market for health insurance

In my last post, I calculated that the average private health plan obtained by ACA marketplace enrollees for 2016 provides coverage equivalent to a gold-level plan, and also roughly comparable to the average employer-sponsored plan.  That is, the average actuarial value (AV) of plans sold in the marketplace is just about 80%, thanks in large part to the Cost Sharing Reduction (CSR) subsidies available to buyers with incomes under 250% of the Federal Poverty Level (FPL) who buy silver plans.

While an 80% average actuarial value sounds pretty good -- better, I think, than most people would guess, based on media coverage -- it covers a lot of unevenness. There are two great divides in the post-ACA individual market. The first is between the subsidized and the unsubsidized. The second is between enrollees with incomes up to 200% FPL -- the phase-out point for strong CSR subsidies -- and everyone else. Let's look at average AVs for the different groups.

First up is a breakdown of the subsidized buyers on the federal marketplace, CSR raises the AV of silver plans to 94% for enrollees with incomes up to 150% FPL, to 87% for those with incomes in the 150-200% FPL range, and to 73% for those from 200-250% FPL.*  More than two thirds of subsidized buyers on the federal exchange get CSR of some kind.

Actuarial value obtained by subsidized enrollees on

Actuarial Value
and metal/CSR level
Percent of subsidized
94  (Silver - CSR1)
90  (Platinum)
87  (Silver - CSR2)
80  (Gold)
73  (Silver - CSR3)
70  (Silver - no CSR)
60  (Bronze)
81.6% (weighted avg)

Compare the distribution for the 15% of enrollees who did not obtain premium subsidies -- or CSR. While the percentage selecting bronze is much higher than among the subsidized, so are the percentages of those who bought gold and platinum.

Actuarial value obtained by unsubsidized enrollees on

% of unsubsidized
90 (Platinum)
80 (Gold))
70 (Silver)
60 (Bronze)
57 (Catastrophic)
68.7% (weighted avg)

While HHS does not provide income breakouts for the state-based marketplaces (SBMs), the results are likely to be similar, as I tried to demonstrate in the prior post. The SBM states are wealthier on average, and have (with one exception) expanded Medicaid, and so have a lower proportion of buyers eligible for AV 94% silver. But that is offset in two ways. First, higher proportions buy gold plans, and second, a higher proportion of those eligible for CSR select silver and so access the benefit, if past state reports are a guide.

If we take the 1.4 million unsubsidized buyers on as a proxy for the entire market of unsubsidized ACA-compliant plans, including the majority that are purchased outside the marketplace, we have a basis for estimating a weighted average AV for the entire ACA-compliant individual market. Admittedly that's a big "if," but the metal levels, and so the AV levels, are the same for all ACA-compliant plans. We also need to rely on other estimates: CBO pegs the off-exchange individual market at 9 million for this year. That includes perhaps 1.5 million "grandfathered" and "grandmothered" pre-ACA plans.

Adding 2.1 million unsubsidized marketplace enrollees to 7.5 million off-exchange buyers suggests a total of 9.6 million unsubsidized ACA compliant plans. At present, there are 10.5 million subsidized enrollees. Here, then, is a rough average for the entire market:

Estimated average actuarial value -- U.S. individual market

Avg. actuarial
% of all indiv.
Market enrollees
68.9% (unsubsidized)
81.6% (subsidized)
75.5% (weighted avg,)

Well, that was fun, but in a way the calculations hide as much as they reveal. What's hidden specifically is the impact of CSR and the extent to which it advantages lower-income buyers.The real divide in the market, as I hinted above, is at 201% FPL, where the AV of CSR-enhanced silver drops from 87% to 73% (and silver plan selection drops accordingly).

In fact the AV gap between buyers under 200% FPL and subsidized buyers over 200% FPL is far wider than between the subsidized and unsubsidized. Here's the breakdown for enrollees up to 200% FPL:

Actuarial value obtained by enrollees under 200% FPL on

% of subsidized
94 (Silver CSR1)
87 (Silver CSR2)
80 (Gold)
60 (Bronze)
86.3% (weighted avg)

The gold allocation here is a guestimate. Among all subsidized buyers, 4% bought gold,and 0% bought platinum. If I'm low on the gold, additions to that total would come off the bronze and so boost the average AV a bit.

