"Not losing, choosing"? - Avik Roy turns up the gaslight on the BCRA

Two weeks ago, I took on the argument of various BCRA proponents that most of the 22 million people forecast by CBO to lose insurance coverage should the bill become law would be "choosing, not losing." That write-off of legions of uninsured was founded on CBO's assertion that most of the first-year coverage loss of 15 million would occur "primarily because the penalty for not having insurance would be eliminated."  Expect to hear more of this, I forecast.

In brief, I suggested that argument depended on  1) conflating the forecast immediate effect of mandate repeal in 2018 with the ten-year effect; 2) ignoring the extent to which, in CBO's own telling, the mandate interacts with other changes in the marketplace and Medicaid; and 3) pooh-poohing the obvious and stated purpose of the mandate, which is to boost the ranks of the insured and reduce premiums by improving the risk pool -- which CBO clearly assumes it has accomplished in some measure.

Now here cometh Avik Roy, chief healthcare apologist for the Republican establishment, making the "choosing not losing" argument in detail.  Roy's addition to the argument hinges mainly on a rather breathlessly presented look at CBO's unpublished estimate of 10-year coverage losses attributable to mandate repeal under the BCRA (provided to Roy by a Congressional staffer). This estimate closely tracks CBO's published December 2016 analysis of the effects of repeal of the mandate alone, with the ACA left otherwise intact. CBO forecast that straight mandate repeal under the ACA would result in a ten-year coverage loss of 15 million, as compared to CBO's forecast that 22 million will lose coverage under the BCRA. Here's CBO's breakout of the losses under straight mandate repeal:

Under current law, about 28 million people under age 65 in the United States would be uninsured in 2026. This option [repealing the individual mandate] would change that number as follows: About 2 million fewer people would have employment-based coverage, about 6 million fewer people would obtain nongroup policies (insurance people can purchase directly either in the marketplaces or from insurers outside the marketplaces), and about 7 million fewer people would have coverage under Medicaid. All together, the agencies estimate, 43 million people would be uninsured in 2026. 

According to Roy, CBO attributes 16 million of the BCRA's coverage losses to repeal of the individual mandate -- a slight bump-up from 15 million under the ACA). He then explains away the rest CBO's estimate of total coverage losses under the BCRA as the result of CBO using its 2016 baseline forecast of individual market enrollment rather than its 2017 estimate, which has 5 million fewer enrollees. This maneuver, deployed by others, was debunked three weeks ago by the Center on Budget and Policy Priority's Aviva Aron-Dine:

As noted, the 2017 baseline shows 5 million fewer people with individual market coverage in 2026 than the 2016 baseline does. But, overall, CBO estimates that 28 percent of people with individual market coverage would lose it under the Senate bill. So even if this were the only difference between the two baselines, CBO’s coverage loss estimate would likely change by at most 1 to 2 million people under the 2017 baseline

Aron-Dine further points out that if CBO used a 2017 baseline (which it eschewed for good reason, in consultation with the House and Senate budget committees), its estimate would have included 2027, which would increase estimated Medicaid coverage losses.

What's left of Roy's argument boils down to this: the only differences in coverage losses between the BCRA and the ACA-minus-mandate are those attributable to repeal of the Medicaid expansion. CBO forecasts that 15 million fewer people will be enrolled in Medicaid under the BCRA by 2026 than under current law. The December report forecast that mandate repeal alone (with the rest of the ACA intact) would reduce Medicaid enrollment by 7 million in the same period. That difference of eight million is apparently further narrowed to six million in the unpublished BCRA estimate that Roy recounts.

Roy's bad-apples-to-bad-apples comparison (ACA-minus-mandate vs. BCRA) ignores several key points:

1) The ACA is designed to be unworkable without the individual mandate. Mandate repeal triggers higher premiums, CBO forecasts, and degrades the risk pool -- until further changes stabilize it  at a lower level by driving out poorer and older enrollees en masse. If all the BCRA's tortured restructuring of the market yields no higher enrollment than the ACA minus the mandate, that's nothing to cheer about -- especially with the ongoing disaster triggered by the BCRA's Medicaid cuts piled on.

The relevant comparison would be between the BCRA and the ACA, with the ACA marketplace bolstered by the stability funding Republicans have showered on the BCRA market. If you want to envision genuine compromise, add in a mandate substitute that CBO deems viable. Or craft a compromise that more thoroughly balances progressive and conservative priorities.

