Stigler’s List and Social Capital

I just reviewed an edited collection on social capital for the Economic Record, and kicked off with what I think might be the standard economic response to most books of this type:

George Stigler once observed that economists’ objections are so predictable, we would be well served to agree on a numbered list of the standard comments that are made in seminars.* To assist, he provided a list of 32 objections that remain eerily pertinent three decades later. These include “The residuals are clearly non-normal, and the specification of the model is incorrect.” (#3), “But what if transaction costs are not zero?” (#14) and “What happens when you extend the analysis to the later (or earlier) period?” (#22).

When it comes to discussing sociological writings on social capital, such a list comes into its own. Before cracking the spine of this 468-page volume, I felt sure that the analysis of social networks and employment would inadequately deal with the issue of causality, allowing me to deploy Stigler’s objection #2 (“Unfortunately, there is an identification problem which is not dealt with adequately in the paper.”). From here, it is an easy step to objection #8 (“Have you tried two-stage least squares?”). If social relationships matter so much to life outcomes, why do individuals not invest in upgrading them? (“The flabby economic actor in this impressionistic model should be replaced by the utility-maximizing individual.”) And of course, any suggestion that stronger social networks could bring down the overall level of unemployment immediately brings to mind #29 (“The problem cannot be dealt with by partial equilibrium methods; it requires a general equilibrium formulation.”)

The rest of the review is here.

* See Stigler (1977). Writing on the Freakonomics blog, Justin Wolfers sought updated suggestions, of which my favourite was Richard Holden’s: “That’s ok in practice, but it won’t work in theory.”

This entry was posted in Economics Generally, Social Capital. Bookmark the permalink.