Friday, January 15, 2021

ACA Hunger Games, Part II

Subscribe to xpostfactoid 

I'd like to revisit the "hunger games" scenario for legislation to improve the ACA posited in the last post. In brief: assume that Democrats, forced to accommodate their most conservative members to get all 51 of their Senate votes on board for any legislation passed through reconciliation, pare back the broad expansion of premium subsidies for ACA marketplace coverage they've proposed elsewhere -- e.g., in candidate Biden's health care plan and in the House-passed Affordable Care Enhancement Act of 2020 (ACEA). If forced to scale back, should they:

a. Concentrate on the currently unsubsidized, for whom coverage is often truly unaffordable, capping premiums for a benchmark plan as a percentage of income (e.g., 8.5% as in the ACEA, but probably higher)?

b. Concentrate on lower incomes, where takeup among the subsidy-eligible has been poor? (See the last post for the ACEA subsidy scale and how it compares to current law.)

c. Spread whatever enhanced subsidy money they can muster across all income groups?

Subscribe to xpostfactoid 

This is not an academic question. President-elect Biden's pandemic relief plan, echoing the healthcare reform plan he put out during the campaign, calls for the 8.5% cap on premiums and unspecified subsidy enhancement at lower income levels. The subsidy scale might be assumed to resemble a scale proposed by the Urban Institute's Linda Blumberg and John Holahan in 2015, which included the 8.5% cap and a gold (80% actuarial value) benchmark proposed by Biden. But Team Biden has presumably elected not to specify a scale for a reason. In the absence of specifics at lower income levels, the 8.5% cap finds its way into journalistic shorthand, effectively foregrounding relief for the currently unsubsidized.

Stan Dorn of Families USA favors concentrating subsidy boosts on lower incomes, because a) that's where the uninsured are concentrated, and b) drawing in younger and healthier uninsured people at lower incomes will improve the risk pool and pull unsubsidized premiums down (the model here is Massachusetts, as described below). In the last post, I rather equivocally took the opposite tack, arguing that a) the unaffordability of unsubsidized ACA premiums violates a core promise of the law, i.e., the "affordable" part, b) unsubsidized enrollment in the ACA marketplace has cratered since 2016 (down 45%), and c) uninsurance at higher incomes has spiked in recent years. 

Of course, all (progressive) parties in this debate would like to see subsidies increased to levels high enough at all income levels to pull in most uninsured Americans who don't qualify for Medicaid or Medicare  (if not replacement of the marketplace with a more cost-effective public program). But a watered-down version of the ACEA scheme is probably what will emerge. Under that premise, where should the water be added?

Dorn is plainly right on the policy merits: concentrate subsidy enhancement at lower incomes. That's where the uninsured are concentrated, of course. Here are the uninsured rates by income bracket in 2019 according to the annual update of the Census' Current Population Survey (CPS ASEC): 

100-199% FPL     14.1%
200-299% FPL     11.0%
300-399% FPL       8.3%
>     400% FPL       3.0%

(These numbers are for all ages: the overall uninsured rate for adults aged 19-64 is three percentage points higher than for the whole population.)

In raw numbers, as of 2019, some 13.6 million Americans with incomes ranging from 138-400% FPL*  were uninsured, compared to 4.4 million uninsured at incomes above 400% FPL. Moreover, notwithstanding marketplace enrollment drops since 2016 at incomes over 400% FPL, the uninsured rate at incomes over 400% FPL was 5.6% in 2013 -- nearly double the 2019 rate. If you had to make a zero-sum choice where to concentrate new subsidy money, on the policy merits you'd have to choose low incomes.

In outreach to states seeking to improve their ACA marketplaces, Dorn, holding up Massachusetts as a model, has recommended that states (which have much more limited resources than the federal government) concentrate supplemental state-funded subsidies at incomes below 300% FPL. Using 300% FPL as a break point complicates the picture. In 2019, there were about 10 million uninsured with incomes in the 138-300% FPL range, and about 8 million at incomes over 300% FPL.  The 300-400% FPL bracket could use some help. In fact they'd get it with a premium cap at 8.5% income, as the current cap at incomes over 300% FPL is close to 10%. But if a cap is enacted, it will likely be higher than 8.5%. 

