When the 1996 welfare reform act was passed, pundits expressed heated opinions about how the legislation would affect the nation's poor. Some claimed it would finally end the "culture of dependency" and help people move up the economic ladder. Others said that it would send single mothers and their children into destitution.
Now almost a decade after the legislation was enacted, Robert Moffitt, a labor economist at Johns Hopkins University in Baltimore, has looked carefully at the data. He has found that the facts are somewhat more complicated than the punditocracy predicted. In general, former welfare recipients have been successful in joining the labor force. But there remains a segment of the population for whom regular work has proved elusive.
Moffitt's research also has addressed such issues as income inequality, the effects of marginal tax rates on labor supply, and the economics of marriage. In addition, he has made a number of important methodological contributions, and is a Fellow of the Econometric Society. Since July 2004, he has served as chief editor of the American Economic Review.
Prior to joining the Johns Hopkins faculty, Moffitt taught at Brown University and Rutgers University. He also held visiting positions at the University of Wisconsin and the University of Maryland, and worked at Mathematica Policy Research, Inc.
Aaron Steelman interviewed Moffitt at Johns Hopkins on July 27, 2005.
RF: I want to start off by asking a few questions about welfare reform. How have single mothers who have left welfare fared in the workplace, both in terms of employment rates and the wages they have been able to command?
Moffitt: The employment rates of single mothers who left welfare after the 1996 legislation have been in the range of 60 to 80 percent. Those rates have generally been viewed as remarkably high. While on welfare, these women had not worked more than 20 to 25 percent of the time and were regarded as essentially unemployable. As for their wage rates, they have been in the $8 to $10 per hour range, well above the minimum. The jobs held by these former welfare recipients are not necessarily the best jobs in the world — they are still quite unskilled, often unpleasant, and sometimes unstable — but they pay decent wages. The combined result of their increase in employment and increase in wage rates has been an increase in their earnings as well. But their total incomes have not risen much, on average, because their lost welfare benefits have been close to their increased earnings. But one of the goals of the welfare reform legislation was to get welfare recipients into employment and out of dependence on federal support, and from that standpoint it appears to have worked.
RF: The welfare reform act was implemented during a period when labor markets were unusually strong. Were the women who moved from welfare to work disproportionately affected by the recession of 2001?
Moffitt: The results on the recession are pretty surprising. The employment rates of single mothers and former welfare mothers fell only slightly during the 2001 recession. So instead of 70 percent, they were about 68 percent — still far higher than they had been in 1996 when the legislation was enacted. I'm less familiar with the data on wage rates, but I would suspect that they didn't fall much either. Consequently, the recession does not appear to have driven many former welfare recipients back into unemployment or back onto welfare.
This finding really reinforces the first one. The first major concern about welfare reform was the fear that a substantial share of women on welfare would not be able to handle the reform and would be unable to find jobs and would possibly become homeless, resulting in widespread destitution. That did not happen and, in fact, the majority of women successfully negotiated the reform. The second concern was that this success was only a result of the good economy, and that the women who had left welfare would be in trouble once the economy slowed. That appears not to have happened either.
So, in retrospect, it appears that many of these women did have sufficient skills to work and that welfare reform gave them the push to enter the labor market and establish themselves. Again, however, it is always necessary to caution against too rosy a picture. The jobs held by former welfare recipients are not particularly good jobs, and most of the evidence shows that those jobs do not lead to much wage growth or advancement. Moreover, many former recipients have serious problems of other kinds. But, on average, welfare reform has been a success story. I don't think even most of the legislation's original critics would object to the gist of this conclusion. Virtually everyone agrees that we should stick with the basic approach of the legislation, though many would argue that we should fine-tune it to better protect the most vulnerable people in the system and perhaps to go back and do a bit of patching up of the holes in the safety net.
RF: For several decades — in fact, almost since the inception of Aid to Families with Dependent Children (AFDC) — there were calls for reforming the welfare system. Yet we didn't see major legislation until the mid-1990s. What were the constellation of events that made such legislation possible?
Moffitt: The public's attitude toward welfare and many other transfer programs has become more conservative over time. The 1960s were probably the high watermark of government activism and strengthening of the social safety net, and since then confidence in government's ability to deal with social problems has generally diminished. Congress attempted a series of welfare reforms in the 1970s and the 1980s that either didn't pass or that weren't particularly strong. But all those reforms were in the same direction: an increased emphasis on work and stronger work requirements. In addition, during the period when Congress was unable to craft a major overhaul of welfare, it allowed the states to experiment on their own. The states started reforming their welfare systems in a more conservative direction, before there was final federal action. So even though the 1996 legislation was a landmark, I think we need to see it as the culmination of a series of smaller steps which eventually led us to a tipping point.
