Hi,
It has become almost an habit for me to review M4A cost studies, time and again, when as a foreigner I don’t talk about alternatives to it. This to excuse yet another diary on the subject.
Two more excuses to write about it are SHIFT, an interactive tool for people to understand M4A costs, and Robert Pollin himself complaining about the misinformation out there. So just like I compared the PERI study with the Urban Institute’s (it’s the ‘again’ above), I will here compare the PERI study (Pollin et al. 2018) with the Lancet’s (Galvani et al. 2020). Using SHIFT.
tl;dr (too-long;didn’t-read) Go check SHIFT.
0. What is SHIFT?
The Single-payer Health-care Interactive Financing Tool (SHIFT) is a simple web tool with sliders for people to test M4A costs by themselves.
It is the best scientific vulgarization the Lancet could come up with to explain to people how, yes, Medicare-for-All reduces costs. Not just showing numbers and a rationale for it, but letting people change those numbers to better fit their own expectations and see the result.
It also conveniently divides that question into three parts: (1) how much M4A would cost, (2) how much M4A would save and (3) how M4A could pay for it.
1. How much M4A would cost
In short: a maximum of 12% more, or $3.911 trillions in 2017.
Both Pollin et al. (2018) and Galvani et al. (2020) use a 2017 baseline to calculate costs. However, their baseline differs: the full $3.492 trillions for Lancet and a $3.267 trillions sub-part for PERI. Reason being that the remainder is investments and that part shouldn’t be affected by higher health care utilization.
“Health care utilization” is what SHIFT calls “expansion in health care use”, and it is exactly that. M4A costs more because more people get access to health care, whether previously uninsured or underinsured. It also accounts for the more generous plans, notably the lack of cost-sharing.
Each study uses slightly different methods to reach the same conclusion.
PERI is relying on Thorpe (2016) and Brot-Goldbert et al. (2017) for a high-end 12% cost increase (Pollin et al. 2018: 32) while the Lancet is relying on the RAND study (Liu et al. 2018) with an 8% cost increase (Galvani et al. 2020: 528). In other words, between $275-400 billions for 2017. Applying 12% to $3.492 trillions gives a $412 billions increase, so that’s the actual maximum with a vast, vast consensus among studies:
M4A would cost, at a maximum, 12% more, or $3.911 trillions in 2017.
|
Baseline |
increase |
proportion |
Total |
PERI |
3.267 |
0.390 |
12% |
3.630 |
Lancet |
3.492 |
0.280 |
8% |
3.769 |
High-end |
3.492 |
0.412 |
12% |
3.911 |
Fig.1: How much M4A would cost
SHIFT only lets you go up to $3.769 trillions, and the PERI study actually says $3.630 trillions, but let’s go conservative (and blind) and seek the high end here.
2. How much M4A would save
In short: a minimum of 14%, or $3.406 trillions in 2017.
If SHIFT doesn’t let you jack up the price as much as you would want, as far as cost savings this is where it, and Medicare-for-All, start to truly shine.
The Lancet proposes 5 categories of cost savings: (a) hospital rates, (b) physician rates, (c) drug prices, (d) administration costs and (e) fraud detection. The PERI study acknowledges all five, including fraud detection (Pollin et al. 2018: 55). Note that there can be more mechanical savings (reducing waste) but we’ll consider them negligible.
On categories (a-d), both studies agree, using high-end savings (Medicare rates and Medicare administrative costs notably). Pollin et al. go with a 3% fraud reduction while Galvani et al. estimate a 4% one. Eventually, their results are comparable, with $650 bn and $700 bn respectively, or 18-19% in cost savings. Note that in figure 2 below I don’t know exactly to what those percentages relate, so rely on totals for accuracy.
