My rating of BernieCare claims: plausible

1/19/2016 10:08:00 AM
In my previous post, I criticized Bernie Sanders for having a policy specific—always a vote-losing thing to have in an election—without actually having any of the specifics needed to defend it from attacks. As an example, I cited Clinton's charge that the plan raises taxes on the middle class, to which Sanders was only able to respond that it actually saves middle class families an unspecified "thousands of dollars."

At long last, the Sanders campaign has released some more detail on his plan. It would, he now claims, save the average family $5,000 per year after taxes.

That estimate is roughly what you'd get if you naively apply Medicare payment rates and administrative costs to the average existing private insurance plan. Medicare pays hospitals and doctors about 80 percent of what private insurance does, and has substantially lower administrative costs of 3 percent compared to 17 percent for private insurers. Average health spending is $10,000 per person per year, and the average health plan (ESI) has an actuarial value of around 85 percent. Applying the 80 percent lower Medicare prices and the 14 percent administrative savings, I get roughly $7,000 in health spending per person per year, a $3,000 savings per person (that is, after paying Sanders's payroll tax, you'd still be $3,000 ahead in savings because of the lack of private insurance premiums and lower out-of-pocket prices). So for a family of four, Sanders's $5,000 estimate actually has a bit of wiggle room—it's smaller than the most optimistic possible scenario.

However, even though Medicare pays lower prices and has substantially lower administrative costs, it is not a slam-dunk case to argue that Medicare-for-all represents a cost savings. For the full story, you also have to look at utilization. It would be nice if you could just compare average Medicare spending per beneficiary to average private spending, but obviously Medicare beneficiaries are on average quite a bit older, sicker, higher cost patients so we'd expect Medicare spending to be higher even under the null hypothesis that it is identical to private insurance. Indeed, that's exactly what we find: Medicare spends about $11,200 per beneficiary per year (including administrative and other costs), compared to the average private insurance premium of $6,010 per person per year (note: Medicare spending per beneficiary is less than the total health spending per beneficiary due to cost sharing; this figure is most comparable to insurer's revenue per beneficiary).

So the raw numbers aren't conclusive at all—they're consistent with both the claim that medicare is cheaper but current medicare enrollees are higher-cost patients (yes, they are), and the alternative claim that medicare is more expensive, because of higher unnecessary or fraudulent use, in addition to serving an older higher-cost group of patients.

There are two ways in which Medicare can still be more expensive than private insurance inspite of lower administrative costs and lower payment rates: more fraud and more "moral hazard" (by which I mean beneficiaries' non-fraudulent but voluntary choice to use more healthcare than they would with other kinds of insurance). The FBI puts the rate of fraudulent health insurance payouts at 3 to 10 percent, but does not differentiate between private insurance and Medicare (see also). Even assuming no private insurance fraud, this is simply not large enough to even offset the higher administrative costs of private insurance (It appears that private insurers' fraud prevention programs are cost-ineffective!).

Maybe there's a lot more fraud going on, in spite of the lack of evidence. Or maybe private insurers really are getting enrollees to avoid unnecessary costly care. In either case, those effects would have to be quite massive to offset the lower prices and administrative costs of Medicare. So, in the end I do believe Sanders's medicare-for-all plan would represent a net cost reduction on average—the opposite hypothesis, while not impossible by the numbers, is pretty far fetched. Moreover, I think Sanders's $5,000 figure is a reasonable (but highly uncertain) estimate of the average family savings, though it should be noted that a large share of "middle class" people with above average incomes will in fact pay more because the payroll tax is a flat tax while the health benefit is lump-sum.