As for the source of the increased conservatism on the part of voters, I think that the increased labor force participation of middle-class women was part of the cause. That transformation really changed the attitude of voters. Once a large percentage of middle-class women were working and putting their children into day care, the public began to question why we shouldn't expect the same thing from poor women. There was no longer the support for paying women to stay at home with their children, which was the goal of the original legislation in 1935.
Another turn against welfare, I think, has to do with the changing composition of the welfare caseload. In the 1960s, the caseload was largely composed of divorced women. One could imagine that members of the middle class, while not looking favorably upon divorce, understood it because many of them were getting divorced too. But by the 1980s, the caseload started to become composed largely of young women who had never been married and were having children out of wedlock. That is a completely different group, and the middle class had a great deal less sympathy toward those women.
A final factor is that I think the attitudes among women receiving welfare changed. If you look at attitudinal studies from the early 1990s, many welfare recipients said that they didn't like welfare, that they thought other women were gaming the system to stay on welfare, and were not really trying to improve their lives. Welfare recipients had incorporated the social norms of the middle class. And once the legislation led some recipients to move off welfare, it had a snowball effect. They began to exert social pressure on other women to find work. I think that increased stigma within poor populations made it easier to overhaul the welfare system. But it took a major shock; incremental reform would not have done it.
RF: What's the share of former welfare recipients, in your estimation, who have very significant employment problems?
Moffitt: I mentioned earlier that about 70 percent of former welfare recipients are working at any time. One could take this to mean that the other 30 percent are the ones with significant employment problems. But that would be an overestimate, because some of those women aren't working at the time and yet have some jobs skills which make it possible for them to find work in the private sector. Instead, there is a smaller group composed of women who have one or more serious problems such as cognitive limitations or poor education which make it difficult for employers to want to hire them, or serious health issues, or substance abuse, or domestic violence.
Many women have multiple problems of this sort, and it is difficult for them to get help if they are off welfare — an irony is that if you are on welfare, you get identified, you have a caseworker, and you are directed to relevant programs. But once you are off welfare you have to find that help yourself, and many women simply don't have the wherewithal to do that effectively. There have been studies trying to count the number of women with each of the problems I listed, how many have multiple problems, and so on. My estimate would be that the share of women who are in this category is no less than 5 percent and no more than 15 percent. Some might think that the percent is slightly higher.
I think that most observers agree that the next step in welfare reform is to address the problems facing this segment of the population. That is going to be very difficult, especially given the fiscal situation the states have faced recently. Many of the social programs that were designed to help these women were conducted at the state level, and when state budgets got tight, those programs were often among the first programs to have their funding reduced. Perhaps their funding will be restored now that revenues are growing again at the state level.
RF: You have done some work on the growth in wage inequality. Please tell us a little bit about your findings.
Moffitt: My work, mostly with Peter Gottschalk at Boston College, began by addressing the perception of all analysts that the increase in wage inequality arose entirely from a change in what we may call permanent earnings, driven by such factors as increased demand for high-skilled labor. But when I and Gottschalk used the panel data and calculated changes in individual earnings over time in the 1970s and 1980s, we found that 50 percent of the increase in the cross-sectional variance was due instead to increases in the transitory component of earnings. In other words, earnings have become more volatile over time and this has been a major contribution to the cross-sectional increase in inequality, yet it has a very different interpretation than the usual one.
Increases in the transitory component have occurred at all levels of the skill distribution but your view of increases in volatility might be different depending on whether you are examining the top or the bottom. If you examine the very top, much of the increased volatility comes from skilled workers who earn big bonuses in one year but don't the next. These workers are doing well and their average earnings are rising, but they are in occupations where their earnings can fluctuate widely from year to year. This strikes me as something that shouldn't worry us a great deal and, in fact, could be argued to be a sign of a more productive and more competitive labor market. But at the middle and the bottom of the distribution, I think there is more of a concern. Earnings on low-wage jobs have become more unstable at the same time that average earnings on those jobs have been declining, making it seem as though the jobs themselves are getting worse in multiple dimensions.
I don't have a conclusive answer for the cause of this change in the labor market, but I wonder if there has not been an increase in the general volatility of many industries in the United States today. Competition is stiffer today in many industries than it used to be, with firms rising and falling faster than they used to, which can produce large peaks and valleys in the earnings of those who work for those companies.
RF: In the early 1960s, Milton Friedman pushed for the implementation of a negative income tax. That program was never implemented in the form that Friedman advocated, though certain elements have shown up in policies that subsequently have been enacted. How do you think the negative income tax has influenced the policy debate regarding aid to low-income people?
