Saturday, January 20, 2018

Subsidy-eligible, but unsubsidized

Updated 1/22 with California discussion corrected - please see note 2 at bottom.

Reading about Gallup's latest finding that 3.2 million more Americans were uninsured in 2017 than in 2016, I wondered about a particular (small) group of ACA marketplace enrollees: those with incomes in subsidy range who are nevertheless unsubsidized. Such people might be expected to have a particularly hard time with the steep premium increases of 2017 (not to say 2018 and 2019), as premiums take a huge share of their income.

I wanted to look at this group because Gallup's findings raised some riddles. Gallup found the sharpest increases in the uninsured among people with incomes under $36,000 -- but also found no change in the percentage of people insured by Medicaid. Among types of insurance obtained, the sharpest drop was among those who say they bought their own insurance (although Gallup's numbers for this population are weirdly inflated).  In the ACA marketplace the large majority (84%) who obtained premium subsidies were largely insulated from the year's steep premium hikes -- though reduced competition may have made choices less viable for some. Overall,  marketplace enrollment was down about 5%, or 500,000 in 2017.

The drop in off-marketplace individual market enrollment was doubtless steeper.  But most of those who buy off-exchange are presumably more affluent. It doesn't all seem to quite add up -- hence my interest in the almost invisible low-income unsubsidized individual market enrollees.

Most people in this situation are rendered subsidy-ineligible, I assume, because of an affordable offer of insurance from an employer. (The Kaiser Family Foundation estimates that 3.7 million uninsured people were in this category in 2016.) By 2017, it's hard to imagine unsubsidized marketplace coverage being more affordable than an employer-sponsored plan offering minimum essential coverage, but a few hundred thousand people do find it so. Other marketplace enrollees may have incomes under 100% FPL in states that refused to expand Medicaid. Eligibility for marketplace subsidies begins at 100% FPL.

There is scanty definite data about unsubsidized enrollees with income in subsidy range (or below it). Every year, somewhere between 6% and 8% of enrollees on have been recorded as "income unknown" (with some deliberate vagueness in 2017 -- see note at bottom). Most, presumably, have incomes high enough to disqualify them from subsidies, but not all. (An enrollee who is not seeking a subsidy does not have to enter income information.) There were about 740,000 enrollees in this category in 2016. In 2017, when overall enrollment was down about 5%, there were were just about the same number (see note).

California, which keeps its own enrollment stats, opens a window on unsubsidized low-income enrollment. In addition to "unsubsidized applications" (i.e., those that don't report income) that account for a bit over 6% of enrollees, the state tallies unsubsidized enrollees in each income category (e.g., below 100% of the Federal Poverty Level, 100-138% FPL, etc.).

As of March 2017, 1.7% of all enrollees were unsubsidized but had reported incomes under 250% FPL, which means that their incomes definitely would have qualified them for subsidies had some other factor not rendered them ineligible. Another .06% had incomes in the 250-400% FPL range and were unsubsidized. Many of them were probably subsidy-ineligible on an income basis. That is, the unsubsidized premium for a benchmark silver plan cost less than the percentage of income deemed "affordable" (9.6% for those over 300% FPL).

In California, then, during open enrollment for 2017 about 2% of enrollees reported incomes in subsidy range but were unsubsidized.  If one sixth of the 6% who did not report income were in subsidy range, that would peg the subsidy-eligible unsubsidized population at about 3% of enrollees. Nationwide, that would translate to about 300,000 enrollees annually, after the the annual shake-out of those who never pay their first premiums. Some fraction of the 6-7 million who buy health insurance off-exchange is doubtless also in subsidy range.

That's a pretty small population, and it couldn't account for much of the 3.2 million drop in the ranks of the uninsured alleged by Gallup (which does not match NHIS findings - see note below). Still, if the low-income unsubsidized enrollees on-exchange were halved, and the same were true off-exchange, that would constitute a reduction of perhaps 300,000 low-income insured.

But it didn't happen -- at least, not in California. There, the percentage of enrollees who were low-income and unsubsidized rose in 2017. As of March 2016,  just 1.1% of enrollees had incomes under 250% FPL and were unsubsidized -- again, compared to 1.7% in 2017.

