Dear all,
I am doing a research on the effect of exchange rate on FDI in the US with a panel data with Individuals being countries. I am having some trouble in R when using the fixed effect code. R drops the variables that do not vary across countries such as GDP in the US and dummy variable controlling for certain global crisis. Is there a way to remediate to that ? I have done the Hausman test and the Random effect model can be used. Does that mean I can only model it with the Random effect ?
My current code is :
plm(Y - X , data=pdata, index=c(“Country”,”Date”), model=”within”)
Thank you.
note : I am not a panel expert and my research for my bachelor.
Dear Charlotte,
In principle, to employ fixed-effects models has advantages and disadvangates. Usually, we consider that the pros are larger than the cons and, nowadays, we rarely use Hausman tests and we simply employ fixed-effects models. Nevertheless, among the disadvantages of the use of fixed effects, one of the most relevant ones is that we cannot estimate the effect of time-constant variables. If you want to test a certain hyphotesis not related to time-constant variables, just proceed ignoring time-constant covariates. If you are really interested in the effect of those variables, I would advise you to employ the so-called Hausman-Taylor model (correlated random effects).
In Stata, it is easily implemented using xthtaylor and I guess that it is also simple in R ( https://rdrr.io/cran/plm/man/pht.html ).
For an explanation, the ones provided by Pudney in his slides are very intuitive (see https://www.iser.essex.ac.uk/files/teaching/spudney/ec968/downloads/Lecture%20Notes/EC968_lecture3.pdf ). The other lectures on panel data (just change "lecture3" by "lecture1"-"lecture4") mean a very good material. Also, Cameron and Trivedi's book (Microeconometrics. Methods and Applications), quite easy to find, is a good reference for applied work.
I hope it helps. If this is the case, just recommend the answer, please.
All the best,
José-Ignacio Antón
If you take FDI % gdp you will face a lot of problems, but I founf that (log value number FDI) is better than another variable.
Yes, you can use fixed effect panel using individual invariant regressors. A fixed effects regression is an estimation technique employed in a panel data setting that allows one to control for time-invariant unobserved individual characteristics that can be correlated with the observed independent variables