A lawyer and an economist get together to write a book on how to make social programs work better. The book, called Targeting in Social Programs: Avoiding Bad Bets, Removing Bad Apples, is published by Brookings and includes blurbs on the back from Larry Summers, William Kristol, Robert Reischauer and Chrstopher Jencks. That's a lot for a book that only policy wonks can love, but it is a fine book for that group indeed. It is a book aimed at both intelligent non-specialists and at academics interested in a low-tech overview and introduction to ideas about how to better target social programs. One can think of this book as an economics book, a law book or a public administration book as it reflects influences from all three literatures.
As the second part of the title suggests, much of the focus is on avoiding bad bets, which means directing services at individuals likely to have large positive benefits at relatively low cost, rather than at other groups. Put differently, avoiding bad bets means choosing to treat those in the upper part of the net impact distribution. Of course, locating such people is not trivial as the literature on job training programs - see e.g. Heckman, Heinrich and Smith (2002) Journal of Human Resources or Bell and Orr (2002) Labour Economics - amply demonstrates. This is the goal that is treated in much of the scholarly literature on statistical treatment rules.
The other focus of the book is on bad apples, which the authors define as participants who impose negative externalities on other participants - think of the bomb-building public housing resident. This aspectof the problem looms larger in the legal literature than the economics one, though it is reminiscent of Ed Lazear's model of class size effects wherein smaller classes are better because they are less likely to contain a disruptive student and thereby limit the range of destruction brought about by such students. The authors make the case that even lefties should want to weed out the bad apples, subject of course to reasonable procedural safeguards and the provision in most cases of some sort of bad-apple-specific alternative treatment.
The one truly odd thing about the book is that the leading current examples of sophisticated statistical targeting systems - namely the Worker Profiling and Reemployment Services (WPRS) system embedded in the U.S. Unemployment Insurance system and various related systems used in a criminal justice context to help make prison versus probation decisions and also parole decisions - are nowhere mentioned in the book. Also missing are any references to the broader scholarly literature on targeting in active labor market programs by Manski and others (including yours truly), to the decades-0ld literature on "selective incapacitation" in criminology and to related literatures in education (on targeted curricula) and in mental health (on targeted treatment regimes).
Highly recommended, nonetheless.
Who was my favorite student this term?
7 years ago
