Identifying restrictions in limited information analysis of the schooling coefficient in a wage equation

Identifying restrictions underlying limited information estimates of the coefficients of a wage equation are considered from a Bayesian point of view. Within this framework ‘exclusion’ restrictions need not be imposed exactly, and it becomes possible to consider the marginal densities of interesting coefficients as functions of the tightness of these restrictions. In the application considered here the posterior mean for the schooling and test-score coefficients in a wage equation are examined as identifying restrictions are relaxed. The paper also serves as an example of the feasibility of Bayesian limited information analysis of a current economic issue.