Analysis of Chapter 3 of Economic Analysis of Medicare for All by Robert Pollin et al.
By Kip Sullivan for Health Over Profit for Everyone, January 2019
In a report released on November 30, 2018, Robert Pollin and his colleagues at the University of Massachusetts Amherst concluded that S 1804, the Medicare for All Act of 2017 introduced by Senator Bernie Sanders, would achieve universal coverage and reduce total US health care spending by 7.2 percent. The 7.2-percent figure is the net of a 12.0 percent increase in spending due to better coverage and fewer uninsured, and a 19.2 percent reduction in spending due to improved efficiency, lower prices, and reduced fraud.
The chapter in which the 19.2-percent savings estimate is explained is Chapter 3. I analyze that chapter in this paper. My conclusion is Pollin et al. have greatly exaggerated the savings S 1804 can achieve in its current form. The total gross savings is closer to 9 percent (not 19 percent), which means the net effect of the bill will be an increase in total spending of roughly 3 percent.
In Table 1 have listed the components of the savings reported in Chapter 3 (they are expressed as percentages of total US health care spending) for the year 2017, the first year in which Pollin et al. assumed S 1804 would be operative. I have italicized the two components that are grossly exaggerated – the 9.0 percent savings from reduced administrative costs and the 1.0 percent from a category Pollin et al. call “service delivery reform.” Pollin et al. claim the 1.0-percent figure would climb even higher over the succeeding decade, and by the tenth year after implementation would equal 5.1 percent. The total savings by Year Ten, then, would be 19.2 plus another 5.1 percent, or 24.3 percent. (1)
The evidence indicates the 9.0 percent administrative savings and the 1.0 percent savings attributed to “service delivery reform” will both equal approximately zero (or might even add to total spending). I turn to the administrative savings estimate first.
Because of two defects in S 1804, the administrative savings will be much less than 9.0 percent, maybe even a negative number (that is, S 1804 might drive up administrative costs).
Defect 1, Section 611(b): The first and most important defect is Section 611(b). That section requires the Secretary of Health and Human Services (HHS) (the person placed in charge of the new system by S 1804) to impose on the entire population “accountable care organizations” (ACOs) plus three or four dozen other complex “value-based payment” (VBP) programs currently imposed only on the fee-for-service (FFS) Medicare program. Expanding the reach of these programs from the 40 million people currently enrolled in the Medicare FFS program to 325 million Americans of all ages will add substantially to administrative costs. Estimating the impact of these VBP programs on administrative costs is very difficult because Section 611(b) authorizes so many of them, because research on VBP administrative costs is almost non-existent, and because it is not clear how far and wide – in terms of geography and people covered – these programs would spread.
Defect 2, no hospital budgets: A second defect that will reduce administrative savings below Pollin et al.’s projected 9.0 percent is that S 1804 does not authorize HHS to use hospital budgets, but instead authorizes HHS to use only the current method used by Medicare (the “diagnosis related group” method), which requires hospitals to keep track of expenses for each patient.
Despite these two defects, Pollin et al. relied, inexplicably, on research based on the Canadian system to derive their estimates of administrative savings under S 1804. This would have been appropriate if the object of Pollin et al.’s study had been HR 676, the Expanded and Improved Medicare for All Act. HR 676 does not contain a section like Section 611(b), and it does authorize the use of budgets for hospitals. Nor does the Canadian system contain these two defects; the Canadian system does use hospital budgets, and it does not impose the VBP programs authorized by Section 611(b). Pollin et al. aggravate their misuse of research based on the Canadian system by failing to notify readers they did that.
For two reasons, Pollin et al.’s conclusion that savings from “service delivery reforms” will amount to 1.0 percent in Year One and grow even more over the next nine years is also inaccurate. First, Pollin et al. fail to cite a section in S 1804 that authorizes these “reforms” (see appendix for further discussion of this defect). Second, the document they cite to support their claims, a 2011 report by the Institute of Medicine (IOM), is sprinkled with warnings that the report’s conclusions are speculative. To take one example, the IOM report says its estimates “should not be construed as consensus or recommendations on specific numbers or actions.” (p.599)
In the remainder of this paper I discuss in more detail each of these inaccurate savings estimates – the 9.0-percent administrative savings estimate and the 1.0-percent “service delivery reform” estimate.
