Medicaid, the government program that provides health insurance coverage to low-income and disabled Americans, is the largest payer for health care in the United States in terms of enrollees and the second-largest payer (behind only Medicare) in terms of spending. Escalating health care costs, a growing federal budget deficit, and fiscal challenges in many states have led to calls to reform the program to decrease spending growth. Recent federal reform proposals from House and Senate republicans would change the current financing system in which the federal government guarantees a share of total program spending to states to one limiting federal cost exposure by setting a per capita cap on federal payments to a state.

A change in the Medicaid program to a per capita cap financing system is included in the House-passed American Health Care Act (AHCA) and in the Senate-proposed Better Care Reconciliation Act of 2017 currently under consideration. With the Congressional Budget Office estimating that the Medicaid proposals in the AHCA will cut federal Medicaid spending by 25 percent by 2026, much attention has been given to the effects of such cuts on decreasing the number of individuals enrolled in Medicaid and increasing state budgets. Much less attention, however, has been given to a related but critical question: How do the reforms affect who enrolls in and gets care under Medicaid? From the lens of economics, we draw an analogy to per capita payments in health insurance markets and explain how the currently proposed reforms threaten the core programmatic purpose of Medicaid by incentivizing states to limit care and coverage to the states’ most vulnerable residents.

Federal funding for Medicaid creates a national safety-net health insurance program. Without federal funding, one might expect a classic “race to the bottom” among states to reduce state spending (and the accompanying taxes) by weakening their Medicaid programs. Federal policy for Medicaid prevents the race to the bottom by conditioning funding on both state spending and on the fulfilment of certain safety-net requirements, such as eligibility for statutory categories of individuals and benefit and access requirements.

But while the need for federal funding is clear, what form should it take? Is it better to share costs at the margin or limit federal exposure with a per-person cap? Currently, the federal government pays a percentage of every program dollar, the Federal Medical Assistance Percentage (FMAP). FMAPs are determined by a formula incorporating the ratio of the state’s per capita income to national per capita income, to ensure low-income states get higher FMAPs. For example, the FMAP for Connecticut is 50.0 percent (the FMAP floor), and the FMAP for Mississippi is 75.7 percent. This effectively results in the federal government giving states at least a $1 for every $1 of state Medicaid program spending. The federal government pays the FMAP regardless of the previous year’s spending or budget projections, paying less when the program spends less than projected and paying more when the program spends more.

Under the House-passed AHCA, the mechanism by which the federal government would share in program costs would change dramatically to a per capita spending cap. Under such a system, the federal government would pay a maximum preset amount for each Medicaid enrollee, regardless of actual spending in a given year (Note 1). These payments would vary across five broad Medicaid eligibility groups, with five different per capita payments (Note 2) based on historical within-group-spending averages. If a state spends more than the total cap, the federal government provides no additional funding and thereby shifts all spending growth risk to states. With a new set of financial incentives and with less federal reimbursement at stake, states would need to determine how to make up for the risk of future funding losses. To do so, policy experts consistently cite that states could either increase their current funding commitments, cut payments to health care providers and health plans, eliminate covered health care services, limit eligibility for enrollment, or (where possible) make the program more efficient. But a per capita cap system also provides strong incentives to states to pursue programmatic policies, many of which would not require legislative changes, to change the state’s enrollment mix toward heathier individuals.

Per Capita Caps Alter State Economic Enrollment Incentives

To think about how the new incentives in the per capita cap system affect who enrolls in Medicaid, it is useful to draw an analogy to plan payments in health insurance markets. Think about the state as an insurer. Under the current structure of the program, states are assured a percent reimbursement for every dollar they spend on Medicaid. Under this system, the additional cost of enrolling a sick individual is shared between the state government and the federal government, dampening the state budgetary impact of enrolling particularly high-risk individuals, in a given year.

In contrast, consider the incentives inherent in a federal per capita spending cap system. Similar to a (partially) community-rated health plan premium, under this system states receive additional per capita payments with higher enrollment, but the risk of spending above the per capita allotments falls squarely onto the state (Note 3).

These two dramatically different funding mechanisms alter incentives to states when it comes to the question of who enrolls and receives care under Medicaid. As with a community-rated insurance market, it is straightforward to see that under per capita caps states have a much greater incentive to avoid enrolling less healthy individuals within a given payment category because, instead of sharing the added costs of a sick individual with the federal government, 100 percent of the additional costs of the sick individual must be paid by the state.

Consider the question of enrolling a healthy young adult or a sick older (but still younger than age 65) adult in the same eligibility category. Under the per capita proposal, the state much prefers to enroll the young healthy person, given that enrolling this person generates a federal payment that more than covers the cost to the state of enrolling him or her, while enrolling the older, sick person generates a federal payment equal to the cap, far below the cost imposed on the state. Additionally, enrolling the healthy adult helps insure against high-cost spending in other categories. In contrast, under the current financing system, states are less concerned about whether they enroll sick individuals because while these individuals are more expensive for the state to enroll in Medicaid, they also generate larger federal payments.

Insurance markets with community-rated premiums such as the Medicare Advantage program provide a useful analogy to this problem. In these markets, insurers face incentives similar to the incentives faced by states under per capita spending caps. For many years, Medicare Advantage insurers faced a strong incentive to enroll healthy Medicare beneficiaries, and in spite of “open enrollment” provisions, analogous to the state having to accept otherwise qualified Medicaid applicants, plans succeeded in drawing a healthier case mix. Only recently with improvements in risk adjustment of plan payments, limits in plan switching, and other reforms, has the selection problem been substantially reduced. The proposed Medicaid per capita payments would come with none of these aggressive enforcement efforts.

