4 Million Medicaid Enrollees Under Obamacare? Doubtful.
The White House has propagated numbers regarding sign-ups for the Affordable Care Act, claiming that the program is responsible for 4 million people gaining coverage through Medicaid expansion; 3 million enrollees in their early 20s who remain on their parents’ policies; and 2 million who have purchased insurance through the exchanges.
This claim has significant ramifications, both in terms of policy and politics. If 9 million people really are new beneficiaries of Obamacare at this point, then there is a decent chance that the GOP’s “repeal” mantra will be a net negative by the fall.
All of these numbers have issues. Objections to the latter two -- as well as the rejoinders to these objections -- have been discussed extensively, and I will not rehash them here. Suffice it to say that at least some of these people had insurance before the ACA went into effect, and would obtain insurance if the ACA were repealed.
But I haven’t really seen much discussion about the Medicaid figures. The 4 million new beneficiaries seems to be taking on near-canonical status, even being used by the fact checkers at the Washington Post for evaluating GOP claims.
This is odd, because after looking carefully at the numbers cited, the Medicaid figures are the weakest of the bunch. It’s a virtual certainty that the number of enrollments attributable to Obamacare is an order of magnitude less than the 4 million sign-ups implied, and the number of people [on Medicaid] who would actually lose their insurance if Obamacare were repealed is probably around 200,000 to 300,000.
The problem is identified in this Ezra Klein column (emphasis mine):
“Meanwhile, in October and November alone, more than 4 million people signed up for Medicaid coverage. This number will be much higher when December’s totals are released. It’s hard to say exactly how many of those Medicaid enrollments Obamacare is responsible for -- the government’s numbers don’t distinguish between people who signed up through Obamacare’s Medicaid expansion and those who entered the program through pre-existing channels. But the fact remains that Medicaid enrolled well over twice as many people as signed up for private insurance through the exchanges."
(Note: Klein has just published more to this effect.)
This is really important stuff. The statistics tell us how many people signed up for Medicaid, period, in October and November. The problem is that people are always signing up for Medicaid. Even without the ACA, we would have had people signing up in October and November. Lots of them, in fact: Medicaid is a program that services 60 million citizens, so the number of monthly enrollments that keep a relatively stable population is pretty substantial.
So, how many of these 4 million sign-ups are attributable to Obamacare, and how many of them are just people who entered the system under the earlier rules? The “4 million” number derives from this document, published Dec. 20 by the Centers for Medicare & Medicaid Services. It reports on the November enrollment numbers. There’s a similar document for October here.*
Let’s turn to this chart, which tells us the number of applications for financial assistance received by states in October and November (4,209,742, of which 1,736,809 applied in November), and the number of people who were determined eligible for both Medicaid and the Children’s Health Insurance Program (3,926,068, of which 1,741,011 were determined eligible in November). These numbers are where the “4 million” comes from.
CMS has published the numbers by state. It has also helpfully -- and critically -- sorted these numbers by states that have implemented the Obamacare Medicaid expansion and those that have not (the charts are big; I’ve relegated them to the end of this piece for the sake of readability. To see slightly larger versions, look here and here).
The above chart tells us that 1.7 million people were determined eligible for Medicaid in November of this year alone. The charts at the end tell us that 780,000 of these enrollees were in states that have undertaken the Obamacare expansion, while 960,000 of them were in states that have not done so.
So, of the November enrollees, 55 percent are in states where the Obamacare expansion of coverage didn’t occur and the ACA is therefore very unlikely to be directly responsible for their coverage. If we look at the October numbers, a little less than half (49.82 percent) were in states that didn’t expand coverage. Therefore, in total, of the 3.9 million individuals newly covered by Medicaid in October or November, only about 1.9 million are from states that expanded Medicaid.**
The next question is: How many of these 1.9 million are eligible directly because of Obamacare’s Medicaid expansion, and how many were just “normal” Medicaid enrollees? As Klein notes, we’d love to have October and November data on enrollment from last year, but unfortunately we don’t.
To get a better handle on this, let’s start with a hypothetical. Imagine two states, both of which are enrolling 100 people per month in July, August and September. Then, in October, one of them expands coverage. It starts enrolling 110 people per month, while the other state continues to enroll 100 people per month. It would be reasonable -- probably our best guess, in fact -- to assume that we have 10 enrollees per month due to the expansion.
