How the States Rewrote Welfare Before Washington Finally Caught Up

Washington has a favourite scheme: let a problem fester for thirty years, wait for the states to quietly invent a fix, then pass a law, hold a signing ceremony and take the credit. Welfare reform in 1996 is a textbook case. The photographs show a President, a pen and a room full of smiling legislators. The real story is in the governors' offices of Madison and Lansing, where people spent the previous decade doing the work and filing the paperwork.

I'd like to tell it in order, because the order is the point. The idea that you can rewrite a national program from the bottom up, by letting states try things and watching what happens, is the best argument for federalism I know. And it has a second use: it gives us a fair way to judge whether the reform worked.

What the old system was

The program everyone argued about was Aid to Families with Dependent Children, or AFDC. It began as Aid to Dependent Children in the Social Security Act of 1935, meant to help widows with young children keep their households together, and it was renamed in 1962. It was a joint federal-state program: states ran it, and Washington matched what they spent. Anyone who met the eligibility rules got the aid. It was an entitlement, which is a technical word with a big consequence: there was no cap on the number of people who could enroll, and no clock.

The program had a design flaw that critics on both sides eventually noticed. The rules made it hard for a recipient to take a job without losing the benefit, so a mother who started working could see her grant shrink nearly dollar for dollar. The structure rewarded staying put. Over time the program also came to serve a different group than the one it was built for, as fewer recipients were widows and more were never-married mothers, and the question of what we expected of them kept getting louder.

Conservatives argued that the system trapped families in dependence and undermined work and marriage. Liberals argued that cutting aid would punish children for the situation of their parents. Both were arguing about real people, and neither could prove anything, because every state was running the same federal rulebook.

The waiver: how a state breaks the mold

Here is the hinge of the whole story. Section 1115 of the Social Security Act gave the federal government authority to waive certain program rules so that a state could run a demonstration project. A governor who wanted to try a different approach to AFDC could ask the Department of Health and Human Services for permission to depart from the rules for a set period, and if approved, could test his idea in the real world.

It was the laboratory Justice Brandeis described, and I take up that concept in Brandeis Called the States Laboratories and Conservatives Took Him at His Word. What made it powerful was that the requests came from governors of both parties, were approved under administrations of both parties, and gave the country something it rarely has: actual results from actual experiments.

In 1988, Congress also passed the Family Support Act, which President Reagan signed. It created a job training and work program for AFDC recipients and was a bipartisan attempt to move toward work. Its reach was modest, and many reformers concluded it did not go far enough. The reformers were just getting started.

Wisconsin: Tommy Thompson's workshop

Tommy Thompson was governor of Wisconsin from 1987 to 2001, a long tenure that gave him time to build on each experiment. Wisconsin used waivers repeatedly. It tried Learnfare, which tied part of a family's grant to teenagers' school attendance, a policy that provoked criticism and some skepticism about its effects. It tried programs that stressed work as the expectation for able-bodied recipients.

The capstone came after federal reform. Wisconsin replaced AFDC with a program called Wisconsin Works, or W-2, in 1997. The central idea was that cash should not be given with no expectation attached. A participant was placed on a rung of a ladder according to readiness for work, with jobs at the top, and the program's design stressed that participation was a condition of aid. The state also invested in child care and other supports that make going to work practical for a single parent, a feature that critics of the straightforward "work requirement" often overlook.

Notice the structure. This was not a theory worked out on a chalkboard. It was a decade of trial, error and adjustment in a state that had the freedom to do it. I wrote about the Wisconsin political context in Wisconsin's Act 10 Fight From the Capitol Crowds to the Recall Vote, and the same state's reform instinct shows up there too.

Michigan: John Engler's cut and his reform

John Engler became governor of Michigan in January 1991, and one of his first major acts was to end the state's General Assistance program, which gave cash to low-income adults without dependent children. It was a state-funded program, not a federal one, so Michigan could end it on its own authority. The move was controversial and made Engler a lightning rod.

He also pursued AFDC changes under a waiver, a package that stressed work and changed the rules on earnings so that a recipient who worked could keep more of the benefit. A recipient who took a job would not see the grant fall dollar for dollar. The principle was straightforward: make work pay more than staying home.

Engler was one of a number of governors, Republicans and Democrats alike, who in those years asked for waivers. Many states sought permission to try some form of work requirement, time limit, family cap or earnings rule. By the mid-1990s the waivers had become a patchwork, and officials in Washington were reading the results.

Washington's politics: from slogan to bill

Now the national story. In 1992 Bill Clinton ran for President promising to "end welfare as we know it" and to impose a time limit on benefits. Whatever you think of him, that was a significant shift, because it came from a Democrat. In 1994, Republicans won control of both houses of Congress for the first time in forty years and made welfare reform part of their agenda, in the Personal Responsibility Act, part of the Contract with America.

