I have a data file (panel) with a time, country and industry dimensions. I presume there are all dimension-specific effects present. Does anyone know what is the correct way to treat such panels and which software to use? The model is Panel VAR or DPD. Maybe there is some good reference literature? Can Stata, R handle such calculations?
Dear Ilya Bolotov,
Article Three-dimensional panel data models with interactive effects...
This article may help you.
Best wishes,
Check the book The Econometrics of Multidimensional Panels: http://www.springer.com/de/book/9783319607825
To close this question. The data is called multi-dimensional panel data (MDPD).
The book recommended above is a good reference. Basically, there are three approaches to estimation of MDPD: 1) LS with dummy variables (LSDV) for FE / Within estimation (can bet substituted with OLS with corresponding dummies, probably, the easiest solution + Hausman tests can be applied), 2) Feasible GLS (FGLS) and 3) IV / GMM (the most general one) for endogeneous variables and dynamics (models with lags).
If someone uses STATA, please, check
https://www.stata-journal.com/article.html?article=st0143 for FE.LSDV.REG
for 3-dimensional panel data.
A trivial solution, if effects are not required, is to group i...j...k-1 dimensions (everything except time) as 1 cross-sectional ID and estimate a standard 2D panel model.
- Badges
- Science topic
- Similar topics
- Philosophy
- Esthetics
- Interpretation