“Equitable Growth in Conversation” is a recurring series where we talk with economists and other social scientists to help us better understand whether and how economic inequality affects economic growth and stability.

In this installment, Equitable Growth’s Executive Director and Chief Economist Heather Boushey talks with Hilary Hoynes, an economist and Haas Distinguished Chair in Economic Disparities at the Richard and Rhoda Goldman School of Public Policy at the University of California, Berkeley. From 2011 to 2016, she was a co-editor of the American Economic Review.

Each of these parts of their interview have been edited for length and clarity. [Editor’s note: This conversation took place on March 8, 2018.]

Heather Boushey: Welcome, Hilary. It is such a delight to have you here in Washington, D.C. today, and I’m glad you could come from the West Coast to hang out here in our much colder weather. I’m going to get right into it. I know you have done a lot of work over your career looking at the social safety net. One of the things that policymakers and economists often talk about is the presumed trade-offs around social safety nets, and they’re often talking about the short-term costs and benefits. What I’d like you to talk about and help us understand is how should we be thinking about the safety net, as a short-term set of policies or are there longer-term implications that we should really be much more aware of in our policymaking thinking?

Hilary Hoynes: Thank you, Heather. It’s a great question to start out with. I think you’re right that we often think about these policies in the short term—that is, they’re anti-poverty policies. So, they’re very much motivated by the fact that people have low incomes, and we want to remediate those income declines or income shortfalls. And so the way that, historically, I think, economists have thought about anti-poverty policies is within the short run, thinking about the trade-offs between the desire to help increase incomes against the potential trade-off in the sense of does it discourage work on the margin? And does that create poverty traps and disincentives to make the transition into the labor market? And so on.

Boushey: Can you just explain what you mean by “on the margin”? When you say encouraging or discouraging work on the margin, what do you mean there?

Hoynes: One of the major goals of a social safety net or anti-poverty program is to try to increase incomes to provide protection against income shortfalls. So, for example, you might lose your job. One of the wage earners might leave the household, there might be other circumstances that create perhaps short-term movements in earnings, declines in earnings. One aspect of a social safety net is to provide a floor—to provide protection against those shortfalls in a way that perhaps the unemployment insurance system doesn’t fully insure against. And so, the concern is that the very protection that a floor provides creates increases in income—and that’s what it’s aimed to do—but along the way, it also means that by providing that protection against income shortfalls, it provides an implicit tax toward increases in income. It almost has to be that providing an income floor essentially provides an implicit tax against increasing income.

Boushey: That makes a lot of sense.

Hoynes: Thanks. That’s a very common framework that, historically, economists would look at for the social safety net. And I think two facts really suggest that we need to and have started to move away from that framing.

Observation number one is that over the past 20 to 30 years, our social safety net and our resources that have been targeted at lower income households, particularly families with children, have very much moved away from providing basic income floors—such as, say, traditional cash welfare—and moved much more toward work-contingent types of activity. So, for example, the Earned Income Tax Credit and the Child Tax Credit are two major growth areas of the social safety net over the past 25 years. And those policies provide no protection in the state of the world that you’re not working and actually are very much geared toward trying to provide income supplementation conditional on work. And that sort of expansion of the social safety net occurred at the same time that we were restricting traditional cash welfare income protection in the postwelfare reform of the 1990s. That’s observation one.

Observation two is that over the past 10 years, there’s been an emerging research agenda and research studies that allow us to do some quantification or estimation of the question: Does providing more assistance through the social safety net, particularly aimed at children, lead to differences in where children end up in adulthood? This question very much feeds into a longer calculation, a more informed calculation about the long-run benefits of the social safety net—at least the part of the social safety net that’s targeted at children. And I think that’s a cleaner examination of your question, about whether we should be thinking about the long term. So, there’s two pieces to unpack.

Boushey: Yes.

Hoynes: We can talk about each one. One piece is this kind of short-run trade-off, which is just not as relevant today, as we’ve moved into work-contingent benefits. That’s Thing One, just put a pin in that. And Thing Two is, it just so happens that we’ve started to accumulate a body of evidence that all points to the fact that indeed, providing more assistance when children are young seems to lead to important improvements in where they end up in adulthood, that allow us to say something—while it’s still a young and emerging literature—about the potential benefits of protection in the long run.

