For those who had counted on Clinton to “end welfare as we know it,” the Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA) was not a disappointment. AFDC was gone in its entirety, along with the entire federal apparatus for directly administering welfare. In its place was a new system called Temporary Assistance for Needy Families (TANF), which completely transformed the country’s public benefits system.

TANF ended the guarantee that poor mothers could receive assistance from the federal government. Instead, it imposed a five-year lifetime cap on benefits and instituted a work requirement. Nobody could receive benefits for more than two years at a time. Immigrants (even lawful permanent residents) were barred from receiving benefits.

Most importantly, it decentralized the administration of benefits. Rather than being a federal program, TANF was delivered as a series of block grants to state governments, who were relied upon to make sure that the funds made their way into the hands of recipients. And the total amount spent by the federal government would be far less than its previous AFDC expenditures.

This move was not just radical. It was an obliteration of the social guarantee to poor families that Democrats had spent decades building.

When Clinton signed the bill, the New York Times reported that in a “sweeping reversal of Federal policy, President Clinton today ended six decades of guaranteed help to the nation’s poorest children” and thereby “eliminated a pillar of Franklin D. Roosevelt’s New Deal social welfare program, delighting the Republican-controlled Congress in this election year and incensing many of his fellow Democrats.” Clinton’s labor secretary, Robert Reich, said Clinton had outright “ended the promise of help to the indigent and their children which Franklin D. Roosevelt had initiated more than sixty years before.”

Many Democrats found this impossible to endorse. Three senior officials in the Clinton administration resigned almost immediately. Even the conservative deregulation advocate Robert Rubin opposed Clinton’s signing of the bill. Peter Edelman, one of the officials who had resigned, took to the pages of the Atlantic and called the bill “the worst thing Bill Clinton has done.” (Barack Obama said he found Clinton’s signing of the bill “disturbing.”)

Bill Clinton was open about the fact that his decision was at least in part motivated by political considerations (although George Stephanopoulos had told him that he was assured of reelection even if he did not sign the bill). “After I sign my name to this bill,” Clinton declared, “welfare will no longer be a political issue.”

But if Clinton thought he would appease Republicans with welfare reform, he was mistaken. The moment the bill was signed, Bob Dole declared that Clinton was transparently engaging in an “election year political calculation” and that “[b]y selling out his own party, Bill Clinton has proven he is ideologically adrift.” With AFDC dismantled, Republicans simply moved on to find other social programs, such as food stamps, to attack next.

After the enactment of the PRWORA, its supporters almost instantaneously declared welfare reform a success. “We now know that welfare reform works,” Clinton declared just a year after signing the bill. But often one of their main criteria for “success” was whether people had been successfully removed from the welfare rolls.

The Department of Housing and Urban Development claimed that switching from welfare to employment meant achieving “self-sufficiency,” concluding that “welfare reform increased the rate at which families living in public housing or using vouchers at baseline became ‘self-sufficient’ in the sense that they were employed, no longer received welfare, and no longer had housing assistance.”

A summary by the Republican House Ways and Means Committee claimed that “since it replaced the New Deal–era Aid to Families with Dependent Children (AFDC) program in 1996, Temporary Assistance for Needy Families (TANF) has been successful at cutting welfare dependence as caseloads have declined by 57 percent through December 2011.” Others, such as the Heritage Foundation, similarly phrased the decline in caseloads as a “decline in welfare dependence.” Conservative writer Heather Mac Donald wrote that:

Congress’s wager paid off handsomely. Asked to look for work in exchange for their welfare checks, hundreds of thousands of women found jobs. From 1996 to 1999, employment among the nation’s never-married mothers rose 40 percent. In 1992, only 38 percent of young single mothers worked; by March 2000, 60 percent of that group were employed. Another large portion of the caseload, faced with new participation requirements, simply decided that welfare was not worth the hassle. The result: a 52 percent drop in the caseload since August 1996, when TANF passed, to June 2001. Nearly 2.3 million families have left the rolls.

Hillary Clinton herself said she was proud that former AFDC recipients were “no longer deadbeats — they’re actually out there being productive.”

All of these triumphant declarations by Republicans and the Clintons were curious. By conflating “no longer receiving welfare” with “no longer needing welfare,” they avoided asking the question of how the actual life situations of the people who were removed from the rolls had changed.

