From 1975 to 2015, social spending by federal and state governments quadrupled in constant dollars, to more than $1 trillion. America now spends enough to give every person in poverty more than $20,000 per year. And yet lamentations for a collapsing safety net are rising in both volume and pitch.
In the past year, an essayist for The Economist told of how, after the 1970s, “the welfare state ceased its expansion and began to retreat.” A staff writer for The Atlantic warned that the net is “thin and getting thinner,” the New York Times’s economics columnist bemoaned a “tightening of the screws,” and The Week titled a column “The Grotesque Injustice of Starving 1 Million Unemployed Americans.”
Because of Medicaid, the safety net feels weaker as it grows heavier, exposing bigger gaps even as it spreads. Bad design and political pressure have allowed this one program to dominate our ever-expanding anti-poverty system. Medicaid now accounts for most of what we spend to aid the poor, even though the program aligns badly with the needs of low-income households and offers stunningly little value for its cost. The opportunity to improve support for the poor without increasing spending by reallocating funds from Medicaid to better uses is great.
This should be an obvious area of bipartisan compromise. While the battle rages over total spending levels, all sides should want to maximize the results each tax dollar produces. Instead, virtue is too often measured by who will write the largest check, and questioning Medicaid is equated with callous indifference to the poor.
As is typically the case with misallocated resources, removing distortions and allowing choice would offer a remedy. If the federal government eliminated the incentives that reward Medicaid spending over other anti-poverty strategies and gave flexibility to states, localities, and even individual households to direct anti-poverty resources, the trillion-dollar safety net could accomplish much more than it does now.
The best way to understand the safety net’s evolution is to track total government spending aimed at low-income households relative to the total number of people in poverty. Social Security and Medicare don’t count, because they are “earned” entitlements paid largely independent of the recipient’s income.
Of course, not every safety-net dollar reaches someone below the poverty line. But the population in poverty gives us an idea of the level of need in society against which to judge the resources society dedicates to that need. When social spending increases in response to an increasing number of people living in poverty, the safety net has widened but there has been no increase in support per person. When social spending increases faster than does the number of people in poverty, the safety net has thickened and each person can receive more support.
In the past 40 years, the safety net has both widened and thickened dramatically. The population in poverty and the amount of spending per person both doubled; total spending quadrupled. The extraordinary thickening — from $12,000 per person to $23,000 — consumed a bonanza of resources that should have been sufficient for an effective strategy to help low-income households move out of poverty. Instead, more than 90 percent of the increased spending per person went to health care — almost entirely Medicaid, for which total annual spending rose from $55 billion to $568 billion (all figures are in 2015 dollars).
That allocation of resources might make sense if health care were the top priority for low-income households, if it were the keystone of economic opportunity, and if Medicaid delivered it well. None of those things is true. Obviously, low-income households without access to Medicaid would not allocate 90 percent of their own resources to health care, or even 50 percent. In fact, households that consumed $10,000 to $20,000 in 2012 and were ineligible for Medicaid or chose not to enroll allocated only 8 percent of their spending to health care. Housing accounted for 42 percent, food for 24 percent, transportation for 10 percent.
Nor did households that consumed slightly more prioritize health care when allocating their additional resources. Households with $20,000 to $30,000 of annual consumption spent only 9 percent of their additional funds on health care. This low priority for health care continues further up the income ladder, where it has frustrated implementation of the Affordable Care Act: Even among middle-class households eligible for subsidies to buy their own health insurance, the majority have chosen to remain uninsured.
But might it be that government knows best, wisely allocating funds to health care that low-income households lack the foresight or discipline to allocate well themselves? It would seem not. Studies repeatedly find that Medicaid recipients not only achieve worse outcomes than do people with private insurance, but they also achieve worse outcomes than the uninsured do. In a randomized controlled study in Oregon in 2008, uninsured residents were assigned to receive Medicaid or not; the study concluded that “Medicaid coverage generated no significant improvements in measured physical health outcomes in the first 2 years.” A subsequent analysis found that each dollar of Medicaid spending generated only 20 to 40 cents of value for the recipients based on what they showed a willingness to pay for and how their health was affected.
Critics of the Oregon study complain that it was “underpowered,” meaning that it followed too few individuals for too short a time. (Such concerns were few and far between when the study’s preliminary results hinted at a rosier outcome.) But longer-term studies have corroborated the finding. In April, for instance, Stanford’s Raj Chetty and his colleagues published a study in the New England Journal of Medicine that found no significant relationship between health-care access and life expectancy for the poor in a given area.
Asked that same month by Vox’s Dylan Matthews about the efficacy of public health-care subsidies for low-income households, Jason Furman, chairman of the White House’s Council of Economic Advisers, offered three defenses, each less coherent than the last. First, he said, was health: “The Oregon study gave us pretty clear evidence of certain health benefits, both self-reported, how people felt about their health, and in areas like mental health.” Fair enough, though defining Medicaid’s primary value as helping people feel better about their health is faint praise.
