Endogeneity, Reverse Causation and Simultaneity in Ordered Probit Models

[Carlo Galimberti]

Dear all,

I am working with a panel data (unbalanced) in the context of an ordered probit model. I suspect that there are problems with endogeneity (especially reverse causation or potentially simultaneity). Therefore I would like to determine whether there are issues of reverse causation or simultaneity between my dependent variable and one of the independent variables (VAR1). I am new to econometrics and I wonder what are the simplest options to investigating these issues in Stata. What I am particulary interested is a way to easily test whether DEP and VAR1 suffer from those problems and if there is a way to estimate the oprobit accounting for those issues. The structure of my data is DEP, VAR1, VAR2, VAR3, etc, ID (unique identifier for panel data), Year

I thank you already for your help

[Joro Kolev]

There are no simple ways to deal with endogeneity.

You need to have a valid instrumental variable for VAR1.

If you have one or more valid instrumental variables for VAR1, you can estimate recursive maximum likelihood -oprobit- taking care of the endogeneity by the user written command -cmp- (by David Roodman).

However this is fairly advanced, and requires good understanding of the problem, and good understanding of what you are doing.

For an example of -oprobit- with instrumentation via -cmp- you can see my paper, about page 1617: Cupák, Andrej, Gueorgui I. Kolev, and Zuzana Brokešová. "Financial literacy and voluntary savings for retirement: novel causal evidence." The European Journal of Finance 25, no. 16 (2019): 1606-1625.