Kicked off insurance? Fact-checking the Fact Checker

To what degree do GOP health care reform proposals kick people off health insurance?

Is there a Democrat who has failed to repeat the talking point that one or another such bill will kick some millions off their insurance? The claim appears ubiquitous, yet fact checkers like the Washington Post Fact Checker and PolitiFact have pretty much ignored the deceptive nature of the claim. But Avik Roy (Forbes) and Ramesh Ponnuru (National Review) did some admirable work in exposing the depth of the falsehood: People forgoing insurance because they will not get penalized for lacking insurance are not rightly described as getting kicked off of insurance. And the number of those is quite high.

Roy and Ponnuru noted that a major factor in the Congressional Budget Office’s projections for loss of insurance was the repeal of the individual mandate. The individual mandate applies a tax penalty to people who do not purchase insurance if a formula shows they capable of affording it, generally speaking. Roy and Ponnuru also found that the CBO associates repeal of the individual mandate with lower Medicaid enrollment. The mainstream fact checkers have failed to emphasize this fact.

Ponnuru later elaborated on his key point in a critique of PolitiFact:

In late July, the CBO estimated the effects of a “skinny bill” that made no reforms to Medicaid but got rid of the fines for people without insurance. CBO found that the bill would cause 7 million fewer people to enroll in Medicaid in 2026.


As Ponnuru also noted, this means that the CBO thinks repeal of the individual mandate keeps people from signing up for a health insurance program that is, for them, effectively free.

Ponnuru gave PolitiFact a blistering critique. But what of the Washington Post Fact Checker?


Michelle Ye He Lee

Glenn Kessler

The Washington Post Fact Checker

We called out the Washington Post’s Glenn Kessler on Twitter, resulting in some changes to two stories. The stories carry the byline of Michelle Ye He Lee. The texts of the two stories vary little from each other on the issue of people kicked off their health insurance. One was published on March 17, 2017 and the other days later on March 20, 2017.

The original version of Lee’s March 17 story gave a very ambiguous estimate of the number of persons getting kicked off their insurance (bold emphasis added):

But this does not mean all of the 14 million or 24 million will be “thrown off” health insurance or “lose” health insurance.

Some of the people who would be uninsured would choose not to have insurance, because they had decided to obtain health insurance only to avoid a penalty under the ACA’s individual mandate; the replacement bill eliminates the mandate. Others, such as elderly Americans, would not get insurance because the premiums are too high. (The replacement bill would allow the elderly to be charged five times as much as the youngest insured, compared with a 3:1 ratio under the ACA.) Many of the uninsured people would lose insurance because of reductions in Medicaid enrollment — after some states discontinue the Medicaid expansion under Obamacare.


Lee wrote that “not all” of the 14 million or 24 million (short and long-term estimates) would be kicked off their health insurance. “Some” would decide not to buy insurance because the bill would end the individual mandate penalty.

But how many is “some”? Fifteen? Three hundred? Five thousand? Seven million?

Using “not all” and “some” gave readers virtually nothing on which to judge the degree of deception. And the Fact Checker awarded no Pinocchios, its symbol for falsehoods, for the claim. A mildly deceptive claim might receive one Pinocchio from the Fact Checker. A highly deceptive claim might receive as many as four Pinocchios.

The “kicked off” insurance claims were not rated. Neither received any Pinocchios.

Updated Versions

As mentioned earlier, the Fact Checker updated the wording of its stories following our Twitter exchange.

But this does not mean all of the 14 million or 24 million will be “thrown off” health insurance or “lose” health insurance.
Many of the people who would be uninsured, at least initially, would choose not to have insurance, because they had decided to obtain health insurance only to avoid a penalty under the ACA’s individual mandate; the replacement bill eliminates the mandate.


