In panel data, cross-sectional dependence also creates the problem of correlated errors or endogeneity. While lag term or series with first difference can easily resolve the problem of time-invariant effect, it is quite implausible to use the same method in the case of cross-sectional dependence.
Even if we use this, we will simply lose information on one country. Moreover, choosing that country is an even more difficult task.
Kindly share some literature supporting reliable methods to resolve this problem.
Finding the correct dummy independent variables should help on this issue. I don't see this as a problem, when there is an endogeneity issue we apply IV. Any textbook on econometric usually explains this.
If cross sectional dependence is due to units within the same cross-section you should correct for group dependence (standard error correction is provided by many softwares)
To provide you some literature you can check out this work among others
https://scholar.google.it/scholar?q=correcting+cross+sectional+dependence+in+panel+data&hl=en&as_sdt=0&as_vis=1&oi=scholart#d=gs_qabs&u=%23p%3Dd_4_OCy6B7MJ
Dear Tiwari,
You can use various methods which efficient for cross sectional dependence and endogeneity. Instrumental Variables and Hausman Taylor (1981) Approach are two of them. But the impact of cross-sectional dependence in dynamic panel estimators could be higher. Specifically in short dynamic panel-data models, if there is cross-sectional dependence in the disturbances, all estimation procedures that rely on IV and the generalized method of moments are inconsistent as N grows large, for fixed T.[1],[2] For this reason, it is very important to choose the instrument variable very well or to perform the analysis with a different estimator.
[1] Arellano, M., & Bond, S. (1991). Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations. The review of economic studies, 58(2), 277-297.
[2]Phillips, P. C., & Sul, D. (2003). Dynamic panel estimation and homogeneity testing under cross section dependence. The Econometrics Journal, 6(1), 217-259.
You can find more information in articles attached.
Best regard,
Merve
@Carlos Mangunsong... it's not the same as a fixed effect problem. Endogeneity may come from several sources and treatment for endogeneity depends upon the sources of endogeneity. Here, I am concerned with the potential endogeneity resulting from regional factors.
I will reply in relation to your last post about regional factors. The within between random effects model can handle measured and un-measured regional differences and can give the same estimates as fixed effects models. Much depends on how many higher level units you have, the more the better. See
Article Understanding and misunderstanding Group mean centering: a c...
Article Explaining Fixed Effects: Random Effects Modeling of Time-Se...
Article Fixed and Random effects models: making an informed choice
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Probability
- Probability Theory
- Reliability