Friday, March 04, 2016

Kaiser: ACA marketplace at 70% capacity among subsidy-eligible

The Kaiser Family Foundation has published an updated assessment of ACA marketplace enrollment to date and what full capacity would look like.  At present, total signups for private plans in the marketplace nationwide stands at 12.7 million, a total that's likely to drift down to somewhere between 10 and 11 million by year's end.  Kaiser estimates that if the national marketplace matched the performance of the 10 best-performing states to date, 16.3 million would be signed up by the end of open enrollment, drifting down to about 14.7 million by year's end.

There's a lot to unpack here. At the outset, I want to highlight a couple of noteworthy findings, discussed in more detail below. The first one is not actually spelled out in this report.
  • According to Kaiser's estimates, nationwide, at present about 70% of the subsidy-eligible potential marketplace customer base have signed up for marketplace plans. That will probably drift down to somewhere around 60% by year's end. Of 12.7 million total enrollees, 82%, or 10.4 million, were subsidized. According to Kaiser's Sept. 30, 2015 estimate, 14.8 million Americans were potentially eligible for premium subsidies at that time.

  • Kaiser's findings as to which income groups are cutting their uninsurance rates most dramatically* raise questions about the prevailing understanding of who finds the marketplace most attractive, at least by implication. Both Avalere Health and the Urban Institute have found that eligible marketplace shoppers with the lowest incomes have high takeup rates, which drop sharply and steadily with income. Kaiser finds that people in the lowest income bracket, below 150% FPL, have the second-lowest reduction in their uninsured rate among the income groups -- just an 18% reduction, compared to 33% reduction at 150-200% FPL, 23% at 200-300%, and 22% for those above 400% FPL (and so subsidy ineligible).
We'll look more closely at those two findings below. First, let's review the basic shape of the current and potential marketplace according to Kaiser.

Kaiser sets out to account for why the marketplace is smaller, currently and potentially, than CBO forecast yearly, with some adjustments, from 2009 through last year. CBO anticipated that average monthly enrollment this year would be 21 million, almost double what it will be, and that it would top out at 24 or 25 million in 2018. 

Kaiser now projects a natural ceiling of 15 or 16 million "over the next several years." Authors Larry Levitt, Gary Claxton, Anthony Damico and Cynthia Cox ascribe the reduced target mainly to two factors:
  • Anticipated reductions in employer-sponsored insurance, estimated at 6 million in 2016 by CBO,  have not materialized. That may partly be due to a good employment market and partly to the employment and individual mandates.

  • Fewer buyers who earn too much to qualify for subsidies than anticipated are buying their plans in the marketplace, rather than off-exchange. That's partly because of the unplanned "grandmothering" of ACA-noncompliant plans purchased after enactment in 2010, though that pool is phasing out, and mainly because, as Kaiser notes, "For people not eligible for premium subsidies, there is little advantage to buying through the marketplace." The only reason to do so is if one anticipates that income might dip into subsidy range during the year. At the same time, ACA-compliant plans bought off-exchange are included in the same risk pools and are of equal quality to marketplace plans. 
The only disadvantage to larger-than-expected off-exchange purchase is that we don't know how many people are in the market, what they're buying, or what their age and income profiles are. They are a partial black box in our improving picture of who has what insurance.

A third point represents something of a shift in emphasis in Kaiser's conception of the marketplace. In the ten most successful state marketplaces, Kaiser estimates that 90% of subsidy-eligible potential marketplace enrollees have selected plans.  The authors seize on that 90% figure as a kind of natural ceiling to project the size of the on-exchange market should the whole country achieve comparable enrollment success. There's a tacit decision there to leave the black box of off-exchange enrollment out of the picture in assessing marketplace success.

Kaiser does also study the off-exchange market, and hopefully will update this May 12, 2015 inquiry into how much the overall nongroup market has grown. But because the subsidy-eligible are less able to take care of themselves and constitute a more natural marketplace constituency than the subsidy-ineligible, it makes sense to focus mainly on the their experience in the marketplace. Hence my ancillary point that nationwide, the subsidy-eligible market is 70% penetrated.

That takes us back to the somewhat disturbing news about relatively low reductions in the uninsured rate for marketplace-eligible people with incomes under 150% FPL. That excludes those eligible for the ACA Medicaid expansion -- that is, citizens and some legally present noncitizens* with incomes up to 138% FPL in the 31 expansion states (including D.C.). It includes those with incomes from 100-150% FPL in the states that refused to expand Medicaid, and those in the narrower 138-150% FPL band in the expansion states.

