Uninsurance goes upscale: The Census's Health Insurance Coverage Update for 2018

Yesterday the Census Bureau* released its health insurance estimates for 2018. The top line showed a rise in the uninsured population of 1.9 million, or 0.5% -- the first increase since the ACA's main programs launched.

Disturbingly, the number of uninsured children increased by 425,000, or 0.6%, raising children's uninsured rate to 5.5%. That spike would appear to be due mainly to a drop in Medicaid coverage, given that  Medicaid and CHIP coverage for children was down 1.2%; and the overall percentage of children with public health insurance dropped 0.8%, while the percentage of children with private health insurance ticked up 0.2%. There was also, however, a sharp spike in the uninsured rate among children in households with incomes over 400% of the Federal Poverty Level (FPL), from 1.9% in 2017 to 2.6% in 2018 -- accounting for almost half of the increase in uninsured children.

Folks at Georgetown University and the Center for Budget and Policy Priorities will probably dive  into the spike in uninsured children, as they have been doing for at least a year. Here I just want to throw some sidelights on the Census numbers generally.

1. Affluent uninsured population spikes. Notwithstanding a drop of 2 million  (0.7%) in Medicaid enrollment, the sharpest increases in the uninsured were at high incomes. At 300-399% FPL, the insured rate dropped a full percentage point, from 92.9% to 91.9%, and at over 400% FPL, the rate dropped 0.8%, from 97.3% to 96.6%. Together, these two income groups account for 55% of the population. Particularly striking, the number of uninsured at incomes over 300% FPL increased 23.8% (from 6.631 million in 2017 to 8.215 million in 2018).  The spike in uninsured children at high income levels seems congruent with this drop.

Census: ACA cut uninsured rate in half in Medicaid expansion states by 2016

The Census Bureau released its report on health insurance coverage in the U.S. for 2016 today. One striking trend was flagged by Matt Broaddus at the Center for Budget and Policy Priorities: the gap between states that expanded Medicaid and those that refused continues to widen:



To this let me add a sidelight: in expansion states, the uninsured rate has been cut in half since the main ACA programs were implemented in 2014 -- from 12.9% in 2013 to 6.5% in 2016.

As more Americans move out of poverty, will some move into ACA marketplace?

Data from the Census Bureau's just-released Current Population Reports for 2015 records a trifecta for the U.S. economy: incomes soaring (after years of stagnation and shrinkage), poverty reduced, health insurance rates rising.

A thought occurred to me while reading this part of economist Jared Bernstein's overview:

The official poverty rate fell from 14.8 percent in 2014 to 13.5 percent last year, a decline of 3.5 million people in households whose incomes put them below the poverty threshold for their family type (the poverty threshold for a single parent with two children was around $19,000 in 2015). 

The thought: Families rising out of poverty could be pushing ACA marketplace enrollment from the bottom up, particularly in states that refused the ACA Medicaid expansion.

Is employer-sponsored insurance superior to ACA marketplace plans? For whom?

So far, the Affordable Care Act has not curtailed the availability of employer-sponsored insurance (though it may have curtailed its growth). On the front page of today's New York Times, Reed Abelson reports that "health care remains an important recruitment and retention tool as the labor market has tightened in recent years." One part of her explanation is "mostly true," as the fact-checkers say -- but hides an important fact about what kind of insurance is available to whom:

The law has resulted in more coverage for low-income people, as expected. But...the plans on the exchanges remain less generous than those offered by many employers, with significantly higher deductibles and a significantly narrower choice of hospitals and networks.

That's probably true for most workers whose employers offer them insurance. It's not true, however, with respect to out-of-pocket costs for workers whose household income is below 201% of the Federal Poverty Level (FPL). As I noted last Friday, for ACA enrollees with incomes below that threshold, the average weighted actuarial value (AV) of the plans they obtain -- that is, the percentage of the average user's costs that the plan will cover -- is 86%.  The average AV in employer-sponsored insurance was estimated by the Kaiser Family Foundation at 82% back in 2011 -- and it's probably dropped since then.  It's likely to be still lower for lower income workers.

