My Research Program

The Origins and Remediation of Human Inequality

James J. Heckman

April 2007

My research focuses on the economics of human development. This analysis includes the study of the development of skills as conventionally measured (education, on the job training), abilities (cognitive and noncognitive skills) and health capacities (both physical and mental health). My research studies develops theoretical models of parental choice and child preference formation as well as intergenerational models of family influence. My research also involves estimation of dynamic models of the evolution of skills and capacities using longitudinal data. In joint work with Cunha and Schennach, I develop dynamic nonlinear factor models to organize large scale developmental data sets and use multiple measurements and multiple equations to identify technologies of skill and health formation. This work determines the origins of human differences and the effectiveness of alternative intervention strategies to foster human skills and capacities, and to remediate disadvantages .

This research is rooted in economics but goes well outside of traditional analyses to integrate research in psychology, demography, neuroscience and biology. There are both theoretical and applied aspects of this work. There is also a data gathering aspect. In partnership with other members of the Pritzker Consortium on Early Childhood Development (which I chair), I am helping to design and evaluate a series of early childhood interventions in Ireland and Northern Ireland. I am actively working with personality psychologists, developmental psychologists, quantitative sociologists, statisticians, and neuroscientists to understand the biology and social science producing inequality in health, in labor market outcomes and in society at large.

I post papers that summarize the main threads of my work at http://jenni.uchicago.edu/human-inequality.

Overview Papers

Three overview papers provide a general introduction to my research program. One appeared in Science (June 2006). A second paper is forthcoming in the Proceedings of the National Academy of Sciences (2007). These papers discuss the evidence on the origins of the gaps in abilities and skills that affect adult health, productivity in the labor market and performance in society. A third paper with Cunha ( American Economic Review , 2007) focuses on the formation of abilities in the labor market. Cunha, Heckman, Lochner and Masterov (2006) provide a more comprehensive analysis with much greater empirical detail.

Studies in Cognitive and Noncognitive Skills

I have a substantial body of research on the effects of cognitive and noncognitive skills on a variety of outcomes. An early paper (Heckman, Journal of Political Economy , 1995) discussed the importance of cognitive skills on earnings. Later work (Heckman and Rubinstein, American Economic Review , 2001; Heckman, Stixrud and Urzua, Journal of Labor Economics , 2006) discusses the importance of both cognitive and noncognitive skills in a variety of areas of economic and social life. The 2006 paper shows that a few low-dimensional latent traits explain a wide variety of human behaviors. I am actively extending this work to include a more diverse array of noncognitive measurements.

A key technical issue in investigating the impacts of cognitive and noncognitive skills is the problem of reverse causality. Tests used in empirical work on the impacts of cognitive and noncognitive skills on outcomes may be a consequence of behaviors (schooling may raise test scores; depression may affect self esteem, etc.) A paper with Hansen and Mullen (Journal of Econometrics , 2004) develops and applies methods for solving the problem of reverse causality for estimating the effects of cognitive skills. These methods are applied in Heckman, Urzua and Stixrud (2006) to analyze the effects of noncognitive skills. Research with Cunha and Schennach (2007, not attached) extends this line of work to more general classes of skill formation technologies.

The Technology of Skill Formation

Most research in psychology and economics treats cognitive and noncognitive traits as invariant characteristics. Indeed, some psychologists require these traits to possess the property of temporal stability. In joint work with psychologists and economists (Borghans et al., 2007, not attached), we show that stability is not a proper characterization of these traits. Cunha and Heckman (2007, not attached) present systematic evidence on thie issue and show how parental investments and environments affect the evolution of cognitive and noncognitive traits. In the PNAS (2007) and AER (2007) papers, frameworks for the development of human abilities and health stocks are presented. These papers feature both genetics and parental investment. The econometric framework underlying our work rests on two papers (not attached): Cunha and Heckman (Journal of Human Resources , 2007) and Cunha, Heckman and Schennach (under revision at Econometrica; not attached). In these papers, we generalize previous work on linear factor analysis to (a) allow for dynamic evolution of latent skills and (b) allow for nonlinear technologies for investment and self-productivity. Both features are required to understand and evaluate the possibilities of remediation and compensation, and the benefits of early intervention for disadvantaged children.

Cunha and Heckman (2006, attached) use these technologies to (a) interpret the enriched early intervention Perry Preschool Program in terms of a well-posed model of human development and (b) evaluate the effectiveness of alternative intervention strategies concerning early versus late investment in the lives of disadvantaged children.

Reanalysis and Synthesis of Existing Experimental Data

Another project being conducted by Heckman and his partners in the Pritzker Consortium is reanalysis of the major early childhood interventions for disadvantaged populations. The Consortium includes most of the developers of the major experimental data sets with long term follow-up. We are reanalyzing the data (a) accounting for the small sample sizes of each study using exact small sample inference; (b) accounting for the effects of multiple hypothesis testing (there are many outcomes for each intervention); and (c) using Bayesian methods to pool evidence across studies.

We are also analyzing many new outcome variables never previously analyzed for each study. We have found major problems with previous analyses and have corrected them. We are conducting a revised cost-benefit study for each major intervention. We are estimating models of the technology of skill formation to interpret the evidence across studies in a unified framework.

New Data

In conjunction with my partners in the Pritzker Consortium, I am also actively engaged in conducting new experimental studies on disadvantaged children in Ireland. These studies are based on acknowledged best practices. This work synthesizes evidence from previous studies. It addresses statistical issues on the design and evaluation of experiments and the design and implementation of questionnaires.

Policy Analysis

The goal of the scientific analysis is to provide a better framework for the evaluation of public policy directed toward health and skill formation. The plan is to evaluate the effectiveness of expenditure to remedy disadvantage by category at different stages of the life cycle. A recent paper by Heckman and Masterov (2007, attached) communicates some of the issues for a general audience. See also the paper by Cunha and Heckman (2006, attached).

Aspirations

I seek collaboration with economists, psychologists, neuroscientists, sociologists and statisticians doing serious research on the origins of human inequality and the effectiveness of remediation. There are many diverse aspects of this work. They can be unified under the rubric of the technology of capacity formation (Heckman, PNAS, 2007, attached). Many details need to be worked out, but the vision is clear. A strong interdisciplinary effort is required.