Hi,
I'm running a model to see the growth effects of the exchange rate.
The coefficient that I got for the real effective exchange rate is positive significant under the fixed effect model and under IV it turned negative and significant. How can I justify this change? Anyone could please refer to some similar evidence in the literature?
Thanks
Nilu
To my knowledge, if you have run the fixed effects model then it should take care of the endogeneity problem to some extent. However, a change from positive to negative sign indicates a serious problem in your estimation implying the existence of a serious endogeneity. Try looking at the endogeneity test results and see whether you have endogeneity issue in your model or not. If the p value is significant then you have an endogeneity issue that you should take care of. Depending on you data type (long panel or short panel), you can use gmm, iv(2sls), liml, ec2sls, or g2sls estimators to take care of this issue. Nevertheless, it is important to be careful in instrument selection (you should try to look at the correlation between the instrument and the endogenous variable). You can contact me if you need further guidance. Pannilage Niluka Suranji Perera Ekanayake
- Badges
- Science topic
- Similar topics
- Biological Science
- Kinesiology
- Running