While nobody will complain about going high-end on cost, people will and do complain about going high-end on cost savings. Which is why I used SHIFT to test my own, more prudent assumptions. On (a-b), I tried to mirror 110-115% Medicare rates (translating to 4.5% for (a) and 6.5% for (b)), but feel free to go lower. I kept (c) as is as few if anyone seems to be challenging that one. And I go with a higher 6% administrative costs. Oh and while that’s inaccurate, I for some reason decided to eliminate all fraud detection gain — it’s mechanical, it should happen but eh. My proportion is relative to the baseline, $3.492 trillions.
M4A would, at a minimum, save roughly 14%, or $3.406 trillions in 2017.
|
SAVINGS |
PROPORTION |
TOTAL |
PERI |
0.7 |
19% |
2.93 |
LANCET |
0.649 |
18% |
3.034 |
Low-end |
0.505 |
14% |
3.406 |
Fig.2: How much M4A would save
I called my own estimate “low-end” although I suspect some would want to go lower on categories (a-c). That would assume M4A has failed. Not that the suspicion would be groundless but if politics enter the calculus I can as well guess cost-savings would be reintroduced by Congress, or speculate freely.
That low-end is reasonable enough, and I won’t go into the politics of how they would actually be helping hospitals (uninsured costs, Medicaid rates, more demand...). Especially because I have no number on that.
3. How M4A could pay for it
In short: however you want, it will be cheaper for individuals (and companies).
The PERI study suggests an 8% payroll tax, coupled with a 3.75% sales tax and a 0.38% wealth tax, essentially. The LANCET study, and SHIFT therefore, let’s you play with an array of options but mainly: payroll tax (companies), income tax (households) and wealth tax (see SHIFT for what that means). They go with 8%, 5% and 0% respectively.
Using SHIFT and my $3.406 package, I went with an 11% payroll, 8% income and 0.6% wealth tax. Paid for. Companies and individuals still pay less on average, and progressive taxation would shift the burden even more away from the lower-class. I’ll put a table there too just for balance:
|
payroll |
income |
sales |
wealth |
total |
PERI |
8% |
0% |
3.75% |
0.4% |
$1.08 |
LANCET |
8% |
5% |
0% |
0% |
$0.811 |
estimate |
10% |
8% |
0% |
0.6% |
$1.145 |
Note how low the totals are. That’s because of previous revenues still being in place (the $2.263 trillions block in SHIFT). This too would be subject to discussion but it would still be just more money-shifting.
Using the Lancet’s cost estimate (see §1), my budget would actually be $3.264 trillions, and I would only need an 11% payroll and 7% income taxes. The point SHIFT is making here is that, really, you can go for any combination you want: as long as your sliders are lower, companies and/or households will pay less on average (no premium, co-payment, etc). You are just not allowed to make any of them pay more (on average).
And since even this high-end/low-end estimate still comes up with an health care cost ~$100 bn lower for 2017, anyone and their golden fish can meet the goal.
4. Stop the misinformation
Medicare-for-All won’t cut health care costs by half. All it can do is use a monopoly to negotiate prices down (and rationalize the administration). But unless you bet on its failure, or use the kind of science climate change deniers are used to, it will lower costs.
Yes, it will double the federal budget. But that is just shifting money. Individuals (and companies) will pay less. And the nation will pay less as a whole.
Yes, it will disrupt the private market, and a 4-years transition might be too short, with so many jobs on the line. But that market is broke, broken and bloated and any serious plan has to disrupt it over ten years — or fail.
And no, health care utilization to the best of our scientific knowledge isn’t going to shoot up magically by 20.5% (hello Urban Institute, page 37). The consensus is actually 9%, and we’re using 12% here. Anything above is koolaid.
I repeat. Warren based her plan on the Urban Institute’s “questionable utilization” (they can explain their calculus any time of the year. Even Blahous admits he can’t tell how they reach their number). John Oliver said noone could tell how much it would cost. The NYT is relying on the Urban Institute too. Biden and Buttigieg routinely use those sky-high figures. None of it checks out. Actual science, backed up with detailed methodology and references, agrees on 9%, 12% if like me you are really risk-averse.
We have reliable studies (do rely on PERI more than the Lancet though...) and we now have a web tool to, while limited, let people understand how M4A costs actually work.