Hilda Abraham and her daughters, Jailyn Izabella Lliguichuzhca and Jocelyn Abraham, 23, all of Dumfries, Va., receive information about the new health-care law. (Jahi Chikwendiu/The Washington Post)

Healthcare.gov’s troubled rollout brings new attention to competing proposals to alter health coverage for low-income Americans. One group of Republican analysts would largely replace Medicaid with  some combination of primary care supports and catastrophic coverage. Others wouldn’t go that far but would make greater use of patient cost-sharing to provide incentives for disciplined use of medical services.

Ross Douthat has put the argument well in some thoughtful columns.

[C]omprehensive health insurance is, at its heart, a deeply wasteful use of resources: Modern people, and especially modern Americans, are much more likely to overconsume health care than to underconsume it…  This doesn’t mean that social insurance shouldn’t protect people against adverse medical outcomes and unaffordable medical bills, but it suggests that there are better ways to allocate our resources than comprehensive coverage, and that most people would be better off if public policy didn’t push so much money into that direction.

Economic intuition suggests that Harold Pollack and Ross Douthat alike would use care more efficiently if we had more skin in the game, if we were less comprehensively insured. Sure enough, non-poor participants in the RAND Health Insurance Experiment used roughly one-third less care when they were enrolled in something akin to a catastrophic plan than did their counterparts who were enrolled in more generous plans. Despite their reduced service use, participants in the catastrophic plan also appeared, on average, to be just as healthy. Such findings provide a powerful argument for catastrophic plans. (Recent results from the Oregon Medicaid experiment are another matter.)

The cracks in this argument become more noticeable when one shifts attention from the typical insured person to the typical insurance dollar spent for patient care, particularly when one considers the vulnerable populations overrepresented among public insurance recipients.

Most people are light users of medical care. As Aaron Carroll regularly emphasizes, expenditures are concentrated within a group of costly patients with complex conditions, many of whom would hit any reasonable catastrophic coverage cap and who are poorly placed to manage the practicalities and financial risks associated with their medical care.

Although I’m curious to see how healthy, affluent professionals would purchase (say) knee and hip replacements if they faced the full costs, this won’t save much money. Most people in a position to make such choices don’t consume much health care. The real money is spent on patients like my relative who is recovering from a nasty stroke. For all sorts of reasons, giving him something other than comprehensive coverage seems unwise.

These distributional realities hold especially true among poor people. Figure 1 shows 2012 data for Illinois’s 3.2 million Medicaid recipients, ranked by percentile from lowest to highest expenditure. The top line shows cumulative Medicaid spending.  The bottom bars show average annual expenditures in dollars. (If you look closely, you’ll notice straight lines where I interpolated between available data points.)

The bottom 72 percent of Illinois Medicaid recipients account for 10 percent of total program spending. Average annual expenditures in this group were about $564, virtually invisible on the chart. We can’t save much money through any incentive system aimed at the typical Medicaid recipient. We spend too little on the bottom 80 percent to get much back from that. We probably spend too little on most of these people, anyway. For the bulk of Medicaid beneficiaries, cost control is less important than improved prevention, health maintenance and access to basic medical and dental services.

The real financial action unfolds on the right side of the graph, where expenditures are concentrated within a small and incredibly complicated patient group. The top 3.2 percent of recipients account for half of total Medicaid spending, with average expenditures exceeding $30,000 annually.

Many of these men and women face life-ending or life-threatening illnesses, as well as cognitive or psychiatric limitations. These patients cannot cover co-payments or assume financial risk. In theory, one might impose patient cost-sharing with some complicated risk-adjustment system. In practice, that is far beyond current technologies and administrative capabilities. Even if such a system were available, we couldn’t push the burden of medical case management onto these patients or their families.

The Affordable Care Act's Medicaid-expansion population won’t include traditional dual-eligible and nursing home patients who make up much of that 3.2 percent. The new group will still include a mix of basically healthy adults alongside more costly recipients with complicated problems who will account for the bulk of actual spending.

In my view, the best way to improve Medicaid is to tackle three inter-related challenges.

First, states lack the fiscal capacity to implement this program well. The ACA’s framers were wise to have the federal government assume virtually the entire cost of expanded Medicaid eligibility under the law. I would have the federal government assume an even larger share of the costs, particularly for the most costly and complicated subgroups such as disabled individuals who are dual-eligible for Medicare and Medicaid.

Second, Medicaid is underfunded. Medicaid-funded care is the most constrained segment of the entire medical economy. Per-recipient cost growth has actually been modest. Projected increases in Medicaid expenditures are quite manageable when compared with the projected growth in Medicare or in expenditures for care provided through private insurance. Many of Medicaid’s most egregious flaws could be addressed by spending more money. We should increase spending in some key areas, including care coordination and adult dental care.

Third and most fundamental, Medicaid recipients (and potential recipients) are politically marginal, so they are poorly placed to tackle the above two challenges. To take the most obvious example, both Medicare and Medicaid squeeze providers. This is part of the game of constraining spending. Yet if Medicare policymakers squeeze too hard, they will face a  political backlash among seniors experiencing access problems. The political feedbacks are much weaker in Medicaid, with predictable results. Many conservative policy wonks lament Medicaid-related access problems. At times, Medicaid critics exaggerate these problems. As Jonathan Cohn and the GAO remind us, consumer satisfaction is notably higher among Medicaid recipients than among individuals enrolled in private individual or small-group plans. Depressingly few of Medicaid’s critics take the next step and press states to raise provider payments or to take other steps to make Medicaid a more attractive payer.

Medicaid’s shortcomings can be mutually reinforcing. State government’s ambivalence and fiscal incapacities aggravate underfunding. Underfunding erodes quality and thus erodes political support for required expenditure.

If we want to provide more cost-effective care to poor people, we should proceed in the same way that we should proceed in other parts of the medical economy. We must do the hard work of improving the quality and economy of care provided to the concentrated group of extremely costly patients. There is no short cut. Under any financing system, this requires the hard work of clinical-care coordination, quality improvement and social services to address life circumstances that undermine health.

It’s unrealistic and unwise to expect that financial pressures on or incentives for patients will play a large part in this work. Of course, patients and their families should be effective partners. Giving patients more skin in the game may sometimes help control costs at the margin. To expect more than that -- especially when we’re talking about Medicaid and covering the uninsured - -mainly distracts us from the real work that must be done.