Dynamic Panel Estimation with Fixed Effects

[matze0786]

Hey EViews,

i've got some trouble with estimating a dynamic panel. I want to check, if there are any opportunistic budget cycles and if there is any influence on political behaviour during divided government.

my dynamic panel equation is (according to Brender and Drazen (2005)):

y(i,t)= y(i,t-1) + ∑β x(i,t) + γWD(i,t) + δDIV(i,t) + μi + μt + ϵ(i,t)

- y is the vector of the 3 time series balance, total expenditures and total revenues
- x is the vector of the 5 control variables gdp per capita, trade, population 15 to 64, population 65 and above and business activity
- WD is an election dummy (election=1, no election=0)
- DIV is a dummy to control if there is divided government at the election date or not
- μi is the individual fixed effect
- μt is the time fixed effect
- ϵ(i,t) is the error term

Some additional information

- 8 cross-sections (8 OECD countries); time period from 1960 to 2001
- unbalanced panel
- i've got the data for all variables
- i'm not really familiar in working with eviews
- i'd like to run 2 estimations (one OLS and one GMM because OLS is not consistent (lag-variable and time fixed effects))


Does anybody of you have some advices for me
- which tests do i have to run BEFORE estimating the equation?
- where do I have to pay attention for what?
- does anybody of you have any expert knowledge concerning dynamic panel estimation (dynamic panel estimation step by step; starting with e.g. unit-root-test, test for multi-collinearity, test for heteroscedasticity, and so on)


Thank your very much for your help and your advices!

(Please excuse my english language performance!)