How to run Panel Data regression analysis(Steps)

[Azizjon Rakhmonov]

Hello every one I wander if anybody could give me some instructions
I am analyzing determinants(Macroeconomic and Political determinants) of FDI with economic union dummy. Panel data (10 countries with 15 years(2002-2016), balanced data)
Benchmark paper I have chosen used Fixed effect model.
I need help with initial steps. from what to begin before choosing method (Fixed or random) and running regression
I mean about(Check for stationarity(unit root) which method best(should I do it before converting natural logarithm or after), auto-correlation, heteroscedasticity, serail-correlation or if there are any others as well could you give some instructions or refer to any manual). Furthermore If there is unit root for each explanatory variable should I convert them all into first difference or not?
Thanks in advance
Azizjon Rakhmonov

[Phil Bromiley]

You didn't get a quick answer. You are unlikely to get a useful answer to this general a question and this many questions. I would suggest you look at how folks analyze similar data in your field.

Read the xtreg documentation that comes with Stata. You might also look at Hausman.

[Carlo Lazzaro]

Azizjon:
as Phil highlighted, list of general queries about such a difficult topic like panel data econometrics, usually left unreplied.
If you weren't taught in (panel data) econometrics and you plan to use Stata for you analysis, you can benefit from reading https://www.stata.com/bookstore/micr...metrics-stata/.
I would also recommend to go through your queries with your teacher/supervisor/professor and/or a more expert colleague.
As an aside, as you have a T>N panel dataset, -xtgls- entry in Stata .pdf manual is probably worth reading.

[Marcos Almeida]

As already commented, you question seems to be much overarching. It would practically take an entire book. The best approach, IMHO, would be to start by reading the Stata Manual on longitudinal studies. The remaining doubts would eventually be solved by creating a thread to the specific issues.

[Azizjon Rakhmonov]

Thank you for all your kind answers much useful.
With urgency, one more request i wonder if you give me some directions
after checking for hausman test fixed effect turned to be more appropriate but i run it before stationarity test and got significant values for my veraibles.
But after checking i found out that some of them are not stationary. And run again regression with variables after converting them into first difference(to make stationry) but got insignificant values for my first differed varibles
my questions as follows:
1. Which stationarity test method do you suggest me to check it ???( since i am not profi yet i checked using fisher type)
2. Should i run regression with variables that were transformed into first difference(since they are not all first differed)
could you give or refer to useful source.
Thanks in advance