“Many” at least counts as an improvement over “some,” but the new phrasing suggests, by implication, that “many” quickly drops to something less than “many” over the course of the 10-year projection. And, in fact, the new phrasing creates a jarring contrast with the Fact Checker’s subsequent quotation of the CBO (this occurring only in the March 17, 2017 version of the story):

“Most of the reductions in coverage in 2018 and 2019 would stem from repealing the penalties associated with the individual mandate,” the CBO found.


If the Fact Checker agreed that most of the reductions in coverage for 2018 and 2019 come from the repeal of the individual mandate, then why not say “Most of the people who would be uninsured, at least initially, would choose not to have insurance …”? The Fact Checker’s use of “many” instead of “most” loses much of the meaning of the CBO’s statement and creates an embarrassing internal discrepancy in the fact check.

CBO’s Insurance Loss Shell Game

We were bothered by another aspect of the Fact Checker’s fact check. It failed to properly describe the dynamics of insurance loss:

The CBO estimated that the GOP bill would lead to 14 million fewer people insured than under Obamacare by 2018. Six million of 14 million would be people who now have coverage in the individual insurance market; 5 million would be people with coverage under Medicaid; and 2 million would be people with coverage through their employers, who also would no longer be required to provide insurance. (The remainder come from other insurance shifts.)

Lee’s description fails to capture the fact that a 6 million drop in the number of people carrying nongroup insurance does not necessarily mean that those people lost insurance. They might move to the categories of people having either Medicaid or group insurance. Claiming that the losses come from “people who now have coverage in the individual insurance market,” for example, shows that Lee fails to account for people moving from one category of insurance to another. Perhaps that aspect of the CBO report contributed to fact checkers overlooking the role of the individual mandate in lowering Medicaid enrollment.

Are long-term insurance losses mostly driven by repeal of the individual mandate? We look at that question next.

Could the Individual Mandate Account for Most of the 24 Million?

We see a two-pronged argument for saying the individual mandate counts for less than half of long-term losses to the ranks of the insured. First, the CBO does not emphasize it. Second, when the CBO said the short term losses were mostly because of the mandate, the quotation of the CBO from the Fact Checker we noted above, it specified only the first 2 years of the projection.

Why would the CBO use that phrasing if the individual mandate was the main driver of insurance loss over the entire course of the estimate?

On the other hand, why didn’t the CBO reveal the factor that supplanted the individual mandate as the main driver behind insurance losses? And what drives insurance losses in 2017 under the new law if not the repeal of the individual mandate?

We admit it is possible the CBO was saying the individual mandate only drives insurance losses in years two and three of the projection.. However, the CBO’s phrasing remains logically compatible with the idea of the mandate driving insurance losses over the entire 10 years.

  1. The repeal of the individual mandate drives insurance losses for 2 years of the projection.
  2. The repeal of the individual mandate drives insurance losses for the last 7 years of the projection.

Taken individually, each sentences vaguely implies that the other sentence is not true. At the same time, the sentences do not contradict each other. Taken together, they mean the repeal of the individual mandate drives losses for 9 years of the projection. Note that we are not saying the CBO states No. 2. We are pointing out that the first statement does not rule out the second in any strict sense.

We conclude the argument for only near-term dominance of the individual mandate effect is not a bad argument, but that it is also a relatively weak argument. The CBO used equivocal language.

Now let us examine the contrary argument.

Ponnuru’s Skinny Repeal Argument

When Ramesh Ponnuru wrote his rebuttal of PolitiFact for National Review, he supplemented his original argument with observations about the CBO’s scoring of the GOP’s “skinny repeal” bill, H.R. 1628.

That bill made two main changes affecting the rate of insurance: repeal of the individual and employer mandates. The bill did not change Medicaid law. The CBO’s scoring of that bill helps us winnow out the CBO’s view of the individual mandate’s effect on insurance, particularly Medicaid. The CBO’s Table 3 from that report tells the story:

Starting in 2018, repeal of the mandates results in 3 million fewer Medicaid enrollees. That number climbs to 7 million by 2023 and stays there through 2026.