Boushey: Now, that is just absolutely stunning, that economists can study how we’re supporting or not supporting children and what happens over their lifetimes. This is very much a new area of economic research in the past 10 or so years.

Hoynes: Absolutely.

Boushey: Tell us a little bit more about why we’ve been able to do this. I understand that it’s both because we have new sorts of methods but also new data. And are there any compelling examples that you could point to in terms of the data that we didn’t have 10 or 20 years ago? Why do you think that such research has emerged at this moment in time?

Hoynes: That’s an excellent question. And I always say, doing the long-run analysis requires three things. It requires a research design, and we always need a research design to do causal identification, so that’s nothing new. We always have that as part of our evaluations.

We need to have data that allows us to either follow people longitudinally or at least know something in the current period that allows us to apply these designs. That’s number two. But the other thing we need, which is easy to forget, is we need time. To evaluate the long-run effects of something, you need time to have passed. So, I think that why we have more evidence on this today is a bit about the second and third categories—that is, we do have new and emerging sets of data such as government administrative data, the use of tax data, and so on, that really allows us to ask questions that we haven’t before.

But I think it’s also the case that enough time has passed from some really critical policy expansions such as the expansions of the Earned Income Tax Credit and welfare reform in the 1990s. Another example is the massive expansions of Medicaid that were started in the late 1980s and early 1990s targeted at children. Well, now those children are 25 years older than they were then.

So, I think the work that Diane Schanzenbach [the director of the Institute for Policy Research, and the Margaret Walker Alexander Professor in the School of Education and Social Policy at Northwestern University] and I have done—which goes back a little bit further in time to the expansions of the Great Society period [in the 1960s]—means we can actually look at people in sort of middle adulthood. And I think it’s the combination of the data and time, along with maybe this kind of advanced mindset about the benefits of the social safety net, that has led us to where we are today—which is still early days, but I think they’re starting to be generating a consistent set of findings that help us learn more about the potential benefits in the long run.

Boushey: We’re really able to tap into this new body of research. So, that’s very exciting. You know, you’ve mentioned this a little bit, the importance of how the social safety net has moved increasingly toward supporting work. Another side of that is that even the programs that don’t necessarily support work have work requirements, so that they’re increasingly requiring folks to work—and, of course, the most recent example at this point in time has been the Trump administration allowing states to add work requirements to Medicaid. Kentucky was the first state to do that.

I would think that the introduction of work requirements would change the composition of who gets these benefits and how many benefits people get. But can you tell me a little bit about what we know about the effects of these work requirements on the beneficiaries and on the programs? Also, are these actually getting people into those kinds of stable jobs that allow them to get off these programs? Kind of two separate questions: Are these work requirements really as important a tool as policymakers think they are? And are they working?

Hoynes: In a way, these work requirements that you mentioned are being advanced now through waivers in Medicaid, and I just was looking at a Wall Street Journal article today about how Wisconsin is implementing work requirements for food stamps.

Boushey: Oh, wow.

Hoynes: So, I think that’s also kind of on the agenda. And I think that that gets us back a little bit to this framing of the trade-offs. That is, we want to provide protection, which is why these programs are in place, but we’re worried that it’s somehow providing incentives that prohibit individuals from moving into the labor market.

I have a couple of thoughts about that. One is that the evidence we have seems pretty convincing to me that the programs themselves are not the reason that individuals aren’t working. That is, the best evidence that looks at when programs change and how much people are working within Medicaid and food stamps doesn’t tell me that these work disincentives are very large. So, that’s the first observation. The second observation is that we actually have fairly limited evidence about what work requirements do. The most and the best evidence we have is from welfare reform. And the challenge there is many things changed at once.

Work requirements were added in the transition from Aid to Families with Dependent Children to Temporary Assistance for Needy Families [in the 1990s] but so were lifetime limits and many, many other changes that states implemented. So, it turns out, that makes it a little bit difficult to isolate the effect of work requirements from the other aspects of welfare reform. But I think what we learned there overall, taken as a package, is that welfare reform—combined with the expansion of the Earned Income Tax Credit that happened around the same time and the strong labor market—did lead to some transition of women from welfare to work.