Second, he said, “It’s a way to get more money or more resources in the hands of people [who] need it, addressing the inequality that we were talking about before, the progressivity.” This is no defense of health-care spending at all. It argues for a generous safety net, but not for emphasizing one type of benefit over another.
Third, Furman made the following, almost incomprehensible statement: “Giving someone a dollar versus giving them insurance. If they really need it, they get a lot more than a dollar and if they didn’t, maybe they don’t get anything is more valuable even if the average cost is just a dollar than giving someone a dollar.” He might mean that insurance coverage with an expected value of one dollar appears wasteful to those who don’t get sick, but that one dollar is very valuable to those who do get sick; so, overall, the insurance is worth more than one dollar.
If this is what Furman means, the claim is not correct. Presumably insurance with an expected value of a dollar is worth about the same as a dollar and, given the choice, an individual would rather receive a dollar with which to buy the insurance or anything else than automatically receive the insurance. More important, the argument begs the question. Yes, if the insurance is worth a dollar, then the insurance is as good as a dollar. But the evidence suggests instead that a dollar spent on Medicaid is worth far less to the recipient than a dollar. Under those circumstances, one cannot defend it by suggesting that maybe it is worth a dollar after all.
The missing link in Furman’s logic is opportunity cost — what else could a dollar spent on Medicaid have achieved? If Furman is defending Medicaid against outright cuts, not against reallocation to better uses, he begins to make more sense. Something, he believes reasonably, is better than nothing. Yet that reasoning underscores the real flaw of the Oregon study, and of Medicaid boosterism generally, which is that Medicaid gets compared only with no Medicaid. A study that better reflected the trade-off facing policymakers would assign Medicaid to half the participants while giving the other half housing vouchers and wage subsidies of comparable cost, which might promote not only better health but greater economic opportunity as well. Would Medicaid then look like an effective strategy, even for health outcomes alone?
Researchers from the Yale School of Public Health attempted to answer that question by comparing health outcomes in each state to that state’s ratio of health-care spending to other social spending. Sure enough, as they explain in the May issue of Health Affairs, the states allocating a smaller proportion of their social spending to health care achieved better health outcomes on measures from obesity to mental health to mortality.
None of which suggests that Medicaid should be eliminated. Health care obviously has value and belongs in the suite of government services for the poor. But it is only one support among many, with no special claim to improving health — let alone delivering the economic opportunity that should form the core of an effective anti-poverty strategy. Understanding how this program came to dominate the safety net is the first step toward restoring balance.
The problem with America’s safety net is less that it’s an ineffective anti-poverty strategy than that it’s no strategy whatsoever. Its major programs were created at different times by different laws, they are run by different agencies with different funding structures, and each new addition is simply piled atop those that came before. Its thickening looks more like an accumulation of random clumps and tangles than careful additions of support.
Medicaid became the center of gravity by an accident of badly designed incentives. The federal government establishes a minimum level of mandatory Medicaid benefits that each state must provide, but it also invites the states to expand from there, providing additional services to additional groups of recipients. Each state must pay a share of these additional costs itself, but for every dollar it spends, it receives matching federal funds — one federal dollar per state dollar for wealthier states, but a ratio as high as five to one for poorer states.
Unsurprisingly, states have responded by massively expanding their Medicaid programs, bringing in the federal dollars. These new programs — which are above and beyond the minimum Medicaid required by federal law — now account for most of Medicaid spending nationwide. Even in those states that refused the optional expansion of Medicaid under the Affordable Care Act, most Medicaid spending is already optional.
In theory, a state should continue expanding its Medicaid programs until the next dollar it spends produces less than a dollar of value for the recipient (or, given constraints on total spending, less than the dollar could produce if spent otherwise). But a state dollar spent on Medicaid brings with it the value of the matching federal dollar or dollars, too. So a state will act rationally by continuing to expand Medicaid long after its dollar returns less than a dollar, and in most cases even after it returns less than 50 cents.
For instance, a state that receives two federal dollars per Medicaid dollar spent will pursue an expansion that is worth only 40 cents on the dollar, because each dollar it spends will trigger three dollars in total spending and thus $1.20 of value. This will be the rational choice even if it knows that some other use of the money (unmatched by federal funds) would be twice as valuable. The federal government might also know that the state’s allocation is senseless, but it has tied the leash around its own waist and is now along for the ride.
Consider the Oregon experiment in this context — specifically, the finding that additional Medicaid spending yielded approximately 30 cents of value on the dollar for recipients. At that time, Oregon received a federal match of $2.65 per state dollar spent. So to generate a dollar of spending, the state had to commit only 27 cents. Of course, it expanded Medicaid to the point where each dollar spent yielded only 30 cents in value. And while the rest of the money may not have done much good for the poor, it still sloshed lucratively through the state’s medical system.
