This is a contribution to Prospect Debate: The Cost of Sanders's Single-Payer Health Plan.
Based on Gerald Friedman’s work, Bernie Sanders’s campaign has published implausibly low estimates of the additional federal expenditures needed to fund the senator’s single-payer health plan. In what follows, I explain the differences in methods that Friedman and I have used and the sources of the gap between his estimates and mine. I have updated my analysis based on more recent information available in the Friedman estimates.
The Differences between Friedman’s Method and Mine
In a two-page, January 17 memo to the Sanders campaign, Friedman uses a simple method in estimating the cost of the plan. Single-payer covered services would have no cost-sharing, but some services not typically provided through insurance (cosmetic and LASIK surgery) would not be covered and would be outside the single-payer system and paid out of pocket. He estimates the cost of new single-payer coverage under the plan by using Centers for Medicare and Medicaid Services projections of national health expenditures over the period of 2017 to 2026. From that, he subtracts estimated net savings. He then adds in the cost to the federal government of assuming the cost of Medicare Part B premiums. From that total, he subtracts projected public spending under current law (assuming that state as well as federal spending rises at currently projected rates) and subtracts continued out-of-pocket spending not included in the single-payer plan. So the estimated new public funding is a residual of these tabulations. The net result is his estimate of new federal spending—$13,773 trillion over ten years or an average of $1.377 trillion a year. (See Table 1.)
My approach to estimating the costs of Sanders’s plan is altogether different and reflects long-used and well-recognized methods among health-care economists. I consider the costs associated with the move to a single payer for each of the major populations of health-care patients—Medicare beneficiaries, Medicaid beneficiaries, the privately insured, and the uninsured. Using standard techniques, I then estimate the effect of the Sanders plan on the costs associated with each population. (See Table 2.)
Sources of the Gap in Estimates
There are three major factors that result in Friedman’s underestimate of new federal revenues needed to pay for the Sanders single-payer plan.
Ø Friedman underestimates the additional spending that would result from the elimination of cost-sharing on single-payer covered benefits.
Ø He overestimates the amount of government spending under current law that can be redirected to pay for the new single-payer plan.
Ø Finally, Friedman’s savings assumptions are not only implausible, but in some cases could eliminate entire sectors of our health-care system.
Underestimates of the New Spending Associated with Single-Payer
Friedman underestimates the change in national health-care spending through the virtual elimination of cost-sharing. He assumes only a 3 percent increase in spending with the elimination of cost-sharing. The increase would be much higher. The privately insured alone would face a 22 percent reduction in out-of-pocket spending, resulting in a much greater increase in spending. (See Table 3.)
Moving to a single-payer system that covers all health-care spending would substantially increase total spending among Medicare patients and the uninsured as well as the privately insured. These dramatic reductions in cost-sharing change individual behavior resulting in higher spending on health care (this is known as “induced utilization”).
As part of the Affordable Care Act, the Department of Health and Human Services released induced-utilization factors for states to use to estimate the increase in health-care spending (for the cost-sharing reductions) among the currently insured population. These are presented in Table 3. So for those with insurance today, for example, moving from a plan that covers 70 percent of costs to the no-cost-sharing single-payer plan would increase spending by 12 percent.
Additional information on induced utilization can be derived from the total spending (premium and out of pocket) in the silver, gold, and platinum plans offered in the exchanges. Nationally, moving from a gold plan (80 percent actuarial value) to a platinum plan (90 percent actuarial value) results in a 13 percent increase in total spending. Since some of the difference is likely due to self-selection into different plans, I used a more conservative induction factor in my calculations.
In the private insurance market, the estimated average plan actuarial value was 83 percent in 2010 (that is, plans covered 83 percent of expenditures).
This percentage has likely declined since then given the precipitous rise in high-deductible health plans (plans with a single deductible of $1,300 and a family deductible of $2,600). Since 2010, enrollment in high-deductible employer-sponsored insurance plans has doubled from 13 percent to 26 percent. Tabulations from the Medical Expenditure Panel Survey (MEPS) showed an actuarial value of 80 percent for those insured nationally by private health insurance.
Spending on the previously uninsured would also increase with the single-payer plan. The literature on the incremental costs of covering the uninsured shows an average increase in spending of 70 percent. This is likely low, however, since the uninsured were assumed to receive a “typical” private insurance benefit far less generous than the single-payer plan—so the change in spending would be greater than 70 percent. (Nonetheless, my average increase for this group was 70 percent.) In contrast, Friedman estimated new spending among the uninsured receiving single-payer coverage as only approximately 45 percent more than at present. He claims this is due to the age distribution of the uninsured, but the age distribution of the uninsured mirrors that of the insured population. We also modeled the change in spending using the MEPS, to pick up both part and full year uninsured using methods similar to those outlined by Hadley and colleagues, to develop our results.
