Dear experts,
I use a model with 06 independent variables and 02 interaction variables as follows:
Y = a0 + a1X1 + a2X2 + a3X3 + a4X1*X2 + a5X1*X3 + a6X4 + a7X5 + a8X6 + v
(a: coefficients, v: error term)
My panel data: N = 35 (Asian countries), T = 20 years.
I have some questions as follows:
1. Which estimation methods I should use?
2. When T < 25 and N > 25 like my case, should I use GMM?
Can I ignore the cross-sectional dependency and slope heterogeneity issues if I use GMM?
3. Are there any books or papers suggesting "Panel Framework Selection" (how to select estimators for panel data)?
Thank you very much indeed!
In your case as N>T, you may employ GMM method for estimation. You may refer Econometric Analysis of Panel Data by Baltagi in addition to the Arellano Bond and Blundell Bond original papers.
Hi,
I agree that in your case GMM would be fine, just make sure not to use too many instruments. As for CSD, your N/T might allow using the CCEMG estimator, although you probably have too many variables for that (up to 4 variables it would be good).
In the GMM/IV framework you can control for CSD by adding time dummies in a model in first differences.
I beleive that you are awae that GMM is for dynamic models and that by construction it drops the individual hetregeneity so you don't have to worry about this issue. However if you don't need dynamics use simply fixed and random models and Hausman test permits to choose between the most adequate for your data. Just keep in mind that fixed effexts cannot estimate fixed variables and they will be eliminated. So if you have important fixed variables that you are interested in, just use random effects.
Papers of Arellano and Bond are the reference but the papers of Baum in the STATA journal are very important with explained applications.
Dear Cong
In this case, I will also suggest you use fixed and Random effect first before considering GMM technique.
Regards
Shreya Biswas Piero Esposito Chiraz Karamti Kehinde Mary Bello : Thank you so much for your useful advice!
Dear Cong, could you please have a look to the footnote number 5 in my paper: Determinants of Operational Efficiency in the Arab Banking Sector.
I hope it will be helpful.
- Badges
- Science topic
- Similar topics
- Papers