Editor’s note: Today marks the 20th anniversary of welfare reform. The law replaced America’s old cash-welfare system, Aid for Families with Dependent Children (AFDC), with a new program called Temporary Assistance for Needy Families (TANF). The new welfare came with a time limit and work requirements. The law was a milestone in America’s transition to a work-focused safety net—a transition that also entailed other benefits for the working poor, for example through the Earned Income Tax Credit.

Today, TAC Managing Editor Robert VerBruggen will be discussing the law at a Cato Institute event, which will take place from 9 a.m. to 12:45 p.m. ET and can be viewed online. Below is the text of the speech he will deliver.

Good morning. First of all, I want to thank the Cato Institute for inviting me to speak today. It’s truly an honor to be on a panel alongside think-tank scholars who have done such respected, detailed, high-quality work on this issue, and one who even helped to design the law we are discussing. I feel both flattered and intimidated.

For I am not a think-tank scholar. I am a journalist who knows just enough about statistics to get himself in trouble. But I’d like to begin by discussing a simple analysis I published in Family Studies that I think illustrates two basic truths about what happened after welfare reform.

The first truth is that when welfare benefits became contingent on work, a lot of people sought work and found it. Here you can see a chart I made with Census data. It focuses on the children of single mothers, the demographic most affected by the law, and it plots their poverty rate against the nationwide unemployment rate so we can see the effect of the economy.

Chart1

The poverty metric doesn’t even include things like food stamps or health benefits, both of which have risen over the past two decades. It also doesn’t include tax credits, most importantly the Earned Income Tax Credit, which was hiked around the same time that welfare reform took effect and was similarly designed to encourage work. It does include cash welfare, which the reform cut. And yet even by this measure, these kids were better off after the law than before it, including in the wake of the Great Recession. The post-reform blue years fall below the previous red years almost categorically.

Before welfare reform passed, critics said vulnerable kids would be thrown into poverty. When the opposite happened, critics said it was just the economy. Two recessions later, that doesn’t seem to be the case either. There are still many poor children in this country, but welfare reform made the problem better, not worse. These major changes to the safety net—both the 1996 law and the boosted Earned Income Tax Credit—succeeded in moving the poor from welfare to work.

Having a job comes with its own expenses and challenges, of course, but it’s the first step toward the middle class, and welfare reform helped people take it. This success is something we can’t forget, even as we consider newer and more nuanced allegations that the reform was harmful to some.

But speaking of those allegations, here is another chart that is almost identical—except that the poverty threshold is divided by 10. In other words, these folks would still be at or below the poverty line if they had 10 times as much income as they do. You might call that “extreme poverty.” And the trend is reversed: After the law, there was a sharp increase in kids living in families that reported very little income, and this became even worse with the recession.

Chart2

In some ways, these results are unambiguous and unsurprising. The reforms were designed with a carrot-and-stick approach, where those who worked received additional aid, while those who didn’t work were cut off from cash welfare. It is hardly shocking that some people got the stick, and that the number of people without cash grew, even as work lifted many others above the poverty line.

But in other ways, we are in murky territory here. For one thing, people who claim to live on virtually no cash are not always telling the truth. For another, as I mentioned before, my simple analysis using the official poverty measure doesn’t include non-cash benefits like food stamps.

This means we have to dig deeper. And the picture we uncover isn’t very clear at all.

There is one thing everyone agrees on, which is that the situation looks better when you correct the data to account for non-cash benefits and underreporting. What they don’t agree on is how much better.

To pick just one example from the left, the Center on Budget and Policy Priorities recently reported that incomes fell for the bottom tenth of single-mother families, and did so immediately and dramatically after welfare reform. In 2005, they were down about $2,800 a year in cash benefits and $250 in food stamps, with private income, tax credits, and other programs replacing only about $700 of that loss. Increasing food-stamp benefits helped incomes rebound after that, but only partially.

Here is a striking chart from that report. Note especially the steep drop between 1996 and 1997 for the poorest single-mother families.

Chart3

But in another new paper, Scott Winship, who will speak next on this panel, argues that even the very poorest might not be worse off at all. Obviously, I’ll leave the details to him. But for now, suffice it to say that a lot depends on how exactly you process the data, including adjusting for inflation and putting a dollar value on health benefits. In fact, researchers don’t even agree as to whether health benefits should be counted at all.

