Ken Chay, Jonathan Gurvan and Bhaskar Mazumdar have an interesting working paper looking at the impacts of improved access to health care during infancy and cognitive test scores. The paper follows black cohorts born before and after desegregation of hospitals in the 1960s and tracks their performance on different cognitive test scores. The main result of the paper is that the convergence in black-white test scores can be explained in large part by improve access to health care. As expected, this convergence is greatest in areas which stood to gain most from the Civil Rights Act. This paper is related to an earlier study, which looked at the birth weights of babies born to black mothers before and after desegregation.
The main message of this piece is that early childhood experiences matter and have long-run benefits. While the previous research in this area has tended to look at the impacts of shocks (famines, recessions, droughts, etc), this paper looks at the impacts of an actual policy and is therefore one step closer to policy relevance. I have a forthcoming working paper that takes a similar approach, looking at the long-run impacts of public health investments in Mexico. Like the study discussed here, I, too, find substantial benefits of such interventions. The next step is to quantify these benefits and assess how this affects resource allocation decisions, which is of great interest and utility to policy makers.
Another thing I like about this paper is its attention to the problem of inference. The study is 72 pages long for a reason: establishing causality is difficult, especially with so many things going on during the time the affected cohorts were born. The authors engage in an impressive number of robustness checks and econometric fixes to rule out competing hypothesis. The authors also present many of their results graphically, which helps put everything on the table up front.
I'm almost tempted to print out their paper and use it as a program evaluation reference...
3 comments:
I really like this and am tempted to replicate this study for other things that increase access to health care. . . introduction of Medicaid?
Hm, military applicants is sort of a weird sample of the population, not sure how generalizeable this is. But I guess its a good data source for the measures and cohorts that they needed for this analysis.
They use both military cohorts and a more general sample to verify their claims.
Post a Comment