How Employer-Based Health Insurance Actually Is Cheaper Than Government-Sponsored Insurance

By Chris Conover, Forbes Magazine

Princeton health economist Uwe Reinhardt this week posted a very informative and entertaining (no surprise there) response to Sally Pipes’ claim that Employer Health Insurance: A Bargain Compared to Government-Sponsored Coverage (read: Medicare and Medicaid). It’s hard to disagree with most of what Prof. Reinhardt has offered in his rebuttal.  But at the risk of having my own post become a “platform for a homework assignment in undergraduate health-economics courses,” let me suggest that Sally Pipes may actually be right for reasons she never even addressed.

Comparing the cost of employer health insurance to the cost of either Medicare or Medicaid is a completely stacked comparison even if we fully adjust for every iota of age and health status differences between these three populations and use an apples-to-apples comparison of plans having the identical benefits and actuarial value.  The premium cost for employer coverage essentially embeds the fully loaded cost to society of providing such coverage, inclusive of profits, insurance company tax payments and administrative costs. In contrast, the Medicare and Medicaid cost figures used by Prof. Reinhardt include only a portion of the administrative costs incurred by both programs and none of the hidden costs of taxation known as excess burden or deadweight losses. Once these hidden costs are taken into account, even Medicaid (whose provider payment rates are so abysmal that one third of primary care physicians refuse to see Medicaid patients) costs more than employer-sponsored coverage.

How Much Does it Cost To Raise $1 of Taxes?

Every federal tax imposes deadweight losses (also termed “excess burden” by economists): whatever we tax essentially shrinks the economy by some amount. Tax labor and people work less. Tax capital and people save less. Tax gasoline and people buy less gas, etc.  An excess burden of 25{ea7427a437b28cbad7e7d899824905928e2417ff8c20d55929235e4c32efaa1c} essentially implies that the economy shrinks by 25 cents for every dollar of taxes collected. The specific manner in which economic activity shrivels obviously depends on which taxes are increased to pay for new federal spending. One would hope that every Princeton undergraduate who has taken health economics is aware of deadweight losses and why the goal of any optimal tax system is to minimize them.  While there may be some debate about the size of such losses, there is no dispute among economists about their existence.

Unfortunately, these hidden costs are rather sizable. For federal taxes, we know that deadweight losses amount to at least 25{ea7427a437b28cbad7e7d899824905928e2417ff8c20d55929235e4c32efaa1c} of revenues. How? Because in 1992 the Office of Management and Budget first began requiring federal agencies to assign a shadow cost of 25 cents to every dollar of expenditures financed out of tax revenues (this 25{ea7427a437b28cbad7e7d899824905928e2417ff8c20d55929235e4c32efaa1c} figure was based on the estimated size of excess burdens from the then-available literature).[1]  But excess burdens increase with the square of marginal tax rates. Thus, the deadweight loss associated with $1 of new federal spending (assuming these are tax-financed) will be much larger than the average excess burden spread across all federal spending paid through taxes.  Elsewhere I have calculated that the marginal excess burden on a “typical” federal tax dollar is 44 cents per dollar of revenue.

The True Social Cost of Public Insurance is Much Higher Than Advertised

Prof. Reinhardt points to an Urban Institute study that essentially shows that the total cost of Medicaid (inclusive  of out-of-pocket spending) would be 35{ea7427a437b28cbad7e7d899824905928e2417ff8c20d55929235e4c32efaa1c} higher if it low income Medicaid beneficiaries instead had employer-sponsored health insurance.[2]  But once we add in the hidden costs of taxes, the story changes considerably.


Using OMB’s figure of 25{ea7427a437b28cbad7e7d899824905928e2417ff8c20d55929235e4c32efaa1c} for average deadweight losses, this huge cost difference shrinks to only 8{ea7427a437b28cbad7e7d899824905928e2417ff8c20d55929235e4c32efaa1c}. Moreover, using my more recent figure of 44{ea7427a437b28cbad7e7d899824905928e2417ff8c20d55929235e4c32efaa1c} for the marginal excess burden makes ESI about 6{ea7427a437b28cbad7e7d899824905928e2417ff8c20d55929235e4c32efaa1c} less expensive than Medicaid.  And even this comparison is not exactly fair. The sad truth is that nearly one-third of doctors are unwilling to take new Medicaid patients. Thus, despite having cards that purport to provide them with coverage whose actuarial value is 99{ea7427a437b28cbad7e7d899824905928e2417ff8c20d55929235e4c32efaa1c}, many Medicaid beneficiaries are second-class citizens, which helps explain why in som studies emergency room use has been found to be 50 percent higher for Medicaid patients compared to those with private coverage (or, for that matter, those without coverage). Even the study cited by Prof. Reinhardt shows ER use among low income adults on Medicaid is at least 20{ea7427a437b28cbad7e7d899824905928e2417ff8c20d55929235e4c32efaa1c} higher than for their statistical twins on ESI.

But if the full social costs of ESI are at least 6{ea7427a437b28cbad7e7d899824905928e2417ff8c20d55929235e4c32efaa1c} less expensive than Medicaid (and probably a few percentage points higher were we to count the cost of tax collection/tax compliance associated with the federal and state taxes used to bankroll Medicaid), then we can be sure that a statistically fair comparison of the cost differential between ESI and Medicare is much greater since Medicare does not underpay providers to nearly the same extent as Medicaid. Medicare and Medicaid payment rates are quite comparable for hospital care, i.e., about one third below payment rates of private insurers (Figure 1).  In contrast, while the ACA has temporarily boosted Medicaid physician payment rates to comparable levels as Medicare, historically there has been about a 20 percentage point difference (Figure 2). That is “Medicare payment levels in 2009 were about 80 percent of private health insurance payment rates, and Medicaid payment rates in 2008 were about 58 percent.”

In short, Sally Pipes probably was right, but not for the reasons she outlined. I hope my post makes a modest contribution towards reducing the massive confusion among the public that Prof. Reinhardt so deeply laments.