Perhaps the following passage from chapter 7 of Goldhill’s Catastrophic Care is worth a conversation.
[H]ealth care experts attribute some of the rise in private insurance premiums to cost shifting, by which hospitals require private insurers to bear some of the costs of public patients not fully reimbursed by Medicare and Medicaid.
It’s an interesting (and endless) discussion, but there’s only one problem: none of this makes any sense. Hospitals lose money on their huge number of Medicare patients? Then why is it that more hospitals have opened than have closed each year for the past decade? Medicaid beneficiaries are unable to find providers? Then how has the number of Medicaid treatments skyrocketed? Private insurance rates have to bear the cost shifting from Medicare and Medicaid patients? Why would private insurers agree to pay for someone else’s patients?
It may be true that many health care insiders obsess over these phenomena, but that doesn’t mean they really exist across the board.  [E]xperts assume that the health care industry has responded to the three-price system by merely accepting its objectives, by simply pricing the same service at three different levels. But here’s a more plausible view: the industry has responded to the three prices by offering three different tiers of service, with each one structured to be profitable to the industry.
At first I thought the hypothesis was that a given practitioner treated Medicare, Medicaid, and privately insured patients differently. Many physicians have told me this isn’t plausible, that they don’t consider (or even know) payment source when treating a patient. But, reading further, what Goldhill really means is that providers sort themselves into those that (mostly) will or won’t care for each type of patient. A Medicaid doctor is not the same as a private pay doctor, etc. They treat patients differently because they have different business models. One can’t do the same for patients at private pay rates as one can at Medicaid rates, for example. That’s both plausible and would lead to different tiers of service for different types of patients.
It is hard to imagine any system remotely acceptable to Americans that doesn’t lead to stratification like this. After all, my lifestyle is not the same as the poorer family in the next town over or the richer one down the street. I suppose the point is, to what extent does the driver of stratification distort behavior relative to some theoretical ideal (e.g., what might exist under perfect competition)? A follow-up would be to acknowledge that the theoretical ideal cannot exist. (This is always true.) So, a more relevant question would be, to what extent does the driver of stratification distort behavior relative to the best, feasible option?