Which Is More Efficient: Employer-Sponsored Insurance or Medicaid?


An old disagreement between Uwe Reinhardt and Sally Pipes in Forbes is a teachable moment. There’s a dearth of confrontational debates in health policy and education is worse off for it.

Crux of the issue is the more efficient system: employer-sponsored insurance (ESI) or Medicaid. Sally Pipes, president of the market-leaning Pacific Research Institute, believes it is ESI. Employers spend 60% less than the government, per person: $3,430 versus $9,130, per person (according to the American Health Policy Institute). Seems like a no brainer.

Pipes credits “consumerist and market-friendly approaches to health insurance” for the efficiencies. She blames “fraud,” “improper payment,” and “waste” for problems in government-run components of health care.

But Uwe Reinhardt, economist at Princeton, counters that Medicaid appears inefficient because of the risk composition of its enrollees. Put simply, Medicaid recipients are sicker. Sicker patients use more health care resources. Econ 101.

The points of tension in their disagreement are instructive.

Is ESI free market?

The term “consumerist” instinctively appeals to competition and choice, elements we value in free market. However, health care can’t be compared to shopping for single malt in airport duty free, deciding between Talisker 18 and Glenlivet 21.

ESI is hardly an assortment of private units functioning autonomously and competing with each other. ESI has been carved by so many regulations that the government figuratively runs through its veins.

Do you wonder why insurers in ESI don’t surcharge a family with a child with Tetralogy of Fallot? That is increase their premiums astronomically or deny coverage because of a pre-existing condition.

Goodness of heart? No, it’s because of the government.

This means that young fit joggers are subsidizing the costs for the unfortunate child’s complex cardiac surgery. Insurance is redistribution.

Risk adjustment: Comparing apples and oranges

Failure to adjust for comorbidities makes it difficult to make comparisons in quality, value and performance.

Not only are Medicaid enrollees sicker, they are poorer and less empowered. A priori they are a more inefficient group to deal with than the employed middle class.

I’ll hazard a guess that Sovaldi (medication for hepatitis C) won’t increase Microsoft’s health care bill as much as the state of Illinois’. One, of course, would not credit Microsoft’s cost savings to greater efficiency through clever free market insurance design.

However, in policy discussions comparisons between apples and oranges are commonplace.  Life expectancy and infant mortality are used to compare U.S. health care to countries such as Cuba or France, when adjudicators well know, or should know, that there is more nuance. Using metrics which can be affected by social determinants of health is misleading.

Is Medicaid an island?

There are no islands in health care.

It’s important not to make the same logical errors with Medicaid as with ESI.  Medicaid is not an autonomous government unit. Its recipients aren’t sent solely to safety net hospitals. For most parts Medicaid recipients share the same system as folks on ESI; a system which, arguably, has been sculpted by ESI, for better or worse.

This means there’s interdependence between ESI and Medicaid, or between a government-regulated/ government-subsidized system and a government-regulated/ government-funded system.

Interdependence would be suggested by cost shifting, where costs of seeing Medicaid patients are shifted to ESI. Even if there is no convincing evidence of cost shifting, as Reinhardt cautions but Pipes disagrees with the caution, this interdependence is not diminished. Providers, or hospitals, might happily see Medicaid patients knowing they can still enjoy good returns from ESI, without purposely shifting costs to ESI, or other forms of insurance.

Politics, Ideology and Medicaid

Medicaid is more than a system of reimbursing physicians. It has become an ideology. Any criticism of Medicaid leads to the unfortunate conclusion by some well-intentioned individuals that the purpose of critique is to send the poor to workhouses and let them die – de facto eugenics. No rational discussion can be had when people shout “Republican reforms kill.” The mob clouded the judgment of Pontius Pilate – and that was before Twitter.

Good intention does not mean access, though. Medicaid recipients have a problem of access. This is because Medicaid pays providers far too little whilst simultaneously imposing far too much red tape. Poor access is fiercely countered by some policy analysts and their fierce counter is fiercely countered by practicing doctors who actually see patients on Medicaid.

Regardless, paying providers the least when caring for the sickest, poorest and most disenfranchised section of society does no favors to that section of society.

Medicaid pays a cardiologist, with years of training, $25-40 for a consultation to manage a complex patient with multiple comorbidities, on polypharmacy, where the cardiologist must indulge in shared decision making and also ensure the patient adheres to statins.

