You can go for correlation and regression.

You can go for ANOVA analysis.

And Paired T-test.

Farjin Hssan

Thank you for the suggestion.

Perform all the diagonistic tests for panel data. Check for cross section dependence. Do stationarity test.

Hi

Based on the various properties of the data you have in terms number of years (T) and cross-section (N), you can go for FIXED EFFECT, RANDOM EFFECT, GMM, POOLED OLS, PANEL ARDL, PANEL FMOLS, POOLED MEAN GROUP ETC.

YOU CAN READ AND GET MORE INFORMATION ON THESE FROM ANY GOOD ECONOMETRIC TEXT BOOK, ESPECIALLY THOSE ON PANEL.

THANK YOU

Based on my own experience, I can indicate that, GMM, pooled OLS, even cross country OLS are the most available not difficult panel data analytics.

It depends on the features of your data. The first step is to explore data. You can use some visual tools for that.

Let me recommend our poster describing some visual tools that may prove useful for your task:

Poster A Visual Framework for Longitudinal and Panel Studies (with ...

Table 1 summarizes the most important questions for the first step of panel data analysis. The tools described aim to get insights for further modelling.

I think correlation, regression or efficiency can do

GMM, and pooled OLS can do the work

I don't think that anyone could answer your question given the little information that you have provided. Books such as Wooldridge (2010), Econometric Analysis of Cross Section and Panel Data, MIT Press, or Baltagi (2020), Econometric Analysis of Panel Data, Springer, or Biorn (2017) or Cameron &Trivedi (2005) or similar describe a multitude of panel data models all of which are appropriate in the particular cases "to examine the impact of a variable (X) on another variable (Y) ". It is not a question of finding the best technique but of finding the appropriate technique in your case. If you use Stata you would find Cameron and Trivedi (2010), Microeconometrics Using Stata, Stat Press useful. For R Croisant (2019), Panel data econometrics using R, Wiley, would be useful.

It may be the case that standard fixed-effects or random-effects model may be sufficient. Your economic theory and common sense should be the prime determinant of the type of panel model(s) that you estimate

Professor, John C Frain

Thank you so much for your informative suggestion. The data are monetary aggregates over some economies. I intend to use STATA for the mentioned analysis. I have downloaded the book just now. I will consult it. And I am not sure about the application of the standard fixed-effects or random-effects models in my case; the extant literature lacks it so far. However, it is obviously a valued point.

Thank you all for your recommendations.

Mousumi Bhattacharya , Samuel Tawiah Baidoo , Ibrahim Niftiyev , Andrey Davydenko , Yusuf Tanko , Adeyemi Oluwole

Instead of choosing econometric model to test penal data, you'd better check unit root test and co integration test. Base on the result of URT, you just follow the methodology to get best model.

Provash Kumer Sarker

You might have a look at the European Central Bank web site https://www.ecb.europa.eu/home/html/index.en.html . If you enter panel data in the search panel you will find a lot of research working papers that utilize panel data with macroeconomic data. Some of these may give you an idea of what can be done, You might even replicate some of the ECB analysis with your data set.

You need to firstly, carry out pre-estimation tests (panel unit root tests, diagnostic test) and cointegration test. If your variables are found to be integrated of the same order, sayI(1), then the appropriate method to use for cointegration is the Fisher or combined Johansen’s procedure upon which you will perform panel ECM to determine the short run effects and speed of adjsutment. But if you have mixed order say I (1) and I (o), you may perform Hausman test to know if fixed effect or random effect is most suitable. Double check your result. E.g, if fixed effect is most suituitable from Hausman test, use WALD test to figure out which is better to use between Fixed effect model and pool OLS. You may check the link below, it may help. https://www.isroset.org/pdf_paper_view.php?paper_id=1806&5-WAJM-03121.pdf

Your choice of the model depends on the nature of your topic as well as your data. Based on this view, you can apply linear, multiple, logistics or other regression analysis methods. For further information you can read econometric books in the topic of regression and you can search in google like 'how to choose an appropriate model to x topic'?

This question is not straightforward. In principle, I would carry out a fixed-effects estimation (using the within-group estimator) to deal with time-invariant unobserved endogeneity. You can also perform a random effects estimation and to compare them using a Hausman-type test, but, in some areas, essentially only fixed-effects results will be considered.

If you have additional endogeneity problems, you have to deal with them in the best way you can irrespective of panel data (e.g., a potentially endogenous time-varying variable).

I would not go to GMM unless you have a very specific model with no other problem (e.g., a dynamic model with lags of the dependent variable but no other problems).

I hope it helps.