Savings from reduced administrative expenditures
S 1804 initially appears to be a single-payer bill because it contains one of the most important hallmarks of single-payer legislation – a section (Section 107) which prohibits insurance companies from selling insurance that duplicates the coverage authorized by S 1804. It is the replacement of multiple insurance companies with one government payer, plus the use of hospital budgets and uniform fee schedules by the one payer, that reduces administrative costs to Canadian levels in single-payer legislation such as HR 676. But by adding Section 611(b) and not authorizing hospital budgets, S 1804 offsets the administrative savings achieved by Section 107. (2)
The two most comprehensive and complex of the VBP programs authorized by Section 611(b) are CMS’s ACO program, authorized for Medicare by the ACA, and its Merit-based Incentive Payment System (MIPS), authorized for Medicare by the 2015 Medicare Access and CHIP Reauthorization Act (MACRA). Of these two programs, the ACO program is the more destructive of S 1804’s ability to cut administrative costs. It is the single most important reason why S 1804 will not cut administrative costs much or at all, and might even raise them. Section 611(b) will insert a layer of ACOs (probably several thousand) between the US Treasury and the nation’s providers. ACOs are simply HMOs with a new name. They generate overhead costs that are almost identical to those insurance companies generate, for both themselves and providers. Thus, substituting a layer of several thousand ACOs for a layer of insurance companies cannot reduce the administrative costs created by America’s current multiple-payer system.
None of this matters to Pollin et al. Although Canada does not impose ACOs or any of the other VBP programs nationwide as S 1804 does, and although Canada controls hospital spending with budgets while S 1804 does not, Pollin et al. ignore those facts and, inexplicably, use research on Canada’s system to estimate the reduction in provider administrative costs S 1804 would allegedly achieve. (3) Although the system S 1804 will rely on multiple-payers (i.e., ACOs), Pollin et al. cite research on the fee-for-service American Medicare program conducted prior to 2012 (including a paper I wrote) when that program was still a single-payer program (CMS did not inject ACOs into that program until 2012). This is akin to using research on a fuel-efficient car to estimate the mileage for a gas-guzzling car but not warning your readers you did that. (4) Moreover, Pollin et al. do not discuss and do not cite what is probably the best study of hospital budgets, a 2016 paper by David Himmelstein et al. This study concluded that hospital administrative costs in Canada and Scotland, both of which use hospital budgets, are half those of US hospitals. (5)
To give you some idea why the extension of numerous VBP programs from the FFS Medicare program to the entire country will make it impossible for S 1804 to achieve a substantial reduction in administrative costs, consider the administrative costs of just the ACO program (and ignore for now other complex VBP programs like MIPS). Imagine a system in which half the US population is enrolled in 3,000 ACOs (which is roughly three times the number of ACOs in the US today), and the other half stays out of ACOs, which is to say they remain in the FFS sector (their doctors and hospitals are paid directly by HHS). Let’s ask now, What happens to administrative costs in the new insurance sector (the 3,000 ACOs plus HHS) and what happens to administrative costs for hospitals and clinic?
