Interview with James Heckman
Nobel prize-winning economist James Heckman on discrimination, job training and early childhood education.
Published June 1, 2005 | June 2005 issue
It's 10 a.m. on a bright Tuesday morning, and University of Chicago economist James Heckman is at his professorial best, lecturing on the impact of taxes on labor supply, education and skills acquisition. Distilling years of research into a brief presentation, he speaks quickly and forcefully, gesturing vigorously, throwing himself into the talk.
Heckman is so eager to share knowledge that his students are soon caught up in his enthusiasm. Initially distracted, they become visibly more attentive as Heckman proceeds—scribbling copious notes, asking pointed questions.
Afterward, one shakes Heckman's hand energetically and calls the lecture a "tour de force."
As it happens, the forum is not a college class but a presidential commission; his students are former U.S. senators, heads of federal agencies, tax attorneys and academic economists. And Heckman's ability to enthuse these Washington insiders and fellow scholars speaks compellingly about his qualities as an economist: penetrating scholarship, masterful grasp of principle and detail and, most especially, a passionate belief in the importance of the work.
These traits led in 1983 to the John Bates Clark award, the profession's biannual recognition of the top American economist under 40. And in 2000, Heckman was honored with the Nobel Prize for his analysis of selective samples. He had shown that much of the data available to social scientists were rendered useless by selection bias—simply put, samples usually aren't random. And he then had developed an innovative technique to correct for the bias. This statistical alchemy—now known as the Heckman two-stage method—has enabled scholars ever since to generate valid estimates from skewed data.
In Heckman, analytical brilliance is partnered with implausible vitality. "Almost unlimited energy" is the description he garners from University of Chicago colleague Gary Becker. This drive finds its source in a sense of personal responsibility for research into meaningful questions that range from the esoteric reaches of econometric methodology to practical strategies for childhood education—topics fielded with characteristic gusto in the following conversation.
Region: Some of your work has focused on discrimination, specifically relative to African Americans, and I understand that one source of motivation for that research was a road trip you took over 40 years ago.
Heckman: Yes, that's true. That and the fact that I lived in the border South for about two years, in Kentucky and Oklahoma. I was born in Chicago, Hyde Park actually, and I was really quite innocent about racial disparity when my parents moved to Lexington, Kentucky. I must have been about 12 years of age. My father worked for Armour and Company, once a big meatpacker here in Chicago, and he was transferred there.
When we went to Lexington, for the first time in my life, I saw organized segregation. My sister and I were dumbfounded to see blacks riding in the back of the bus. In fact, that's how we first encountered segregation. When we got on our first bus in Lexington, we went to the back because we liked the big bay windows—you could see everything. We were told, "This isn't right. You're not supposed to ride in the back. That's for these people." And I remember seeing water fountains and park benches marked "whites only" or "colored only."
The dichotomy in treatment was so strange. It definitely left an impression, and it was reinforced when my parents moved to Oklahoma City, which was also segregated at that time.
What fascinated me, among other things, was the deep resistance to change or even experimentation. When my family arrived in Lexington, a group of people came to our house and said, 'We know you're from Chicago. We want you to understand this is the South. We have different ways down here.'
Region: Quite a welcome wagon.
Heckman: It wasn't said in a hostile way. But they made a very definite point that there were Southern ways. So I really felt I was in a different culture, and it became very interesting to me, as a kid of that age, to see this racial code in action. There were no Klan activities. I saw no lynchings. But the open racism was amazing to me. So were the segregated schools. I attended segregated schools. I saw no blacks in my school except in menial capacities. The black students were all in a school somewhere far away. I'd see them going to school but would never interact with them. There were no occasions for interaction.
Late in 1963, I took a trip South with my college roommate from Nigeria. We drove to Lexington to see some friends of mine, then into Birmingham, Alabama, and then through the Mississippi Delta area, very close actually to where [civil rights workers] Cheney, Goodman and Schwerner were murdered the next summer. Eventually, we arrived in New Orleans. This was just after the bombing of the church in Birmingham where the little girls were killed, so there was an atmosphere of great tension.
In Birmingham, we stayed at the black YMCA, and the people there were frightened to death because I was breaking the local Jim Crow law. When we went into Hattiesburg, Mississippi, the next day—it happened to be New Year's Eve actually—the police who stopped us said, "Are you guys integrating us? Are you trying to make trouble?" And they carefully arranged for us to stay in separate hotels. It got a little scary because they were monitoring us so closely.