The Counter-Upshot: Obamacare is quite as egalitarian as it appears

Tyler Cowen draws a rather odd conclusion from the spike this year in premiums for health plans sold in the nongroup market: The ACA is not as egalitarian as it appears.

Cowen legitimately spotlights the weakness of the ACA value proposition for uninsured people with incomes above 250% of the Federal Poverty Level (FPL) and so for a large group of the remaining uninsured. But his discussion of its "egalitarian" impact is limited by his leaving the Medicaid expansion out of the equation (if not entirely out of the discussion).

Cowen bases his case that "by some measures, the Affordable Care Act has had only a limited impact on economic inequality" mainly on a recent study* by Wharton School researchers led by health economist Mark Pauly.* Pauly et al found that slightly less than half of the still-uninsured who are eligible to buy health plans on the ACA exchanges would not experience "welfare improvements" if they buy health plans. Cowen summarizes:

Who leaves employer-sponsored insurance on the table?

The Kaiser Family Foundation's latest estimate of the still-uninsured population in the U.S. includes about 4.8 million whose employers offer health insurance that they decline to buy.

Some are locked out by the ACA's so-called "family glitch," which denies marketplace subsidies to those for whom an employer's insurance offering would cost less than 9.5% of income for individual coverage, even if the family coverage offered by the employer costs far more than that. But many low income workers find even individual employer-sponsored insurance (ESI) unaffordable. Probably the bulk of Kaiser's estimated 4.8 million who forgo ESI could do well in the ACA marketplace if the ESI offer didn't disqualify them from subsidies.

The New York Times' Stacy Cowley has a good story that spotlights the plight of those who can't afford insurance offered by their employers:

Hey, Jeb! Who's getting those "huge new subsidies" under the ACA?

Touting his new ACA replacement plan, which would wipe out ACA coverage rules for insurers and replace means-tested ACA private plan subsidies with tax credits for catastrophic coverage available at any income level, Bush asserted:

Obamacare created huge new subsidies for low-income Americans, but it left middle-income Americans facing higher premiums and higher out-of-pocket costs.

There are some grains of truth to that.  People who 1) get their insurance in the individual market,  2) earn too much to qualify for subsidies (that is, over 300-400% of the Federal Poverty Level (FPL)), and 3) don't have pre-existing conditions pay more  than they would have pre-ACA.* You could argue, too, that the ACA has driven up out-of-pocket medical costs, if not premiums, for people with employer-sponsored insurance (ESI) -- or at least that it will do so once the Cadillac Tax kicks in, if it ever does. The claim is highly contestable, though, as neither premiums nor out-of-pocket costs have risen faster in ESI than in pre-ACA years and myriad factors are at work. Really, it's simply too early to tell.

But Bush's statement, like most Republican claims to speak in defense of "the middle class," reveals a top-heavy view of what "middle class" means. His sneer at "huge subsidies" is also a sneer at huge swaths of the U.S. population -- where the uninsured are concentrated.

A recent Census report** indicates that in 2014 the ACA caused large drops in the uninsured rate among Americans with incomes under 100% FPL ( a 4.2 point drop, from 23.5% to 19.3%), 100-199% FPL (a 5.3 point drop) and 200-299% FPL (4.2 points.).

ACA exchanges in 2016: targeting just 6.4 million subsidy-eligible uninsureds?

In a speech at Howard University College of Medicine today, HHS Secretary Sylvia Burwell laid out a few facts about the target market for the ACA exchanges -- those still uninsured and eligible for private plan coverage. A couple of key points:

• About 10.5 million uninsured Americans are eligible for Marketplace coverage in the upcoming open enrollment.


• Almost 40 percent of the uninsured who qualify for Marketplace plans are living between 139 and 250 percent of the federal poverty level (about $34,000 to $61,000 for a family of four).

