Informational Content of Factor Structures in Simultaneous Binary Response Models

We study the informational content of factor structures in discrete triangular systems. Factor have been employed a variety settings cross sectional and panel data models, this paper we formally quantify their identifying power bivariate system often treatment effects literature. Our main findings are that imposing structure yields point identification parameters interest, such as coefficient associated with endogenous regressor outcome equation, under weaker assumptions than usually required these models. In particular, show non-standard exclusion restriction requires an explanatory variable equation to be excluded from is no longer necessary for identification, even cases where all regressors discrete. also establish models more general structures, situations one has access at least two continuous measurements common factor.