Clinton and the Republican Congress then fought over it. Clinton vetoed Republican welfare bills twice, once in December 1995 as part of a larger budget package and again in January 1996. Republicans argued that he was dodging the commitment he made in 1992. Clinton argued that the bills were too harsh, particularly in how they treated children and on the elimination of safety nets. The standoff mattered because of a political fact: the Republican Congress could not pass a bill over a veto, and the President could not pass a bill without the Congress. They had to agree.

They did, in the summer of 1996. On August 22, 1996, Clinton signed the Personal Responsibility and Work Opportunity Reconciliation Act, known as PRWORA. The story of those months is told in more detail in two vetoes, one signature and the 1996 welfare law that put marriage in statute, and the broader lesson about divided government is in divided government gave us welfare reform and a balanced budget.

What the 1996 law did

Here, in plain terms, is what PRWORA changed.

1. It ended AFDC and replaced it with TANF. Temporary Assistance for Needy Families is a block grant to the states. Instead of Washington matching whatever each state spent, it gave each state a fixed amount, with $16.5 billion nationally per year in the basic grant, and let the state design its own program within federal guardrails.
2. It ended the individual entitlement. The old rule that every eligible family was guaranteed aid was gone. States could set their own rules and decide how to use the money.
3. It set a federal time limit. Federal TANF funds could generally not be used to give cash assistance to a family for more than five years in total, with room for states to exempt a share of cases.
4. It set work participation expectations. States had to meet targets for the share of recipients engaged in work or work activities, or face penalties.
5. It required states to keep spending. A "maintenance of effort" requirement meant states could not simply pocket the federal money and cut their own contribution to zero.

Look at what that did to the relationship between the levels of government. The waivers had been states asking permission to depart from the rules. TANF made the experiment the rule: the states now held the pen. The accumulated experiments, in Wisconsin, Michigan and elsewhere, supplied the ideas and the confidence that it could be done.

The argument over what it did

Now I owe you honesty, because this is the part of the story that gets distorted from both sides.

What the supporters point to: caseloads fell sharply in the years after the law, and employment among single mothers rose through the late 1990s. They argue that the combination of work expectations, time limits and support services moved many families from the rolls to the payroll.

What the critics say: that the strong economy of those years did much of the work, along with an expansion of the Earned Income Tax Credit enacted in 1993, and that the caseload decline cannot be credited to the law alone. They point out that some families lost aid without finding stable work, and they point to the group in deep poverty as a place where the safety net thinned. They also note that the TANF block grant was set at a fixed dollar figure, which has not been adjusted for inflation, so its real value shrank over the years, and that states used the flexibility in different ways, some spending on things other than cash assistance.

My honest assessment is this. The two sets of claims are not mutually exclusive, and any single-cause story, whether "welfare reform cured poverty" or "welfare reform threw families on the street", is too simple. The employment gains were real, and so was the economic boom, and so was the EITC. The fixed grant was a genuine weakness, and flexibility was both an opportunity and a temptation. If you want the best summary, I'd say the reform succeeded in changing the expectation attached to aid, and it is less clear that it did as much for the lowest-income families.

Not every critic was a bystander. Some officials in the Clinton administration, including Peter Edelman and Mary Jo Bane at the Department of Health and Human Services, resigned in protest after the President signed the bill. Their resignations are a reminder that the objections were serious and held by people who knew the program well. The best argument for reform deserves an equally fair hearing for them.

What conservatives should take from it

First, the bottom-up route worked. The national law drew on a decade of state experiments, which is the reason it carried a degree of confidence that a theory from Washington could not. When a principle has been tested somewhere, the legislative debate becomes about evidence and not just ideology.

Second, expectations matter as much as dollars. The core idea of the reform, that aid should be linked to work and that those who can work should be expected to, rests on a moral conviction as much as an economic one. The conviction is that work is dignifying, and it is the old question of the role of the family and the community, which I explore in The Family Is the Original Welfare State and the Cheapest One.

Third, do not oversell. The reform deserves credit, and it deserves scrutiny. A movement that wants to be trusted should be the first to admit where the block grant eroded or where flexibility was misused.

Fourth, keep the laboratories open. Some of the most interesting work on welfare since 1996 has happened in states and localities, not in Washington. If you want the next reform, watch the statehouses, as I described in Interstate Competition Is Why States Cannot Afford to Be Lazy.

The bottom line

• Before the 1996 law, governors in states like Wisconsin and Michigan used federal waivers to test work requirements, earnings rules and time limits, and what they found shaped the national bill.
• President Clinton signed the Personal Responsibility and Work Opportunity Reconciliation Act on August 22, 1996. It replaced AFDC with TANF, a block grant to the states, ended the individual entitlement, and set a federal five-year limit and work participation targets.
• The honest record is mixed. Caseloads fell and work rose, but the strong economy and the Earned Income Tax Credit also played roles, and the fixed block grant lost real value over time.
• When someone tells you reform either solved poverty or abandoned the poor, ask which evidence they are using. Both claims are too simple.
• If you want to change a program, start in your own state. Ask your legislator what waivers and pilot programs exist, and what they have shown.