Spending among Medicare beneficiaries would also increase. Approximately 15 percent of Medicare beneficiaries have just the 80 percent actuarial value Medicare plan. Providing single-payer coverage would increase their spending. There is substantial published literature showing that Medicare patients with supplemental coverage have Medicare expenditures that are approximately 20 percent higher than those without additional coverage. The average Medicare patient pays about 15 percent of costs out of pocket, so moving to a single-payer system would result in higher spending that Friedman does not include.
Providing all seniors a comprehensive drug benefit would also increase spending among all Medicare beneficiaries. About 15 percent of Medicare patients have no drug coverage today, so their spending would increase. Moreover, the average cost-sharing in a Medicare drug benefit exceeds 25 percent and is thus dramatically higher than the single-payer plan. This increased spending is also not included in the Friedman estimate.
Another source of the difference between our estimates is that Friedman assumes a slowdown in health-care costs that is greater than the experience of single-payer countries collectively. The most recent data show that costs in the Organization for Economic Cooperation and Development (OECD) countries have been growing about 0.5 percent more slowly than in the United States. Friedman projects that under a single-payer system, costs will grow 1.1 percent more slowly in the United States than they do now.
Implausibly High Estimates of Public Sources of Funds
The biggest source of funding for the Sanders plan is public funding for existing programs—a total of $27.2 trillion between 2017 and 2026. This total includes projected federal and state costs for the existing Medicare, Medicaid, and CHIP programs ($18.2 trillion). It also includes funds for several programs, such as the veterans health-care system and Indian Health Service. So if these programs are going to continue, their funds would not be available to fund the single-payer plan.
Another issue with Friedman’s estimates is that he assumes the states will make $3.3 trillion in Medicaid and CHIP payments even after those programs have been eliminated. There is a precedent here, though it is of uncertain legal status today. When Congress passed the Medicare drug benefit in 2003, it required states to contribute 75 percent of their previous spending on “dual eligible” Part D beneficiaries (those eligible for Medicaid as well as Medicare). This kind of provision is known as a “maintenance of effort” (MOE) requirement. But in more recent decisions, the Supreme Court has greatly limited the power of the federal government to require states to spend money. In my estimates, I used the MOE approach used in the Part D dual-eligible program requiring the states to spend 75 percent as much as they previously spent. Obviously, financing costs would be substantially higher absent the MOE contributions.
There is an additional aspect to Friedman’s estimates that allows him to project lower new costs for single-payer. He estimates the cost of the new single-payer plan will grow at 4.7 percent per year, but at the same time he assumes the redirected funds for prior programs (including state Medicaid, state general assistance, vocational rehabilitation, state maternal and child health, and school health payments) to fund the single-payer plan would be required to increase at a higher baseline rate of 6.5 percent per year. In other words, the states would have to continue to pay 6.5 percent more each year for their former Medicaid and CHIP beneficiaries, even though the actual cost for those patients would increase, according to Friedman, at only 4.7 percent per year. Under these assumptions, the states subsidize an increasing share of the total cost of the federal program. So not only does Friedman assume a 100 percent MOE, but Sanders’s plan apparently forces the states to pay each year nearly 2 percentage points more than the actual cost of the single-payer program.
The same is true for some important private-sector spending included in the $27.2 trillion “public sector” total. Included in this total are the workers’ compensation programs and worksite health programs. So again the single-payer plan would require employers to contribute an average of 6.5 percent per year for these programs, even when the single-payer plan limits spending growth in these programs to 4.7 percent per year.
Implausible Savings Assumptions
Unrealistic assumptions about potential savings from single-payer account for another $5 trillion of the financing gap.
Friedman assumes that single-payer would generate an unrealistic net savings of nearly 8 percent of national health spending in the first year, rising to nearly 18 percent in ten years. A major transition of this scale and magnitude would require dramatic changes over time, and perhaps be fully implemented over a three-year period. So savings of this magnitude in year one is clearly an overreach and would not occur.
Friedman estimates two types of savings. The first are administrative expense savings traced to the elimination of private health insurance. His estimate is that this would generate a 15 percent reduction in costs. To “capture” these savings, provider payments would be cut by a similar amount. On top of these provider cuts, he proposes additional reductions in provider payments.
