Protection for sale?

New evidence isn’t friendly to Grossman & Helpman’s “Protection for Sale” (AER, 1994):

Unlike existing methods in the literature, our approach does not require any data on political organizations. We formally show that the PFS model predicts that the quantile regression of the protection measure on the inverse import penetration ratio divided by the import demand elasticity, should yield a positive coefficient for quantiles close to one. We test this prediction using the data from Gawande and Bandyopadhyay (2000). The results do not provide any evidence favoring the PFS model.

That’s Susumu Imai, Hajime Katayama, and Kala Krishna in NBER WP 13900. They argue that the past literature testing PFS doesn’t cut it:

In sum, we argue that if we are to structurally estimate the PFS model using the data used by Goldberg and Maggi (1999) and Gawande and Bandyopadhyay (2000), we should not use an arbitrary classification scheme along with the campaign contributions to generate political organization dummies. The structural estimation and testing of the PFS model would require treatment of the political organization dummies to be fully consistent with the prediction of the
PFS model. To our knowledge, this has not been done in the literature.

Of course, the old literature on testing the PFS model of trade protection produced the (puzzling) result that governments cared much more about consumer welfare than contributions. But this attack on the relationship between protection, demand elasticity, and import penetration is even more fundamental.