Checks that might otherwise constrain spending are not present for Medicaid. As an entitlement program, it gets fully funded each year regardless of how high spending goes. (A program such as Section 8 housing vouchers, by contrast, is limited by what Congress appropriates; currently only 25 percent of eligible households receive the benefit.) Meanwhile, the unique emotional salience of health care makes any suggestion of cuts politically perilous.
Perhaps health care has achieved special status because it is a universal worry, and one whose costs can seem imposing even to higher-income Americans. Perhaps, as suggested by a paper published in May in the American Journal of Political Science, people across cultures and of varying political views perceive patients as less responsible for their plight than a group such as the unemployed and therefore as more deserving of assistance. Perhaps it comes down to storytelling: The news report of a single transplant denied delivers a harder punch than the one about 50 families living in substandard housing and consigned to hours-long commutes.
Those reasons might explain the misallocation of resources, but they don’t justify it. Kowtowing to the illogic and to the accompanying accusations that reform will “abuse poor people and particularly black people” — as City University of New York professor David Himmelstein characterized block-grant proposals in a recent CUNY broadcast — only hurts the low-income households that must rely on a less effective safety net as a result.
The key to moving forward is to focus on opportunity cost: the fact that we are accomplishing less than we could. Reallocating money to education, transportation, child care, housing, and wage subsidies will surely do more to alleviate financial distress and will probably improve health outcomes as well. Further, a safety net oriented toward such programs would provide far greater economic opportunity, helping people rise out of poverty and rely less on government support over time.
Medicaid is the largest pool of safety-net resources and the most obviously misallocated one. But it is far from the only one. The same pattern of uncoordinated agencies, distorted incentives, and ineffective programs repeats itself throughout the safety-net programs. And the same kinds of reforms that would allow Medicaid’s funds to flow to better uses would undam other areas as well.
The solution is not to cap and cut Medicaid through a traditional block grant or to reassign its dollars to other federal programs. Neither approach lets resources find their best use. Rather, the goal should be to eliminate the distortions that channel funds poorly and to remove the barriers that hold misallocations in place. Stop telling states they get a bigger reward for one use of funds (health care) than for another; and stop offering other federal funds through a defined set of creaky federal programs on a take-it-or-leave-it basis. Safety-net dollars will stretch further when states have the freedom to allocate resources to existing programs or use them for entirely different programs of their own design. No state should be forced down any new path: If you like your current plan, as the saying goes, you can keep it. But states that think they can do better should be encouraged to try.
The most aggressive reform would be a Flex Fund, as I described in these pages in October 2013. Such a fund would consolidate each state’s share of federal social spending as a lump sum without strings attached. The federal government would play the role of tax collector and redistribute resources to poorer states, but the day-to-day role of safety-net provider would fall squarely to states and localities. That model should be the long-term goal, but two other, more incremental steps could be a start.
One option, focused on Medicaid specifically, would be new waivers that allow states to repurpose Medicaid dollars not only to other health-care programs (an option available today) but also to other programs entirely. For instance, each state could choose to shift up to 25 percent of its Medicaid funds into an expansion of the Earned Income Tax Credit (EITC) to subsidize the earnings of low-income households. President Obama and Speaker Ryan both support an expansion of the EITC that would cost approximately $6 billion, yet the proposal has stalled over whether to pay for it with a tax increase or with funds taken from another program. Surely all sides can agree that the safety net would be stronger with that expansion in place and Medicaid spending 1 percent lower. If not, could they at least agree to let states make their own choice in the matter and learn from the result?
A second option, offering states wider flexibility, could replace the distortionary matching of Medicaid funds with a “universal match” that rewarded states with a comparable ratio of federal dollars for whatever anti-poverty initiatives they pursued — if a dollar spent on Medicaid earned two federal dollars, so should a dollar spent on child care. (For this reform to be budget-neutral, the ratio’s level in each state would have to be lower than its current Medicaid ratio.) The best use of each dollar, not varying ratios of federal largesse to state spending, would define the rational allocation of resources. This mechanism could operate solely within the existing federal programs, or it could apply to matching funds for state-led initiatives as well.
If only half the safety net’s increased benefits per person over the past 40 years had gone to health care, instead of more than 90 percent, Medicaid would still be its largest component. But annual spending on other anti-poverty efforts could be $200 billion higher, enough to provide every household in poverty with options such as child care, a car, a housing voucher, a subsidized job — in other words, a real chance to move out of poverty, not just to survive another year of it. And for those whose most pressing need actually is the health-care access that Medicaid offers? We could still provide that, too.
– Mr. Cass is a senior fellow at the Manhattan Institute and the author of the recent report “Over-Medicaid-ed: How Medicaid Distorts and Dilutes America’s Safety Net.”