Let’s examine the new insurance sector first. ACOs incur nearly all the costs that insurance companies do, for the simple reason that they do almost exactly what insurance companies do:
they enroll people,
they make contractual commitments to provide all necessary medical services to their enrollees for one year in exchange for a per-head fee (aka a premium),
they spend money each year cobbling together networks of clinics and hospitals beyond which enrollees are not supposed to stray, and they print lists of “preferred providers” within their networks (that are out of date by the time they are published),
they monitor “their” doctors, hospitals and nursing homes in order to punish and reward them according to crude cost and quality measures,
they spend money on utilization review and other activities that reduce their exposure to sick patients;
they pay their top executives outlandish sums,
those with shareholders (for-profit ACOs) divert some of their revenues to profit, and
they lobby. (6)
The only activity ACOs clearly spend less money on than insurance companies do is revenue collection. Unlike premium collection conducted by insurance companies, revenue collection for ACOs is done by the insurers the ACO has contracts with (Medicare, a state Medicaid program, a private-sector insurance company, or a self-insured employer). It is also possible ACOs spend less money on sales than insurance companies do. (7)
Now let’s look at what happens to administrative costs for providers. Let’s take a hospital as an example. Hospitals will have to treat patients insured by multiple payers – HHS for patients who remain in the non-ACO (or FFS sector), the multiple ACOs in the hospital’s service area, and, for hospitals with national reputations, ACOs outside their service areas as well. This means hospitals will have to do what they do now in our multiple-payer system: Determine whether patients are in an ACO and, if so, which one, and prepare bills for each patient. Hospitals and doctors will have to spend time on the phone with ACOs seeking permission for admission to hospitals, for an extra day in the hospital, and for access to specialists who are not in the ACOs’ networks.
This exercise gives you some idea why there are no hospital budgets in S 1804. Under a multiple-ACO system, would it make any sense to give each of America’s 5,500 hospitals budgets? That might make sense in a few rural areas where no ACOs set up shop. But for most of the country, authorizing HHS to negotiate with hospitals over budgets as well as ACOs over their premium payments would only make a big mess even messier. It seems reasonable to infer that Senator Sanders (and whoever else wrote S 1804) understood that they had a choice – either authorize ACOs and give up on hospital budgets, or leave ACOs out of S 1804 and use hospital budgets. They chose ACOs. This exercise also helps us understand why Pollin et al. chose not to discuss research on the role that hospital budgets play in reducing administrative costs. (8)
Savings from “delivery system reform”
To estimate savings in this category (1.0 percent of total spending in Year One), Pollin et al. relied on a report issued in 2011 by the IOM, one of the earliest and most prominent proponents of HMOs and other managed care solutions. The report, entitled The Health Care Imperative, was prepared by the IOM’s “Roundtable on value and science-driven health care,” which consists of four dozen of the nation’s health policy elite, including representatives of big business (GE and Microsoft, for example) and the insurance (United Health Group), drug (Pfizer) and hospital (Mayo Clinic) industries. It was financed by the Peter G. Peterson Foundation, a foundation devoted to shrinking the national debt by shrinking Medicare and other social programs. Pollin et al. repeatedly praised this study; at page 6, for example, they called it “a major … study by the IOM.”
The IOM claimed the US could cut health care costs substantially by reducing seven broadly defined, overlapping types of waste and overuse. The IOM gave these categories labels such as “prevent medical errors,” “reduce hospital readmissions,” and “improve hospital efficiency.” Pollin et al. claimed that S 1804 would reduce all seven categories of waste. You can read this claim at page 65 of their report, and you can view the seven types of savings in Table 14 at the bottom of that page. You will see a footnote at the bottom of Table 14 telling you the numbers come from Box 22-1 at page 602 from the IOM’s 2011 report. You will see as well a link to Chapter 22 of the IOM report containing that box.
