Listen to the podcast or read the full transcript here.
Joining me on the podcast today is Kevin Corinth, a senior fellow at the American Enterprise Institute, where he researches economic mobility, poverty, safety net programs, homelessness, social capital, and other issues. He is a co-author, along with Richard Burkhauser, of a recent paper titled “Poverty and Dependency in the United States, 1939 to 2023.”
What motivated you to revisit nearly a century of American poverty trends?
We have a lot of evidence about the extent of economic progress over the past 50 years or so. My co-author, Richard Burkhauser, and I did a paper a few years ago showing that, since President Johnson declared his war on poverty, we’ve seen a dramatic decline in poverty in the US from a 20 percent baseline to less than 2 percent today. In this paper, we wanted to ask, “What happened before that? Were we seeing declines in poverty before we had a large and expanding safety net?”
I also think that the idea that living standards have not improved over time is motivating a lot of today’s political turn toward socialism. And if we lose our understanding of what we’ve gained through our imperfect capitalist system, that could have very deleterious effects on policy today.
Before we get into the findings, why should people care about how we measure poverty?
A lot of people would agree that you should judge a society based on how it treats its most vulnerable members. I subscribe to that. So measuring poverty matters a lot for people’s views about how society is doing, which matters a lot for determining which policies to adopt.
You measure poverty almost the same way that you measure incomes. The only difference is that you have to pick a poverty line because poverty is the number of people with an income below some threshold. A lot of the debate on poverty comes down to where to draw that line. Where you draw the poverty line is a value judgment. It’s scientifically arbitrary, but it’s not arbitrary in terms of our values. As an economist and social scientist, I don’t get to decide that. You tell me where you want me to draw that poverty line, and then I’ll update it with inflation each year to measure our progress.
The official poverty measure is not measuring poverty well. It tries to be an absolute measure in that it increases the poverty line with inflation, but it uses an inflation measure that overstates how much prices go up each year. It also doesn’t include in-kind transfers like the SNAP program, which is our largest non-medical transfer today. It doesn’t include Medicaid or refundable tax credits. It doesn’t adjust for taxes at all.
Based on the official poverty measure, we’ve seen essentially no decline in poverty over the past 50 years because of these problems. When Rich Burkhauser and I looked at it, we found 90 percent reductions. So how you adjust the threshold and which resources you include matters a lot. Fortunately, almost everyone recognizes that the official poverty measure doesn’t make sense.
Among those who do recognize that poverty has come down in the United States, many assume that it has fallen because of government redistribution. Your central finding is that poverty fell dramatically before the War on Poverty even began. Could you tell me more about that result?
We’re not looking at the causal effect of the War on Poverty. I think we’d argue that it’s not possible for someone to estimate the causal effect of such a massive change in society. However, a lot of people want to say, “Poverty has fallen a lot since 1963 or 1964. That must have been caused by the great expansion of the safety net.” We strongly push back against that assertion, because we found that poverty was actually falling in the 24-year period before the War on Poverty began.
In the paper, you distinguish between poverty reduction through government transfers and poverty reduction through rising earnings. Why should policymakers pay attention to dependency as well as poverty?
That’s the exact way that President Lyndon Johnson thought about it when he declared the War on Poverty in 1964. He wanted to reduce poverty, but he cared a lot about the way in which we did it.
You can reduce poverty by more redistribution, providing more resources to people at the bottom. But that comes with some adverse behavioral effects; maybe they work less, and there could be reductions in marriage, and that could counteract some of the benefits. The other way, which is the one that President Johnson wanted, was to reduce poverty by increasing people’s earnings. People are usually better off, materially and psychologically, when they are able to overcome poverty through their own efforts.
Today, with our spiraling federal debt, there’s also some point at which we cannot continue to expand these programs. If we want to see people flourish and rise up the economic ladder, it’s going to require more than just government transfers. Whenever you redistribute resources to people with lower incomes, these programs are all means-tested. That creates an implicit tax on increasing your own earnings, because if you increase your own earnings, you’re going to reduce the amount of transfers that you receive. Those penalties can stop people from moving up the economic ladder. So those are all reasons why it’s typically better, if possible, to reduce poverty through increases in market income as opposed to more redistribution.