Driving across the Black Belt, we were treated as if we were civil rights workers. Black people would wave to us—they'd see a black man and a white man in a car. And when we went into local stores, we were asked to go into separate lines to buy food and other items.
Then we drove down to New Orleans. It was January 1st. It was right after a Sugar Bowl game, Ole Miss versus Alabama. And again, before I arrived I did not realize how strict the race code was. We walked up and down Bourbon Street, and I'll never forget what happened. There are hawkers there trying to get customers to come in to their places of business. When they saw the two of us approaching, they closed the doors. It was universal.
Seven years later, I was at an American Economics Association meeting in New Orleans. This was my first conference as an academic. I went back to the same places. I was amazed. It was completely integrated, totally changed. Same doors, same town.
The issue of why blacks got disparate treatment has deeply interested me for a long time. The sources of the rapid change in their treatment also interested me. These issues still fascinate me today. Trying to understand the sources of black-white disparity will occupy me throughout my life. It's actually what led me to my recent research on early childhood, because when one starts looking at gaps in achievement among racial and ethnic groups, one realizes that despite many efforts to improve the circumstances of the African American population, a lot of progress remains to be accomplished.
Region: Doesn't one of your early papers on discrimination argue that substantial progress was achieved through legislation?
Heckman: The civil rights movement definitely had an effect and, yes, I wrote a paper on exactly that topic. Actually, that research ran into a little controversy. It still does, in fact. There was a lot of resistance among some in the academic community at [the University of] Chicago to the idea that the government had played such a powerful positive role.
The progress that African Americans experienced in the second half of the 20th century occurred primarily in the South. I saw this rapid change firsthand, and I studied it in the Southern textile industry.
Migration trends verify this claim. Blacks were steadily leaving the South from the 1880s, 1890s. Out-migration swelled in the 1940s and '50s. Blacks were coming to Chicago, Detroit, New York, New Jersey and so forth. In the late '60s, though, the pattern of migration reversed, for the first time in the 20th century. Migration is a movement to opportunity, and the reversal indicated that conditions had changed fundamentally in the South.
I studied the textile industry, where I had data going back to 1915, and found tremendous changes occurring in a three-year period after the passage and enforcement of Title VII [of the Civil Rights Act] in 1964. In South Carolina and the South in general, there was massive integration.
It took me a long time to piece this evidence together. And to be honest, I found that some of my colleagues at Chicago were very hostile to this finding, and some remain so. Some want to believe that markets by themselves will solve problems like racial disparity. Markets do many useful things, but they did not solve the problem of race. Not in America. That's probably heresy to admit it as a Chicago economist, but I became convinced that a doctrinaire notion that markets would solve the problem of discrimination is false. Civil rights legislation and civil rights activity played huge roles in eliminating overt segregation in the United States. On the other hand, I also believe that affirmative action in the post-civil rights era has played very little role in elevating the status of blacks. It is important to notice that many blacks are not in the workforce, and the trend of workforce withdrawal even among prime-age black males has increased over the past 20 years. Since their wages are missing, we don't know what true black-white gaps are.
I'm still very interested in the question of black-white disparity, but I think about it quite differently than I used to. The blatant discrimination that existed in the South before Title VII was in large part eradicated by civil rights legislation. It's far less of an issue today. There's still disparity, of course, but it's not now primarily due to discrimination. I now think it's much more due to the differentials in family environments and the fact that the initial life circumstances of racial and ethnic groups are very unequal. And understanding that, I think, is the source of solving the black-white problem, not new civil rights laws, and certainly not affirmative action laws. So yes, the race issue fascinates me.
THE BELL CURVE
Region: In 1995, you wrote a very strong critique of The Bell Curve, Herrnstein and Murray's book about IQ, genetics and ability, which argued that nature far outweighs nurture.
Heckman: My review wasn't as negative as those of others. I think the book was very important. It broke a taboo by showing that differences in ability existed and predicted a variety of socioeconomic outcomes. So, I thought the book was important in raising that issue, but it failed totally when it focused so much on genetic determination of ability. It had no hard evidence on genetics. The youngest person in the Herrnstein-Murray sample was 14 years of age at the start of the sample. By the time they are 14, people are pretty well formed and environments play a big role. The idea that a test score measured at age 14 was a good measure of genetic determinism is absurd.