2) The mandate, as CBO is at pains to specify, interacts with other features of the insurance markets. In 2018, its repeal coincides with a forecast 15-20% rise in individual market premiums. Two years later, when the BCRA subsidy schedule kicks in, the lack of a mandate would work in tandem with a massive rise in cost-sharing, as the ACA's Cost Sharing Reduction subsidies end and the actuarial value of benchmark plans drops from 70% to 58%.  By 2026, when Medicaid losses are forecast to reach 15 million, it coincides not only with the effective end of the ACA's expansion of eligibility but with the BCRA provisions designed to make Medicaid enrollment and re-enrollment more difficult, exacerbating the effects of the perpetual high "churn" in Medicaid, which stems from income and employment instability in low income people.  To the extent that CBO overestimates mandate effect, it probably underestimates the impact of these other factors.

3) Regardless of what baseline it employs, CBO forecasts a reduction of 15 million insured by Medicaid while discrediting the BCRA's bid to provide those who lose Medicaid with access to coverage in the individual market. The BCRA begins subsidy eligibility in the individual market at zero income, as opposed to at an income of 100% of 138% of the Federal Poverty Level in the ACA*.  Roy enthuses that while the BCRA repeals the Medicaid expansion, it "replaces it with a robust system of tax credits and block grants that ensure that every single person enrolled in that Medicaid expansion will get financial assistance to afford private coverage."  CBO calculates that the BCRA's "robust" credits will result in plans with $13,000 deductibles -- and forecasts:

Because a deductible of $13,000 would be a large share of their income, many people with low income would not purchase any plan even if it had very low premiums—on net, after accounting for premium tax credits—CBO and JCT estimate. Under this legislation, in 2026, that deductible would exceed the annual income of $11,400 for someone with income at 75 percent of the FPL. For people whose income was at 175 percent of the FPL ($26,500) and 375 percent of the FPL ($56,800), the deductible would constitute about a half and a quarter of their income, respectively.

4) In CBO's forecast, as the examples above illustrate, the BCRA not only reduces overall individual market enrollment -- it pushes out low income enrollees in favor of wealthier ones better able to cope with high deductibles and effectively pair them with Health Savings Accounts, which the BCRA loads with new tax favors. The BCRA would also exacerbate underinsurance as well as uninsurance. At present, about 5 million ACA marketplace enrollees are in plans with strong Cost Sharing Reduction subsidies that raise the plans' actuarial value to 94% or 87% -- i.e., plans with deductibles averaging $255 (at AV 94%) or $809 (at AV 87%). Under the BCRA, enrollees at these income levels will face 5-digit deductibles. Ditto for the few former or would-have-been Medicaid enrollees who take advantage of premium subsidies to buy AV 58% coverage (perhaps mildly ameliorated by one of the byzantine layers of stability funding piled on to the BCRA, one of which states can use to reduce out-of-pocket costs). Some of those who lose Medicaid eligibility will also find their way into employer-sponsored insurance, which has much higher cost-sharing than Medicaid -- particularly in plans often offered to lower income workers.

5. The BCRA's harshest per capita caps on federal Medicaid spending kick in in 2025. Coverage losses will continue after 2026 and likely spread to all categories of Medicaid enrollment, as federal Medicaid spending is cut by $2.6 trillion compared to current law in 2027-36, as estimated by CBPP's Edwin Park.

In short, discounting coverage losses that CBO does not attribute in some fashion to repeal of the individual mandate is a useless mental exercise -- as is comparing the effects of the BCRA to a disfigured (or rather sabotaged) ACA, which it still comes out far behind. There used to be consensus among healthcare experts that if the U.S. wanted to progress toward universal coverage in large part by means of a reformed individual market, a mechanism would be required to pull people into that market. Those who have decided that the individual mandate is an unacceptable impingement on personal freedom need to put forward a substitute that works within the context of whatever other reforms they devise. The CBO gave the BCRA a failing grade on this front, as on every other.

----

* As originally enacted, the ACA extended Medicaid eligibility to adults with incomes under 138% FPL. In 2012, the Supreme Court made that eligibility expansion optional for states; at present, 19 states have refused to enact it. In those states, subsidy eligibility in the private plan marketplace begins at 100% FPL; those below that level are left out in the cold.