In your view, how much of the post-1960s slowdown in market income growth among lower-income households may be connected to some of those changing incentives from welfare policy itself?
I don’t think it’s the whole story, but it is probably one reason that we’ve seen diminished growth.
There are a few ways that redistribution can slow growth. One of them is by reducing the incentives for people to work, invest in themselves, and get married. It also requires a lot of taxes, which is possibly even a bigger factor. We have high effective marginal tax rates that disincentivize working and investing, which can slow growth. We also borrow money to fund these programs, and the more we borrow, the more taxes we pay in the future. It also means higher interest payments on our existing debt, which slows growth by driving up interest rates and reducing investment.
Many people today are arguing that unconditional cash transfers are the future of social policy in the age of AI.
I worry a lot about the universal basic income debate. My former colleague at AEI, Charles Murray, has talked about it, although for him universal basic income was a replacement for the large safety net that we already have. His idea was to spend the same amount of money, but do it in a way that doesn’t have some of these perverse incentives. The debate has moved on from that idea. Now people are saying,
“Let’s keep the existing means-tested programs and also add a universal basic income for everybody, even middle-class people.”
In the short term, that’s completely unrealistic. We don’t have the budget for it. The cost would be astronomical, and you certainly could not do it just by taxing the rich. You would need vast expansions in taxes for middle-class people too.
However, I worry a lot about universal basic income in the longer term. I don’t know what’s going to happen with AI, but I think in the back of people’s minds there’s this belief, and maybe they’re right, that with AI and all of the benefits it brings, we could see much stronger economic growth in the future. I think that people see that as an opportunity for a universal basic income, and think that once we can afford it, we should implement it. But universal basic income payments could discourage people from working, and especially from entering the labor market. Younger workers may decide, “I may not need to enter into the labor market.” And if you have less labor, you also have less of the augmentation effect of AI, which is where most of the value from AI might be created. So we could see a much smaller economic boom if the labor supply shrinks as a result of universal basic income.
Even if we don’t need human workers and we could actually get by with the machines doing everything, I still think universal basic income would be a terrible idea. People don’t just need work to fuel their own material well-being, but also for the sense of purpose that comes with contributing to society. And there will be ways to contribute to society, no matter what happens with AI.
What policies can we pursue to increase earnings opportunities for Americans without expanding dependency?
We can’t wave a magic wand and get the economic growth rate that we had during the 1940s and 1950s. And we already have a large social safety net. But we can want to design it in a way that reduces poverty and reduces dependence.
Starting in the 1960s, if you measure poverty based only on people’s market income, you see a flatlining of the poverty rate, maybe even a small increase. And at the same time, you see a greater dependence on government transfers. That pattern breaks starting in the early 1990s. Between 1990 and 2000, you see a continued reduction in poverty based on market income and a reduction in dependency.
What happened was welfare reform.
In the late ’80s, there was a lot of frustration about our cash welfare program. It was called AFDC, Aid to Families with Dependent Children. There was concern that it was breeding dependency and reducing marriage rates. In the late ’80s and early ’90s, states started experimenting with other ways of providing assistance, like having time limits on cash welfare, work requirements, and more investment into training programs and education. Then, in 1996, or early 1997, we passed welfare reform legislation that rolled those kinds of policies out nationally: time limits on assistance, work requirements, and, at the same time, expanded refundable tax credits that target families with children, including the child tax credit, which didn’t exist before 1997, and the earned income tax credit, which did exist but was much smaller until the 1990s.
In short, didn’t get any of the benefits unless you worked. And then benefits phased in as you worked more. Eventually there is a phase-out, and that has major problems, but these programs all encouraged people to enter the labor force and work. And after these reforms went into effect, we saw a growth in work efforts. Single moms had something like a 10 percentage point increase in their labor force participation. We saw many measures of child outcomes improving, including test scores and long-term outcomes. And even though we were still seeing a reduction in poverty, we were also seeing a reduction in dependency.
The 1990s example shows that, even with a large social safety net, if we design it correctly and address some of these perverse incentives, we can see reductions in poverty and reductions in dependency at the same time.
