Hi all,
I've tried to read through the forum to see if there were any similar posts, so I apologise if I've missed them and my post is a repeat.
I hope you can help me.

In order to identify the determinants of changes in expectations of future spending on used clothing, I estimated an Ordered Probit, the dependent variable is a discrete ordered variable (1= spending will increase, 2 = spending will remain the same 3 spending will decrease), however I believe that a binary independent variable is endogenous.

How can I test the endogeneity of the regressor?
In order to "solve" the endogeneity problem can I use the "conditional mixed process estimator" (Roodman) ? Also, I think I have some potential instrumental variables that I can use also they are dummies variables.

Thank you for any suggestions you may have.

My dataset is an unbalanced panel (2 periods) with about 2300 observations for the first period and 2070 (who all participated in the first period survey) for the second period.