Estimation GMM

[docamani]

Hi,
Now, am working on the cyclicity of the fiscal policy, but knowing that fiscal variables and growth (as GDP) are auto-correlated, the most of studies of this cas suggest the utilisation of the GMM to overpass this problem while the OLS are useless here (and it is the only method i had used since now). Others methods with instrumental variables are also suggested but the GMM seems the most used.

My equation to estimate is: (wich i put in the equation specification)
S = C + a Y + b Y(-1) + b S(-1) + d X(1.....n)

With S a fiscal variable, Y as GDP and X some other variables (économical and political variables).

Well, after that i must introduce the instrumental variables, but sincerly i don't know what variables i must put here to eliminate the endogeinity of the variables or how can i choose them, even if i read so many empirical studies in this case i can't understant how they choose them and the manuals of econometrics focus much in equations than in practice.

Hope i can found a response here and count on your help so please if someone can give me a hand.....

And sorry for the long text and my weak english.