Now for the folks at the upper end of subsidy range. These are by many accounts the folks who are not buying.  At an income of $29,400, the benchmark silver plan costs about $200 per month, subsidy included, and CSR reduces out-of-pocket costs only marginally, so the plan's deductible might be $1500-2500.  Here's the AV distribution for this group:

Actuarial value obtained by subsidized enrollees over 200% FPL on

% of unsubsidized
80 (Gold)
73 (Silver CSR1)
70 (Silver - no CSR)
60 (Bronze)
69.3 (weighted avg)

Average AV for this group is barely higher than for unsubsidized enrollees.*** The marginal benefit added by weakest-level CSR to silver at 200-250% FPL is mostly offset by the higher percentage of unsubsidized buyers who select gold (and, to a lesser extent, platinum).

Basically, then, we have a market where the average AV is close to 70%, the level for silver unenhanced by CSR, except for those buyers with incomes under 200% FPL, who collectively get a gold-platinum alloy. I'll note again that this is the income group (100-200% FPL) that has been hit hardest by reductions in employer-provided insurance since the late 1990s. The subsidies are well targeted to the neediest, but leave those with somewhat higher but still-modest incomes rather short.

UPDATE 1/11/17: In July 2016, CMS produced a data brief that provided a kind of Rosetta Stone for CSR takeup -- a metal level breakout that included the three CSR gradations. It was based on enrollment as of April, when attrition had reduced overall enrollment by about 12%,and it shows slightly higher levels of CSR takeup than the report produced at the end of open enrollment, and so on the calculations above. It appears that attrition is somewhat lower among CSR eligibles. I consider the implications here. See also the previous post, which notes that the CMS post-attrition breakout bumps average weighted AV for the marketplace as a whole up to 81.4%. [2/1/17  transcribing error corrected: in box for AV for whole individual market, I had 81.4% AV for subsidized rather than 81.6%, corrected. That bumped avg AV for the whole marketplace from 75.4 to 75.5%.]

Update, 6/23/17: I originally titled this post "Two major divides in the ACA marketplace," even though it actually emerged as I was creating it that there's really one fundamental divide -- between those eligible for strong CSR (up to 200% FPL) and those who aren't. Better late than never, I've changed the title today, after sitting on it for too long.

*  HHS breaks out enrollees by income level, metal level selection, and overall percentage obtaining CSR, but not, specifically. by level of  CSR, which is determined by income. (Tables for 2016 enrollees are here and here.)  To get the percentages of enrollees at each of three CSR levels, I took the total number of CSR recipients (59% of all enrollees) and divided it proportionately, according to the income breakouts, among the three income levels (50.6% at AV 94%, 30.8% at AV 87% and 18.5% at AV 73%).  This time around, though, knowing from certain state reports that CSR takeup drops sharply at 200% FPL, where the benefit weakens, I assumed that only 60% of enrollees in the 200-250% FPL band chose silver and so obtained CSR. That added a couple of percentage points to the enrollee totals at AV 94% and AV 87%.  Based on the state data, 60% silver selection for those in the 200-250% FPL band is a generous estimate.

** HHS reports that 35,147 subsidized enrollees selected platinum, which comes out to .04%. While I included that figure in the weighted average for all subsidized buyers, I don't have a means of allocating it between those over and under 200% FPL. If you were to assume .07% of subsidized enrollees over 200% FPL selected platinum, and take the corresponding amount off gold selection, average weighted AV for that group would rise from 69.3% to 69.4%.

***  There is an important divide between subsidized and unsubsidized, but it pertains mainly to premium subsidies. While these phase out gradually for younger, older buyers are faced with a real "subsidy cliff" at the edge of eligibility, especially if they're buying for multiple household members. That's because unsubsidized premiums for the oldest eligible adult buyers are as much as three times higher than for the youngest eligible, but subsidies erase the difference: a subsidized 64 year-old earning $29,000 will pay exactly the same for a benchmark silver plan as a 24 year-old of the same income, and the subsidy will cover the near-tripling of the premium.


  1. Another error:
    --Your second to last table reads: % of unsubsidized

    --Should read: % of subsidized

    1. Thanks - fixed. That's what comes of copying a chart and filling in new numbers.

  2. Thanks for these past two posts. I'd love to see a rough statistical analysis (or set of informed guesstimates) about where lower and middle income people with ESI stand in relation to the marketplace population, in terms of average AV. It would expand the picture and help me get a better handle on just where the growing under-insurance crisis lies in the US population. For example, a lot of the energy among Bernie Sanders supporters emphasizing his single-payer plan seemed to come from people who had ESI but clearly felt underinsured and dissatisfied with the AV they could get for a given price-point. And Sanders' plan really seemed to focus on moving toward a plan in which a very high AV would be offered to everyone. Would this be "over-insurance"? Either way, supporters of Sanders seemed to have a different set of guiding priorities in their vision of improving US health insurance in the existing population.

