Dynamic Panel Data estimation

[cind du bois]

Dear community,

As my dependent variable is dynamic in nature, I want to use a dynamic panel data regression. However, when trying the different available options (xtabond, xtabond2, xtdpd), I get diverging results. How do I choose between the different models? Any tests I can use? What if my data is not suitable (low number of groups)? Thank you very much for your help!

[Jeff Wooldridge]

It’s up to you what model you use. Data are neither static nor dynamic. With N = 26, you probably shouldn’t use a dynamic model. What’s the purpose of the analysis?

[cind du bois]

Dear Jef, thank you for answering. The idea of the analysis is to study determinants of military spending (the dependent variable will be the % of military spending to personnel). The independent variable of interest is government ideology where we also include other control variables such as GDP, unemploymentrate, ... . Indeed, I only have a small number of groups as my dataset relates to NATO member countries (for which I have the data). My idea was to include the lag of my dependent variable...