Region: What have you found in your own research about the effects of schooling on test scores?
Heckman: Very strong effects, much stronger than what Herrnstein and Murray claim in their book. In a paper published last year with Kathleen Mullen and Karsten Hansen in the Journal of Econometrics, we found substantial effects of an extra year's schooling on the Armed Forces Qualifying Test, the same test they used. The point is that the test they used is an achievement test. It embodies knowledge that people acquire through experience.
I thought the book played a very important role in raising the issue of differences in ability and their importance. It stimulated discussion if only by being a target of attack. There's an awful lot of convention in academic life. And The Bell Curve was important precisely because the topic of ability had become off-limits to "right-minded" people. It forced scholars to confront important facts about differences among people. I think that was the contribution of the book. So, actually, I'm a bigger fan of it than you might think.
Before the book was published, psychologists studied test scores but never matched them to observations on behavior. These guys did; they took the test scores, and they showed that they predicted who was committing crimes and who was dropping out of school. The brilliant feature of that book was that most of the analysis in the first part of the book is for whites. They show that the AFQT is very powerfully predictive of a whole range of behaviors for whites.
They were lightning rods. Everybody attacked them. My review didn't attack them on the grounds frequently raised. I did say that they misinterpreted their own data by focusing on genetics. Ability was very powerfully predictive. It played a major role in determining outcomes. But there was nothing in the book that justified any theory of genetic determinism. Everything we've learned since then suggests that the traditional way people measure genetic effects ignores interactions, and ignores a growing body of important literature about
Work in psychology suggests a genetic predisposition to violence among a small proportion of boys. It's not specific to ethnic groups. Different socioeconomic groups respond to this predisposition differently. If you have an aggressive child in an affluent family, the child will get treatment, will be taken to a psychologist. Parents will take special care to override or suppress the behavior. An aggressive kid in a poor family is far less likely to receive such treatment. If his parents themselves are aggressive, they may in fact exacerbate such problems. So there are genetic predispositions, but they're always manifested through an environmental interaction.
Region: So it's not nature versus nurture, but rather nature with nurture.
Heckman: Exactly. It's an interaction. Epigenetics is the field that studies this. There are a lot of recent books and scholarly articles on this topic. I was just at the National Institutes of Health last weekend, and part of the discussion we had there was about this. It's a fascinating field.
The people who favor genetic explanations of social phenomena need to be careful about two things. The methods they use for determining heritability assume additivity. They don't allow for interaction. Secondly, when one does the standard additive analysis for different socioeconomic groups, one finds that the socioeconomic status critically affects the so-called heritability coefficient.
A paper published in Psychological Science (2003) by Eric Turkheimer [et al.] shows very strong family background effects on a number of heritability coefficients. Richer families are providing ways for children to override some defective genes and enhance those genes that are productive. We are just beginning to understand these mechanisms. They are very important.
Region: Which leads us to early childhood education. At the Minneapolis Fed, as you know, we rely very heavily on your research in making the case for early childhood education as an economic development strategy. Could you tell us a bit about your research and how economists have reacted to it?
Heckman: There's a very strong bias among economists against some of the basic findings of the child development literature. Many economists assume that family effects operate primarily through cognitive child ability. A lot of formal economic models view the development process solely in terms of raising IQs. Or else they assume that IQ is purely heritable. Neither view is correct.
Enriched early intervention programs targeted to disadvantaged children have had their biggest effect on noncognitive skills: motivation, self-control and time preference. We know that there's a scientific basis for this finding. The prefrontal cortex, which is a center of these noncognitive skills, matures late. The executive function, the very definition of ourselves as people, the way we motivate ourselves, these things are malleable until quite late stages—into the 20s, according to research by neuroscientists. This means that in principle we can modify these behaviors. Noncognitive skills are powerfully predictive of a number of socioeconomic measures (crime, teenage pregnancy, education and the like) as I show in a recent paper with Jora Stixrud and Sergio Urzua.
Kids in the Perry Preschool Program, an early childhood intervention, are much more successful than similar kids without intervention even though their IQs are no higher. And the same is true of many such interventions. There is a lot of research on such programs. I'm trying to integrate studies of early childhood into a common economic framework for policy analysis. This is the goal of my recent Handbook of the Economics of Education chapter coauthored with Flavio Cunha, Lance Lochner and Dimitriy Masterov.
