Last week, a group of scholars from the Urban Institute (UI) released cost estimates for several alternative approaches to health care reform, including an updated estimate of the cost of Medicare for All (M4A) in essentially the form proposed by Democratic presidential candidates Sens. Bernie Sanders (I-VT) and Elizabeth Warren (D-MA). The UI team found that M4A would add trillion to federal spending over its first tenyears. Several people have asked me how the UI team’s estimate compares with my own that M4A would add at least .6 trillion to federal budget costs over its first ten years, and most likely substantially more (between .6 trillion and .8 trillion). To help with comparing the two estimates, I will first offer some general observations and then some specific
Charles Blahous considers the following as important:
This could be interesting, too:
Tyler Durden writes OECD Sees Global Growth At Decade-Low As WTO Warns Of “Doomsday Scenario”
Tyler Durden writes Martin Armstrong Warns Of The Coming “Big Freeze”
Tyler Durden writes Fake Growth? China “Adjusts” 2018 GDP 2.1% Higher Due To Census
Last week, a group of scholars from the Urban Institute (UI) released cost estimates for several alternative approaches to health care reform, including an updated estimate of the cost of Medicare for All (M4A) in essentially the form proposed by Democratic presidential candidates Sens. Bernie Sanders (I-VT) and Elizabeth Warren (D-MA). The UI team found that M4A would add $34 trillion to federal spending over its first tenyears. Several people have asked me how the UI team’s estimate compares with my own that M4A would add at least $32.6 trillion to federal budget costs over its first ten years, and most likely substantially more (between $32.6 trillion and $38.8 trillion). To help with comparing the two estimates, I will first offer some general observations and then some specific numbers.
Observation #1: Scholars’ estimates of M4A’s costs are remarkably similar. Although the UI team’s methodologies differ from mine, it is striking how closely ours and other projections match one another, once adjusted for what they are measuring, which years are analyzed, and for certain key assumptions. Policy preferences do not enter into the estimating process, which simply draws upon the best available data to quantify the likely costs of M4A.
Observation #2: The UI team’s methodologies are different from mine, enabling them to analyze a wider range of policies and outcomes. The UI team makes use of a sophisticated health insurance policy simulation model (HIPSM), as well as another model for simulating future individual and household income (DYNASIM). I did not have access to such models. The potential scope of my analysis was therefore limited relative to the UI team’s. For example, because I was working primarily with national aggregates for various categories of health spending, and did not attempt to project how individuals will respond to different options and incentives, I could only estimate the cost of a universal system lacking significant individual choice. I was able to do so for Sen. Sanders’ M4A proposal because it would provide first-dollar coverage of nearly all health expenditures for nearly all Americans. The UI team produced estimates for a wide range of policy proposals in addition to M4A, and many of these estimates required a capacity to project individual participation decisions in addition to other key variables. As we have seen with the Affordable Care Act, it is very difficult for even the most knowledgeable experts to accurately project such outcomes.
Observation #3: My oft-cited figure of $32.6 trillion is a lower-bound estimate, whereas the UI team’s $34 trillion is toward the low end of a range. The $32.6 trillion number often referenced from my study is the number that results from assuming M4A fully delivers all the cost savings that its supporters aspire to achieve. Accordingly, my study makes clear that actual costs would almost certainly be substantially higher than that lower bound. The UI team surrounds its top-line estimate with various sensitivity analyses exploring possible deviations in either direction—both more expensive and less expensive. The UI team’s analyses taken together make clear that their top-line projection is more likely to underestimate costs than to overestimate them, but they also indicate that it’s not the lowest possible estimate.
Observation #4: We now have additional information indicating that actual costs under M4A would likely be substantially higher than under either of our headline estimates. Both my lower-bound estimate as well as the UI team’s estimate assume substantial cost savings from cutting health provider payment rates down to or near Medicare levels. Developments throughout this year further substantiate that M4A is highly unlikely to be implemented with such dramatic payment cuts.
