Addressing the 35% "Phantom Enrollee" Zombie Lie...again

"In other words, Republicans’ insistence that they have a superior alternative to Obamacare is a zombie lie — a claim that should be dead after having been proved false again and again, but it is still shambling along, eating people’s brains."

--Paul Krugman, NY Times, 6/29/20

This is perhaps the 3rd or even 4th time that I've written a post debunking this particular Republican zombie lie, but usually I've included it as part of a larger post debunking other silly claims as well. Since it's making the rounds again I figured I should write something up which addresses it standalone.

Over the weekend, Assistant U.S. Attorney General for Fraud Colin McDonald went on FOX News and, while discussing & defending the Trump Regime's announcement last week that they've kicked ~760,000 ACA enrollees off of their coverage due to suspected (not proven, mind you) fraud, he made the following claim:

"Apparently 35% of people have never even used their Obamacare account for any health services. That is a statistical impossibility and it's because these are not real people seeking real care. These are fraudsters who have exploited their names to put them on the Obamacare roles."

This claim is based on an analysis done by Brian Blase of the Paragon Institute. However, there's some major problems with the underlying data released from the Centers for Medicare & Medicaid Services (CMS).

First up: Is this really a "statistical impossibility?"

Let's listen to Matt Fiedler, Senior Fellow at the Center on Health Policy at the Brookings Institute. He addressed this very issue last August in a brilliant Twitter thread which I'm reposting below (reformatted for easier readability in a blog format); I hope he's OK with my doing so:

To start, it’s common for insured people to have no claims in a year; not everyone needs care. Is the 35% figure for individual market enrollees unreasonably high, as claimed?

Using the MEPS [Medical Expenditure Panel Survey], I estimate that about 15% of non-elderly people with 12 months of group* coverage in 2022 had no medical spending, even though no one thinks phantom enrollees exist in the group market.

*Group coverage = Employer-sponsored insurance, as opposed to individual market (ACA exchange) insurance.

So, right off the bat we've already accounted for over 40% of the so-called "phantom enrollees." Moving on...

And, as others have noted, there’s good reason to expect the individual market rate to be *much* higher: individual market enrollees are typically enrolled for only part of the plan year. Fewer months enrolled mechanically means fewer chances to incur claims.

Concretely, other CMS* data show Marketplace enrollment spells averaged only ~8 months in the FFM states in 2024. That’s enough to explain much of the difference between CMS’ estimate of the individual market no-claims rate and my 15% group market estimate.

*CMS = The Centers for Medicare & Medicaid Services...aka the very division of the Health & Human Services Dept. which Dr. Oz is in charge of.

In other words, the average person with employer-sponsored insurance is far more likely to be enrolled in that policy for the full 12 months than the average person with ACA coverage, who is generally someone who's either self-employed or who works for a small employer which doesn't provide coverage for them...both of whom tend to be less pinned down to a specific home address, have much more variable incomes from one month to the next, etc.

Also, as another health insurance industry colleague (who prefers to remain nameless...they work for a large insurer & are an expert on health insurance risk adjustment) put it after reviewing the Paragon analysis:

The short version is that the EWOC [Enrolled Without Claims] “total enrollees” counts enrollees multiple times if they change plans. It is impossible to say how many EWOCs there are in a state, but Paragon/WSJ’s numbers are inaccurate and really a measure of enrollee/plan combinations that double-counts anyone who changed plans during the year.

Of course there are plenty of other problems with Paragon’s analysis, but this one basically means the numbers they’re publishing are wrong and the phenomena driving the trends they identify may have very little to do with fraud.

In other words, if you enroll in a Blue Cross plan for 6 months but then move out of state and switch to, say, a Molina plan for the other 6 months of the year and don't happen to visit the doctor once the entire year, guess what? They're counting that as 2 enrollees "not filing a claim all year" instead of one.

Getting back to Fiedler's thread:

Here’s the math. Suppose enrollees have a 14.6% chance of generating a claim each month. I’ve chosen this percentage so that if a person was enrolled for 12 months, they’d have a 15% (= [1-0.146]^12) chance of having no claims, matching my group market estimate.

Now suppose that the exact same enrollees were enrolled for only 8 months (the FFM average). In that case, we’d expect 28% (= [1-0.146]^8]) to incur no claims.

There are lots of ways one could refine this calculation, but refinements could actually make the effect of enrollment duration differences larger. What’s clear is that duration matters *a lot*. Plus there could be other big differences between the individual & group markets.

We've already accounted for at least 80% of the supposedly "statistically impossible" 35% "phantom enrollee" figure, and this is just the low-hanging fruit.

