Hello,
I am sorry for the long post.
I am trying do a research on evaluating the performance of stock (comparison between 3 countries)
Period of study is 10 years ( annual data).
I choice the whole bank sector in each country and I get 13,14 and 15 in each one respectively.
Number of the variables: 11 independent variables and 1 dependent variable ( micro var)
I am trying to do panel data I ended up with short panel data:
And I read a lot of references in different languages and I am lost since the methods of the researches ( steps of the analyze) they use are vary from one to another like: Badi H.Baltagi, Peter Kennedy, Wiliam H. Greene, Philips and moon, Gutierre (they didn’t pay any attentions to the pre-test like unit root, cointegration, serial correlation …etc..), others they did these tests and some of them just did and ended up choosing between ( POLS, FEM,REM) and others choices to go with the ARDL( depending on the pre-test), and some of them got a macro panel but they used a micro panel analyze steps, and others choice the model and then did these test.
So my questions are:
  1. Is there any pre-test in short panel that I can do except for ( heteroskedasticity, multicollinearity and to deal with unbalanced data) in other words does the existing of these issue in the short (micro) panel really effect ( has power on the estimations )?
  2. Can I do the comparison between the 3 countries using FEM or should I study each country separately?
  3. Is those steps I did choice are right ( analyze each country separately) : correlation matrix ( or vif ) to test for multicollinearity, test for heteroskedasticity after running FEM, choice between ( POLS, FEM,REM) ?
Please help me.
Thank you.