There's also the Abecedarian program, an intensive child enrichment program targeted toward disadvantaged children, that starts at 3 or 4 months after the children are born. It intervenes up to age 8 in some cases. So it provides a very solid dose of intervention. And when you look at the IQ scores? Lasting substantial differences in IQ are found between those in the program and those not. IQ is raised, on average, four to six points—more for girls than for boys. And this increase lasts at least into the mid-20s. Thus, if we start early enough and offer enriched environments, we can raise the IQs of disadvantaged children.
Flavio Cunha and I have organized this evidence around the twin concepts of complementarity and self-productivity. We develop a multistage model of childhood development and distinguish early versus late investments in childhood. The standard model developed by Gary Becker and Nigel Tomes implicitly assumes that early and late childhood investments are perfect substitutes, that one can make up later for what disadvantaged families neglect early. They also assume a single market skill.
For the study of early childhood investments, these are bad assumptions. First, skills are multiple in nature. A proper accounting of human skills recognizes both cognitive and noncognitive skills. Second, investments raise the stock of later skills through self-productivity and complementarity. Early advantages reinforce each other through self-productivity and complementarity, reducing the cost of future learning. Because of these life-cycle dynamics, the substitution between early and late investments in children is low. The most economically efficient way to remediate the disadvantage caused by adverse family environments is to invest in children when they are young.
We have found that for severely disadvantaged children, there are no levels of later childhood skill investments that can bring the children to a level of social and economic performance attainable from well-targeted early investments. We find that both social and emotional skills are essential in producing successful people. These findings change the way economists think about the human capital formation process.
If we don't provide disadvantaged young children with the proper environments to foster cognitive and noncognitive skills, we'll create a class of people without such skills, without motivation, without the ability to contribute to the larger society nearly as much as they could if they'd been properly nurtured from an early age. Neglecting the early years creates an underclass that is arguably growing in the United States. The family is the major source of human inequality in American society.
I was led into this line of work not just because of my interest in racial gaps, which is real and continuing, but more recently by my work on job training programs. Public job training programs try to improve late adolescent dropouts. Many cannot read and cannot write. They aren't motivated to learn or achieve. American society puts its faith in public training programs to make up for 17 years of neglect. It's nuts. The success rate is even worse for convict remediation programs. How many such programs actually rescue people? The success rate is low.
I am finishing a book on the GED, another remediation program. Twenty percent of all high school degrees these days are GEDs. And it's even higher in New York, Florida, California, Illinois.
Region: I would have guessed 4 or 5 percent.
Heckman: It was 2 percent in the 1960s, and it's risen to 20 percent. And you ask, what do the GEDs earn? They earn what high school dropouts who do not get GEDs earn, once you adjust for their somewhat higher cognitive ability. And what's the difference between GEDs and high school graduates? Well, GEDs have the same Armed Forces Qualifying Test (an achievement test) scores as high school graduates who do not go on to college, so they're just as smart as those people. But they lack something. They're missing motivation, self-control and forward-lookingness. I call these noncognitive skills. In recent work with Jora Stixrud and Sergio Urzua, we find strong evidence of pervasive importance of noncognitive skills as the key to explain success and failure in socioeconomic life.
These findings have major implications for American educational policy—for example, the No Child Left Behind Act, and all the related policies which are predicated on the assumption that we succeed with an educational intervention if we improve on test scores. Such policies are at best misleading. The achievement test scores of these GEDs show they're as smart as high school graduates, but they don't earn anywhere near what high school graduates earn because they lack persistence and motivation.
Most macroeconomists think of human capital as education, measured by years of school. Or if they're a little more sophisticated, they measure human capital by test scores like IQ or an achievement test. Neglected are all the noncognitive abilities that are produced by healthy families. Deficiencies in these skills can be partially remediated, as we know from the early intervention programs. Not completely remediated, but certainly gaps can be closed. The things we used to think of as soft and fuzzy have a real effect on behavior.
Economists tend to dismiss such soft concepts because they want to measure hard concepts. You know, things like a monetary aggregate, three basis points on an interest rate, freight car loadings. These sound like really hard numbers. But if you start talking about a mother's love or giving people emotional support, well, that sounds like you've been watching....
Heckman: Yes, Oprah. [Laughter]
UNIVERSAL OR TARGETED?