The IOM’s Chapter 22 is not based on research. It is based on the equivalent of a focus group. The IOM asked ten “knowledgeable authorities” (including two insurance company representatives) to participate in a “round table” and opine, based on their “personal experience,” on how to reduce each of the seven categories of waste by an amount that would equal 10 percent of total US spending. Chapter 22 contains numerous warnings that the estimates derived from the ten experts’ conversation were too speculative to be used to make estimates or set policy. Here are four of those warnings:
This chapter summarizes the discussions, insights, and perspectives offered by the individual attendees at the meeting, and it should not be construed as consensus or recommendations on specific numbers or actions. (p. 599)
[S]ubstantial work is needed to refine and strengthen the analytics…. (p. 603)
As this table summarizes the discussions and ideas offered by the individual attendees at this meeting, it should not be construed as consensus or recommendations on specific numbers or actions. (Footnote to Table 22-1, p. 607)
[P]articipants pointed out that the estimates discussed had not accounted for implementation and overhead costs. (p. 607)
Pollin et al. ignored all these warnings and, moreover, offered their readers no hint that the IOM thought the “insights” of their ten “knowledgeable authorities” were so imprecise and so poorly documented they could not support policy recommendations. To the contrary, Pollin et al. made it clear they took the IOM guesstimates at face value. For example, Pollin et al. state, “The IOM study offered a range of estimates that the various analysts contributing to the project believed was realistic within a 10-year time frame.” (p. 65) (9)
Because so little research has been done on the administrative costs of even the largest of the VBP programs authorized by Section 611(b) (ACOs and MIPS), never mind the dozens of smaller programs, it is impossible to say with any precision what impact Section 611(b) will have on total administrative spending in the US, which now constitutes about one-third of the $3.5 trillion we spend every year on health care. As I noted above, there are reasons to think a multiple-ACO system paid by one federal agency will cut administrative costs slightly. But there are also reasons to believe ACO cost-control tactics will create new administrative costs. The only defensible position to take on such a complex issue with so little evidence is to attribute zero reduction or increase in administrative costs, and simply to delete from Table 1 above the 9.0 percent savings Pollin et al. attributed to reduced administrative costs.
For three reasons, Pollin et al.’s 1.0-percent estimate of savings due to “delivery system reform”requires the same response:
(1) Pollin et al. do not cite a section in S 1804 that would mandate “delivery system reform”;
(2) they are vague about the “reforms” they’re talking about, but because they mention ACOs and cite research that explicitly discusses ACOs and other VBP programs, we may infer they have ACOs and VBP programs in mind; and
(3) the small body of rigorous research on ACOs and other VBP programs on costs suggests they either break even or raise costs.
The only appropriate assessment, then, is to strike the 1.0-percent savings estimate for Year One, and the 5.1 percent savings for Year Ten.
Pollin et al.’s estimate of savings from reduced drug costs (5.9 percent), use of Medicare provider rates (2.8 percent) and less fraud (0.5 percent) strike me as reasonable, especially given S 1804’s four-year phase-in period. The new total savings, therefore, is now 9.2 percent, not 19.2 percent (see Table 2 below). The authors’ estimate of a 12.0 percent increase due to covering the uninsured and covering everyone with first-dollar coverage is reasonable. The net effect of S 1804, then, will be a slight increase in total spending of about 3 percent.
1 Pollin et al. reported neither the 5.1 nor the 24.3 percent figure. I derived the 5.1 figure from Pollin et al.’s statement that the savings from “service delivery reforms” would equal 1 percent of total US spending in Year One, 1 percent in Year Two, and 0.5 percent in each of the succeeding eight years. Adding 5.1 to 19.2 yields 24.3.
2 Section 611(b) states: “Any payment reform activities or demonstrations planned or implemented with respect to such title XVIII [this is the title in the Social Security Act that created Medicare] as of the date of the enactment of this Act shall apply to benefits under this Act, including any reform activities or demonstrations planned or implemented under the provisions of, or amendments made by, the Medicare Access and CHIP Reauthorization Act of 2015 (Public Law 114–10) and the Patient Protection and Affordable Care Act (Public Law 111–148).”
3 In an inaccessible endnote (note 35, p. 44), Pollin et al. indicate they understand that their use of research on the Canadian system was inappropriate, and that Section 611(b) damages S 1804’s ability to reduce administrative costs. “It is critical to recognize here that the large-scale potential savings through Medicare for All in the area of BIR [billing and insurance-related costs] are achievable only through operating Medicare for All with a simplified financing system,” they write. “A large share of the potential savings will be lost to the extent that, for example, Medicare for All retains much of the complexity of the current Medicare system. We return to this issue below.” The phrase “much of the complexity of the current Medicare program” appears to be a reference to the VBP programs authorized by Section 611(b) and S 1804’s failure to authorize hospital budgets. If that’s accurate, I totally agree: If Senator Sanders does not delete Section 611(b) and add hospital budgets, a “large share of the savings” will be sacrificed. Why is this warning so abstract and left so unexplained (it takes knowledge many people don’t have to understand what is being said here), and why is it tucked away in an endnote and not repeated anywhere else? Contrary to this endnote, Pollin et al. did not “return to this issue below.” This warning to themselves went unheeded; Pollin et al. did assume that “large share of savings” would be achieved.
