The Congressional Budget Office (CBO) had to deal with two separate worlds when it analyzed the American Health Care Act (AHCA), the version of "repeal and replace" put forward by two House committees. While the CBO exists to provide numbers that relate to spending, deficits, and the economy, a different number got the most attention: the assessment that by 2026 24 million fewer people would have health insurance.
That is not the same thing as saying 24 million people will lose health insurance. Rather, it is a statement that, compared to how many people CBO expects to have health insurance in 2026 under current law (the Affordable Care Act, also known as the ACA or Obamacare), 24 million fewer people will have health insurance. The largest contribution to that projection is 14 million fewer people enrolled in Medicaid, the federal-state program that pays for health care for low-income people.
Ordinary English and budgetary baselines diverge over the meaning of the word "lose." In ordinary English, you must have something before you can lose it. In the world of comparisons to a budgetary baseline, you can lose something you don’t have.
The rules of congressional budgeting look out over the next ten years. CBO projects how things will play out if nothing happens: no new laws get passed, nothing bothers the economy, and people keep on behaving as they have in the past. Any policy change, such as the AHCA, gets measured against that current law baseline. Here, CBO’s assumptions about what state governments would do under current law and what they would do under the AHCA become crucial.
Back when the ACA became law, there was no requirement for assumptions. Every state would have to expand Medicaid eligibility. The federal government would set the line dividing who had income that was too high and who had income that was low enough to qualify. Then the Supreme Court spoke. States could not be forced to expand eligibility; they suddenly had flexibility.
What would they do with it? CBO had to assume something. We know that 31 states (and the District of Columbia) went for some form of expansion. Altogether, 50 percent of the population lives in these states. CBO assumed that by the year 2026 more states would expand Medicaid eligibility — enough that 80 percent of the population would live in states that chose to expand.
Some share — CBO doesn’t say exactly how many — of the 14 million fewer people who have Medicaid won’t lose it. Rather, they are people who would have gotten Medicaid at some point between now and 2026 and now won’t. But because they would have become eligible under CBO’s current law baseline, they are among the 14 million fewer Medicaid enrollees in 2026.
Another CBO assumption says that states that expanded Medicaid eligibility in response to the ACA will now reduce it. CBO assumes the share of the population in states that stick with expanded eligibility will decline from 50 to 30 percent of the population.
While the CBO bases its nuanced model of private insurance on academic studies of how people behave when prices change, this assumption about how political actors will behave is ad hoc. The extant academic literature predicts something else. Public choice economics literature of the "ratchet effect" variety claims that politicians aren’t likely to take things away from voters. If nothing else, CBO assumptions are based on "out of sample" projections: taking relationships found over some range and then using them to project what will happen in a different range of numbers.
These assumptions about how states will behave provide the basis for much of the 14 million fewer people who will be enrolled in Medicaid, the single largest contributor to the "24 million fewer" number that has dominated coverage of the CBO report.
Much of the CBO scoring of the American Health Care Act produced by two House committees has the backing of years of studies and analysis. But some of it does not. This must be kept in mind especially when evaluating grave predictions about millions of Americans losing Medicaid coverage.