Moffitt: The negative income tax is a topic that has fascinated economists for a long time. The basic idea of a reduction in the implicit tax rate on working for welfare recipients appeals to basic economic principles. When I left graduate school in the 1970s, for example, it was the hope among many economists that the United States would adopt something similar to a negative income tax. It has been extremely popular among academics and scholars, but it has turned out to be less popular among policymakers.
There was a reform of the AFDC system in the late 1960s which lowered the marginal tax rate on earnings for women in the AFDC program. But the reduction was pretty small, so the effects on labor force participation were limited. It wasn't really until the welfare reform of 1996 that major reductions in the marginal tax rate have occurred; for most of these last few decades, the rate has been at or very close to 100 percent.
The negative income tax had a number of other features which have likewise fared pretty poorly in the public domain. Perhaps the most notable was that the negative income tax was supposed to replace all other transfer programs — that there should be one simple program that covered all needs. That was an essential part of the idea according to Friedman, but it is an idea that has been decisively rejected by Congress and the voters.
We still have numerous special programs for food, medical care, housing, child care, and the like. Indeed, the pure general-purpose cash programs have become very small relative to everything else. From an economist's standpoint, this is problematic. We generally believe that the most efficient way to help people is to give them a direct cash transfer, because they know better than us what their needs are and how they should allocate their resources. But the public, I think, believes that the poor have demonstrated that they do not make good choices on their own, and that we should give them the services we believe they need rather than allowing them to purchase them on their own. The voters, I believe, are basically paternalistic toward the poor and this has shaped the policies we have adopted.
There are only two cash programs of any significance today, and they both reinforce this conclusion. One is the Supplemental Security Income program, which goes to the aged, blind, and disabled people with low incomes. Those individuals are not viewed as having bad preferences or bad habits. Instead, they are in their situation through no fault of their own, and the public believes we ought to treat them as such. The second program is the Earned Income Tax Credit. This program is targeted at earners; by definition, you have to be working in order to qualify for the program. So with both of these programs, the recipients are seen as having good preferences, and that they can be trusted to wisely spend money that is transferred to them.
RF: Have you looked at the effect recent reductions in marginal income tax rates have had on the labor supply of high-income male workers? If not, can your work on the effects of the 1986 tax reform act shed light on the responsiveness of such workers to alter their labor decisions when faced with lower marginal rates?
Moffitt: I haven't done work on the recent reductions in marginal income tax rates, but I do have some thoughts on the topic given my previous research. It is an interesting topic because there has been a sharp change in the way economists view the evidence. In the 1960s and 1970s, most economists did not think that lower marginal rates would have much of an effect on the labor supply of high-income workers. The first person who really challenged that view was Jerry Hausman, who found that there were some nontrivial effects at the top that led to significant deadweight losses. But when other analysts looked at his data more closely, it wasn't clear that the evidence supported such a conclusion because sample sizes at the top were very small.
Then Marty Feldstein began examining this question by using tax data with much larger samples of high-income workers, and he, too, argued that there were significant effects on the labor supply of high-income males — as marginal tax rates went down, the amount of taxable income reported increased. Another new point made by Feldstein was that much of the responsiveness was in increased wage rates that came from increased incentives to find better employment opportunities and even entrepreneurial activities; the response was not entirely in hours of work.
In my work on the issue, conducted with Mark Wilhelm, we went back and looked at hours of work following the 1986 tax cuts, focusing again on the very top of the income distribution. We found no effect. The reason turned out to be fairly straightforward: The workers at the very top were already working extremely long hours (3,000 to 4,000 hours per year) and really couldn't work much more. There simply wasn't room for them to respond to lower marginal tax rates in the way that others had argued or that theory might predict. So how can we square our findings with Feldstein’s findings? One is simply that the response was in earnings, not hours of work, as Feldstein pointed out. But I think that there is another factor at work as well, which relates to the way that workers at the very top structure their compensation packages.
We have fairly good evidence that high earners pay close attention to the tax treatment of their income. When tax rates were high, they received more of their compensation in nontaxable forms. But when tax rates fell, there was less reason to do this, so more of their compensation began to move into taxable forms. This process can make it appear that lower marginal rates increase earnings when in fact it was just a change from nontaxable to taxable form.
As I said, I haven't looked at the data from the most recent rounds of tax cuts. But I would be surprised if the evidence was much different from what we found in the 1980s. I would doubt that the biggest earners in the economy have increased their labor output much as a result of lower marginal tax rates.
