RAND Corporation: Net increase of 16.9M insured via ACA (UPDATE: With several caveats)

OK, I just got this and it just went live, so I'm just now reading it as I'm typing this...call it "liveblogging" if you will...(Update 6:00pm: OK, pretty much done now)

4 p.m., ET, Wednesday
May 6, 2015


Insurance coverage has increased across all types of insurance since the major provisions of the federal Affordable Care Act took effect, with a total of 16.9 million people becoming newly enrolled through February 2015, according to a new RAND Corporation study.

Researchers estimate that from September 2013 to February 2015, 22.8 million Americans became newly insured and 5.9 million lost coverage, for a net of 16.9 million newly insured Americans.

Among those newly gaining coverage, 9.6 million people enrolled in employer-sponsored health plans, followed by Medicaid (6.5 million), the individual marketplaces (4.1 million), nonmarketplace individual plans (1.2 million) and other insurance sources (1.5 million).

OK, the initial net numeric increase (16.9M) is actually higher than my own estimates (14-15 million net), which is great news...but it's still in the same ballpark.

The first thing which jumps out at me is the breakout of the newly insured; the ESI portion is higher than I would've figured. If memory serves, the big survey released by RAND last spring also claimed an unexpectedly high increase in ESI (employer sponsored) policies, so at least that's consistent.

The study also estimates that 125.2 million Americans -- about 80 percent of the nonelderly population that had insurance in September 2013 -- experienced no change in the source of insurance during the period, according to findings published online by the journal Health Affairs.

“The Affordable Care Act has greatly expanded health insurance coverage, but it has caused little change in the way most previously-covered Americans are getting health insurance coverage,” said Katherine Carman, the study’s lead author and an economist at RAND, a nonprofit research organization. “The law has expanded coverage to more Americans using all parts of the health insurance system.”

Imagine that! "Obamacare" didn't cause the earth to open up and swallow the entire nation whole after all!! Awesome!

Here's where the numbers are broken out:

The RAND study estimates 11.2 million Americans are insured through new state and federal marketplaces created under the Affordable Care Act, including 4.1 million who are newly covered and 7.1 million people who transitioned to marketplace plans from another source of coverage.

I'm not happy about saying this, but I actually find that 11.2 million figure difficult to believe.

The RAND study only ran through "the end of February 2015" (first page of the full report). The total number of QHPs selected via the ACA exchanges as of February 22nd was 11.7 million...yet only around 9.5 million of those had actually selected their policies (or had them renewed) as of January 15th (in time for February 1st coverage). The additional 2.2 million enrolled between 1/16/15 - 2/22/15, and their policies wouldn't actually kick into effect until March 1st (after the RAND study).

Even assuming that all 11.7 million were "counted" as being insured as of the end of February, not all of them paid their first premium. As regular readers know, my general estimate has been that around 88% of exchange enrollees pay up. If so, that means only about 10.3 million of these folks should be counted. Of course, I have reason to believe that the 2015 payment rate is higher than last year (perhaps as high as 94%), but even that would only bring it up to 11 million even. If 11.2M really did pay up by the end of February, that'd be a 96% payment rate, which would be fantastic but I seriously doubt.

My own estimate of the number of people who had actually paid as of the end of February was more like 10.3 million, although it should be around 10.8 million or so as of now (early May).

I'm not sure where RAND is getting their 11.2M figure from; seems to me it should either be higher (if they're counting everyone) or lower (if they're counting paid only). If 96% really did pay their first premium, however, that'd be phenomenal.

UPDATE: I just received clarification from Katherine Carman of the RAND Corporation, which confirms my thoughts above and which reiterates the "Limitations" points listed at the end of this entry:

Our estimates are based on survey questions that ask people what type of coverage they have.  We estimate what fraction of our sample reports each type of coverage and then multiply by the population between 18 and 64.  It is  possible that some of the people we surveyed reported coverage, but weren’t actually covered yet (if their coverage starts in March) or they won’t end up paying their premiums. 

I should also add that the numbers you cite (11.7 and 9.5 million) are within the confidence intervals of our estimates.

As for the 4.1M newly covered / 7.1M previously covered, I'll have to read the whole thing to get what they mean by this. Last spring the Kaiser Family Foundation estimated that around 57% of the 8 million 2014 exchange enrollees were newly insured (around 4.6 million). Of those, around 4 million paid up, and of those, around 3.8 million were still enrolled by the end of the year.

The actual breakout this year was around 6.3 million renewals from 2014, plus another 5.4 million new to the exchanges. It sounds like RAND is saying that 4.1 million of those 5.4 million are newly insured, which again, would be fantastic (76% vs. last year's 57%)...but it gets a bit confusing.

In addition, among the 12.6 million Americans newly enrolled in Medicaid, 6.5 million were previously uninsured and 6.1 million were previously insured.

Just yesterday the CMS Dept. released the February report which gave the net gain in Medicaid/CHIP as 11.7 million...except that you have to tack on an additional 950K for the "bulk transferrees" as I noted. This brings it to 12.6M, which matches perfectly with the RAND study.

The RAND study is the first to examine insurance transitions since the end of the second open enrollment period under the Affordable Care Act.

It analyzes information from the RAND Health Reform Opinion Study, a survey that has followed a representative sample of about 1,600 Americans aged 18 to 64 from September 2013 through February 2015. Participants are questioned periodically about whether they have health insurance, their source of insurance and other related questions.