But welfare reform was not the whole story there. It was clearly a part of the story. But I think what we’ve learned—and this feeds a little bit into thinking about recessions and concerns about the poorest of the poor—is that some fraction of individuals were simply left behind. They were not able to make that transition to work. The social safety net really shrunk for them. And it seems as though the very bottom of the distribution, as the poorest of the poor, are worse off than they were prior to welfare reform.

So, our knowledge about work requirements is still surprisingly thin. I think that what we now know is that there’s a group that is a subset of the population for whom perhaps providing more incentives to make transitions might help. There’s probably a large part of the population that doesn’t probably change their life course very much. Most people are on welfare for short-term periods of time.

Boushey: Yes.

Hoynes: And they’re going to make the transition regardless of the system. And then there’s a group that I think is harmed because they need more protection. And the challenge here might be that the population of who’s on Medicaid and who might be really affected by this is quite different from the population on cash welfare, primarily single mothers. And really, there’s a much bigger issue about low employment rates of men than there is low employment rates of women right now, and Medicaid really spans those populations.

Boushey: And so, do you think—to put more of a fine point on it—that the Medicaid population is more like the population that needs just a little push to get a job, assuming that they would get a job with access to health insurance coverage? That’s a separate question. Or do you think that they are more likely those who demographically or economically look more similar to the group that is more likely to be harmed? Do you have any sense? Or maybe that’s not knowable at this time.

Hoynes: I don’t have a lot of sense on this, but I do know that this group of nonemployed men seems to be growing into a very unique population and that we can’t really quite get a handle on what is limiting their transition into the labor market.

Boushey: Interesting.

Hoynes: And I think the other piece we’re learning a lot about right now is how hassle costs or transaction costs are really affecting who ends up on public assistance. And there’s a lot of very interesting sorts of randomized experiments going on right now in the field, where people are trying to understand why everybody doesn’t participate in food stamps if they’re eligible.

One of the things that comes out of that kind of set of studies is that the bigger the hassle cost, the more likely you are to screen out those who are the most disadvantaged. For one reason or another, whether it’s like this behavioral economics view of poverty as being partly about bandwidth and scope to get what you need done or because of the stress burdens of poverty, hassles may decrease the targeting of programs.

The bottom line is, putting these work requirements in place requires people to think about this population. They’re transitionally working in and out [of the labor market]. It’s a reality of the low-wage labor market. And you’ve got to stay in compliance to continue to get Medicaid as your work, which is kind of erratic and not fully under your control.

My prediction would be that it’s going to be very hard [for this population] to keep up with the paperwork that is necessary to stay insured. And, as you point out, the jobs that this population are likely to be employed in don’t have health insurance. So, they lose their jobs, they fall out of compliance, they’ve got to go back up that hill and through those hassle costs to get back in compliance, and my concern would be that that would just lead to a real decline in who ends up getting Medicaid.

Boushey: Yes, or getting health care at all.

Hoynes: Or getting health care at all.

Boushey: Right. Because with the Affordable Care Act, they should be able to get health care once they get a job, a subsidized plan, but many states—and Kentucky in particular—have cut back on the availability of those plans. So, it seems like it’s a one-two punch that is making it harder to get access to health care if you’re at the very bottom but not more likely that you’re going to get it if you’re in that low-wage job.

Hoynes: Exactly.

Boushey: And, of course, as we talked about a little bit earlier, there are long-term implications we know for children. But, one would also imagine, adults who are denied access to these basic needs have implications for economic growth and stability, because those folks are not able to participate in the labor market as much. So, this is seems to be a policy that is pushing away from the kind of growth we want to see.

Hoynes: Right.

Boushey: So, thinking about the growth question for a minute, one of the things about the social safety net is that there are a lot of folks who think about it as a set of policies to support people in poverty. That’s one super important function that [the social safety net] serve[s]. But then, of course, if you go to a macroeconomics conference, or if you talk to people who are thinking about macroeconomic stability, then the conversation about these social safety net programs will often be about so-called countercyclical effects—that in bad economic times, in times of high unemployment, the number of people on these programs should automatically increase, so that they “automatically stabilize” the economy. And then that adds to demand, and then that helps make the recession weaker.