Bottom line: it’s unsurprising that the individual market has a sizeable no-claims rate, especially given its relatively short enrollment durations. Given that, these data do not provide persuasive evidence that “phantom” enrollments are widespread.

Some have also highlighted the fact that the share of enrollees without claims rose substantially from 2020 to 2024. But that’s exactly what we’d expect in a period where enhanced subsidies attracted many (real) new enrollees into the market.

Prior research has found that subsidy expansions tend to pull in enrollees who use less care—and are thus less likely to incur claims—than those already in the market. For particularly high-quality evidence on this point, see here

In the next portion of his thread, Fiedler addresses another fascinating, if ironic, flaw in the argument being used by the Trump Regime: That insurance carriers are supposedly pocketing billions of dollars thanks to all of this alleged "fraud" via "phantom enrollees:"

There’s also another serious flaw...even if there were lots of phantom enrollees, health insurers wouldn’t “get the benefit,” and it wouldn’t impose the claimed large costs on the federal government.

In particular, the “tens of billions” cost estimate they both cite appears to be derived by assuming that the presence of phantom enrollees would result in additional federal subsidy payments and no other changes. That assumption is indefensible.

To see why, suppose there were a surge of phantom enrollees into the market. Total claims spending in the market would remain the same, but *per enrollee* claims spending would fall. Premiums would follow, and, by design, Marketplace subsidies follow premiums.

So the gov’t would pay for more enrollees but pay less for each one. On net, federal costs would likely rise only slightly, and mainly because there would be a modest cost shift from unsubsidized enrollees (whose premiums fall) to the gov’t, not because insurers got a windfall.

For the math on this, I'm going to flip Fiedler's example around:

Let's suppose you have an insurance carrier with exactly 100,000 enrollees whose policies cost an average of $750/mo apiece at full price and who receive an average of $570/mo in federal tax credits, leaving them paying an average of $180/mo apiece (the actual national averages for 2025 were $741, $563 & $178, but I'm using round numbers for illustrative purposes). Let's say that 35% of them (35,000) were indeed Phantom Enrollees.

To simplify it even further, let's assume that 100% of these enrollees remain enrolled for all 12 months, and none of the Phantom Enrollees were being double-counted.

If so, this means:

• Premiums for all 100,000 enrollees would total $750 x 12 x 100,000 = $900 million for the year
• Federal tax credits would total $570 x 12 x 100,000 = $684 million
• Net premiums for the enrollees would total $180 x 12 x 100,000 = $216 million

The first adjustment we have to make is that these are averages only--the actual full price premiums vary widely depending on the enrollee, plan, carrier etc. Again, to keep things simple, I'm going to assume that all of them are enrolled in the same policy, are the same age, and so forth.

The actual tax credits per enrollee also vary widely, however, and I can't ignore that, since the entire premise of the "phantom enrollee" theory is that they're among those who pay nothing in net premiums after tax credits are applied...because if they received an invoice for even $1/month and failed to pay it, they'd eventually be kicked off of their policies.

In other words, in this example, all 35,000 of these (alleged) Phantom Enrollees are getting a full ride: Mr. & Mrs. U.S. Taxpayer are shelling out the full ~$9,000/year to insurance carriers for policies which no one is actually enrolled in!

The Trump Regime is arguing that these "phantom enrollees" mean that insurance carriers are pocketing a stunning $315 million per year in taxpayer money for enrollees who don't exist. If so, it stands to reason that terminating all 35,000 policies would save taxpayers that amount, right?

Well...no, actually.

What you have to remember is that under the Affordable Care Act's Medical Loss Ratio (MLR) rule, insurance carriers are legally required to spend at least 80% of gross premium dollars on actual medical claims (it's actually 85% for large group plans, but we're sticking with the individual market here).

If they spend less than 80% on claims (on a 3-year rolling average), they have to pay back the balance to the policyholders in the form of a rebate check. This is a real thing which I've written about many times before: Every year, up to several million ACA enrollees receive rebate checks which can range from just a few dollars to several thousand dollars! In 2025 alone, over 5 million enrollees received nearly $1.2 billion in rebates averaging $233 per enrollee!

Why is this relevant? Because it means that a minimum of $720 million of that $900 million total has to go to pay for actual medical claims, leaving just $180 million for the insurance carrier's gross (not net) margin.

Do you see where I'm going with this?

If 35,000 of the enrollees don't actually exist, that means that the 65,000 who do exist really did rack up $720 million in medical claims that year (give or take). I mean, from the insurance carrier's perspective, those 35,000 "phantoms" are the equivalent of a healthy 20-yr old college kid who never does anything dangerous and who blows off his mother whenever she nags him to "at least go in for a checkup!"