Intensity of need at different income brackets is hard to measure. As noted in my last post, the Urban Institute's Blumberg and Holahan found that among individual market enrollees, people just over the subsidy line, in the 400-500% FPL bracket, were paying the highest percentage of income for coverage and care. The subsidy cliff is very steep: the federal government foots an average of 75% of the premium bill for subsidized enrollees, and none of it at incomes over 400% FPL. The political imperative to make coverage at least marginally affordable for all is powerful. 

I suspect that if watered-down subsidies pass, they'll be watered down across the board: there will be something for every income bracket, albeit probably not enough.  That raises the question of how much additional subsidy at different income levels will attract substantially more enrollment. And for that we have some state data, with more on the way.

Two states that recently invested in supplemental subsidies, California (2020) and New Jersey (2021), have had to grapple with them. California concentrated 75% of state funding on subsidies for people with incomes ranging from 400-600% of the Federal Poverty Level, i.e. those fairly close to the current subsidy cutoff of 400% FPL, with minor boosts to subsidies at lower income levels. The upper-income subsidies cover about a quarter of the premium on average (in contrast to about three quarters of premium covered by federal subsidies at incomes under 400% FPL). The effect on enrollment has been marginal. 

New Jersey focused subsidies at incomes below 400% FPL, with bigger subsidies at higher incomes (supplemental subsidies start at $20 per month per person for those with incomes up to 150% FPL and max out at $95 per month). The effect on enrollment is not yet public information -- and may be offset by the glitches associated with establishing a new state exchange, which Jersey did this year. [Update, 1/21/19: Through Jan. 2, marketplace enrollment in New Jersey was 2% higher than at the end of Open Enrollment for 2020 (Dec. 15, 2019).  OE in New Jersey's new state-based exchange continues until Jan. 31.]

In marked contrast are the supplemental state subsidies that enrich coverage at incomes up to 300% FPL in Massachusetts, where the subsidized state exchange predates the ACA and served as its blueprint. The Massachusetts program, launched in 2006, subsidized coverage only up to 300% FPL, but subsidized it more richly than the ACA does. In advance of the ACA marketplace launch, Massachusetts obtained waivers and some federal financing to maintain the state's generous subsidization level to the 300% FPL threshold.  Massachusetts effectively runs a separate program (ConnectorCare) for enrollees with incomes up to 300% FPL, with standardized benefits and coverage far more generous than in the standard ACA design:

The regulatory structure of ConnectorCare, the program offered at incomes below 300% FPL, incentivizes insurers to pay low rates to providers, keeping program costs down. Importantly, the plans offered by each insurer at incomes below 300% FPL are in a single risk pool with the plans offered above that threshold, and relatively high takeup at low income levels has kept unsubsidized premiums low (either lowest or second lowest in the nation, depending on the measure). Massachusetts accordingly has the lowest uninsured rate in the country. 

ConnectorCare bears a family resemblance to managed Medicaid programs: insurers compete for enrollees, they pay relatively low rates to healthcare providers, and enrollees' premium and out-of-pocket costs are lower than in the marketplace. That suggests an alternative model for the ACA generally. It also suggests, more narrowly, that improved subsidies at lower income levels yield the best bang for the buck.


* A note about the 100-199% income bracket: in the 36 states and D.C. that have enacted the ACA Medicaid expansion, Medicaid is available at incomes up 138% FPL (and Oklahoma and Missouri will enact the expansion in July of this year). In nonexpansion states, marketplace subsidy eligibility begins at 100% FPL rather than 138% FPL.  For policy purposes, ending the "coverage gap" for people with incomes below 100% FPL in nonexpansion states -- either by enticing those states to expand, or offering some kind of replacement coverage, as Biden's campaign plan proposed -- is a vital goal. But it's somewhat separate from what we're considering here: in the narrow frame of ACA amendment, where to concentrate subsidy enhancement. About 38% of the population in the Census' 100-199% FPL bracket have incomes below 138% FPL, and most of them are Medicaid-eligible. I've therefore estimated the uninsured population from the 138% FPL starting point in considering the needs of different brackets here.

Update, 1/19/21: See Charles Gaba's deep dive into various Democratic plans to enhance the ACA on both federal and state levels.