HHS has confirmed for me that the 10.5 million estimate is not limited to the subsidy-eligible. At present, according to Kaiser estimates, about half of those who have bought plans in the nongroup market are ineligible for subsidies, and most of the subsidy-ineligible have bought their plans off-exchange. 

Does that, then, suggest a target market of just 5 million subsidy-eligible uninsureds? Not quite.* HHS's estimate of the target market between 139% and 250% FPL (4.2 million or a bit less) provides a basis for estimating the size of the subsidizable target market.  In 2015, about 76% of private plan buyers on the exchanges had incomes under 251% FPL.*  According to HHS's most recent enrollment snapshot, 83.7% of all exchange enrollees qualified for premium subsidies. Thus 91% of subsidy-eligible buyers were under 251% FPL. Not all of them, however, fit HHS's "40 percent" category of 139-250% FPL - because in states that refused the Medicaid expansion, eligibility for subsidized marketplace coverage begins at 100% FPL.*** In 2015, about 15% of all enrollees**** (and 20% of those under 250% FPL) were under 139% FPL and so outside HHS's category. Those between 139% and 250% FPL thus constituted about 60% of total enrollment. If that percentage holds in 2016, that would suggest that about 7 million of the 10.5 million in the target market are subsidy eligible.

But the percentage of potential buyers under 139% FPL will probably be considerably lower this year. Takeup among that group was disproportionately high: Avalere Health estimated that 76% of eligible buyers from 100-150% FPL did in fact enroll, and the percentage was probably still higher under 139% FPL (and dramatically lower for all higher income bands). Moreover, two (small states) have accepted the Medicaid expansion for 2016. If, in 2016, 10% rather than 20% of those under 250% FPL are also under 139% FPL, then that suggests about 580,000 fewer subsidizable targets. Since we're now in the realm of educated guesswork, let's say that HHS's estimate of the 139-250% market suggests that about 6.4 million of their 10.5 million overall target market is subsidizable.

At present, 83.7% of 9.9 million exchange enrollees are subsidized. If there indeed are only about 6.4 million subsidy-eligible uninsured still out there, then about 57% of the subsidy-eligible target market has been enrolled. That doesn't sound right. Kaiser has pegged the percentage of potential exchange population enrolled through June 20 at 35%.Though that estimate is not limited to the subsidy-eligible, uninsured rates are much higher in lower income brackets.

The 10.5 million estimate excludes the Medicaid-eligible and -- I assume but have not confirmed -- those in the Medicaid gap, who theoretically could buy unsubsidized plans on the exchanges. Those who earn too little to qualify for premium subsidies are unlikely to pay full price for plans on the marketplace -- although, confusingly (to me), the just-released census report on health insurance showed greater gains in private coverage than in government insurance for those whose incomes should qualify them for Medicaid under the ACA expansion.

The still uninsured: can't afford coverage or don't know what's on offer? (Or both?)

With respect to reaching the still-uninsured, Burwell cited findings from a PerryUndem study that reflect a tension between two key factors:

• About half of the uninsured have less than $100 in savings.

• Nearly three in five of the uninsured are either confused about how the tax credits work or don’t know that they are available.

The first point is a proxy for several points highlighting the financial precariousness of the uninsured. Another: 58% of respondents report having less than $100 left over each month after paying bills. In other words, many would have a very tough time with the average premium paid for marketplace plans, net of subsidies: $101 per month. At the same time, most who consider coverage unaffordable do not know what's on offer.

I have noted the same tension in data from the Urban Institute, and in other surveys of the uninsured. If the still-uninsured who qualify for premium and cost-sharing subsidies knew what was on offer, would they still consider it unaffordable? Doubtless some would and some would not. The proportion in each camp will go a long way toward determining how viable the ACA private plan marketplace will prove over the long haul.