For comparison, my personal trainer charges me $80. There’s no shared decision making – he tells me to do “burpees” and I must abide or face his wrath.

Serve and volley at the margins

Both Reinhardt and Pipes cite several studies supporting their point of view. One wonders whether policy wonks truly can form opinions solely from evidence since it’s so easy to cite evidence to support one’s prior convictions and subconsciously disregard or criticize the methodology of studies which refute our convictions.

For example, outcomes are often used to adjudicate the efficacy of treatments and healthcare systems, and the same constituency which flags poor outcomes when comparing the US healthcare to Sweden’s asks that these outcomes not be used to assess the efficacy of Medicaid. I agree with them as strongly as I disagreed with their use of life expectancy to judge American healthcare.

Disagreements are common because economics is not a hard science such as physics. It does not so much get us to the objective truth as it does to the action at the margins through methodology that is not as robust as the physical sciences, yielding different results on different occasions.

Who is correct, Reinhardt or Pipes?

In a sense both.

Reinhardt is right. Medicaid recipients are not the same as those enjoying ESI.

Pipes is right. Medicaid has structural issues. It pays physicians too little compared to ESI.

This begs the question which reimbursement corresponds to the fair market price in health care: Medicaid or ESI. We will never know because health care has not operated as a free market, and never will. And ESI does distort the price signals as do mandates and virtually everything else.

But here is the important point: ESI is going nowhere. Neither the most left-leaning Democrat nor the most right-leaning Republican has the courage to rid health care of ESI.

What’s the objective truth? Which system really is more efficient?

The truth lies in the answer to this: Would ESI deliver the level of care enjoyed by ESI recipients with paucity of cost sharing that Medicaid recipients face to Medicaid enrollees at a lower cost than Medicaid?

For Medicaid recipients cost sharing should be zero otherwise it defeats the purpose of a safety net. But remember we want them to have the same level of care as ESI for a true apples-apples comparison.

It’s practically impossible to conduct a randomized controlled trial to answer this question. Nor does empiricism suffice. All quantitative analyses have assumptions. With regards to assumptions I can do no better than paraphrase Groucho Marx: “Those are my assumptions, and if you don’t like them … well I have others.”

Importance of disagreements

The current system does not have many genuine alternatives. Single-payer is out as is a genuine free market. As politicians don’t wish to talk about costs because of political expediency, all we are arguing about is which part of health care has the most administrative cost/ informational loss. This is at best a marginal argument. To resolve this argument I would encourage more dialectic between partial truths.

But if Medicaid truly is a high risk pool, and I believe it is, then it should be treated as the other high risk pool – Medicare. Which means that the poor and sick, the uninsurable, should be covered by the Federal government through general taxation. I would suggest a “Medicare for the Poor” which offers the same benefits as traditional Medicare. This would allow the states to balance their budgets better and concentrate on local infrastructure, such as parks, police and public libraries.

Summary of key points

  1. It’s more cost-efficient treating healthier patients.
  2. Accurate adjustment for comorbidities and social determinants of health is key for any comparisons in health care. This is (never) seldom achieved.
  3. There’s interdependence between employer-sponsored insurance and Medicaid.
  4. No one knows true market prices in health care because it’s not a free market.
  5. Economic analysis yields information about the margins, until the next analysis.
  6. The poor should be covered by the Federal government through general taxation.

About the Author:

Saurabh Jha is a contributing editor to THCB. He can be reached on Twitter @RogueRad

12 replies »

  1. Many good points and of course relevant to the current political debate, since part of what Rs wanted to do in BCRA is shift people out of Medicaid and into ESI and exchanges. Famously, the Obama administration did a quote a bit of modeling/analysis on Medicaid vs exchanges for low income people and concluded Medicaid was cheaper–for some of the reasons you cite.
    For better or worse (mostly worse) we have a patchwork hybrid gov/private system. Medicaid right now is indispensable and the Rs were dead wrong in AHCA and BCRA to try and gut it. That said, Medicaid ain’t perfect and long-term solutions are needed…which would pay providers more for primary care etc.
    The Rs in advance of AHCA and BCRA did no analysis, held no hearings and near as anyone can tell never talked to a range of outside experts on the impact of their attempt – now dead — to shear Medicaid of 35% of its projected funding over the next 20 years.