The good news is that CMS provides something like these data. It gives us the July-through-September average number of applications, as well as the preliminary number of applications for October and November. Presumably if there are additional people being enrolled in the states that implemented the Medicaid expansion due to that expansion, there will be more applications being submitted as well. If we look at the difference in application rates between states that expanded Medicaid and those that did not, we can get a decent sense of how many of the new enrollees are due to the Medicaid expansion. It isn’t perfect, but it will provide at least a useful heuristic.
The bad news is that the real-world data are nowhere near as clean as in our hypothetical. Also, not every state provided data for July through September, while others didn’t provide October and/or November data. Making matters even worse, these data are only applications to state Medicaid and CHIP agencies, making them simultaneously over-inclusive and under-inclusive.***
These problems insert real uncertainties into our estimates. But I don’t think they’re insuperable problems, either. We just have to keep in mind that these are estimates, and may be off by a fair amount (in either direction). To begin, I (necessarily) eliminated states that didn’t submit data for both the July-to-September timeframe and the October/November timeframe.
States that engaged in the Medicaid expansion saw, on average, 39,203 applications for Medicaid and CHIP in November. That is actually down 4.2 percent from the 40,936 applications from July-September (average per month). That sounds bad. But states that didn’t expand Medicaid saw, on average, 35,777 applications in November, down from 42,168 in July-September, a 15.2 percent drop-off. So, we can estimate (with some uncertainty!) that had Medicaid not been expanded, enrollment in these states would have been 11 percent lower.
In October, the raw numbers look better -- applications were up from the summer. But remember, what we really care about is the difference between states that expanded Medicaid and states that didn’t. States that expanded Medicaid saw an uptick in applications of 14 percent on average. States that didn’t expand? Up about 4 percent. Once again, applications were 10 percent lower in states without the expansion than in states that had it. That consistency from two different months’ worth of data is reassuring.
Which brings us back to our number above: 1.9 million total enrollees approved in October or November in states that actually expanded Medicaid. If we are correct in our assumptions above—that is, if 10 percent of these enrollees are due to the Obamacare expansion—then we have an actual estimate for Medicaid enrollment due directly to the ACA’s expansion of Medicaid: 190,000.
We should remember that there are people who applied through the exchanges and were found eligible for Medicaid who aren’t included in this estimate. We should also emphasize that the numbers for November are still preliminary. On the other hand, we (necessarily) lumped together people who applied for Medicaid and CHIP in our data.
More importantly, several states that expanded Medicaid advertised this fact heavily and probably attracted a disproportionate number of people who were already eligible for Medicaid but didn’t know it. While it’s probably fair to attribute these enrollees to Obamacare, they would keep their coverage in the event of a repeal, and might not figure heavily in November. Regardless, even if you double the estimate -- which seems awfully generous -- that’s only 380,000 new Medicaid enrollees due to Obamacare, a far cry from 4 million.
This makes intuitive sense. The states that expanded Medicaid were blue states, many of which already had more generous Medicaid programs. In other words, their baselines were higher, so we’d expect the number of newly qualified recipients to be lower than if, say, Texas had participated.
Again, this doesn’t tell us much about how things will look in November, especially now that the federal exchanges are functioning. I suspect that as more people utilize the exchanges, we’ll see increasing numbers of “Obamacare Recipients” entering the system. But for right now, the “4 million enrollees” number looks like a significant exaggeration.
* It’s important to note upfront that all of these numbers are preliminary. There are 700,000 individuals, for example, who weren’t discussed in the October enrollment numbers but who gained eligibility afterward in a subsequent revision. We don’t know how they were distributed. But as we will see, there’s quite a bit of stability between the October and November enrollment numbers, so it seems unlikely that their distribution is different from those included in the preliminary assessments.
** Certainly some of the enrollees from states that didn’t approve a Medicaid expansion are “woodworkers,” meaning the people who were already eligible for Medicaid under the old rules and hadn’t enrolled, but were expected to “come out of the woodwork” when they heard about the Medicaid expansion. These folk were only expected to account for 5 to 10 percent of the actual “Obamacare expansion” number, even when every state was expected to undertake an expansion of Medicaid. I suspect there are maybe 20,000 of them in states that didn’t expand the program. More importantly, their political impact is minimal, since they wouldn’t lose coverage if the ACA were to be repealed.
*** CHIP wasn’t expanded significantly as a part of Obamacare, so enrollments in it are probably mostly not due to the new program. At the same time, we don’t have good numbers on the number of Medicaid applications via the federal or state marketplaces -- we have the numbers for some states (column four in the charts), but these numbers include people applying both for Medicaid and tax credits for private insurance. Fortunately, given the state of the marketplaces in October/November and the relatively small number of CHIP applicants, these are pretty small figures.