Researchers at the Urban Institute and healthcare reporter Jed Graham have argued that the ACA subsidy structure is too skimpy to meet the needs of large percentages of the uninsured.  To whatever extent that's true overall, it's increasingly true as you move up the income scale -- as Avalere Health's analysis of takeup rates at different income levels indicates. Avalere estimated that about 76% of the subsidy-eligible uninsured with incomes under 150% FPL bought subsidized private plans, compared to 41% for those from 151-200% FPL, 30% for those from 201-250% FPL, 20% of those from 251-300% FPL, and so on down.  The threshold for really strong Cost Sharing Reduction, 200% FPL, is one dividing line between strong and weak aid.  The ACA works best for those up to that income level. At the same time, that's where most of the uninsured are concentrated.

P.S. The PerryUndem study, conducted in May 2015 and surveying some 1,270 adults, is full of interesting info about the finances and priorities of the uninsured.

UPDATE, 10/14/2015: Kaiser yesterday estimated that 7.1 million uninsured people are eligible for subsidized private plans in the ACA marketplace.

----
* I originally oversimplified this calculation, simply taking the percentage of exchange buyers under 250% FPL. Correction is in this paragraph.

** I calculated the percentage of exchange enrollees with incomes under 251% FPL in this post.

*** Thanks to Jed Graham for pointing out that HHS's estimate began at 139% FPL (rather strangely, when you consider the doubtless still-sizable number of uninsureds between under 139% FPL in states that refused the Medicaid expansion. Also among subsidized buyers under 139% FPL: legally present immigrants who are time-barred from Medicaid; they are eligible for premium subsidies even if their income is under 100% FPL (in nonexpansion states) or 139% FPL (in expansion states).

**** See this post for a calculation of 2015 exchange customer with incomes under 139% FPL. In this post, I've slightly dropped the estimate, from 16% of all buyers to 15%, in light of the recent purge of those who failed to verify their state income when asked.

Why does the Census show such small gains among the poor in government-provided health insurance?

If I may reiterate: it seems quite strange to me that the Census health insurance surveys show a net gain for 2014 of just 1.3 million people with incomes under 138% of the Federal Poverty Level enrolled in government insurance plans -- whereas according to HHS, Medicaid enrollment increased by over 9 million in 2014, thanks to the ACA expansion. 138% FPL is the eligibility threshold for Medicaid under the ACA expansion.

The second half of this post provides more detail. According to the Census surveys, those with incomes under 138% FPL showed much stronger gains in private insurance than in public, while those in higher income brackets showed stronger gains in public insurance than in private. That's odd.

The difference may in large part be due to differences in how households are defined: the ACA marketplaces determine eligibility according to who is included in a household tax filing, while the Census surveys (CPS and ACS) consider who lives under one roof. Many other factors are in play, including a modest acceleration in Medicare eligibility and higher income thresholds for children in CHIP than for adults in Medicaid. But none appear on the face of things (as far as I can see) to explain the very low recorded gains in government insurance among adults under 138% FPL (and even more strikingly, under 100% FPL).

Again, there's more detail below the second subhead here (along with some updates added over the weekend). This post is simply to unbury the lead a bit, as I look into the state data.

The Census on health insurance gains: who got what and how?

This week the Census reported on changes in Americans' health insurance rates from 2013 to 2014, based on results of its two yearly surveys, the Current Population Survey and the American Community Survey. The two together show what is probably the most dramatic drop in the percentage of people without insurance since Medicare and Medicaid were implemented. The drop in the ranks of uninsured was steepest among the roughly one third of the population living in households with incomes under 200% of the Federal Poverty Level (FPL) -- where lack of insurance is most concentrated.

In my grand personal tradition of burying the lead, I discuss an apparent oddity in the data under the second subhead below. Feel free to skip! If you're well-versed in these matters, it may be no mystery to you.

The near-poor gain most

The ranks of the uninsured dropped more steeply for the near-poor than for those below the poverty line, and for the part-time employed than for the nonworking population,  The pattern does not hold for educational level: the uninsured rate dropped most for those without a high school diploma and next most for high school grads, with smaller drops at each level of educational attainment.