Region: Many issues about early childhood research remain unresolved. For example, how early is early? How do we judge quality in education? But I'd like to ask you about one issue in particular. Should public funding go for universal early childhood programs, or should funding be targeted for at-risk children?
Heckman: There's been a lot of discussion about this point. I think the evidence is very strong that family background is a major predictor of future behavior of children. So a disproportionate number of problem kids come from disadvantaged families. The simple economics of intervention therefore suggests that society should focus its investment where it's likely to have very high returns. Right now, that is the disadvantaged population.
Unfortunately, in discussions of early childhood interventions, people often bundle political issues with economic issues. Part of the appeal of universal early childhood intervention is that it provides universal day care, so some groups favor universal early childhood education because it effectively subsidizes women's working. But bundling in this way also creates an opposition group saying, "Why should we subsidize affluent working women?" I'm not going to get into that debate. But I do think the issues of support for working women and early childhood development need to be separated analytically.
Functioning middle-class homes are producing healthy, productive kids. We don't measure their output very well in the national income and product accounts, but it's very well documented that professional working women spend an enormous amount of time after work in child development. It is foolish to try to substitute for what the middle-class and upper-middle-class parents are already doing.
I think that the evidence suggests that we can target pretty well, and we can certainly deal with the major problems, by starting first with children from disadvantaged families. As an economist, I would argue, go where the returns are highest. At some point, diminishing returns will set in, and you might want to fund early childhood education for other groups. Right now, there's plenty of room for intervention in disadvantaged families.
Now you say, Do I have really hard evidence on this? The answer is no. Do we need it? Yes. Am I trying to develop it? Yes. And it's a tough but not impossible empirical problem. And I think I can pull it off by putting together different data sources on different types of interventions with children of different degrees of disadvantage to see which inputs are truly effective, which interventions are effective and which are not.
Region: You referred earlier to job training and, of course, you've done a great deal of research on evaluation of job training programs. The Nobel committee called you the world's leading scholar in the field. Your findings are rather pessimistic, I believe. Is that a fair characterization?
Region: Is job training effective in any form? Are there any types of programs that do work?
Heckman: Most job training is actually being done in private companies, not in the public sector. And who is more likely to get private job training? People who have higher cognitive and noncognitive skills—the same abilities that helped them get the job in the first place. These people earn high returns to private job training.
Region: So, it's a selection issue.
Heckman: Yes, it's a selection issue. It's a consequence of the dynamics of skill formation that we talked about earlier—that skill begets skill. Public training programs are aimed at the bottom of the barrel, at people who've been put out of schools, or maybe they've been given high school degrees through social promotion but can barely read or
write. And the typical, short-term job training program tries to remedy lifetime deficits in a few months. They are based on the hope that society can solve 17, 18 years of neglect of a child with a
short-term program that can transform people who have grown up in extreme disadvantage.
The dynamics of human skill formation are such that high levels of skills acquired early in life make it relatively easy to acquire later skills; they put the child in a position to benefit from later interventions. It's the self-productivity idea we discussed earlier. There is a growth trajectory, and people with different early conditions start to diverge in their development. Those who receive skills early on are more likely to get jobs and enhance those skills through private job training. Those who don't get those early skills are unlikely to benefit much from
short-term public training programs later on.
Public job training also has a bad record because it does not use market incentives. A lot of public programs train people at tasks and skills that are obsolete.
If you look at private job training and say, what are the returns to
on-the-job training for educated workers, you see a very high return. Skill begets skill. In a paper I wrote with Lance Lochner and Chris Taber, published in the Review of Economic Dynamics in 1998, we looked at the effects of ability and formal schooling on job training after school and found that they are very powerful. Job training per se is very important in determining life-cycle skill formation. We find that about half of all the human capital accumulated after people leave home is actually on-the-job training, not just human capital embodied in schooling. Thus, in that study we also found that skills beget skills.
If a child starts out with low levels of cognitive and noncognitive ability, it becomes much less profitable to invest in the young adult. That's the notion of complementarity. If a child has a low level of ability at age 17, then productivity of investment in that person is much lower than it is in somebody who has ability and motivation. The major contributors to the college-going gap by child family income class have to do with child ability. Richer families are much more likely to send their kids to college, but once one conditions on the ability of the child at age 17, virtually all of the income effect goes away. It's all about the ability that's embodied in the child from a lifetime of early investments. So families play a huge role, but it's in making the kid college-ready. It's human ability, or rather, abilities. This is one place where Adam Smith was wrong, actually. He has a passage in The Wealth of Nations which I used to believe and used to quote in classes. And then I realized that Smith was dead wrong.