4 Pollin et al.’s discussion of their estimate of administrative costs under S 1804 appears at pages 43 and 44. They rely exclusively on research on the Canadian system for their estimate of reductions in provider administrative costs, and on research on Medicare’s overhead and insurance overhead in other countries (not just Canada) for their estimate of reductions in insurer-sector overhead. Here is an example of their discussion of the research on provide overhead in Canada: “A 2011 study finds … that these costs in the U.S. are 73 percent higher than those in Canada. See Morra, Dante et al. ‘US physician practices versus Canadians: spending nearly four times as much money interacting with payers’ Health Affairs …. http://bit.ly/2OoaJlA.” Here is a link to another study they cited that relied on Canadian data.
5 Hospital budgets as they are used in Canada and several other countries, and as they are described in HR 676, reduce administrative costs by reducing paperwork associated with billing (the hospital no longer has to keep track of every aspirin given to every patient). The positive effects of hospital budgets extend beyond reducing administrative costs. Reduced administrative costs mean doctors and hospitals have more time to spend with patients. Because hospital budgets are divided into capital and operating budgets, they give society control over capital spending, not only the level of capital spending but the allocation of capital assets (where MRIs and emergency rooms will be located, for example). Control over the level and allocation of capital spending may also contribute to lower spending by reducing overuse of some types of assets, such as MRIs. Because hospital budgets are set in advance, hospitals are guaranteed a certain amount of money each year. That greatly reduces the incentive to concentrate on profitable patients and avoid sicker patients compared with incentive created by the crudely risk- adjusted, per-patient payment method proposed under S 1804. Finally, it is important to note that hospital budgets are applied to individual hospitals, not hospital chains, or hospital-clinic chains. Setting a budget for an entire chain of hospitals and clinics dilutes social control over both capital and operating expenditures. For example, a budget for a chain turns control over where ob-gyn facilities will be located to the chain management, and gives to the chain executives authority over how much to pay doctors.
6 This description of ACOs is based on the definition of ACOs contained in the Affordable Care Act and the regulations promulgated for ACOs by CMS in 2012 and later years, which is the definition Section 611(b) references. Strictly speaking, Medicare beneficiaries do not “enroll” in ACOs; they are assigned to them by CMS using a complex algorithm that begins by determining which primary care doctor the beneficiary saw the most often in the two or three years prior to the “performance” year. But ACOs are not happy with CMS’s “attribution” method. CMS is under pressure to switch to enrollment or something resembling enrollment called “attestation.”
7 Note that I’m not counting the extra payments ACOs make that help some patients, such as the cost of hiring extra nurses to stay in touch with congestive heart failure patients after discharge, or the cost of giving free transportation to enrollees to ensure they keep their appointments. These costs are not administrative costs – they are services that insurance companies typically don’t pay for that ACOs are expected to pay for out of the savings they allegedly achieve, which is to say out of thin air.
8 Clinics will face billing hassles just like those hospitals will face in a multiple-ACO system. They will have to determine for each patient which entity – an ACO or HHS – should be billed. ACOs will use the bills they get from their doctors to determine how much of the ACO’s profits or losses at the end of the year must be shared with/inflicted upon which doctors.
9 To their credit, Pollin et al. do note in their discussion of VBP programs that they can worsen disparities because they punish providers who treat a disproportionate share of the sick and the poor (see discussion at page 58). In my view, they should have gone beyond that statement and noted that this well known threat induces providers to game ACO and other VBP programs by multiple means, including avoiding sicker patients or denying them necessary care. According to recent research, the Hospital Readmissions Reduction Program (HRRP), one of the VBP programs authorized by the ACA that Section 611(b) would unleash upon the entire country, may be killing up to 10,000 elderly people with congestive heart failure annually. Whether the harm being done to at least some patients by the HRRP is offset by benefits provided to others is not clear. What is clear is that Senator Sanders is proposing to expand a program that is harming at least some patients.