How does removing them from the risk pool change the math? Well, before, you had $720M / 100,000 = average medical claims of ~$7,200 apiece. Now you have the same $720M divided across just 65,000 people...or average medical claims of over $11,000 apiece!

The risk pool has instantly become 54% more expensive on a per-enrollee basis.

What does this mean for the insurance carrier? Well, for starters, it means that they're gonna have to jack up their premiums the following year.

Remember, this year they're only charging an average of $750/month at full price. They'll have to charge at least $923/mo next year...and that's just to cover the actual medical claims (note that in this example I'm not even considering inflation or other factors). In order to make a 10% gross margin, they'll have to charge $1,026/mo on average, and in order to make the highest margin they're legally allowed to they'll have to charge each enrollee $1,154/mo on average...again, a 54% rate hike.

OK, so the unsubsidized enrollees are screwed, but what about the taxpayers? Certainly they'd save money by doing this, right?

Well, no...because the federal subsidies would also have to increase by a similar amount in order to keep the net premiums for subsidies enrollees at roughly the same thresholds they're at this year.

Remember, ACA subsidies are structured on a sliding scale based on your household income. Even without the enhanced subsidies, the standard ACA subsidy formula states that if your household income is up to 150% of the federal poverty level, you only have to pay around 4.2% of it in premiums for the "benchmark" Silver plan; if it's ~200% FPL you only have to pay around 6.6% in premiums, and so on.

Under this formula, a single adult who earns $30,000/year (192% FPL) only has to pay $155/month for the benchmark plan regardless of how much that plan costs at full price.

If it costs $750/mo this year, they receive $595/mo in tax credits

If the premium jumps to $1,154/mo next year, they'll still only have to pay (roughly) $155/mo...which means they'll receive around $999/mo in subsidies.

The exact dollar amounts and percentages would vary a bit for a variety of reasons, but the point is that from a strict federal budgetary POV, the federal government would likely end up paying--and the insurance carriers would receive, in aggregate--roughly the same amount in tax credits as it does now...just spread out across fewer enrollees.

Now, does this mean that actual cases of fraud should be shrugged off? Not at all: If any of these "zero-claim" enrollees really do turn out to be nonexistent, then yes, their policies should indeed be terminated and those involved in the fraudulent enrollment should absolute be held accountable.

For one thing, the reason why this type of fraud exists in the first place is because of unscrupulous insurance brokers who do it in order to pocket the commission. That really is fraud which really does cost both taxpayers and the insurance carriers real money, since brokers can typically earn around $15 - $20 per member per month...which is paid by the insurance carrier, which of course bakes that cost into their premiums as well.

If you assume $18 PMPM on average, that amounts to around $216/year per enrollee, so in my hypothetical example, giving all 35,000 "phantoms" the boot would theoretically save the insurance carrier (and, thus, the U.S. federal government) around...$7.6 million, or roughly 1.1% of the total federal tax credits.

And again, remember that all of this assumes that all 35% really don't exist to begin with, which, as Fiedler explains above, is extremely unlikely. My personal guess is that actual "phantom" enrollees only make up perhaps a tenth of this estimate at most, although the other fraud problem (brokers signing real people up without their knowledge or permission) may very well make up a decent number of cases as well.

THAT type of fraud (by a subsection of insurance brokers/agents, NOT by any actual enrollees themselves) really does cause real harm to people, in two ways:

• First, if someone has been enrolled in a $0-premium plan for months without knowing it and their actual income turns out to be higher than the income projected by the asshole broker who illegally enrolled them, the enrollee may be on the hook to pay a chunk (or even all) of the subsidies back when they file their taxes the following year even though they never touched the insurance they didn't even know they had.

• Second, in some cases brokers were apparently switching enrollees from one insurance carrier to another without their knowledge/permission. This can be absolutely devastating to the enrollee when they try filing a claim with the carrier they're supposed to be enrolled with only to discover that they're no longer enrolled in that policy...and that the doctor/hospital/etc that they're filing the claim with isn't in network with the carrier they've been illegally switched to.

BOTH of THESE scenarios, to the extent that they've actually been happening absolutely are serious fraud and the brokers/agents responsible should not only lose their certifications, but should be prosecuted if possible. At a bare minimum they should be required to be the ones to either repay any subsidies owed or the uninsured medical bills for their victims.

In short, I'm all for cracking down on actual fraud, but "1.1% in savings at the outside assuming 100% of all of those accused really are guilty" is a far cry from the 35% savings implied by this talking point.