  2. Fifty experimental factories all looking at the other 49 in order to improve their own.

  3. That also seems to be coming at the local level. I am peripherally involved in a project where we are looking at adding more social workers and partnering with our local churches to handle those kinds of patients rather than hiring more doctors and nurses.


  4. Moving the Medicaid responsibility to the federal government is worth looking at. This would also eliminate dual cost report submissions for Medicare and Medi-Cal (California) and also eliminating the Medi-Cal cost report audit. Annual cost reporting and the subsequent annual cost report audits are a large administrative burden on hospitals in addition to many other government reporting that must be done which adds to the cost of health care.

  5. You are correct, Steve. ESI is not a free marketplace because it involves third party payers. It is hard to reduce costs with such an arrangement and ESI dumps the sickest into the self pay insurance group which raises their rates and makes everyone in the self pay group look like a high risk.

    We have to go back to what Pauley says. Insure the healthy and the unknown sick while subsidizing those that are sick and require financial help. Most of the unknown sick when ill will remain in their pools.

    I always get into a fight with my good friend Barry pointing out that I don’t want subsidies to help supplement someone’s golf course fees. That is what we do today even when treating the poorest and most needy. We add to that group healthy, more educate and better off people without the infrastructure to treat everyone. Who gets hurt the most? The least educated, poorest and most needy.

  6. I’m skeptical about state level innovation also. There might be some potential to minimize ER visits among diabetics and CHF patients by using intensive case management, including social workers, to assess needs in the home and follow up as needed but that effort doesn’t come cheap either.

  7. Has any state really ever found that more efficient way to deliver care? It looks to me like more efficient just means they find ways to deny coverage. (Am skeptical about the Indiana plan.)


  8. Good, Steve2.

    Medicaid is so much different from ESI that one cannot really compare. EG. 40% of Medicaid’s budget goes to Medicare patients who need help with premiums, copays and deductibles. Two thirds of Medicaid’s budget is spent on seniors or the disabled. Go to Matt Salo on YouTube. He is CEO of the National Association of Medicaid Directors.

  9. I think federalizing Medicaid would be a good idea for several reasons. First, if reimbursement rates were brought into line with Medicare rates, more doctors would presumably be willing to see Medicaid patients. Second, alignment of documentation requirements for providers should ease their administrative burden. Third, to the extent that the feds could invest more to combat fraud, federal taxpayers would reap the benefit whereas the current split payment responsibility between federal and state governments discourages either from investing in better data analytics.

    The downside, of course, is that the federal government would have to find $200-$300 billion per year more in tax revenue to finance a federal takeover of Medicaid. Also, we would lose the opportunity for innovation at the state level to find more efficient ways to deliver care.

    For the states, offloading the responsibility to pay for a portion of Medicaid costs would free up an enormous amount of money that could be used to fully fund their unfunded retiree pension and health insurance liabilities, modernize under maintained infrastructure and maybe even increase funding to localities so they could reduce the hated property tax which must be paid in cash whether the homeowner has sufficient income or not.

    As Steve2 notes, a huge portion of Medicaid costs pay for long term custodial care in nursing homes and home healthcare. The program now also pays for nearly half of all births in the U.S. A disproportionate share of mental illness, alcohol and drug abuse is likely concentrated within the low income population. Trying to compare Medicaid to ESI is ludicrous in my opinion.

  10. Medicaid is so much different than ESI not sure you can really make that good of a comparison. A big chunk of Medicaid is long term care for the elderly. Another big chunk is OB costs for the poor. Yet another chunk is for chronically disabled. I can certainly agree about the low pay as we get zero (0) dollars for taking care of mentally challenged patients in the OR for dental care.


  11. Then ESI is not real insurance by your definition. We, like all the employers of which I am aware, charge every employee the same and pay the same for them. They all get the same benefits. (We have a menu of choices, but everyone who picks the same product is treated the same.)

  12. “This means that young fit joggers are subsidizing the costs for the unfortunate child’s complex cardiac surgery.”

    There is nothing wrong with insuring the healthy, unknown sick and mildly ill patients in a free marketplace while subsidizing a a healthcare program based upon need that the individual can buy out of.

    “Insurance is redistribution.”

    Social programs are redistribution. Real insurance is risk based.