Wednesday, 11 April 2018

The Heckman Curve is flat

It sounded good in theory. Interventions targeted at youths could very plausibly have been rather more effective than programmes targeted at older cohorts. Heckman's foundation thing made a pretty infographic about it.

But it's fundamentally an empirical question. The infographic is based rather more on intuition than on any real lit survey. 

Rea and Burton checked it out. They went into the big Washington State intervention database, sorted interventions by age, and just plotted out the cost effectiveness. There was no curve - just a noisy mess. And the regressions found nothing either. 

Here are the key tables. 





Figure 3's the kicker. Compare it to the stylized Heckman curve. Ocular least squares really are enough here. 

It also links into my other worries about the Perry preschool stuff that Heckman's been pushing and that seems really not to have fared well in broader roll-outs in Tennessee or in Quebec. 

If there is a Heckman Curve, it's hard to see it in the programme evaluation work compiled in the Washington State database

No comments:

Post a comment