How about elsewhere in the income distribution? Here, too, I think the evidence is fairly mixed. For males, it is not clear that lower marginal tax rates induce them to work more. Many males in the labor force are already working 35 to 40 hours a week and they do not have much opportunity to increase that number. Their employers simply do not offer them overtime or flexible hours. In order to increase their labor supply, they would have to find a second job. But for females who are not in the labor force, lower marginal rates can have a significant effect on their decision to work. When rates are high, they may decide that work is not worth it — that there are important things to do at home which are more valuable than working. But when rates are lowered, their calculation changes. They are now able to keep a larger share of their earnings and may decide that entering the labor force is worthwhile.
I should also add that the self-employed face different incentives too. Unlike adult males working for someone else, the self-employed generally have an opportunity to alter their work patterns — that is, to increase their labor supply in response to lower marginal tax rates. So I wouldn't be surprised if this were to show up in the data.
RF: How does increased compensation affect soldiers' decisions to re-enlist?
Moffitt: I have tried to build some simple models which assume that soldiers are forward-looking agents. Much of this work was co-authored with Tom Daula. I have found that the bonuses the military has offered to get soldiers to re-enlist do not have a large effect. They simply are not big enough to change most soldiers' decisions. What is more important is the type of training that a soldier can expect to get in the military. Let's say that you join when you are 18 and that you make a career of it, meaning you spend 20 years with the military. That means you get out when you are 38. Sure, you get a nice military retirement package when you leave. But at 38, you are still a relatively young man and you will want to have acquired skills that will allow you to have a career on the outside at a good job. Those type of issues tend to dominate compensation issues, especially when the compensation takes the form of a one-time bonus payment for re-enlistment.
RF: There are some who have argued that population growth will ultimately lead to severe social and economic problems. I'm thinking of people like Paul Ehrlich, for instance. And then there are others — Julian Simon probably being the most prominent example — who have argued that population growth has unambiguously positive effects. Let us assume that these two positions define the extreme positions of the debate. Which one do you find more consistent with the evidence?
Moffitt: Although I have done work on the economics of fertility, I have not done work on this specific question. However, I have followed the debate fairly closely. As far as I can tell, the best work on that issue can be found in a couple of volumes put out by the National Academy of Sciences that examined how population growth affects a whole host of issues, including the environment, health, per-capita income, and others. And when you look at the data, it's fairly hard to find major negative consequences of population growth. You can build models where this might be the case, but the empirical evidence seems fairly straightforward, and it is closer to Julian Simon's view than to Paul Ehrlich's.
I think that economists have generally been persuaded that population growth, on average, has positive effects — and so, too, have demographers, a group that used to include a pretty large number of population growth opponents. Also, I think most people would agree that we do not face a "population bomb" except, possibly, in Africa, and AIDS has changed things rather dramatically there. Quite the opposite: In many developed countries, population growth is now below the replacement rate.
RF: What do you see as the major challenges facing the American Economic Review and academic publishing in economics more generally?
Moffitt: I took over as chief editor of the AER in July 2004 and have therefore been in the job now a little over a year. There are both specific and general issues with the AER and with economic journals in general. By far the most important specific issue is turnaround time for papers, the time it takes from when a paper is sent to a referee to when I finally get enough reports to make a decision. At the AER, this time is often quite long, six months or more, and that has become a serious problem. It is worse at some other journals but better at some others, but it is a systematic problem at all economic journals. I do not quite understand why this is the case, because you don't see it in the other social sciences. There have been a number of explanations offered — Glenn Ellison has done the best work on this — but none of which I find completely convincing. But one thing is certain: When I talk to those economists interested in submitting papers at the AER and elsewhere, this is their biggest complaint. Working on this problem at the AER is one of my highest priorities.
Another less specific problem one hears from authors is that the quality of the referee reports is low, sometimes occurring when reports express very divergent opinions. Many authors feel as though a paper has not received a fair reading by at least one of the referees. While there are limits to how this perception can be changed, because it is only human nature to feel that way when your work is rejected, the one thing I have done at the AER to address this problem is to increase the number of co-editors on the journal. This introduces more expertise among the editors and allows a better choice of referees and better judgements after the reports come in.
Finally, my major goal for the AER is to see it build upon its niche. The number of journals in economics has been growing sharply, but there are still really only three leading general interest journals: the AER, the Quarterly Journal of Economics, and the Journal of Political Economy. I think that the QJE and the JPE have become increasingly associated with specific schools of thought. Both of those journals are affiliated with university economics departments — the QJE with Harvard and the JPE with Chicago — and I think a many of the papers that appear in those journals reflect the perspectives dominant in those departments.
In contrast, the AER is the official journal of the American Economic Association and, as such, it should be expected to represent the profession as a whole. I would like to see the AER be open to a wide variety of perspectives, methodological and substantive, and to publish the best work that is conducted in all of those perspectives. I think that the AER has been fairly successful in that regard in the past, but there are areas for improvement, and it is a very important objective that deserves constant attention.