Hi there,
I am running a model that sufers from both endogenity and selection bias. I am examining the impact of a particular form of aid on exports. I run 2 models, the first is exporter model that sees how this aid affect recipient countries' exports and the second is a bilateral model which examines donor recipient countries relation to examine whether aid benefit donor countries or not.
In the exporter model I used IV (ivreghdfe) instrunmenting aid with the freedom house index. Results were fine and the instrument is strong and relevant.
However, in the bilateral model, when I tried to use the sam instrument I kept on getting inconclusive results, and the tests indicated the instrument was weak and irrelevant.
I know that my model suffers from selection bias because when I run heckman selection on its own, the inverse mill ratio is significant. Now I was wondering if there is a way that I can account for selection bias and endogenity in one model and maintain fixed effects because fixed effects are crucial for this model.
I initally thought of the following sequence of steps but I am not entirely sure they are correct:
- Probit equation estimating the proability of recieving aid.
- Estimating inverse mill ratio.
- First stage equation with the instrument and inverse mill ratio
- Second stage equation
would this be appropriate?
I am running a model that sufers from both endogenity and selection bias. I am examining the impact of a particular form of aid on exports. I run 2 models, the first is exporter model that sees how this aid affect recipient countries' exports and the second is a bilateral model which examines donor recipient countries relation to examine whether aid benefit donor countries or not.
In the exporter model I used IV (ivreghdfe) instrunmenting aid with the freedom house index. Results were fine and the instrument is strong and relevant.
However, in the bilateral model, when I tried to use the sam instrument I kept on getting inconclusive results, and the tests indicated the instrument was weak and irrelevant.
I know that my model suffers from selection bias because when I run heckman selection on its own, the inverse mill ratio is significant. Now I was wondering if there is a way that I can account for selection bias and endogenity in one model and maintain fixed effects because fixed effects are crucial for this model.
I initally thought of the following sequence of steps but I am not entirely sure they are correct:
- Probit equation estimating the proability of recieving aid.
- Estimating inverse mill ratio.
- First stage equation with the instrument and inverse mill ratio
- Second stage equation
would this be appropriate?