Observation #5: None of the approaches modeled by the UI team would solve both the problems of unsustainable federal health spending and national health cost growth. Experts generally agree that current federal health spending obligations are unaffordable and that national health costs are excessive. Each of the options modeled by the UI team would make at least one of these problems worse, and half of them would actually make both problems worse.
Now let’s try to quantify more specifically how my estimate compares to that of the UI team. The UI team’s analysis is most detailed with respect to year 2020, so my comparisons will focus on that year. As a starting point for comparison, consider that the UI team projects that M4A would add $2.845 trillion in federal costs in 2020.
Adjustment #1: Offsetting tax revenues. If we establish a single-payer system, the federal government will take on more costs, while employers would no longer compensate workers with pre-tax health benefits. To the extent that such worker compensation subsequently takes the form of higher wages, these would be subject to federal taxes. Taking this additional tax revenue into account reduces the UI team’s projected net federal budget effect in 2020 from $2.845 trillion to $2.687 trillion. My estimates accounted for this effect, so UI’s $2.687 trillion figure is the one that more closely mirrors my analysis.
Adjustment #2: Years studied. My study focused on the years 2022-31, the first ten years of full implementation for Sen. Sanders’ 2017 bill at the time I studied it in 2018. Had I assumed the bill was fully effective in 2020, my lower bound estimate for that first year would have been $2.340 trillion.
Adjustments #3-5: Long-term care benefits, administrative costs, and hospital payment rates. The previous Sanders bill I studied did not provide new long-term care benefits whereas the latest Sanders bill, estimated by the UI team, does. My administrative cost assumption was also lower than the one employed by the UI team. The UI team assumed an administrative cost rate of 6 percent for the entire M4A system, whereas I implicitly assumed a lower rate by leaving unchanged the current administrative costs of covering all those now covered by Medicare, Medicaid or other public insurance. The UI team also assumes M4A would pay hospitals at 115 percent of Medicare rates, higher than the Medicare rates assumed in my lower-bound projection. Adjusting my estimates for my best understanding of the UI team’s assumptions would increase my estimate for 2020 from my lower bound of $2.340 trillion to $2.703 trillion—nearly the same as the UI team’s $2.687 trillion.
Adjustment #6: Other technical factors. Not being familiar with the UI team’s models, I cannot know all of the ways in which their assumptions differ from mine. The UI team uses a different data base for national health expenditures than the one I relied upon from the Centers for Medicare and Medicaid Services (CMS). According to the UI team’s paper, reconciling their estimates with CMS data would increase their NHE estimates for both current law and M4A by equal amounts. It’s not fully clear to me whether or how their federal cost estimates would be affected by this adjustment. Their sensitivity analysis also indicates that their headline estimate does not assume a state-level Maintenance of Effort requirement, whereas my projections assume one for long-term care. My best guess is that if I employed the UI team’s assumptions for these factors, my estimates (already adjusted for assumptions #2-5) would shift slightly downward, to $2.673 trillion. This once again is extremely close to the UI team’s $2.687 trillion.
There are undoubtedly many other ways in which my assumptions differ from those of the UI team. It appears that any other differences, however, largely cancel one another.
Bottom line: The UI team estimates that M4A’s net pressure on the federal budget in 2020 would be $2.687 trillion. Adjusting my lower-bound estimates for my best understanding of the UI team’s assumptions would bring mine somewhere between $2.673-2.703 trillion.
Ten-year cost estimate: The UI team estimates that new federal costs over ten years (2020-29) would be $32.0 trillion, net of new federal taxes. My lower-bound estimate for 2020-29 would be $29.2 trillion using the best-case assumptions in my paper, whereas if I employed the UI team’s assumptions as described above, it would be somewhere between $31.8 trillion and $32.2 trillion.
The latest analysis from the UI team is another piece of information substantiating what we already know about Medicare for All—that it would add at least $32 trillion to federal budget costs over its first ten years. We are still waiting to hear from M4A’s advocates how they intend to finance this unprecedented increase in federal expenditures, as well as how they would mitigate the economic damage that would result from imposing this amount of additional taxation upon the US economy.
Photo credit: JIM WATSON/AFP/Getty Images