Region: As well as dead; he won't be able to respond to your critique.
Heckman: You're quite right. [Laughter] Dimitriy Masterov and I actually visited his tomb last year in Edinburgh, where we presented our work on Scottish skill formation.
But anyway, Smith says people are basically born the same and at age 8 one can't really see much difference among them. But then starting at age 8, 9, 10, they pursue different fields, they specialize and they diverge. In his mind, the butcher and the lawyer and the journalist and the professor and the mechanic, all are basically the same person at age 8.
This is wrong. IQ is basically formed by age 8, and there are huge differences in IQ among people. Smith was right that people specialize after 8, but they started specializing before 8. On the early formation of human skill, I think Smith was wrong, although he was right about many other things. And Dimitriy and I said that in the speeches we gave while in Scotland last year. We wanted to be a little titillating. But I think these observations on human skill formation are exactly why the job training programs aren't working in the United States and why many remediation programs directed toward disadvantaged young adults are so ineffective. And that's why the distinction between cognitive and noncognitive skill is so important, because a lot of the problem with children from disadvantaged homes is their values, attitudes and motivations.
Cognitive skills such as IQ can't really be changed much after ages 8 to 10. But with noncognitive skills there's much more malleability. That's the point I was making earlier when talking about the prefrontal cortex. It remains fluid and adaptable until the early 20s. That's why adolescent mentoring programs are as effective as they are. Take a 13-year-old. You're not going to raise the IQ of a 13-year-old, but you can talk the 13-year-old out of dropping out of school. Up to a point you can provide surrogate parenting.
So, coming back to job training and other interventions targeted toward disadvantaged adolescents, mainstream discussions miss the basic economics of the skill formation process. When we understand how that works, that skills build on each other, it's very common-sensical. It's not just IQ, or achievement measured by a test. That's very hard for many economists to understand. There are interactions among IQ, cognitive ability as measured by an achievement test and noncognitive ability.
We tell stories in nursery school, such as the story of the tortoise and the hare and the story of the little train that could. I read these to my kids, and they were read to me. All these folk tales, all these pieces of wisdom, the fact that a mother's love matters and all this stuff, we tend to dismiss them in our formal models of education policy. We economists like to write down specific technologies and make things very precise. That's a useful discipline, and that's what I am doing with various coauthors. We are making this subject precise. But sometimes I have my doubts. Some of what economists do is to explain to fellow economists what most intelligent people already know. A lot of what economists do is explain to themselves what the rest of the world already knows. There's a real risk of being caught up in that.
However, it's important for policy purposes to make ideas precise, to try to understand which interventions are the effective ones. So I don't think our research is just putting simple ideas into the terminology that economists like. I think there is scientific merit in what we are doing. In fact, I've found that in response to some of the papers we have written, where we write about the critical and sensitive periods of child development, and we use the economists' technologies to explain these ideas, the neuroscientists get very excited. I've had several prominent psychologists coming up to thank me at professional meetings of psychologists for helping them to organize their way of thinking about their life's work. Economics can be very powerful in taking ideas, organizing them in the forms of technology, intervention and basic causal factors. That's why I feel there's a lot for economists to do in this area. But there's also a lot to listen to as well.
LOST IN DEBATE?
Region: So much of your work is policy-oriented, and when you've been pulled into policy discussions—and I suspect that sometimes you're being pulled and other times you leap in because of your interest in these issues—are you ever concerned that the careful nuances of your empirical and theoretical work are going to be lost in the political debate? Some of that seemed to happen this morning [at the President's Advisory Panel on Federal Tax Reform]. Are you concerned about that?
Heckman: Deeply. But on the other hand, what choice do I have? I guess I've learned that if I go on television for 15 seconds or 15 minutes even, I really can't communicate very much. I've really come to dislike television interviews for that reason.
I completely understand the risk. In some sense economists help create the risk by using a jargon that's dense and by not trying to communicate. Many ideas in economics are fundamentally simple ideas. Of course, there's a danger that they'll be oversimplified. When you come to empirical estimates, inevitably there is a technical discussion if one is careful. But the main ideas can be said pretty clearly without a lot of jargon.