In 2017, however, there was much sharper attrition from March to September among low income unsubsidized enrollees than in 2016 -- also sharper than in the enrollment population as a whole. By September 2017, enrollment among unsubsidized enrollees with incomes under 250% FPL had shrunk by 24%, from 23,010 to 17,410. Among all enrollees, attrition was just 6% in that period. In 2016, attrition in the unsubsidized 0-250% FPL cohort was just 8%, compared to 6% for the enrollment population as a whole.

That sharp drop from March to September 2017 in low-income unsubsidized enrollment could be statistically insignificant. Or it could reflect increased difficulty paying full freight for premiums that rose 13% in 2017.  Even after the drop, though, unsubsidized enrollees with incomes below 250% FPL made up a higher percentage of total enrollment in September 2017 than in September 2016 -- 1.3% compared to 1.1%.

Other states' experiences may not mirror California's -- and a higher percentage than I'm assuming of those who don't report income may have subsidy-range incomes. Perhaps a related faulty assumption is that the vast majority of those who buy health insurance off-exchange have incomes above subsidy range. Off-exchange enrollment may well have dropped sharply in 2017, and those who dropped out may be disproportionately low income.

One more afterthought: In 2017, premium increases in California were large -- 13% on average -- but lower than in most states. Perhaps the drop in low income unsubsidized enrollment was much steeper in states that saw the worst spikes -- in which case my "negative" results, derived from California enrollment stats, may not hold for the country as a whole.

Update 1/21: As Stan Dorn notes in comment below, the latest National Health Interview Survey (NHIS) early release estimates, published in December, found no overall change in the U.S. uninsured rate in the first six months of 2017, and found that the uninsured rate had decreased in states that have expanded Medicaid and increased in states that did not. Similarly, the uninsured rate decreased in states that run their own marketplaces and increased in states that do not, with some notable exceptions.

Note: In the 2017 public use files recording ACA marketplace enrollment, CMS did not report the percentage of enrollees in states with incomes under 100% FPL or over 400% FPL, as it had in past years. Instead it rolled those enrollees into an "other FPL" category, which included those with unknown FPL, i.e. those who did not report income. Altogether, "other FPL" accounted for 12.2% of enrollees. According to the "Definitions" sheet,  "unknown incomes accounted for approximately two thirds of the total 'Other' category" -- which would be about 8%, compared to 7.6% in 2016. In 2016, when HHS did break out below 100% FPL and above 400% FPL categories, 3% of enrollees had incomes under 100% FPL, and 2% had (reported) incomes over 400% FPL.

My working assumption, reinforced some time ago by Kaiser's Larry Levitt, is that the vast majority of those who don't report incomes have incomes above subsidy range. Other researchers, e.g., Avalere Health, spread the unknowns proportionately through all income categories -- which would yield a much higher estimate of unsubsidized enrollees in subsidy range.

As for the roughly 3% of  marketplace enrollees whose incomes are below 100% FPL, most of them probably are legally present noncitizens subject to the federal "5-year bar" for Medicaid enrollment, or even longer waiting periods imposed by some states. People in this category are subsidy-eligible and pay 2% of income for benchmark silver, enhanced by Cost Sharing Reduction to an actuarial value of 94%. But some are probably citizens in states that refused the ACA medicaid expansion, left out in the cold by that refusal -- as subsidy eligibility in the marketplace begins at 100% FPL.

Note 2, 1/22/17: I erred in the initial post in picking up a data set from Covered California for March 2016 that was not apples-to-apples with 2017 - a report of "Covered California Premium and Subsidy Statistics by Demographics", available on the Data and Research page. I also used a saved version of the March 2017 numbers that's since been updated.  The corrected text above compares the Active Member Profiles (on the right margin of the Data and Research page) for March and September of both years.


  1. I'd be curious about undocumented immigrants, otherwise income-eligible for subsidies, who purchased unsubsidized plans off-exchange (since the California legislature killed the plan to let them buy from Covered CA)? Surely there's at least a non-zero number of families who fall into that category, though we'll never really know how many.

  2. Check out NHIS data showing that coverage losses occurred in states not expanding Medicaid and FFM states. In expansion and SBM states, coverage increased. Time frame: Jan to june 2017.