http://www.nejm.org/doi/full/10.1056/NEJMp1615696?query=featured_home&

Medicaid’s introduction also generated large benefits. Medicaid reduced mortality among infants and children, provided financial protection for their families, and led to better health, higher employment, and lower use of public benefits when they grew up. Moreover, by increasing tax revenue and reducing cash transfers, Medicaid currently saves federal and state governments $21 billion per year.5

How do these historical policies compare with today’s Medicaid-reform proposals? Ryan’s proposed caps apply only to Medicaid spending and recipients, since Medicaid was long ago decoupled from cash welfare. The cap amounts would initially equal average 2016 Medicaid spending by eligibility category and by state, rather than a single statutorily defined amount. Yet the caps would be “set to grow more slowly than under current law,” so over time they cease being related to actual Medicaid costs, thereby limiting the ability of states to adjust to rapid advances in technology, epidemics, or other unforeseen events. Nevertheless, as in the 1950s, discouraging Medicaid recipients from receiving costly care or keeping the highest-cost patients out of the program would be the clearest ways to limit state outlays. Toward that end, the Ryan plan would allow states to impose work requirements, charge premiums, offer a limited benefit package, shift beneficiaries into the individual insurance market, and create enrollment caps or waiting lists.

Medicaid creates a divisive relationship between the federal and state governments. Federal mandates and open-ended federal cost sharing are meant to provide incentives for state spending, but states often balk at the large costs. Both state and federal budgets would benefit if each Medicaid recipient cost less. Unfortunately, a per capita cap on federal Medicaid spending is unlikely to achieve this aim. Rather than “modernize” Medicaid, the historical experience in the United States suggests per capita caps would simply shrink the program.