What do you get for the same premium in the BCRA

In Health Affairs, Joe Antos and James Capretta wrote in support summarizing of the Senate BCRA. I want to reply to a single point that I think is wrong and expensive. It is a smarter version of Avik Roy’s argument that Medicaid is not being eliminated as poor people will be able to buy subsidized private plans with a baseline actuarial value of 58%.

Very low-income households receiving this credit could enroll in health insurance while paying minimal premiums themselves. A person with income at the 2017 federal poverty level would pay a maximum of 2 percent of their income, or $23.10 a month for coverage….

As federal support for Medicaid is lowered in future years, more people are likely to become eligible for this tax credit, which, unlike Medicaid, is financed entirely by the federal government.

There are two things to untangle in these statements. The first is whether or not 58% AV insurance is a near substitute for Medicaid. The second is whether or not this is cost effective.

The Congressional Budget Office does not think a 58% actuarial value plan is an attractive offering for people who are currently on Medicaid or who would be eligible for Medicaid Expansion:

The deductible for a plan with an actuarial value of 58 percent would be a significantly higher percentage of income—also making such a plan unattractive, but for a different reason. As a result, despite being eligible for premium tax credits, few low-income people would purchase any plan, CBO and JCT estimate.

The cost effectiveness is the part that has me scratching my head. 58% AV insurance would presumably be paying providers at or above Medicare rates. More likely in most regions, the policies that would be sold on the Exchanges would be paying providers standard commercial rates. This is a major cost effectiveness issue.

The Kaiser Family Foundation estimates that Medicaid pays providers on average 66% of what Medicare Fee for Service (FFS) pays. As we saw yesterday, the Congressional Budget Office has evidence that commercial group insurance from national carriers pay roughly 50% more than Medicare FFS for the same procedure code.

Let’s work through two scenarios that apply to Antos and Capretta’s argued high point that poor people can buy subsidized private plans. We will assume that some of the private plans will be paying their providers 120% of Medicare FFS on average. These plans will be narrow network plans with significant gatekeepers. The second iteration assumes the plans will pay 150% of Medicare FFS. These plans will be broader networks with fewer gatekeepers.

The federal government will be paying a lot for little actuarial value under both of these scenarios.

The first analysis (AV Raw) is a simple application of algebra. It produces a raw AV increase of 26% and 9% for Medicaid and Medicare FFS pricing benchmarks. This is not quite right. We know that on average, lower out of pocket maximums will lead to more utilization. The ACA uses induced demand factors in their risk adjustment formula to normalize induced demand to create apple to apple comparisons.

Using the induced demand factors of the ACA risk adjustment formula, these estimates are reduced. Medicaid provider reimbursement plans reduce to the equivilent of 78% Gold plans under both the low commercial rate and the high commercial rate scenarios. This is because the Platinum induced demand factor is 1.15 which applies to the raw 90% AV plan under the high commercial rate scenario while Gold has an induced demand factor of 1.08 applying to the 84% AV plan under the low commercial rate scenario.

This induced demand adjusted actuarial value has Medicaid offering a 78% AV plan under either assumption. That is a net 20 AV point bump in value for the same total cost. Medicare would be able to offer either a 65.7% or 71% AV plan for the same total premium cost.

Private insurance is expensive on a per unit basis and thus it is expensive on a per actuarial value point basis.

There is either an ideological justification for private insurance instead of spending money on Medicaid administered through private managed care plans or an accounting justification. If we are to believe the CBO that very few people who earn under 100% FPL will spend 2% of their income to buy a plan with a base AV of 58% which means a deductible of more than half their income, the federal government will not be spending significant sums paying 90% or more of the premium for these individuals who are now uncovered. However if the insurance is made attractive at the same per capita cost by increasing the actuarial value and lowering out of pocket exposure by paying providers less, more people will actually buy the coverage and this will increase federal spending.

Jed Graham at IBD noted this dynamic in the House bill as he wrote on the original CBO score for the AHCA on March 13, 2017:

Because the GOP bill would mostly retain ObamaCare coverage rules, insurance would be unaffordable for lower-income and older adults with the new, smaller tax credits on offer, so some 30 million people wouldn’t claim the GOP tax credit averaging $3,000 in 2020 and rising with inflation. That would add up to more than $600 billion in unclaimed subsidies through 2026, or roughly the same $600 billion amount by which House Speaker Paul Ryan’s plan cuts taxes. Those unspent subsidies go a long way to explaining why CBO found that the American Health Care Act would reduce deficits by $323 billion over a decade.

Offering attractive coverage decreases the number of people who are uncovered but it costs the federal government money. Having private insurers offer low actuarial value plans at the same per capita cost as what much higher actuarial value plans would cost the federal government decreases enrollment all else being equal.