Yes, 

if you average T is ten or higher, you data series may be subject to serious unit root. However, if you data are constructed data from few other variables, the unit root problem may be eliminated. Pre-test the data for unit root is very useful anyway.

of course, it is necessary

If you do dynamic panel data using GMM. it is not necessary. PMG also not necessary.

thank you all for your response.Mohd Adib Ismail sir i have a doubt. which test i have to do for unit root. The Levin-Lin (LL) Tests,The Im-Pesaran-Shin (IPS) Test ??? or ADF test ?? please guide me in this. for FIXED AND RANDOM  EFFECT models.

what about  panel corrected standard error model? is it GLS random?

i need guidance on panel unit root test procedure.

All above panel unit root tests are traditional tests. You may consider CD test to cater for cross dependence problem introduced by pesaran. 

REM & FEM models are suitable for small T panels. For small T panels, unit roots are not big problem.

Check the following papers. The first one does not deal with cross sectional dependence. For a review and application look at the last article. You can find it in the internet /researchgate readly. Good luck.

Im, K., Pesaran, H., and Y. Shin, (2003), “Testing for Unit Roots in Heterogeneous Panels,” Journal of Econometrics, 115, 53-74.

Pesaran, M. Hashem (2004) “General Diagnostic Tests for Cross Section Dependence in Panels” Cambridge Working Papers in Economics, No. 435, University of Cambridge, and CESifo Working Paper Series No. 1229.

Pesaran, M. Hashem (2007) “A Simple Panel Unit Root Test In The Presence Of CrossSection Dependence” Journal of Applied Econometrics, Vol.22, 265-312.

Lopcu, K., & Ateş, S. (2009). Income Convergence between Turkey and EU Regions: A Panel Unit Root Approach. In Anadolu International Conference in Economics.

Kenan Lopcu sir thank you very much

Yes dear, it is necessary, if values are non-stationary then you can not expect good results.

Here is an attached files, that will help you.

If you are using VAR or ARDL model it is necessary Criteria if data series is I(0) or I(1). But I am not sure that for Panel Estimation model

of cause, it is necessary because it put you on the right track of analysis, shows how your data behave

Dr. Mohd Adib Ismail,

Could you explain why GMM does not require stationary test? I have run GMM model with R. I have so many problems to test unit root using a package in R. Does it have to do with being unbalanced panel data?

Thank you.

In the presence of cross-section dependence, 1st generation unit root tests are inconsistent, and 2nd generation unit root tests like CIPS and CADF. GMM and unit root testing are two different things, and preferably its depends upon the study. If it is empirical analysis, then testing unit root will guide which kind of long and short-run analysis and modeling will require further estimations.

In order to use the correct panel data analysis method, the stationary of the series must be taken into account. For example, if you have a 30-year observation interval and do not know that the dependent variables are not I[1] or I[0], you can make a mistake in using panel error correction methods. At the same time, cross-sectional dependence is a serious problem for many of the panel datasets used today. From this care, as Mr. Khan has stated, You can often encounter situations that require you to use first-generation and second-generation panel unit root tests.

If the variables in a panel data set are nonstationary, estimators are inefficient unless they are cointegrated.

It is necessary to test for stationarity because of the time series component of the data

It is necessary to test for stationarity in panel data. The validity of many time series models and panel data models requires that the underlying data is stationary. As such, reliable unit root testing is an important step of any time series analysis or panel data analysis.

Yes it is