Consider a ten-year-old child. He is being raised by a single mother who drinks at night and shouts frequently at her three children. They live in a bad neighborhood, and the kids attend a tough school. In the classroom, the boy does not concentrate very well and is often disruptive in class. At recess he bullies other kids. The school social worker thinks that he is hyperactive with antisocial tendencies, such as poor empathy for others. The pediatrician that sees him would like him to use psychiatric medications. His mother thinks that he is bad. Meanwhile, a welfare caseworker would like the mother to attend parenting skills classes and alcohol rehabilitation.
On such rough terrain policies that aim to help disadvantaged children often fail to take root. In the parlance of social science, this child’s problems are interwoven and multifactorial. He displays levels of psychopathology that call for therapy or medication, but these problems are also clearly linked to a dysfunctional family environment that is probably exacerbated by a dysfunctional school and neighborhood. Considered from a different vantage point, this child displays an acute lack of social and interpersonal awareness that cause difficulties relating to others. If we track this child over time, it is likely that we will see harmful behaviors pass down from parent to child, causing disadvantage to replicate itself over time.
Over the past two decades, economists – particularly James Heckman at Chicago – have formalized models of skill formation over the life course (drawing on ideas that have been known to social psychologists for a long time). We now know that skills beget skills, so that investments early in life accumulate and interact with later investments. Conversely, failing to sustain investments in the skills of children can cause early gains to fade out later on. Critically, not all investments matter in the same way at the same time. Investments in intellectual achievement matter the most in early childhood, so that it is very difficult to compensate later on for a failure to stimulate the cognitive development of a baby or toddler. Investments in “non cognitive” capabilities, on the other hand, are most critical in middle childhood. Our ten-year-old child, for example, may benefit more from social skills training than from investments in his brain development. This investment may benefit the child later on in the labor market, even if the gains are not measurable on school achievement tests.
The concept of non-cognitive skills is a bit nebulous and difficult to formalize. Sometimes it seems to resemble a catchall for various dimensions of personality and character. Being friendly or hard working, are two kinds of non-cognitive skills, although clearly developed through different experience (and the friendliness in particular seems to be mainly an inherited trait). The other difficulty with non-cognitive skills is that a pathological lack of non-cognitive skills may be a reflection of poor mental health, and not just a poor environment or a lack of a generalized investment. Some children need more than socialization and good role models, they may need treatment for mood disorders and behavioral problems. The conceptual space between these two domains is narrow, however (although Heckman does gesture that they are different).
The research of Janet Currie and Mark Stabile on behavioral disorders in childhood emphasizes a few critical points. They show us that a generalized measure of behavior problems exerts a qualitatively large and (often) linear effect on achievement in many domains, including delinquency, school achievement, and grade repetition. In siblings fixed effects models, they find that mental health condition in childhood are linked to lower achievement on a par with having less than $100,000 of family income in childhood. As I said before, I am surprised that they find that the disadvantage of mental health problems is as great for higher income children as it is for lower income children, although the expansion of evidence based treatments over time may yet lead to inequalities.
We need a better sense of how packages of interventions (“general interventions” that help all children to develop non-cognitive skills, targeted interventions that are intended to benefit children with pathological mental health problems, and mixed interventions) address different areas of functioning across heterogeneous populations of children. Specific to the example I raised earlier, we also need to know how parenting skills and other forms of remediation in the environment stack up against child-centric policies. Helping kids that are considered to be “bad” may be partially an effort to help their parents to cope with their challenging parenting roles.