After all, “reducing the welfare rolls” is only a positive outcome if people are leaving the rolls because they no longer need assistance, rather than because they are facing new obstacles. Yet oddly, in the post-AFDC era, “declining caseloads” are often invoked, even by liberals, as a positive outcome. (In a critical assessment of welfare reform in the Washington Post, Dylan Matthews wrote that the declining number of recipients suggests that “some” of the support for welfare reform is justified.)

HUD’s view was that less welfare meant more self-sufficiency. But a decline in caseloads said very little in itself; it was hardly unexpected that a law designed to reduce welfare access would end up reducing welfare access. The question was whether the particular people who would otherwise have been on welfare were better off.

There were early signs that this was not the case, and that while the PRWORA was, as the New York Times suggested, “effective in getting people off welfare, the system proved far less nimble at making them financially sound.” Senator Paul Wellstone of Minnesota was skeptical of the “boosterism” exhibited by the Democratic defenders of welfare reform:

There’s been a flurry of credible reports suggesting that all is not well with welfare reform. But President Clinton and Vice President Gore continue to claim that welfare is “working.” What they overlook is why, at a time when the welfare rolls have been cut in half and the economy is booming, we now are finding that millions more children are going to bed hungry each night; demand for emergency food assistance is growing; millions of poor families are dropping off the Food Stamp rolls faster than economic indicators would predict; and former welfare recipients are losing their medical coverage, cannot make the rent and utilities, and are unable to afford child care. These are not the results of successful reform.

Robert Reich was similarly unimpressed by Clinton’s insistence that he had succeeded. Reich doubted the premise that declining welfare rolls necessarily meant a successful program:

The White House now claims that the 1996 welfare bill has been a huge success, based on the large number of people who have been removed from state welfare rolls since then. But we have no way of knowing how many of these people are in permanent jobs paying a living wage, or are in temporary jobs paying so little that they have to double up with other family members and leave their children at home alone during the day, or are living on the street. And we may never know, even after the economy slides into recession, and the ranks of the unemployed begin to grow once again. The sad truth is that America has embarked on the largest social experiment it has undertaken in this half of the twentieth century without even adequate base-line data from which researchers can infer what has happened, or deduce what will happen, to large numbers of poor people who no longer receive help.

One problem in testing these assertions was that by getting rid of federal involvement in welfare, Clinton had also gotten rid of a core accountability mechanism. The states were being entrusted to administer welfare, but they were not required to demonstrate that they had used the money well. And measuring outcomes across fifty different state systems, and trying to determine the extent to which each program was succeeding or failing, would have been an impossible research task.

Thus, just as Jim Clyburn had noted, while welfare recipients were being saddled with new obligations, states had to meet very few requirements in exchange for their own handouts. TANF dollars flowed in regardless of how well a state’s program was doing. Compounding the problem was the fact that states could actually reallocate TANF money toward other programs if they felt it necessary. Consequently, as the Center on Budget and Policy Priorities reports, the block grants ended up being used to “plug state budget holes” rather than providing cash assistance to the poor.

Other provisions similarly incentivized states to restrict assistance. A “caseload reduction credit” lowered the obligations states had to meet for their work programs if they could demonstrate they had removed people from the rolls. And while states were evaluated on whether they had put people to work, they were not evaluated on whether they had provided adequate assistance to families, encouraging states to move the poor into low-wage jobs to rid them from the rolls as quickly as possible.

What came next was almost amusing in its predictability. Because TANF’s effectiveness depended on the competence and benevolence of state governments, levels of TANF assistance varied considerably across states. In some parts of the country, TANF benefits were sufficient to elevate a family to nearly 50 percent of the poverty line, though nowhere did benefits do any better than reaching half the federal poverty level.

But across much of the South (including many of the country’s poorest states), TANF recipients were lucky if their benefits got them to 15 percent of the poverty line, a sum impossible to subsist on. The states with the highest African-American populations extended the stingiest TANF benefits.

Alejandra Marchevsky and Jeanne Theoharis of the Nation go so far as to say that in the post-welfare era, “a revival of Jim Crow–like practices and exclusions have flourished, as Southern states have largely dismantled their welfare systems and pay some of the lowest benefits in the nation.” And because African Americans face intense job discrimination (they are less likely to receive call-backs than their equally qualified white counterparts), the work requirements are already likely to be disproportionately difficult for African Americans to fulfill.

Nor is the program improving. The value of TANF checks keeps dropping, making its benefits ever more meager. Federal funding has been stagnant for twenty years, and doesn’t change in response to recessions. Thus recipients generally get by during prosperous times like the final years of the Clinton administration, but when the economy contracts, TANF does little to help families survive hard times.