Consider the proposed savings from prescription drugs that Friedman counts on. Medicare and private health insurance—payers for whom “negotiations” could reduce spending—are projected to spend an average of about $350 billion in total per year on prescription drugs over the next decade year. Asked by Dylan Matthews of Vox to provide details on where the Sanders savings were coming from, the Sanders campaign (relying on Friedman) originally cited an impossibly high estimate for savings on drugs—$324 billion. Friedman then cut the savings on drugs to a level that is still implausible—$241 billion. On average over the next decade, Medicare and private insurance plans are projected to spend only about $240 billion a year on brand-name drugs (another $30 billion is spent out of pocket on brand drugs). Since generics don’t leave much room for savings, the Sanders plan was essentially estimating drug savings that would eliminate nearly all spending on brand drugs.
Another reported provider cut is also implausible—home health. Medicare is the dominant payer for home health care in the United States. A typical large provider receives about 75 percent of its income from Medicare, which is virtually single-payer today, leaving few if any opportunities for administrative savings. The electronic billing system linking home-health providers to Medicare is already quite streamlined. Across the entire industry, Medicare and Medicaid account for about 90 percent of home health insurance revenue. To achieve his spending savings, Friedman cuts home health-care payments by nearly 20 percent. The average provider margin for Medicare beneficiaries is less than 13 percent, placing the entire home health-care industry at risk. Since home health is very labor-intensive, staffed by nurses, therapists, and home health aides, the cuts would represent substantial reductions in home health-care workers’ wages and income.
In another dubious calculation, Friedman assumes $6.3 trillion in other health-care savings (administrative savings plus provider payment cuts less the costs of covering the 33 million uninsured). That $6.3 trillion amounts to a whopping 13.3 percent of total spending—an estimate that is nearly three times higher than Friedman’s own estimate of the savings for another recent single-payer bill (S. 676). Why Friedman’s estimate of savings should have tripled from one single-payer proposal to another isn’t clear. The differences between the proposals can’t account for the bigger savings he now finds.
Friedman also reduces the growth in health-care spending for services unrelated to single-payer that would affect patient treatment costs. For instance, he reduces the growth in health-care spending on government public health activities, investment in structures and equipment and noncommercial medical research activities from their projected rate of growth to the single-payer rate of growth. Single-payer would have nothing to do with these investments and research programs. This reduces his estimated federal financing needs by $110 billion over ten years.
Friedman and other supporters of Sanders have criticized my numbers on the ground that I haven’t shown as large of reductions in administrative costs as they do. But they are misinterpreting my estimates. I reduced private health spending by 20 percent, or nearly $250 billion in 2017, which is the net cost of private insurance, so as to reflect the elimination of private insurance plans. Also included in the overall assessment of the change in health-care spending nationally is an additional 4.7 percent reduction in system-wide administrative costs.
Not all of the difference in administrative costs between the United States and Canada is the cost of insurance administration that would be unnecessary under single-payer. In the United States, unlike Canada, payments for health care are now increasingly linked to measures of quality, patient outcomes, and efficiency. Most of the published administrative cost savings estimates Friedman is relying on were completed 10 to 20 years ago in a substantially different health-care system. Our administration is not simply designed to pay bills. Much of it goes into preventive and care-coordination efforts designed to attack the key drivers of rising health-care costs—chronic diseases. For instance, nearly 90 percent of the growth in Medicare spending since 1987 is due to the rise in the prevalence of treated chronic disease issues. We have developed sophisticated measures to track the quality of care and link improvements in outcomes and quality to payments. These quality initiatives are “administrative” costs well worth retaining and improving on in future health-care reforms.
To get to his $1.37 trillion estimate of the cost of the Sanders plan, Friedman makes several dubious assumptions. He assumes that the states will continue to pay for Medicare and CHIP even after the programs are eliminated, and he makes exaggerated estimates of savings from single-payer. He also dramatically underestimates the additional spending associated with the elimination of any cost-sharing. The provider savings estimates are unrealistic—in some cases, such as prescribed brand drugs and home health care, resulting in the virtual elimination of these industries.
As a result, the Sanders single-payer plan is underfunded by an average of about $1.1 trillion a year. To finance it adequately would require dramatically higher taxes than Sanders has provided for—equal to about 20 percent of workers’ compensation. Many more people would consequently end up losers under his proposal.
Next: Paul Starr, "The Larger Problems of the Sanders Single-Payer Plan"