dos In practice it is rather hard to select parameters you to assume migration but not mate kind of. We analyzed the fresh new appropriateness of your model specifications pursuing the mathematical test advised of the Leung and you may Yu (1996) and you may Puhani (2000). They argue that the level of collinearity between your explanatory regressors anticipating lover sorts of and you may extracted from estimating the likelihood of Durham quarters (picture (2)) is the definitive analytical requirement to guage this new appropriateness of one’s design requirements as it catches new the total amount to which this new analytical requirements distinguishes both behaviors. Specifically, if status amount of new matrix of your regressors (the brand new inverse Mills ratio of picture (2) anticipating companion method of) exceeds 20 then design is not powerful in addition to instrumental variables aren’t breaking up the chances of migration out of types of spouse.