The training that professional economists have often leads us to speak a private language. Private languages can be productive in communicating subtle ideas. But simple, clear language can be understood. We should recognize the fundamental intelligence of most people. Some people are smarter than others, of course, but there's such a thing as common sense, and common sense often prevails. If you can appeal to that common sense, you've done your job as a communicator.
An appeal to common sense should not end the discussion. One has to back up any claims, especially empirical claims. Thus, for many, it is "common-sensical" that reducing class size raises test scores of students and is a policy worth funding. When the common sense is tested against data, the effects of classroom size reduction are second order in character, especially compared to the effects of early childhood interventions that cost the same. So we have to be clear and appeal to common sense, and we also have to be empirically rigorous and honest.
Region: I'd like to ask you about heterogeneity. Much early economic theory, since Marshall at least, has relied on the concept of the "representative" consumer or firm. You've argued that it's very important to recognize the diversity of economic agents.
Region: Why? And does current economic theory and modeling do a sufficient job of recognizing heterogeneity?
Heckman: It's beginning to. On empirical grounds, it's critical to recognize heterogeneity because it exists at every level.
Marshall introduced the representative agent into economic theory, and it was a very fruitful idea in 1890. Why did he think of the representative agent? In 1850, a famous Belgian statistician, Adolphe Quetelet, formalized the idea of the "average" man. It was only 150 years ago that we really even had the idea. Reporting averages was big news in 1850. This was a huge advance in knowledge. Macroeconomists are basically still looking at averages. They're very informative on overall trends.
But in the past 50 years, we've collected massive amounts of data on individuals, on firms, on asset holdings of various groups, and we always find tremendous variability among people—heterogeneity. And we never find that the conditions required to represent these aggregates by the representative person maximizing utility defined over the averages hold when tested against data. You can form an average, but it's a statistical entity. It doesn't have any behavioral content. The "stand-in" representative agent does not receive much empirical support, even though it is widely used in conventional macro theory.
When we look at individual data, we see very diverse behaviors. Some people react positively to schooling, for instance, and some people respond negatively. Some people react positively to job training; some negatively. There's a distribution of responses. One of my lifetime research activities has been estimating empirically how important this heterogeneity is. I started this work about 30 years ago at Columbia University.
Over the years, my interest in heterogeneity has placed me in a minority position in the profession. Many economists remain wedded to the representative agent approach, and I'm disappointed that more economists—even here at Chicago—haven't incorporated heterogeneity into their thinking.
However, things are changing. In macroeconomics and in finance, a lot of the frontier work recognizes heterogeneity. If you look at wealth holdings, at who's creating entrepreneurial wealth, it's a very heterogeneous group—the self-employed entrepreneurs, the ones who are actually risk-taking. Whenever risk preference parameters are estimated, we find huge variety in risk preferences among people. We see distributions of initial abilities in many, many dimensions of human behavior. Martin Browning, Lars Hansen and I demonstrated the empirical importance of heterogeneity and its relevance to macro theory and the interpretation of macro evidence in a 1999 Handbook of Macroeconomics chapter. But the message is slow to get out and be accepted, given the large body of macroeconomists weaned on the representative agent model.
Back in 1974, I wrote a paper published in the Journal of Political Economy where I looked at preferences between leisure and labor. I estimated the distribution of the rate at which people were willing to trade off leisure for labor. The variation was substantial. I suggested a Pareto-improving program for child-care benefits tied to work that would exploit that heterogeneity. I have been working on various aspects of heterogeneity ever since.
I was at a March 2004 conference in Tempe, Arizona, that brought macro- and microeconomists together. Many of the young macroeconomists were working on aspects of agent heterogeneity, diversity, the response of individuals to shocks. They were documenting that there is a tremendous heterogeneity in these shocks. So, many younger macroeconomists recognize the importance of heterogeneity.
The potentially enormous payoff for recognizing and modeling agent heterogeneity has already been recognized in parts of macroeconomics, even at the Minneapolis Fed. In 1983, I discussed a paper by a leading labor economist at a Carnegie-Rochester conference. My discussion was published in the Carnegie-Rochester series in 1984. This economist was stating a familiar argument of the day—that aggregate labor supply fluctuations were too big relative to the micro estimates of labor supply elasticities and the observed wage variability. He claimed economists had to abandon equilibrium models and instead work with disequilibrium frameworks to account for "the facts."
