If you’ve been following the health care debate for the past several months you’ve seen the estimates put out by the Congressional Budget Office. All of those estimates have shown millions fewer people would have insurance in 2026 under the various GOP plans. In fact, different approaches to the bill haven’t seemed to make much of a difference in the CBO’s projections. From National Review:

Do you want to repeal every word of Obamacare and replace it with nothing? CBO says 22 million fewer people would have health insurance. Do you prefer replacing Obamacare with a system of flat tax credits, in which you get the same amount of assistance regardless of your financial need? CBO says 23 million fewer people would have health insurance. Do you prefer replacing Obamacare with means-tested tax credits, like the Senate bill does, in which the majority of the assistance is directed to those near or below the poverty line? CBO says 22 million fewer people would have health insurance.

Democrats have pointed to these figures as proof the GOP bills will be a disaster. But why are all the CBO estimates so similar when the underlying reforms are so different. Why does repeal with no replacement have almost the same impact as the Senate repeal and replace bill? It turns out CBO has mostly been looking at just one factor: The individual mandate. From Forbes:

This week, I obtained from a congressional staffer the CBO’s estimates of the coverage impact of repealing the individual mandate, separate from the Senate bill’s other provisions. The estimate was built out of earlier work CBO did to model how repealing the mandate would affect the federal deficit. CBO projected then that repealing the mandate alone would lead to 15 million fewer insured U.S. residents in 2018, and 16 million fewer by 2026, though they did not publish those estimates.

16 million represents nearly three-fourths of the CBO’s estimate of the coverage difference between the GOP bills and Obamacare in 2026.

Much of the remaining difference has to do with the CBO baseline which vastly overestimates the number of insured under Obamacare:

Even if one excludes the CBO’s exaggerated view of the impact of the individual mandate, CBO scores the Senate bill as covering 6 million fewer people than Obamacare in 2026: 2 percent of the U.S. population. But even that number can be partially explained by CBO’s outdated March 2016 baseline, which assumes that enrollment in Obamacare’s exchanges peaks out at 19 million, when it’s more likely to end up below 9 million, if Obamacare stays on the books and premiums continue to rise.

Roy offers this chart showing how the baseline and the individual mandate account for nearly all of the losses projected by CBO:

So any GOP proposal which removes the individual mandate was always going to be scored by CBO as dropping at least 20 million people. As Roy puts it, the impact of all other changes “amounts to statistical noise.” Roy posted a tweet storm about the story this morning. Here’s a portion of it:

(6/X) Claiming GOP bill will "kick 22 million off their coverage" is either lying or ignorant. Voluntary decision for most according to CBO.

— Avik Roy (@Avik) July 23, 2017

(8/X) "BCRA = -22MM covered because of fewer deductible subsidies" doesn't explain why full repeal w/o replace leads to same -22MM result. pic.twitter.com/wv0j5cw6s8

— Avik Roy (@Avik) July 23, 2017

(10/X) CBO is obviously wrong about that. Hence GOP should focus on good policy and not defer to CBO on coverage estimates.

— Avik Roy (@Avik) July 23, 2017