One of the things that characterized the Great Recession was that it was so deep and dark and had so many effects on these various social safety net programs, including the budget cutbacks in the states because of the budget crisis in so many states. Can you tell me a little bit about the strength of the safety net as we think about preparations for a future recession? What are particular things that we should be really concerned about shoring up now, or do you think that everything seems to be working great and we should just keep going? Do you have a sense of how we should be advising the macroeconomic community to think about that piece of the puzzle?

Hoynes: So, to back up a little bit, this massive change in our social safety net toward being more targeted at work-contingent policies (such as the Earned Income Tax Credit) and away from traditional welfare very much tees up the question that you ask. And in fact, I’ve done a lot of work with [University of California, Davis economist] Marion Bitler that looks at the Great Recession as an opportunity to what we call stress test the social safety net. And what we found from that work is that our policies are doing a pretty good job—at least compared to what they had done in the past—for families with incomes around the poverty level.

And so, if you think of, say, a three-person family in the current period, that would mean at income levels of around $20,000 to $22,000 a year. During the Great Recession, what was particularly effective was the Supplemental Nutrition Assistance Program. It’s not block granted. It’s an entitlement. It’s federally funded, so it provided protection against—as you pointed out—the state challenges that were in existence in the Great Recession. And so, it provided quite important protections that essentially insured against incomes of around 100 percent of the poverty level, compared to what we’d done before.

So, you could make a comparison to other countries, but maybe that’s not so relevant. However, what is also very clear in the Great Recession is that incomes below the poverty level—like around 50 percent of poverty and below, which we call deep poverty—were much more affected in the Great Recession than had been the case in earlier recessions. And we can very much draw a straight line between that and the fact that the cash out-of-work social safety net was no longer there.

Boushey: Wow.

Hoynes: Temporary Assistance for Needy Families, the reformed cash welfare system, was never very generous, but it did provide a floor around 30 percent to 50 percent of poverty in most states. And after welfare reform and the block granting and the diversion of payments to other activities, what we saw in the Great Recession was that states’ welfare expenses and number of people receiving welfare was actually decreasing in the Great Recession, and states that experienced a larger economic shock experienced no greater adjustment in their cash welfare assistance, whereas with the Supplemental Nutrition Assistance Program and other programs, you could see in the data that states that experienced a bigger economic shock saw a greater increase in their participation in SNAP, as you suggest, as a standard countercyclical response.

I think going forward, the concern would be for the next recession that we haven’t done anything to fix that gap. So, we’re doing okay—again, relative to historical experience—because we have SNAP and that remains strong, and the Earned Income Tax Credit and the Child Tax Credit are two policies that do a lot of anti-poverty fighting. The CTC is the biggest anti-poverty program for children in America, but you need to have earnings to get the EITC. And so, if you lose your job, you lose your earnings and the earnings subsidy through the EITC.

Boushey: Wow. And if you get unemployment benefits, do you get the EITC?

Hoynes: Unemployment benefits do not count as earnings.

Boushey: So, even if you then are lucky enough to get unemployment insurance benefits…

Hoynes: Exactly. So, that provides another protection—unemployment insurance—but to some degree it would crowd out the Earned Income Tax Credit. So, an interesting side-step to this is that for slightly higher-income families—say couples, where there’s two potential earners—if one loses their job, their income can kind of bring them down into the EITC range. So, the EITC does provide protection—and we see this in the data—against falling incomes for slightly more advantaged, typically two-parent households, but it doesn’t do much to insure against earnings losses and income losses for one-earner households.

Boushey: Well, that’s a lot of food for thought for thinking about the preparations. I want to get into another meaty question here. I feel like these days, a lot of people are bringing up this question of whether or not we should either add on to or replace—which is a huge difference—our social safety net with a universal basic income. And I wondered if you have any thoughts on the pros and cons of a universal basic income.

Hoynes: I think, as with many things, the devil is in the details on a universal basic income. I mean, on the one hand, my own view is that we are indeed providing less protection at the bottom of the income distribution. And I think there’s a lot of reasons to be concerned about that, both in the short term and, to get back to our earlier conversation, in the longer term, because of potential disadvantages of the long-term consequences of having disadvantages, particularly in childhood. So, I think that there is a strong argument that we need to do more to shore up the basic income floor in the United States. On the other hand, the universal basic income is often argued as a solution to the loss of jobs and robots taking over the world and all the rest. And I don’t think that the universal basic income is very well-suited to do that.