Others abandon income data altogether. Some prefer to look at consumption, which tends to be much higher than income at the very bottom. Extreme poverty is vanishingly rare by this measure. We can also look at indicators of hunger, which fell after welfare reform but rose during the Great Recession.

That’s a lot of puzzle pieces, and they don’t all fit together. But at a minimum, I feel comfortable saying two things. One, welfare reform helped a lot of people, and there is no use pretending otherwise. Even in the Center on Budget’s new analysis, nine-tenths of single-mother families were better off a decade after the reform, most of them significantly so.

But two, the very poorest families have less cash than they used to, even if other benefits may have helped to make up the difference, and even if they have more cash than they admit in surveys. This is a nearly unavoidable consequence of cutting off cash assistance, and it’s a problem in itself: As Kathryn Edin and Luke Shaefer document in their book $2.00 a Day, it is hard to get by with little cash even when you have benefits that pay for things like food or housing. It’s tough to look for a job, or make needed car repairs, or buy clothes.

The panel after this one is going to focus on where we go from here. But I would like to briefly lay out a few lessons I think we can learn from this experience.

The biggest lesson is that a work-focused safety net can succeed. We have made important progress against poverty, and any future reforms should be made with an eye toward preserving that progress.

I think there’s also a lesson here, though, for conservatives who would like to charge full steam ahead and apply the ideas of welfare reform to other programs as well. Specifically, they must grapple with the fact that those other programs helped to keep a lid on extreme poverty when cash welfare dried up. Changing those other programs would increase the incentive to get jobs even further, and it could address the fact that over time, work requirements have been gutted in some programs. But it could also severely punish those who don’t find work, whatever the reason.

One way out of this conundrum is for states themselves to provide a way for non-working recipients to earn their benefits. Early proposals for what became welfare reform actually included government-provided jobs. Community service is also an option, one the state of Maine is trying out in its food-stamp program. We should keep an eye on such experiments.

Yet another lesson, nicely explained in $2.00 a Day, is that cash matters even when other benefits are available. This brings up the thorny question of how to get more cash to the poor without recreating the anti-work incentives we had 20 years ago. Broadly speaking, there are three ways to do this that warrant our attention.

First, we can make sure TANF is doing its job. Under the welfare-reform law, states were given block grants and broad discretion over how to spend them. Many states abused that discretion, using the money to patch up their budgets rather than help the poor. This is actually a separate lesson in itself: states may be the laboratories of democracy, but that doesn’t mean you can just throw money at them and expect to get good results. At any rate, this is not what those who designed the program intended, and it’s keeping cash assistance away from people who qualify for it.

Second, we could rethink the paternalism of the rest of the safety net. We make a lot of money available to the poor on the condition that they spend it the way we want them to—on food, on housing, even on their heating bills. Simply converting some of this aid to cash would give them much-needed flexibility without changing the total amount of aid. But it would also provoke pushback from the more intrusive elements of both sides of the political spectrum, often called the nanny state and the daddy state.

And third, we might want to consider some form of broadly available cash assistance. The idea would be to make sure that each family receives a modest amount of cash, staving off extreme poverty, while otherwise maintaining the work-focused incentives that made welfare reform a success.

In particular, we might look at the child tax credit—which is popular with the anti-poverty left and the pro-family right alike. Right now it’s worth up to $1,000 per child, but it’s not available to the poorest parents because it isn’t fully refundable. I have suggested changing that. Other conservatives, including Patrick Brown and Josh McCabe, have made similar proposals, and just last week two liberal think tanks, the Center for American Progress and the Center on Budget and Policy Priorities, released reports doing the same.

A nice thing about the child tax credit is that the middle class already gets it, so we don’t have to deal with the difficult issue of phasing it out as people earn more money, which can discourage them from doing so. And to address concerns that the credit itself would discourage work or encourage nonmarital childbearing, other benefits like food stamps could be cut back to compensate.

Again, the idea is to make sure every family has at least a little cash. It’s not to undo what we’ve accomplished by orienting our safety net around work.

Thank you all so much for coming this morning, and I look forward to any questions.

Robert VerBruggen is managing editor of The American Conservative.