What's up with those Heritage numbers?

10/23/2015 08:41:00 AM
Heritage has a piece arguing that almost all (97 percent) of the increase in insurance coverage in 2014 was just the medicaid expansion, and that there was hardly any expansion of private, non-medicaid coverage due to a large net drop in coverage under employer-sponsored plans (ESI) by 4.5 million. They write
In sum, when it comes to increasing the number of individuals with health insurance coverage, the net effect of the ACA in 2014 was almost entirely a simple expansion of Medicaid.
The implication they're going for, I think, is to suggest that repealing the ACA except for the medicaid expansion would have only very minor effects on coverage levels.

Charles Gaba has already explained that these numbers don't make a whole lot of sense. To be sure, we've known for a while that the medicaid expansion has been among the most effective parts of the ACA in terms of boosting insurance rates. But none of the other data sources can corroborate the extreme disparity in the Heritage numbers.

I was reminded of the on-going RAND health insurance surveys, so I looked up their survey that covers the September 2013-November 2014 period, which should roughly line up with Heritage's numbers. They actually found that almost as many uninsured people gained ESI as signed up for medicaid—7.3 million to 7.5 million respectively. There's plenty of churn though, so counting those who lost ESI, they found a net increase in ESI of 6.6 million, way different than Heritage's -4.5 million. In fact, the gross number of those loosing ESI in the RAND data was just 3.4 million, on the same order as Heritage's net figure. The final sentence in RAND's abstract might as well be a direct rebuttal to Heritage:
These findings suggest that the ACA is expanding coverage through a variety of insurance sources, perhaps because the individual mandate is encouraging people to take up insurance offers that they might otherwise have declined.
Now, the two datasets are looking at slightly different windows—Heritage is looking at, supposedly, January 2014 through December 2014, while the RAND data starts a few months earlier and ends one month sooner. But if what Heritage found is that more than 10 million people lost ESI in December alone—wiping out years worth of gains—they should do a better job writing up their findings!

Heritage and RAND also differ in methods. Heritage took their data from a private consultant, Mark Farrah Associates, who compiled the data from surveys of insurers, while RAND conducted their own survey of households rather than insurers. Methodologically, trying to directly track transitions in a churning market can be error-prone as people will leak through the cracks in your models. So I find it helpful to look at snap-shot pictures like this Gallup survey. Unfortunately, they didn't break out just 2014 for us—to compare to the Heritage data—but their totals for the 2014 and 2015 period don't support Heritage's implication that the ACA is all medicaid and no private insurance: the uninsurance rate dropped 5.5 percentage points, while the proportion covered by medicaid rose only 2.4 percent. Gallup finds that the single biggest contributor to the decline in uninsurance—3.6 percentage points worth—was a large expansion in private insurance not through one's own employer. That sounds like a lot of people signing up for ACA exchange plans, but it's hard to be sure—as I understand their data, the ESI category only counts people covered by their own employer, while those covered by their parent's plan would be in the "self or family member" category along side ACA exchange plans. That's an important caveat because the ACA extended family plans to all under-26-year-olds.

Nevertheless, for whatever it's worth, Gallup's "covered by own employer" category looks less optimistic than the RAND numbers, with a net decrease of 1.2 percent, which corresponds to a net loss of roughly 1.5 million people. So, that's still way below the Heritage numbers. For reference, the prediction before the ACA went into effect were that coverage in ESI would remain relatively unchanged.

Quick update. I meant to include this remark in the original post but forgot: one might reasonably expect ESI to decline, because the population is aging and in particular large numbers of baby boomers have just become medicare-eligible, and even more will in the coming years. Projections were that the ACA would have little impact on this longer-run trend.
Anonymous 11/07/2015 10:22:00 PM
Just an anecdote, but in 2014 my employer switched from one of the major carriers to a self-insurance scheme put together by a hedge fund (or at least, that's how our insurance agent described it). If this were fairly widespread (and the rates were quite compelling for our small business) that might account for the difference between an insurer survey and a household survey, assuming they mostly polled the major insurers.