Though the CBO explained its view on why repealing the individual mandate affects Medicaid enrollment, none of the “elite three” fact checkers, including, cover this critical aspect of the report.

CBO, on the American Health Care Act (bold emphasis added):

Under current law, the penalties associated with the individual mandate apply to some Medicaid-eligible adults and children. (For example, the penalties apply to single individuals with income above about 90 percent of the federal poverty guidelines, also known as the federal poverty level, or FPL). CBO estimates that, without those penalties, fewer people would enroll in Medicaid, including some who are not subject to the penalties but might think they are. Some people might be uncertain about what circumstances trigger the penalty and others might be uncertain about their annual income. The estimated lower enrollment would result in less spending for the program. Those effects on enrollment and spending would continue throughout the 2017-2026 period.


People may naturally assume that lower numbers covered under Medicaid must come directly from lower funding for Medicaid under the GOP repeal proposals. But much of the lower funding stems from lower enrollment tied to the repeal of the individual mandate. That unexpected fact counts as exactly the type of thing fact checkers should explain.

As we pointed out to Kessler on Twitter, the CBO’s assessment of “skinny repeal” suggests it credits repeal of the individual mandate with 9 million fewer insured in 2018. For 2026, that number rises to 13 million.  Nine million would count as “most” of the 14 million considered in Lee’s fact check. Likewise, 13 million would count as “most” of 24 million expected to lack insurance in 2026.

One Possible Objection

The numbers implied by the CBO’s assessment of “skinny repeal” do not pop out in its other assessments. Note Table 5 from the CBO’s assessment of the AHCA.

One might expect the effects of the individual mandate repeal to fall on the nongroup market. But for 2026 the CBO shows only 2 million fewer insured through the nongroup market. Add that figure to the 7 million loss we’re figuring from the CBO’s report on the skinny bill and we have a mere 9 million–less than half of 24 million projected increase in uninsured. However, the drop in the number of insured for the nongroup market likely has some explanation in the higher number moving out of the employment-based market (7 million). Some portion of that group will likely buy insurance in the nongroup market instead, accounting for the smaller change in insured for the nongroup market. And some individuals will choose not to accept an insurance offer from an employer without the threat of penalty, as the CBO also acknowledges.

Unless we find a significant flaw in this argument, it outweighs the semantic argument discussed above. We need some reason why the drops in various types of insurance coverage from repealing the individual mandate are different for the AHCA than they are for the “skinny repeal” bill. Perhaps changes to age-rating or subsidies could account for a lower relative effect from the repeal of the mandates? We do not see how that would work, and the CBO offers no explanation.


The Washington Post Fact Checker’s updated version of its “kicked off” insurance fact check still suffers from at least four problems.

  1. It downplays the fact that the CBO projection states that “most” of the projected drops in insurance coverage for 2018 and 2019 stem from repeal of the individual mandate, calling the number “many” instead of “most.”  That in spite of later quoting the CBO using “most.”
  2. It falsely implies that the individual mandate’s influence on removing “many” from insurance rolls is short-lived.
  3. It neglects to explain the CBO’s contention that the repeal of the individual mandate results in substantially lower Medicaid enrollment.
  4. It incorrectly states that insurance losses in a given category mean that number of people from that category lost insurance.

We would add that the lack of Pinocchios awarded to the “kicked off” insurance claim also supports our charge that the Fact Checker largely ignored the deception of the claim.


We tried in vain to find evidence that Kessler followed the Washington Post’s guidelines for performing a correction or clarification on a published article. The Fact Checker’s verified status as a signatory to the International Fact Checking Network’s statement of principles supposedly relies on compliance with that statement of principles.

A Note of Appreciation for Glenn Kessler

Whatever criticisms we might have of Glenn Kessler’s work, we stand in sincere admiration of his willingness to defend it with argument. Kessler has few peers in this among the mainstream fact checkers, and we count it as one of the best traits a fact checker can possess.

Leave a Comment

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.