Now, this economist was basing his claims on labor supply estimates for workers who were working. Entry-exit decisions are not easily accommodated in representative agent models. He ignored the
entry-exit decision of workers, which I had studied in my work on labor supply in the 1970s.
I pointed out in my discussion that half, or more, of all of the aggregate man-hour variation was due to entry and exit decisions and that the author of the paper I discussed was using the wrong labor supply elasticity to establish his benchmark labor supply elasticity. Choices at the extensive margin (entry and exit) are much more elastic than those at the intensive margin (defined for workers). And the appropriate elasticity needed to capture this empirically important phenomenon is conceptually different from the representative agent elasticity that numerous labor economists and macroeconomists were using to reject equilibrium models.
This simple insight about heterogeneity in labor supply choices and the distinction between the extensive and intensive margin motivated development of a literature in theoretical macroeconomics by Gary Hansen and Richard Rogerson, who formalized these empirical insights into special general equilibrium frameworks that were congenial to Minnesota economists. But irrespective of the particular formal general equilibrium model, it is now widely recognized that heterogeneity is an empirically important feature required to understand aggregate labor supply.
I think the harvest from recognizing heterogeneity has just begun.
Region: Is it empirically feasible to incorporate heterogeneity in macro models?
Heckman: It's feasible. But it requires computational power. When you allow for a lot of heterogeneity among agents, you really have to compute. Economists now have the computational power to do this. The Economics Research Center at Chicago is running an institute in conjunction with Argonne National Laboratory precisely to implement models that will allow economists to incorporate heterogeneity into a variety of economic models for policy evaluation and analysis.
HECKMAN'S LABOR SUPPLY
Region: Let me ask you one last question. It's about labor supply, and specifically about yours.
Heckman: My labor supply? [Laughter]
Region: Yes, because, to put it mildly, you are incredibly productive. You mentioned heterogeneity in labor-leisure choices just now, and I'm guessing that you fall at the extreme end of that scale. Your CV lists over 200 articles and counting, and they cover a wide range of topics. You've written several books, and several more are in the works. You frequently give speeches, seminars and testimony, as you did this morning. How do you explain such productivity?
Heckman: I think I'm a very lucky person. I think what some people would call labor, I would call leisure. It stunned me in college—when I looked at my professors—that people were actually paid to read, discuss and create ideas. My eyes lit up; I'd never dreamed that was possible. My teachers were paid to discuss Aristotle, Plato, Joyce, the metaphysical poets and all the large issues of knowledge, and this was what I liked to do.
And I still find that learning is an enormously satisfying activity. I just came back from Washington. I was at a conference last Friday on gene-environment interaction, and honestly, my greatest relaxation that weekend was listening to a two-hour DVD that somebody had given me from another conference on a related topic. It was fascinating! Watching this put me in a very restful mindset.
I realize that some people would view my pace of work as compulsive in character. But I get an enormous amount of pleasure from my work. Enormous. For me, working with graduate students, working with colleagues, discussing ideas and doing our best to solve difficult, important problems is just enormously enjoyable. The greatest pain is to be cut off, without serious exchange of ideas, or to be in an environment where politics and not ideas rule the daily agenda of academic life. Most of my intellectual life I have had stimulating colleagues, students and environments, and I am grateful for this.
With this pleasure, I feel a lot of responsibility as well. In economics there's a trend now to come up with cute papers in an effort to be cited as many times as possible. All the incentives point that way, especially for young professors who seem risk-averse rather than risk-taking after they get tenure. In some quarters of our profession, the level of discussion has sunk to the level of a New Yorker article: coffee-table articles about "cute" topics, papers using "clever" instruments. The authors of these papers are usually unclear about the economic questions they address, the data used to support their conclusions and the econometrics used to justify their estimates. This is a sad development that I hope is a passing fad. Most of this work is without substance, but it makes a short-lived splash and it's easy to do. Many young economists are going for the cute and the clever at the expense of working on hard and important foundational problems.
I feel it's far more important to do good work, important work on basic problems and to do it very well. That is the work that lasts. Along with the enormous freedom academics have to explore ideas and read and write, we have a huge responsibility to do our work well. So I try to do that. And I'd have to say that if anything, the pace of my research is accelerating because I understand more what is not known and what needs to be known.
Region: We look forward to the results. Thank you so much.
More About James J. Heckman
Awards and Honors
Major Professional Activities