Maybe we should call blue states bronze states

In several posts, I have tried to track how successful the ACA exchanges were in steering lower-income buyers of private plans toward silver plans, the only metal level at which subsidies that reduce deductibles and out-of-pocket costs are available. Low-income buyers of bronze plans, which offer low premiums but sky-high deductibles, forfeit access to these Cost Sharing Reduction (CSR) subsidies, which radically reduce costs for those with household incomes under 200% of the Federal Poverty Level and phase out at 250% FPL.

On the whole, the news is good, though information that correlates metal level selection with buyers' income is sketchy. In New York, the only state that breaks out metal level selection by income bracket, 89% of buyers under 200% FPL chose silver plans and so took advantage of CSR. In the federal exchange, only 15% of buyers who qualified for any kind of subsidy bought bronze plans.

Bronze plan selection varies quite a bit by state, however, for reasons that are hard to tease out.  For starters, more people on average bought bronze in the 14 states (plus D.C.) that built their own exchanges -- most of which are also among the wealthier states.

State per capita income seems to affect metal level selection. interacting with other factors such as premium prices, the quality of outreach to the uninsured, whether the state expanded Medicaid (which takes out potential buyers between 100 and 138% FPL), and the age composition of the buying pool.*

My thanks to a friendly stranger on Twitter who put together a scatterplot correlating state median income and bronze plan selection. That led me to the latest census data on median household income by state. Below, I've matched it up with HHS state-by-state data on ACA buyers' metal level selections.

For ACA buyers at all income levels, here are the bronze takeup rates for the thirteen highest- and thirteen lowest-earning states:

Will-fully Misleading?

UPDATE 4/6/10: David Brooks joins Will in misrepresenting Stephen J. Rose's household income figures as individual income

George Will, slamming the populism of Edwards and Huckabee, uses some pretty dicey-looking numbers to paint a picture of a wealthy electorate:

Economist Stephen Rose, defining the middle class as households with annual incomes between $30,000 and $100,000, says a smaller percentage of Americans are in that category than in 1979 — because the percentage of Americans earning more than $100,000 has doubled, from 12 to 24, while the percentage earning less than $30,000 is unchanged. "So," Rose says, "the entire 'decline' of the middle class came from people moving up the income ladder." Even as housing values declined in 2007, the net worth of households increased.

According to U.S. Census figures, in 2006, 19% of U.S. households earned over $100,000. Households in the top two income quintiles , those with an annual household income exceeding $60,000, had a median of two income earners while those in the lower quintiles (2nd and middle quintile) had median of only one earner. These numbers suggest that a small percentage of "Americans" earn over $100,000. The median household income is $48,201.00 (a handy summary is at Wikipedia*).

This is not to say that Edwards doesn't overstate the economic distress of Americans generally. It's fine to emphasize the continued prevalence of poverty in the U.S., which Edwards rightly calls shameful. It's also fair to stress that middle class income has stagnated, that income inequality is rising, and that there's been a massive risk transfer to households over the past 2-3 decades (e.g. in pensions, healthcare costs, and homeowners' insurance). But when Democrats start speaking as if the majority of Americans are in acute economic distress, it puts them out of tune with much of the electorate.

UPDATE, 5/6/09: What Rose actually said, in a Dec. 23, 2007 Washington Post op-ed, is that between 1979 and 2007,"the number of people in households that bring in more than $100,000... rose from 12 percent to 24 percent." Rose doesn't say whether the 1979 percentage is inflation adjusted. But he's a reputable economist, so let's assume that it is.  His broad point that per capita income has risen in this period holds -- though it's largely offset by the rise of one-income households, the fraying of the safety net, and the disproportionate rises in the cost of housing, higher education and health care.

* While the Wikipedia article places the 19% of U.S. households earning over $100k in 2005, the 2006 Census figures show the same percentage.