Appendix: Sections in S 1804 Pollin et al. claimed authorized ACOs and the “restructuring” of the medical sector
Pollin et al. open chapter 3 with this statement: “Implementing Medicare for All will generate a full-scale restructuring of the U.S. health care system.” At page 54 they use the same phrase, “full-scale restructuring,” but this time they limit the predicted restructuring to the provider sector: They claim S 1804 will achieve “significant savings through the major restructuring of the U.S. health care delivery system.” (emphasis added). These are not statements anyone would make about HR 676. HR 676 “will generate a full-scale restructuring” of only the insurance sector. Unfortunately, Pollin et al.’s statement is an accurate characterization of S 1804. S 1804 will substantially restructure the insurance sector and the medical (provider) sector. Boasting about “restructuring” clinics and hospitals is not a good idea. It gives credence to claims by single-payer opponents that “single payer will give the government control of your doctor.”
But nowhere do Pollin et al. tell us where in S 1804 we might find language authorizing HHS to impose “full-scale restructuring” on the medical sector. The most obvious section is Section 611(b). That section authorizes dozens of VBP reforms (ACOs, MIPS, etc.) all of which apply to, and profoundly affect, the medical sector, not the insurance sector. Those changes are so
profound it is accurate to say they amount to a “full-scale restructuring” of the medical sector. But nowhere in their report do Pollin et al. say Section 611(b) is the section that authorizes this “full-scale restructuring.”
Because Section 611(b) authorizes dozens of VBP reforms (ACOs, MIPS, etc.), and because it is the only section that does that, it would seem to be the obvious section to cite for the “full-scale restructuring” authority. But Pollin et al. give us two reasons to believe they do not view Section 611(b) as the clause that authorizes HHS to unleash “full-scale restructuring.” First, they fail to cite it as the section that does that. Second, they say Section 611(b) must be “reassessed” because VBP reforms have, to date, failed to cut costs and have worsened disparities between income and racial groups. At page 58, Pollin et al. state:
This draft of the bill includes a Section (Section 611(b), p. 47) which mandates that “current and planned payment reforms” established under the ACA and MACRA be carried forward under Medicare for All. Clearly, this feature of the Medicare for All bill will need to be reassessed in light of the evidence that, to date, pay-for-performance programs are reinforcing existing health care disparities according to both socioeconomic and prior medical status. (p. 58, emphasis added).
So if Pollin et al. do not rely on Section 611(b) for their claim that S 1804 will “restructure” the medical sector, what sections did they invoke? The sections Pollin et al. appear to be thinking about, which they discuss at pages 55-56, are three sections that give HHS general authority over the quality and cost of health care in the US. Here are three excerpts from Chapter 3 in which the authors explain their interpretation of these sections:
To begin with, in Title IV of the draft legislation on “Administration,” the measure states that the “general duties” of the Secretary of … HHS include establishing both “methods for determining amounts of payments to providers of covered services,” and “the determination of medical necessity and appropriateness with respect to coverage of certain services….” [pp. 55-56];
The Secretary of HHS is further responsible for formally monitoring and providing an annual report that covers, among other matters “cost containment measures and achievements under this Act” and “quality assurance,”…. [p. 56]
Title VI of the draft legislation, “Health Budget; Payments; Cost Containment Measures,” provides additional details on the proposed budgetary framework for Medicare for All. This section of the legislation begins with defining a “National Health Budget” in these general terms: “By not later than September 1 of each year…the Secretary shall establish a national health budget, which specifies the total expenditures to be made for covered health care services under this Act…..” [p. 56]
Where in any of these provisions would the Secretary of HHS find authorization for “full-scale restructuring” of the medical sector, or anything else for that matter? It isn’t there. All one can infer from these excerpts is that HHS is authorized to set a national budget, to determine how providers will be paid, and to issue annual reports on cost and quality.