Welfare reform did succeed in moving many single mothers into jobs, if only because the disappearance of benefits left little other option than to accept whatever employment was available. But the evidence on whether the increased employment actually increased these mothers’ overall well-being is mixed.

Conservatives and the Clintons tended to speak of full-time employment as if it was in itself the desirable state of being for single mothers; to be “productive” meant no longer being a “deadbeat.” But this was not necessarily so; a mother who had to travel on an hour-long commute to a dead-end job, thus never seeing her child, may have increased her “productivity” while becoming worse off.

If a mother’s increased earnings are spent on child care for the time she spends at work, and her parenting suffers as a result of her mandatory employment and her exhaustion after a day’s work, moving mothers from welfare to work could damage the lives of poor children even as it increases both employment and earnings among single mothers. The fetishization of work, and the corresponding obliviousness to other important life outcomes, can make welfare reform appear more successful than it truly was.

Former welfare mother Diana Spatz explains that welfare reform wrongly assumed that work automatically improves mothers’ lives, when certain types of work are themselves barriers to education and self-improvement:

“Any job is a good job” was the slogan emblazoned on the walls of county welfare agencies across the country, as tens of thousands of low-income mothers were made to quit college to do up to thirty-five hours per week of unpaid “workfare”: sweeping streets, picking up trash in parks, and cleaning public restrooms in exchange for benefits as low as $240 a month. Contrary to “welfare queen” stereotypes, like most welfare mothers, I worked first. Work wasn’t the problem; it was the nature of the work — low-wage, dead-end jobs with no benefits and little chance for advancement — that kept families like mine on the welfare rolls.

In assessing its results, if we examine welfare reform fairly, what we see generally is a bifurcation. Those who were successfully “moved to employment” saw their earnings increase. But for others, conditions not only stagnated, but worsened.

In particular, those at the “bottom of the bottom” suffered. In the years since welfare reform, the percentage of families in extreme poverty increased by 50 percent. And there is evidence to suggest that large numbers of people live in incomes as low as two dollars a day, with many selling plasma in order to survive. As Ron Haskins, a Republican who helped write the welfare bill, notes, because of the structure of TANF “any mom who does not have the ability to maintain her household and work at the same time is going to have trouble.”

TANF can leave such people stuck in deep poverty. Previously, under AFDC, a mother could receive benefits while attending school full-time, thus enabling her to get a degree and gradually move off welfare. But TANF did not count higher education as fulfilling the work requirement, thus women were unable to support themselves while being educated. Consequently, while in 1995, “649,000 student parents were receiving cash assistance while enrolled full-time in education programs, only 35,000 full-time students received TANF aid in 2004.”

Because of TANF’s restrictive conditions, mothers who cannot work or struggle to meet their obligations are unlikely to receive benefits. There has consequently been a decline in the percentage of poor families that receive benefits. While under AFDC, 68 percent of families with children in poverty were receiving some form of benefits, by 2014, that number had declined to 23 percent. And while AFDC had succeeded in lifting 62 percent of children out of deep poverty, for TANF that number was only 24 percent.

Welfare reform was therefore precisely what its liberal critics said it was: little more than the elimination of the safety net.

The Center on Budget and Policy Priorities concludes that “[TANF], the cornerstone of the 1996 reforms, is not the success that some proclaim and should not be used as a model for other safety net programs.” Marchevsky and Theoharis, in the Nation, summarize the best available empirical findings on welfare reform’s success:

[M]ost studies of the effects of this legislation, including our own, have concluded that it has only succeeded in pushing people out of the welfare system — not helped the vast majority out of poverty. Even during the boom years of the late 1990s, when former recipients were job hunting in a robust labor market, research showed that poor mothers were channeled into low-wage, dead-end jobs, and that many who “timed out” or were “sanctioned out” of TANF could not find work because they needed more education and training . . . [TANF] traps poor mothers into exploitative, poverty-wage jobs and dangerous personal situations, deters them from college, and contributes to the growing trend of poor mothers who can neither find a job nor access public assistance.

Bill Clinton strongly disagrees that signing welfare reform was inherently damaging to the interests of the poor. During his testy 2016 exchange with the Black Lives Matter protesters in Philadelphia, Clinton attributed all of welfare reform’s failures to the actions of state governments, who stripped benefits.

