Triangular structural model specification and estimation with application to causality

摘要

I consider the paper by Adams, Hurd, McFadden, Merrill, and Ribeiro (AHMMR). I find many of the conclusions of the paper to be eminently reasonable. However, I would suggest a different approach that would involve structural model specification and estimation. Econometric causality can only be reliably determined within the context of a structural model. I discuss the econometrics of triangular models in the panel data situation. I then consider the findings of the paper. Much of the discussion is based on Hausman (1983).