That's an important point: If I'm understanding this correctly, it doesn't include children, which presumably means that the total numbers had to be corrected for this (otherwise there'd be no way they could hit 11.2 million exchange enrollees or 12.6 million Medicaid/CHIP enrollees, for example). It also doesn't include adults over 64, which is less of an issue since thanks to Medicare (you know, that government-run "socialized medicine" program), the uninsured rate among the 65+ crowd has been around 2% or less for quite some time now.

None of the other surveys with equally current data on changes in coverage caused by the Affordable Care Act provide longitudinal information about coverage transitions, and none of the other longitudinal surveys are as current.

RAND researchers say the findings that the biggest gain in coverage was from employer-sponsored insurance runs counter to predictions that many employers may quit offering insurance in response to the Affordable Care Act and suggests that regardless of whether that occurs, employer-sponsored coverage will remain the nation’s major source of health insurance coverage.

Other study findings include:

  • Most of the gain in the number of Americans with insurance occurred between September 2013 and May 2014, reflecting the first open enrollment period.

Yup, that's about right: A net gain of around 11-12 million thorugh last summer, plus another (net) 4-5 million from last fall through this spring, according to RAND (or 3-4 million according to, um, me).

  • Coverage through individual nonmarketplace policies declined by 1.9 million and coverage from other sources (Medicare, military insurance and state programs) declined by 10 million over the study period.

The first number refers to "Off-Exchange" policies, including Grandfathered and Transitional ones as well as ACA-compatible policies purchased outside of the ACA exchanges. This sounds about right to me; after all the fuss and bother about "OMG!! ELEVENTYBILLION POLICIES CANCELLED!!" back at the end of 2013, in the end only around 400,000 off-exchange individual policies ended up being cancelled specifically due to ACA non-compliance. That suggests that an additional 1.5 million people voluntarily moved from their 2013 off-exchange policy to something else (some moved to the exchanges, some to Medicaid, others to Medicare, ESI or what have you).

The second number (-10M) is interesting; I'd imagine this mostly consists of state employees being given the boot due to budget cuts.

  • An estimated 24.6 million Americans moved from one source of insurance to another source of coverage during the study period.

Ah, gotta love baseline churn, which makes tracking this stuff SO easy...

Support for the project was provided by RAND internal funding. Other authors of the report are Christine Eibner and Susan Paddock.

RAND Health is the nation’s largest independent health policy research program, with a broad research portfolio that focuses on health care costs, quality and public health preparedness, among other topics.

I'll write up further analysis of the full study itself later on, but there's a lot to unpack here.

UPDATE: OK, this may explain some of the somewhat curious numbers in the study; here's their "Limitations" section:

These data provide a unique opportunity to study insurance transitions since September 2013. However, there were some limitations. First, the sample contained only 1,589 observations, which reduced the precision of our estimates. Second, some respondents may have incorrectly reported the type of insurance coverage they had. In particular, Medicaid and non-group coverage were difficult to measure in survey data because of confusion among consumers over the names of these programs. Furthermore, people may have had difficulty distinguishing Marketplace coverage from Medicaid and other nongroup coverage as a result of confusion over the definition of “Marketplace” and because qualified applicants may have been directed to Medicaid through a Marketplace website.

Third, as previously mentioned, the response rate for our survey, around 9 percent, was low. Nonresponse especially in web-based surveys may bias estimates of enrollment in web-based Marketplaces. Despite weighting to match the CPS as closely as possible, this low response rate may indicate that the results were not nationally representative. Fourth, one concern with panel data was that participation in later waves may be influenced by the variables of interest—in this case, that insurance choices may influence the decision to participate in later waves of the survey. To address this concern, our survey weights adjusted for nonresponse associated with factors that are observable in our data. A strength of the longitudinal approach is that it avoids recall bias that might occur when respondents are asked to retrospectively report about prior insurance coverage.

In addition, regarding the unusually large-sounding ESI number, there's this (thanks to Adrianna McIntyre):

Among the 22.8 million people who gained insurance, most enrolled in employer-sponsored insurance, followed by Medicaid and the Marketplaces. Employer coverage is by far the largest source of insurance among Americans younger than age sixty-five, and the ACA creates new incentives for people to take up employer policies. Specifically, while the ACA mandates that most people must enroll in insurance, people are ineligible for Marketplace subsidies if they have an affordable offer of coverage from their employer. Gains in employer coverage were also found following Massachusetts’s health reform.

However, other nationally representative surveys did not show an increase in employer coverage between 2013 and 2014. It is possible that the increases in employer coverage that we observed were idiosyncratic to our small sample, rather than a true representation of changes in coverage at the population level.

In other words, while the total number is in the same ballpark as pretty much every other major survey/study (Gallup, Urban Institute, Commonwealth Fund, etc), there's a lot of room for error here, especially as you drill down into the smaller numbers.

HOWEVER, the Big Picture in the Conclusion remains totally valid:

The ACA has greatly expanded health insurance coverage in the United States with little change in the source of coverage for those who were insured before the major provisions of the law took effect. Furthermore, the law has expanded coverage using all parts of the health insurance system, including employer-sponsored insurance, Medicaid, and the newly created Marketplaces. While these data have limitations, especially due to the low response rate, they provide an early look at how the ACA has affected insurance enrollment.