As mentioned above, the perception has been that penetration in the lowest income bracket has been high. In nonexpansion states, nearly 50% of all enrollees and a majority of subsidized enrollees have been under 150% FPL. In March 2015, Avalere Health estimated  that 76% of marketplace-eligible individuals with incomes under 150% FPL selected plans, with takeup dropping off steeply at higher income levels:

An Oct. 2015 Urban Institute study showed a similar pattern:

Those results have seemed intuitive to me, in that subsidized premiums for benchmark silver ACA plans rise steadily with income. Below 138% FPL, they're capped at 2% of income; at 200% FPL, they're over 6%; at 250% FPL, over 8%; and at 300% FPL, 9.6%. For those under 150% FPL, moreover, silver plans have an actuarial value of 94%. A single person in a red state with an income of $12,000 (a bit over 100% FPL) in Mobile, Alabama pays $14 per month for the cheapest silver plan available, with a $0 deductible. If her income is $23,000 (a bit under 200% FPL), cheapest silver is $113 per month, with an $800 deductible.  Yet Kaiser finds that 150-200% FPL is the sweet spot for insurance gains among marketplace-eligible uninsured (at least in 2014):

I asked Larry Levitt, lead author of the assessment and SVP for special initiatives at Kaiser, for his take on the apparent discrepancy between Kaiser results and those of Avalere and Urban. He wrote as follows:
I was somewhat surprised by our result. But remember, those take-up rate calculations can be very misleading. Yes, the marketplace take-up rates among the potential market (uninsured plus non-group purchasers) is much higher among lower income people. But it makes all the sense in the world for those folks to buy in the marketplace. As you go up in the income scale, it matters less and less, and many of those people are insured, just outside the marketplace. That’s why we looked at the change in the uninsured among marketplace-eligibles…it captures a true take-up rate of insurance.
The suggestion is that  a good number of low income marketplace buyers were previously insured in the nongroup market -- more than I would have thought. Perhaps some were also semi-insured by employer-sponsored "mini-med" plans that don't provide comprehensive coverage.  But I would expect the percentage of marketplace buyers who were previously insured to rise with income.

One other possible partial factor might be a discrepancy between how people report their income in surveys vs. how it's recorded in marketplace applications. The latter defines a household as those who file taxes together; the former (the Current Population Survey) by who lives under one roof. Perhaps some of those in the 150-200% FPL bracket here are under 150% FPL on their exchange applications.

To add somewhat to the muddle: as I noted last September, the Census report on insurance gains from 2013-2014 seemed to show an anomalously high pickup in private insurance among the poorest groups, and an anomalously low pickup in pubic insurance among same: 
The data shows greater percentage increases in private insurance among those below 100% FPL (or below 138% FPL) than for those between 100% and 199% FPL or 200% and 299% FPL. Conversely, there are greater increases in government health insurance among those in those in the higher two ranges cited above than in the lower two. That seems counterintuitive, as those under 100% FPL (or 138% FPL in Medicaid expansion states) are ineligible for subsidized private coverage, and those over 138% FPL gained no new access to government insurance. Private insurance does include employer-sponsored insurance, but it seems unlikely that large numbers of people in households under 100% FPL would have gained access to that.

Specifically, the Census reports that government-provided insurance increased 2.6 percentage points for people with incomes from 100-199% FPL and 2.7 points for those in the 200-299% FPL range, compared to just 1.3 points for people under 100% FPL and 1.4 for those under 138% FPL. Conversely, private insurance is reported to have increased 3.7 percentage points for those under 100% FPL and 3.6 points for those under 138% FPL,but just 3.2 for the 100-199 % income band and 2.9% for those between 200 and 299% FPL.

The apparent discrepancy is reflected in raw numbers: almost as many people under 100% FPL gained private insurance as did those in the 100-199% FPL range (1.81 million vs. 1.87 million), while the number of people who gained government insurance in the 100-199% FPL income band (1.47 million)  exceeded the number of people below 138% FPL who did so (1.30 million)* -- despite starting from a lower 2013 baseline (29.2 million vs 40.4 million).
On the other hand, as the same post notes in an update, NHIS data on public and private insurance gains is more in line with what you'd expense.

