Jonathan Ernst / Reuters

What do a Christian overnight camp, abstinence-only sex education, and pro-marriage advertisements all have in common? They’ve all been funded with money that used to provide cash assistance to low-income families.

In the United States, the federal Temporary Assistance for Needy Families program—often known simply as “welfare”—is administered by the 50 states, which have considerable leeway in how to spend the money. The choices states make are unmistakably correlated with race. The higher the proportion of African Americans in a state, the more likely officials are to try to change the way poor families run their lives, rather than simply help them with basic expenses.

Many know TANF as the nation’s primary cash assistance program for low-income families. But depending on which state you live in, TANF may provide barely any cash assistance at all.

In a new study published in the journal Socio-Economic Review, I find that a state with a higher share of black families is less likely to allocate TANF funds toward the provision of cash assistance, but more likely to allocate TANF funds toward efforts to “encourage the formation of two-parent families” and “reduce the incidence of out-of-wedlock pregnancies.” The stated assumption behind these initiatives is that strengthening the family unit has greater long-term benefits than simply giving money to needy people.

In practice, though, the diversion of TANF funds away from cash support and toward programs meant to influence family formation has likely exacerbated racial differences in poverty. A clear pattern emerges: A black family in poverty is more likely than a white family to be offered advice via a “Healthy Marriage Initiative” in place of direct cash support.

These racial inequities in states’ use of TANF funds turn out to have important consequences for racial differences in child poverty. I find, for example, that closing the racial differences in states’ use of TANF funds would narrow the black-white child-poverty gap by up to 15 percent.

Arkansas’s experience epitomizes these findings. The state has a large African American population and no shortage of poverty. Yet in 2017, Arkansas spent only 4 percent of its total TANF budget on cash assistance. Instead, the state allocated two-thirds of the budget for the “formation of two-parent families” and the “reduction of out-of-wedlock pregnancies.”

But Arkansas is hardly alone in spending less on cash support. From the introduction of TANF, in 1997, to 2017, total spending on cash assistance across all states declined from $14 billion a year to $7 billion. (In constant dollars, the amount of money spent on cash assistance has fallen by two-thirds.) While cash support has waned, however, states’ TANF budgets haven’t changed: The federal government provides states the same chunk of money each year to run their TANF programs. Thus, every dollar that a state does not spend on cash assistance should generally be spent on another program or service that, at least in theory, will support low-income families.

So where is the money? In Mississippi, a state that spends comparatively little on cash support, TANF funds have been used to finance an abstinence-only sex-education curriculum that “teaches the social, psychological, and physical effects of engaging in sexual activities.” Officials in Louisiana, where only 4 percent of poor families with children receive cash support, redirect TANF funds toward an Alternatives to Abortion program that discourages low-income families from terminating pregnancies. Southern states aren’t the only ones using the money for moral uplift. In Maine, children were bused off to a TANF-funded Christian summer camp. In Connecticut, TANF money was diverted toward compulsive-gambler assistance.

To be sure, not all of the excess TANF funds are spent on summer camps and sex ed. Many states use the resources to fund child-care programs, earnings subsidies for the working poor, and other efforts to promote or incentivize employment. Some states, such as Georgia, have spent the largest chunk of their TANF funds on child-welfare services and foster-care support. Both are worthy purposes, but other states pay for these services out of general revenues rather than TANF funds.

Meanwhile, fewer and fewer families in need are receiving direct cash support—an essential resource for reducing levels of child poverty.

To understand why a state’s racial composition is the strongest predictor of how it allocates TANF resources, consider that academic research has shown time and time again that many Americans tend to view black families as lazy, unworthy of help, and receiving “more than they deserve” from the state. Whether researchers are exploring Why Americans Hate Welfare, as Martin Gilens did in his 1999 book, or asking “Why Doesn’t the U.S. Have a European Welfare State?,” as Alberto Alesina and colleagues asked in 2001, the answer nearly always begins and ends with evidence of racialized perceptions of the beneficiaries of social assistance. These perceptions have crept into many policy-making decisions, including those relating to TANF. Indeed, my results show that—unlike race—the share of single mothers in the state, a state’s wealth, or which political party has control of its legislature explains little of the variance in states’ cash-assistance spending.

Removing the racial inequities in states’ TANF allocations would mark a large step in reducing racial differences in child poverty. For context, the estimated 15 percent reduction in the black-white child-poverty gap is comparable to the effect of moving all children from single-mother households into two-parent households (while keeping all other characteristics of the households as is).

This is not to say that single motherhood is unimportant—more children growing up in two-parent households would surely be a good thing. But if a state’s purported goal is to reduce the number of its residents living in poverty, offering cash support coupled with employment incentives is likely to be more effective than sex-education courses or ad campaigns “promoting the value of healthy marriage.”

TANF is not the only social program in the U.S. that stratifies low-income families based on state boundaries and skin color. In Jamila Michener’s new book, “Fragmented Democracy: Medicaid, Federalism, and Unequal Politics,” the Cornell political scientist depicts in vivid detail how state differences in Medicaid policies “tether health policy even more deeply to race and poverty.”

Most concretely, Michener points out that eight of the 11 states with the largest share of the nation’s black population are among those that have failed to implement Medicaid expansion. Even among Medicaid beneficiaries, access to dental treatment, hearing or vision support, and end-of-life services vary by state. If a person eligible for Medicaid “got sick in the wrong state during the wrong year,” Michener writes, “the consequences of policy fragmentation could be life altering.”

Other examples of “policy fragmentation” abound. Today, roughly half of the 50 states offer a supplement to the federal earned-income tax credit, but the average black family is less likely than the average white family to live in one of those states. States vary in levels of minimum wage, paid-leave policies, and investments in early-childhood education, but again, regional and racial inequities are large. And perhaps most damaging, states have long varied with respect to their criminal-justice policies. In 11 states, at least one in 20 black men is incarcerated. Ironically, the states most likely to chide black women for raising children alone—and to promote marriage as the key to poverty reduction—tend to be the same states that incarcerate the largest share of black men. Including the incarcerated population in our estimates of poverty in the U.S. would only widen racial divides.

“Federalism,” as Michener poignantly writes, “is a primary channel through which the geography of opportunity is shaped in America.” The question of in which state a low-income family lives is associated more and more with how well that family is able to live. And more often than not, low-income families of color tend to find themselves living within the borders of a more punitive state.

We often perceive policy makers as defenders of economic opportunity, and social policy as a set of tools to alleviate inequalities. But as Michener’s work shows, and as the data on TANF suggest, state governments often function as a source of inequality rather than its cure. Instead of narrowing gaps between the advantaged and disadvantaged, social policy can, when deployed unevenly across the country, act to deepen them instead.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.