Looks like I may owe the Heritage Foundation HALF of an apology...sort of.

Last October, I posted a piece in which I blasted the Heritage Foundation for posting a widely-disseminated article in which they claimed that only 3% of the net increase in 2014 insurance coverage was "due to" private ACA healthcare policies, with the remaining 97% being found via expansion of Medicaid via the ACA.

It's a long, wonky piece, but Heritage's claims fell into two main areas:

• First, they, like so many other right-wing organizations, constantly operate under the presumption that Medicaid "doesn't count" for some reason or another.

Apparently people enrolled in Medicaid aren't human beings deserving of healthcare coverage, and can therefore simply be subtracted from the "real people" total. This is the same type of logic which, to this day, has Republicans stating that "Barack Obama only won because so many black people voted for him"...which, even if true, is utterly irrelevant: "So many" black people did vote for him, and their votes count just as much as anyone else's. You could just as easily argue that most other presidents "only won" because so many white people voted for them, and it would have exactly the same relevance to the fact that they were still, you know, elected president.

More to the point, though, the other main claim was this:

• Yes, the Heritage Foundation says, the private individual market did increase by about 4.8 million people, however...
• Over 4.5 million people who had been covered by employer-sponsored insurance lost or moved off of their plans at the same time, meaning that...
• ...the net increase in private policy coverage only went up about 260,000 people, which, in their view, means that...
• only about 3% of the 9.25 million net increase in covered individuals in 2014 was via private policies, with the remaining 97% being via Medicaid expansion.

I countered by pointing out a whole mess of other reliable sources which collectively concluded that the "4.5 million ESI reduction" was likely way off base for a variety of reasons, including cherry-picking multiple sources, misinterpreting datasets and so forth.

My conclusion was that yes, it's true that most of the increase in coverage in 2014 was indeed due to Medicaid expansion, but that it was more along the lines of 70% instead of 97% (leaving around 30% via private insurance, fully 10x the 3% that Heritage claimed).

Well, today, Paul Houchens of Milliman, one of the largest and most respected actuarial firms in the world, just released a conclusive report regarding full-year 2014 private health insurance industry financial results. There's a whole mess of stuff in there about the "3 R's" programs (Risk Corridors, Risk Adjustments and Reinsurance), etc., but the part which caught my eye was this:

• While enrollment in the individual health insurance market increased from 10.9 to 15.0 million covered lives (nearly a 40% increase), overall enrollment decreased in the commercial health insurance markets from 2013 to 2014 by approximately 1.5% as a result of a 5.3 million decline in enrollment in the fully insured group markets.

Oops. Color me embarrassed...looks like Heritage was right about the individual market increase being cancelled out by ESI losses...and in fact, this makes it sound like there was a net reduction in private coverage (of around 1.2 million), right? Well, not quite; the very next sentence says...

...This decline may be attributable to insurers moving business to self-funded platforms, which avoid adjusted community rating rules (small group market only) and the ACA’s Health Insurer Fee (HIF). For 2015, insurers subject to the tax owed amounts equivalent to approximately 3% of their 2014 premium (grossing up for federal corporate taxes).

I asked Mr. Houchens about this, and while he wasn't able to provide an exact figure, he did say that he estimates that "most of" the ESI "reduction" was indeed due to companies switching over to "self-funded" platforms, which simply don't show up in the numbers. In other words, those people are still presumably fully insured, but their data just doesn't show up in the official "small group" or "large group" private insurance categories.

@charles_gaba @xpostfactoid1 I would estimate most is switch to self-funded, particularly large group (1/2)

— Paul Houchens (@PaulHouchens) March 21, 2016

He went on to provide links showing that the percentage of private employers offering healthcare coverage barely dropped from 2013 to 2014 (from 49.9% to 47.5% for companies of all sizes, and from 95.7% to 94.8% for businesses with 50 or more employees). This suggests a pretty nominal reduction.

Of course, "most" is pretty vague. Half of 5.3 million is around 2.65 million; let's assume that 2.8 million of the total "reduction" (53%) was actually just a move to self-funding. If so, that means that there was only an actual net loss of 2.5 million people covered by ESI in 2014.

That, in turn, would mean, using the Heritage Foundation's other numbers:

• 4.5 million increase in indvidual market enrollees
• 2.5 million decrease in ESI market enrollees =
• Net increase of 2.0 million private market enrollees in 2014.

The Heritage Foundation agreed that there was a total net increase of 9.25 million covered in 2014. That means that instead of a 97/3 Medicaid/Private split as they claimed or a 70/30 split as I estimated, the actual ratio was more like 78% Medicaid, 22% Private Coverage (which would still be closer to my estimate than theirs). The higher the actual "self-funded" portion actually was, the closer to my 70/30 split was actually the case.

Of course, none of this includes anything that happened in 2015, when Medicaid expansion started slowing down, nor does the 2016 Open Enrollment period come into play. In addition, the Milliman report is based on full year monthly average enrollment, whereas some of the other figures are based on "point in time" or partial-year results, so there's still a lot of room for error one way or the other. I should also point out that shifting from ESI to the individual market can actually be a good thing since it gets rid of the "Job Lock" problem, but that's a whole other discussion.

Anyway, there's a whole bunch of other data-geeky info in the Milliman report. Read the whole thing.