Urban and Avalere used American Community Survey data, Urban from 2012 and 2013, and Avalere from 2013 [corrected 3/7], while Kaiser probably used more recent Current Population Survey data that reflects a questionnaire revamped to better capture respondents' insurance status (and tends to show more insured than past surveys). Even accounting for the fact that Urban and Avalere counted takeup among those eligible for marketplace coverage rather than those uninsured when they obtained marketplace coverage, their conclusions don't seem to me quite to align. If anyone has any further insight, please let me know.

* Corrected, 3/5. Originally, this sentence read "Kaiser's findings as to which income groups are taking up the marketplace offerings and which are leaving them on the table flip the prevailing understanding of those patterns on their head. " That speaks to an initial misreading of this finding that I kind of corrected by degrees  while writing (and corresponding with Larry Levitt). Lower income shoppers are buying marketplace plans at higher rates than higher income shoppers, but at the same time, it would appear that a) more lower income marketplace enrollees were previously insured than might seem obvious, and b) higher income (including subsidy-eligible) insurance seekers are picking up insurance by other means as well as in the marketplace. It makes sense, as Kaiser notes, that 300-400% FPL would be the weak point, since many in that range are eligible for no subsidy at all, and even those who qualify for subsidies will pay almost 10% of income for benchmark silver. Kaiser further notes that the relatively sharp drop in uninsurance for those over 400% FPL may be due in part to the fact that those with pre-existing conditions are no longer shut out or priced out of the market.

P.S. (3/5): In keeping with Kaiser's reduced estimate of the overall potential size of the ACA marketplace, it's worth noting that between Nov. 2013 and Sept. 2015 Kaiser cut its estimate of the subsidy-eligible potential marketplace population from 17.2 million to 14,8 million. Kaiser's estimate of the entire potential marketplace population dropped less far in the same period, from 28.6 million to 27.4 million. Again, the marketplace enrollment among the subsidy-ineligible doesn't have much meaning -- what matters is enrollment in ACA compliant plans, on- or off-exchange.

UPDATE 3/7: Below are thoughts about the differing impressions left by Avalere's and Kaiser's research from Avalere Chris Sloan manager Chris Sloan:
A few thoughts on Kaiser’s report. Obviously, one of the biggest differences between our analysis and theirs is our inclusion of those individuals who were previously purchasing coverage on the individual market. We were looking at the total population of eligible individuals who could have been expected to enroll in exchange coverage, they’re more focused on the uninsured population and the change over time...

Additionally, we used 2015 exchange plan selections data as opposed to Kaiser’s 2014 numbers and used American Community Survey data (ACS) rather than CPS data [I had this wrong and have corrected - AS]. However, all that said, it appears from your article that this is a difference in analysis. I think our analysis more directly deals with the size of the exchange market and where future growth needs to come from. Kaiser’s appears to focus more on the uptake of the uninsured specifically and where growth from just the uninsured may come in future years...

I also agree that it is a good question whether the proportion of low income QHP buyers who were previously uninsured could’ve been enough to account for the difference. I don’t think it would be enough to explain the whole difference, particularly given the relatively small numbers of individuals purchasing non-group coverage in 2013 who were in the 100-150% FPL group (individuals in that income level frankly would’ve been very unlikely to be able to afford coverage).
Avalere will take a new look at marketplace plan takeup among different income groups when HHS publishes more detailed data about 2016 enrollment in current weeks, so that's something to look forward to. 


  1. Kaiser finds that marketplace-eligible uninsured in the lowest income bracket, below 150% FPL, have the second-lowest reduction in their uninsured rate among the income groups -- just a 14% reduction, compared to 33% reduction at 150-200% FPL, 23% at 200-300%, and 22% for those above 400% FPL (and so subsidy ineligible).

    SHould 14 be an 18 as per KFF Fig 1 (<150)? You lost me on your citation if not.

    1. Oops, yes, thank you, Brad. Just a typo, or a misremembering between eyeballing and typing.

  2. Btw, another theory. The heterogeneous and "harder to reach" nature of <150 may make inroads into that group highly variable. In one region, you might get blanket exposure because of interest group/navigator push (region 1 45% reduction) vs nada (region 2 5% reduction). Politics and resources. Whereas >150 have better means, receptivity, and have more across the board, smoother acceptance and uptake. Geography thing. Just thinking out loud.

    1. That could be. There are certainly a lot of areas where penetration is very poor.