“We left them with enough money to take care of all the poor people who couldn’t go to work, on welfare,” Clinton insisted. “We left them with the money they had before the welfare rolls went down 60 percent. The Republicans took it away and they’re blaming me.” The fault therefore lies not with the federal government, says Clinton, but with the states to which it handed over authority.

But this is a bit like saying “Don’t blame me, I just handed the baby to the dingo. I didn’t know it was going to eat it.” One would have to be spectacularly unfamiliar with the centuries-long history of Southern legislatures to believe they would cheerfully, efficiently, and effectively administer the distribution of welfare benefits to poor black mothers.

Clinton may pronounce himself shocked that once given the ability, the Southern states kicked people off the welfare rolls en masse, but as Eric Levitz of New York magazine wrote, “[o]ne problem with Clinton blaming Republicans for ‘taking away’ people’s welfare is that, without his law, they wouldn’t have been able to.”

By simply blaming the states, Clinton is also forgetting the extent to which he was the architect of the unforgiving time-limited benefit restrictions, and his own promotion of the idea of a “dependency crisis.” Clinton himself publicly deployed the “declining caseloads” metric of welfare reform’s success, without considering whether those declining caseloads could be concealing the existence of deprivation. As Peter Edelman, himself an old friend of the Clintons, says of them, “[t]hey don’t acknowledge the number of people who were hurt. It’s just not in their lens.”

The welfare bill was never in any sense a compromise for Clinton. The elimination of welfare was something Clinton introduced into the 1992 race, something he fundamentally made an issue in the first place. In 1996, Bob Dole was so flustered by Clinton’s “tough welfare talk” that he accused Clinton of engaging in “petty theft” of Dole’s platform.

And it’s not as if Clinton came to regret what he had wrought; ten years after it passed, free to speak his mind without fear of political consequence, Clinton proudly wrote that “the 1996 Welfare Act shows us how much we can achieve when both parties bring their best ideas to the negotiating table and focus on doing what is best for the country.” Clinton said that “we never betrayed our principles and we passed a bill that worked and stood the test of time.”

Clinton therefore stood proudly behind his repudiation of the kind of robust social-safety-net programs passed under Roosevelt and Johnson. Most disturbingly, he refused to acknowledge the part that race played in turning welfare reform into a winning political issue.

As Neubeck and Casavere write, “[t]he ease with which political elites abolished the Aid to Families with Dependent Children program — the primary safety net protecting poverty-stricken mothers and children — would have been impossible had not many politicians, along with policy analysts and the mass media, spent decades framing and morphing welfare into a supposed ‘black problem.’”

Clinton rode this “black problem” to the presidency in 1992, then worked with Newt Gingrich to intensively address it by stripping away the public benefits that most benefited African-American women. As a result, he made life harder for millions, and eroded the core of the Democratic Party’s commitment to supporting the poor.

Lillie Harden was a black Arkansas mother who had spent time in the AFDC program. When Bill Clinton met her at an event, he was instantly impressed by her story. Harden had gone from being on welfare to having a job at a supermarket. For Clinton, she was the perfect poster woman for his welfare reform plans. Clinton cited her repeatedly in his pitches for the personal responsibility bill.

He dwelled particularly on something she had told him about why she was proud to have a job instead of living on welfare: “When my boy goes to school and they say what does your mama do for a living, he can give an answer.” Clinton invited Harden to the signing ceremony for the welfare bill, where she was seated next to him as he eliminated AFDC.

But while Lillie Harden made for a useful welfare-to-work anecdote in Clinton’s speeches, her life did not turn into the kind of success story Clinton spoke of.

As Marchevsky and Theoharis note, “Had the Clintons maintained an interest in Harden, they would have discovered that her ‘success’ was short-lived.” She continued to struggle with poverty, and her son ended up in prison. In 2002, she suffered a stroke. Harden’s application for Medicaid was denied, though she had been given access to Medicaid when she was on AFDC.

As a result, she could not afford her $450 prescription medication, and Harden died at the age of fifty-nine in 2014.

Heartbreakingly, when Harden was interviewed by a journalist, she requested that he relay a message to Clinton asking the former president to help her get Medicaid. Of her job, she said only: “it didn’t pay off in the end.”

During a debate with Bob Dole in 1996, Clinton declared that “I want to make more people like that woman, Lillie Harden. So I’ve got a plan to do it. And it’s just the beginning.”

Clinton was not wrong. He did indeed make more people like “that woman,” Ms. Harden.