Or, it would be a better world if people who pronounced on policy measures knew something about policy efficacy
Head Start will cover 57,000 less children come November as a consequence of the sequester. Just as well, says David Brooks, on Meet the Press, since Head Start doesn’t work. (transcript doesn’t quite get it right, but you can tell that he is dismissing Head Start if you watch the video.)
Why did Brooks assert that Head Start “didn’t work”? I suspect it’s a particular reading of a set of findings. As Cunha and Heckman (2010) observe:
Currie & Thomas (1995) document a decline in the performance of Head Start minority participants after they leave the program, return to disadvantaged environments, and receive the low levels of investment experienced by many disadvantaged children.
That’s where I think Mr. Brooks stopped reading. Cunha and Heckman continue:
It is essential to invest early to get satisfactory adult outcomes. But it is also essential to invest late to harvest the fruits of the early investment. Such dynamic complementarity helps to explain the evidence by Currie & Thomas (1995) that for disadvantaged minority students, early investments through Head Start have weak effects in later years if not followed up by later investments. … Our explanation is in sharp contrast to the one offered by Becker (1991), who explains weak Head Start effects by crowding out of parental investment by public investment. That is a story of substitution against the child who receives investment in a one-period model of childhood. Ours is a story of dynamic complementarity.
One conclusion would be, well if it doesn’t work permanently, then don’t do it at all. An alternative conclusion would be that Head Start needs to be buttressed by subsequent investment in human capital.
Cunha and Heckman (a Nobel laureate, by the way ) also note that test scores are not the only basis for evaluating an investment in human capital:
The controversy over Head Start fade-out may have been a consequence of relying only on cognitive measures to gauge performance. The Perry Preschool Program had an IQ fade-out but a lasting effect on a variety of participants through age 40. They work harder, are less likely to commit crime, and participate in many fewer social pathologies than do control group members.
This conclusion is echoed in a paper published in American Economic Journal: Applied Economics, by David Deming, entitled ”Early Childhood Intervention and Life-Cycle Skill Development: Evidence from Head Start”.
This paper provides evidence of the long-term benefits of Head Start for a recent birth cohort of children. While my results rely on nonexperimental comparisons between siblings who differ in their participation in the program, I find little evidence of systematic within-family bias in preschool assignment, and the results are robust to sensitivity checks and alternative specifications. I estimate that the long-term impact of Head Start is about 0.23 standard deviations on a summary index of young adult outcomes, with larger impacts for African Americans and relatively disadvantaged children. This gain is about one-third of the size of the outcome gap between the bottom quartile and the median permanent income in the CNLSY sample, and is about 80 percent as large as the gains from the Perry Preschool and Carolina Abecedarian model preschool programs (Carneiro and Heckman 2003; Anderson 2008).
I find an initial age 5–6 test score gain of about 0.15 standard deviations that fades out to about 0.05 by ages 11–14. Fade-out is particularly strong for African Americans and for very disadvantaged children, and yet they experience the largest long-term gains. This does not rule out, for example, an increase in latent cognitive skills that is more poorly measured by the same test as children age—but it does imply that a projection of future benefits for these children based on test score gains alone would greatly understate the impact of the program.
I’ve just touched the vast literature on the subject of early investment in human capital (after all, I’m a macroeconomist — but heck, in an era of Econlit and Google Scholar, one can learn a lot quickly if one wants to). In sum, there’s a lot of other research indicating that Head Start is well worth the investment. Instead of wailing over inequality, perhaps we should start increasing, rather than cutting, investment in the areas where we know we’ll get a payoff.