First generation panel data analysis often assume cross sectional independence, i.e the shock to a variable in a country will not have any effect on the other countries variables. However, as a result of globalization and other related cross nation interlinks, it is apparent that a problem in country A can affect country B. Most of the conventional panel analysis like fixed effect, Random effect, Panel ols, among others fall into this category. In order to correct the bias in the estimate of 1st generational panel analysis, the 2nd generational panel analysis was invented. This methods appropriately incorporate the cross sectional dependence in the modeling. This includes methods like ccemg, cs-ardl, cup-FM and so on..

Hamid Muili ....thank you very much for your detailed answer. Now, if i want to use GMM ( as my data is appropriate for it) , should i worry for cross-sectional dependency or how to proceed then?

Basically, Gmm is not robust to cross sectional dependence as with other first generational panel data methods. Other dynamic model that corrects for cross sectional dependence is the bias corrected fixed effect panel data method uses bootstrap method to correct the cross sectional dependence problem and can be used for dynamic analysis, even when th N>T. Although, it comes with its own problem as it assumes strict exogeneity which implies that it is not robust to endogeneity in the panel

2nd generation panel data econometric methods are largely characterised by the rejection of the cross-sectional independence hypothesis. Examples of such 2nd generation methods include CIPS unit root test, ECM panel co-integration, Pedorni Dynamic Ordinary Least Square (PDOLS) and Fully Modified Ordinary Least Square (FMOLS), Common Correlation Effects Mean Group (CCEMG), Augmented Mean Group (AMG) and Average Correlation Coefficient (ACC) estimators as proposed by Pedroni (2007), Westerlund (2004) and Pesaran (2006).

Within the 2nd generation of tests, two main approaches are implemented. The first approach is the factor structure approach (Bai & Ng, 2001; Choi, 2002; Phillips & Sul, 2003; Pesaran, 2003; Moon & Perron, 2004; etc). The second approach imposes either none or few restrictions on the residuals covariance matrix, as adopted markedly by Chang (2002, 2004) who proposed the application of nonlinear instrumental variables methods or bootstrap approaches to solve the parameter problems arising from cross-sectional dependence.

As opposed to the 1st generation panel data methods which are anchored on the assumption or condition of cross sectional dependence, the 2nd generation methods are robust enough to account for cross-sectoral dependence, and have the potential/power circumvent problems relating to the presence of endogeneity between dependent and independent variables and serial correlation between co-integrated panels.

Kehinde Mary Bello Hamid Muili Elvis Munyaradzi Ganyaupfu ............thank you very much for your feedback.....indeed its help me a lot to clarify things....

1st generation tests assume that there is no interunit correlation.

At the same time, Tests are examined in two groups according to the assumptions about autoregressive parameter..

In the first group, it is assumed that the autoregressive parameter does not change from unit to unit.In the second group, the autoregressive parameter is assumed to be heterogeneous.In other generations, inter-unit correlation is taken into account. Generally, this is the difference. The features deepen according to the tests included in the generations.