Many people are talking about the universal basic income as a substitute for other parts of the social safety net that we know are fairly targeted, and so my concern would be that either this would be a very expensive policy or it wouldn’t be funded at a very high level, and that would make many individuals much worse off with the resources that are currently more targeted at the bottom of the income distribution. So, I think there’s a strong argument for trying to shore up the income floor, but the universality of it doesn’t make sense to me.

Boushey: Yes.

Hoynes: The universality of having this available all the way up the income distribution or even far up the income distribution is so costly that I don’t know how we do that while maintaining sufficient support for those who are most disadvantaged.

Boushey: It’s a great set of questions that we need to think about, and I know that a lot of folks are focused on a research agenda around that. Which leads me to my last question, which is that Equitable Growth is a grant-making institution, and so I’d like to ask you what, in your view, are a couple of the most important research questions that we should be funding that are out there that really would push our knowledge further?

One of the things [we’ve] learned at Equitable Growth is that data can be expensive, and it may be that we should be prioritizing more data purchases for scholars, but I’m just throwing that in as an option for you to think about in your answer as well.

Hoynes: I think one of the things that’s on the top of my list that fits into what’s important for Equitable Growth is another dimension of equity, which is regional. That is one of the things that we’re really seeing and, you know, we haven’t really talked about. It’s not that prominent and present in the work that I do, though I think it is incredibly important that we’re starting to see many more differences across areas in the United States in basically any measure you might look at—employment, well-being, health, and all the rest. And our lens has typically been to look at differences in disadvantage across socioeconomic status groups. I think what’s becoming clear is we need to really focus our lens and start thinking a lot more about differences across areas and about whether our policies are in order to try to remedy those regional differences.

I think for a long time, economists have thought that we don’t need to worry about regional differences, that people will migrate, firms will migrate, costs will equilibrate, and things will work themselves out in the long run. But one of the things that we’re seeing is that migration rates have really fallen in the United States, particularly out of these disadvantaged areas. We don’t really understand why. There’s a kind of theory out there that part of it has to do with the social safety net, which gets back to where we started the conversation—though I don’t think the evidence is all that well-developed at this point. So, I think that high on my list would be to try to do more to understand both the root causes of these differences across place and to think about how our policies might be adapted to better address the issue.

Boushey: Well, that brings up a question, in economics, about our unit of analysis as the individual.

Hoynes: Right.

Boushey: When economists make policy recommendations, they’re making recommendations, as you said, across socioeconomic lines. It’s basically about individuals’ incentives, and in a lot of my work, I’ve tried to blow up the frame a little bit to think about the unit of analysis being the family.

Hoynes: Absolutely.

Boushey: The family being an economic unit. But how you put that into our economic framework is complicated. Thinking about the regional differences kind of actually asks us to blow that up another step and think about the unit of analysis maybe as a community. We don’t have families in our toolbox, and we certainly don’t have communities in our toolbox. One line of thought could be that we should be doing more to encourage economists to partner with people in other disciplines that have more of an understanding of how to think that way. Or if we need to go back and rethink our econ toolkit. I agree with you that this is urgent and important, but it brings up, for me, a profound question on economics. I don’t know what your reaction to that is—or, if you have good answers, I’d love to hear them.

Hoynes: Well, it may be that we don’t need to change our toolkit to think of place as the unit of analysis, but it may be that we need to rethink our policy toolkit to try to address what seem to be the facts on the ground. So, for example, one thing that I’ve heard advanced is whether or not we should have an Earned Income Tax Credit or have a little bit of an additional top-up in areas that are more urgently needing transitions into work.

So, in the same way that our unemployment insurance system automatically kicks in for longer periods of time if your state is experiencing an above-normal unemployment rate, should we be thinking about other aspects of our social safety net that are trying to encourage work as having additional generosity in some dimension or another in areas that are having above-normal levels of unemployment or nonemployment? I’m not sure whether that’s the end of the story, but it certainly seems to be a way that we could stay within our paradigm but think about reorienting our policy toolkit a little bit more toward the facts on the ground.

Boushey: That sounds great. Well, this sounds like a great place to end. I want to thank you for taking the time today. It’s just always a delight to talk to you and to learn from your research. Thank you.

Hoynes: Thank you very much.