It is not hard to envision Medicaid, under a per capita spending cap system, working as poorly as the early version of Medicare Advantage. While states are not for-profit insurers, as they grow to understand that it is financially difficult (or even impossible) for them to enroll a disproportionate number of sick individuals in Medicaid, they will likely shift enrollment efforts to less “risky” groups, such as the healthy and the young. States may be less eager to enroll or reenroll sick Medicaid recipients. Such incentives could, for example, manifest in the form of fewer enrollers at and less streamlined enrollment procedures for safety-net hospitals. At the same time, states may be much more likely to send enrollers to schools or community health centers where they can find young relatively healthy children and families. Alternatively, states could take a page out of the Medicare Advantage playbook and modify the services they offer to make them more attractive to young families and less attractive to older, sicker individuals. This could be done fairly easily even within existing federal benefit rules by boosting payments for primary care while lowering payments for specialty care, especially care for mental health and substance use disorders. Considering the relatively low rates of Medicaid take-up among eligible individuals, such a shift in enrollment efforts or benefits could have a substantial effect on the mix of individuals enrolled in Medicaid.

A key question is whether or not such economic incentives are strong enough for states to act upon. The case seems clear that they are. While per capita cap proposals do specify different payments for each eligibility group, there is enormous variation in spending across and within eligibility groups in a given state that creates obvious “winners” and “losers” from the state’s point of view. For example, the Kaiser Family Foundation reports that while average annual spending per adult in Ohio is $4,498, spending for adults varies widely with per-person average costs of $530 for those in the first quartile to $20,143 for those in the top 5 percent of the spending distribution. Similarly, for children in Oklahoma, average spending is $2,724 per child but varies from $131 in the first quartile to $24,571 in the top 5 percent of the spending distribution. The spread for the elderly and people with disabilities categories are even greater, with, for example, spending varying from $1,051 for those in the first quartile to $116,515 for those in the top 5 percent for those with disabilities in Pennsylvania.

The within- and across-category incentives for a state are clear—a state can limit its risk of breaching the caps by enrolling healthier individuals and avoiding those most in need. It is not a stretch to imagine that states may be enticed to act on these incentives, either directly or indirectly, to reduce their financial risk. States have both direct and indirect methods for selection even in the elderly and disabled populations. First, some states require a separate Medicaid application from Social Security Disability Insurance. Second, but to a greater and more important extent, there are benefits and providers that can be tweaked that will make using such Medicaid coverage more or less attractive to healthy versus unhealthy disabled or elderly individuals.

The incentives are especially stark for individuals with a number of chronic conditions, such as mental illness. Individuals with mental health and substance use diagnoses, particularly those with serious mental illness or an opioid disorder, represent some of the most expensive Medicaid recipients. This holds within each of the eligibility categories. Under a per capita spending cap scheme, states will thus clearly have an incentive to avoid enrolling these individuals. This may be fairly straightforward for the states to do. States could easily limit enrollment of the mentally ill by making enrollment more difficult at psychiatric facilities, limiting payments for mental health services even lower than they already are and driving more mental health providers out of Medicaid or lowering payments to specialty behavioral health managed care organizations.

Potential Solutions To Avoid Risk Selection

An obvious potential solution to this problem is sophisticated risk adjustment of the federal caps within each eligibility group. The implementation of risk adjustment in Medicare Advantage had important effects on insurer selection incentives and behaviors (Newhouse and McGuire 2014). The risk adjustment program in the Affordable Care Act Marketplaces also improves insurer selection incentives (Geruso, Layton, and Prinz 2017). Sophisticated risk adjustment thus seems necessary for any reasonable per capita caps proposal, yet feasibility challenges, both administrative and temporal, loom large.

Another potential solution might be a middle ground. A “mixed” payment system consisting of a fixed per capita component and a “matched” component could better accomplish the goals of increasing state incentives for cost control while limiting the selection-related issues highlighted here. Under such a system, states would receive a fixed payment per beneficiary but a modified matching payment system, such as the FMAP, would be used for spending above caps to limit state risk. Importantly, such a system would recognize the trade-off between incentives for cost control and selection, and shift the debate away from two extremes and toward the question of how much weight should be given to the fixed component of the payment and how much weight should be given to the variable component.

Medicaid financing reform is difficult and complex. Shifting one set of incentives (for example, cost control) may have important unintended consequences (for example, enrollee mix), particularly in light of the fact that the Medicaid program already has lower per-enrollee costs and slower per-enrollee cost growth than private health insurance. While the potential outcome of a per capita cap system that we highlight above (improved enrollment of and services for the young and healthy segments of low-income and disabled populations) is clearly not a bad potential outcome, if these improvements come at the cost of enrollment of and services for sicker more vulnerable groups, it runs counter to the ultimate goal of a safety-net health care system—to provide access to health care for society’s most vulnerable individuals. These groups already suffer from significant disparities in access to care and health outcomes making it imperative for policy makers to consider the consequences of their actions not only for the number of individuals enrolled in Medicaid but also for who enrolls and the type of care they get.

Note 1

Under the current American Health Care Act proposal, a maximum federal funding contribution would be calculated based on enrollment. However, unlike a block grant, if a state spends less than the cap, the state does not get to keep unused federal dollars.

Note 2

The five categories are the elderly, disabled people, children, newly eligible adults, and all other adults.

Note 3

Dissimilar to a per capita health plan premium, states do not get to keep any federal dollars should they spend less than the federal cap. Under a per capita cap system, states only face downside risk.