  3. Sanders' supporters are tired of the endless means testing and subsidy cliff. It makes no sense that a single 55 year old has a $350 monthly subsidy if he/she earns $44,000, but sometimes no subsidy if they earn $45,500. (numbers approximate but not far off)

    This happens in Medicaid all the time too. It is repulsive how many people go ond off Medicaid in the course of just a year or two, due to varying incomes.

    Sanders' supporters want universal programs, paid for with broad based universal taxes.

    1. Hi Bob,
      Thanks for the reply! Yes, I agree, the subsidy cliffs are a big problem and incoherent from a means-testing standpoint.

      I pre-emptively apologize for speaking generally about my sense of a "Sanders" left-liberal/democratic-socialist/social-democratic vision for health insurance (and health care provision). But if you'll indulge my good faith effort to describe matters, I'd be interested in yours and Andrew's reaction/corrections/etc.

      The nugget of an emerging left-of-center debate that I'm intrigued by is the following: Given the evidence of Sanders' single-payer plan outline and of many Sanders' supporters' expressed dissatisfactions about the US health-care system, the left wing of the Democratic Party wants to make a greater priority of improving the insurance for lower-middle and middle income Americans, both in terms of lower cost and greater coverage of actuarial value (comprehensiveness). In a certain way, I think the ambition of the vision is to do away with what we'd meaningfully describe as "insurance" all together: that is, the goal should be to socialize (NB: not a dirty word for me with any bad connotation) all costs and provide nearly or completely full coverage (0-minimal premiums, co-insurance, deductible), not gear health "insurance" to managing the risk of large costs due to extraordinary health risks. On this accounting, even "Medicare for All" is found wanting. Sanders pointedly does not like the idea of gradually expanding Medicare downward to 55+, then 45+, etc. His plan calls for universal comprehensive coverage with little to no premiums or co-insurance, something Medicare has never had.

      On the other hand, the current mainstream of the Democratic Party, extending to past neoliberals like Hillary Clinton who evolved left to fit with mainstream liberalism since ~2006/2008, has maintained as a priority -- and agrees with Sanders liberals -- that universal coverage is morally paramount, and so is also seeking to close the gap on that last 8-12% of the population. It would also ideally like to keep improving subsidy levels (directly or indirectly, straightforwardly or piece-meal) across the board from 0-400% FPL, mainly by relying on taxes on very high income Americans or on the changed revenue picture that would come from jumpstarting the economy. This mainstream of the Democratic Party, building on the PPACA, also has diffuse plans about trying to continue to "bend the cost curve" ideally on the provider side, and less so on the consumer dissuasion of needed care (a love of the GOP). BUT it doesn't really have that Sanders' vision that foresees turning the everyday employer-sponsored insurance of lower/middle/upper-middle-income Americans into something with an across the board actuarial value of ~90-100% -- something which I think that really captures the imagination of the Sanders liberal supporters I know who have very good middle-class and upward jobs in things like higher education / media and the arts / etc.

      So I feel like this is the emerging intra-liberal debate, and it does to a substantial extent turn on priorities and what we mean by "insurance," and thus by what is "under-insurance" and what could conceivably be "over-insurance."

  4. The Sanders group has the vision, but the Clinton group has practical politics to consider.

    All of us would like a health care system with no insurance premiums, co payments, or deductibles. In fact many elderly persons on Medicare with a Plan F Medigap policy see virtually no bills during the year.

    The trouble is that this little slice of health care heaven costs:

    $200 billion in payroll taxes for Part A
    $400 billion in income taxes for the bulk of Part B
    Senior's premiums of $120 a month for Part B
    Senior's premiums of $200 a month for Medigap policy
    $50 billion roughly in income taxes to subsidize Part D

    The Clinton group recognizes that extracting this much and more for a single payor plan is not possible today.

    The ACA with all its limits is funding with a 3% extra tax on capital gains and very large incomes, plus cats and dogs of other revenue extracts. This was hard enough to pass with Democrats controlling both Houses and the Presidency.

    But a $1 trillion annual tax package for single payer is not going to happen. Not even if the slippery Sanders projections of mysterious 'savings' commes to pass.