Subscribe to xpostfactoid 


  1. Great article, thanks! It is not hard to see why people at 300% of poverty are not too excited about enrolling in an ACA plan that costs $305 a month, (not so bad), but has a deductible of $4500 and OOP max of $8100.........this is a crappy plan!!!

    Incidentally, why can't other states get the "waivers and federal financing" that is helping Massachusetts? I realize this may be a very complex question.

    1. Crappy plans all over the place, unfortunately. It's really a shame that no fixes to get sound, affordable coverage for all are likely.

      I calculated in 2019, for instance, in a not-uncommon bad case just over-the-400%-FPL-subsidy-cliff, a couple, 63, in a state with ACA max 3:1 ratios old to young:

      I got in Cook County, IL zip 60617, couple 63, just over the cliff (income $84, 731)

      second lowest cost bronze plan (bronze to avoid overprice from silver loading)

      premium $18,513 (21.9% of income)
      oop max $15,800, (19% of income)

      If anything, for this year, 2 years later, this must have worsened due to higher oop max.

      With this, and remember, as of two blogs ago, some of the 91% that the Census counts as insured actually have only a loan, until death, for whatever medical expenses occur (no insurance--though many think that's what they have), it's pretty pathetic.

      I often compare the US, with our current ACA, to the the Ontario Canada system, which caps total designated premium plus out-of-pocket costs at about 5% of post-tax income. (Note post-tax is less than pre-tax, so it's not just 40% vs 5%. It's worse!)

      This comes about as follows:

      There is a small designated tax (like a premium) of at most about 1% of income.

      The main Ontario plan covers everything except out of hospital Rx's, with negligible tiny copays.

      Out of hospital Rx's are not covered at all until about 4% of after-tax income is hit, but then are covered fully up to tiny copays of like $2.

      The designated tax for health care, equivalent to a "premium", is here:

      (The exact amount of the premiums are on the last page of the form 6006-c-19e that you can click on.)

      Main system (everything virtually free, covering in-hospital Rx's, but missing out-of-hospital Rx's):

      and, the "Trillium" program, capping expenses for out-of-hospital Rx's at about 4% of after-tax income:


      Great work, as always, Andrew.

      I enjoy your getting into details on cutpoints and subsidy amounts, etc., which the Times, etc., avoid, always assuming their audience will not be able to understand them. (It creates a very blurry presentation at those places.)

      As a math-backgrounded person, I am also impressed that your main academic credential is a Ph.D. in English literature, which does not imply any sort of quantitative skill. But you apparently have such skill in substantial amount, and put together very useful info using that skill. (All the way down to knowing about, and understanding, the silver loading!)

    2. Let me offer these resources, if anyone wants them, for comparison to pre-ACA.

      CT, which I lived in for a long time, had a high risk pool.

      (It was sound, at least became sound after I pointed out some defects, in the sense that anyone who had maintained continuous prior coverage could always get the high-risk-pool, without any waiting period to cover pre-existing conditions.
      Before I pointed out the defects, in certain cases such a person, with full continuous prior coverage, could have a 6 month exclusion for pre-existing conditions.)

      I saved the rates for 2012 on the Wayback Machine:

      This was for the $7500 annual oop max plan:

      and this

      is for the $1000 annual oop max plan (which would be an obvious bad choice of the two in most cases, unless you were in the special low-income group.)


      Since at least Andrew, Bob, and I are all pretty wonky, let me throw in another reference to help research the pre-ACA system, if the need arises.

      This is the old Georgetown statewise health insurance consumer help site.

      You can click on states, and then sub-click, and get further details.

      (If you need to try and recover exact rates or eligibility conditions for a high-risk-pool, say, you can try plugging the links on the Georgetown site into the Wayback Machine, to recover them. I've done this a few times myself with success.)

      Who would have guessed, over 10 years since the ACA was passed, that there would be a need to go back to the details of the miserable pre-ACA system to fend off attacks by people trying to take us back to it?

  2. thanks for comments...

    is the subsidy cliff for two people really close to $84,000 in income? When I was selling plans two years ago it was about $64,000.....

    I understand that Medicaid benefits are like a loan..(Ron Rayburn of Minnesota lived in my town and I knew him.) But how are you figuring that 91% of benefits are like a loan? Just curious.