Testing for endogeneity is extremely difficult in ordered choice models. One way to proceed is to ignore the endogeneity or simultaneity issue and access its magnitude from the linear IV model. This can be done if the endogeneity bias is relatively small and its sign is clear. Alternatively, a two-stage routine can be applied as in the linear case: In the first stage, the predicted value for consumption is determined like in the linear model. In the second stage,this predicted value enters the ordered probit model. I rely on the results of linear IV methods,which are more robust. The literature on endogeneity treatments in ordered choice models is not fully developed yet. A formal treatment of endogeneity and the corresponding IV regressions (conducted in the linear model) is performed in section 5.3.