Hi!
I was planning on doing fixed effects on panel data information, but then I realised that my model has probably some endogeneity issues (explanatory variables correlated with unobserved error term), first how do I check if this assumption is true? If so, how do I solve it? Since fe only works under the omitted variable assumption, I was wondering if I can still do fixed effects but correcting for this problem.
Thank you
In a fixed effects panel study of repeated measures within a person, the fixed effects are used to dummy out each individual thereby removing the unmeasured effects of each individual, that is removing one form of endogeneity.
You may want to look at Article Explaining Fixed Effects: Random Effects Modeling of Time-Se...
Endogeneity is a tricky issue. Most of the time, there are no magic fixes or statistical solutions, your theoretical model, and deep area knowledge will help immensely. There are also quasi-experimental designs such as instrumental variables (IV) that are designed to deal with the issue but they also have their own share of problems (you can combined fixed effects with IVs too, but finding good IVs is another problems in itself). If you are interested in a "causal" effect, I would highly recommend two books. They cover fixed effects with panel data as well.
https://www.theeffectbook.net/
https://mixtape.scunning.com/
- Badges
- Science topic
- Similar topics
- Engineering
- Electrical Engineering
- Signal Processing
- Image Processing