Dear All,
I would ask you to give me some suggestions about two problems:
a) I need to estimate a system of simultaneous dynamic panel such as:
y1(t) = y1(t-1) + y2(t-1) + X(t)
y2(t) = y2(t-1) + y1(t-1) + Z(t),
in which each dependent variable (in first difference to eliminate the correlation with fixed error term) depends from its lag and from past lag of the other dependent variable. I think I have 4 endogenous variables: y1(t) y2(t) y1(t-1) y2(t-1). I want to use as instrument lagged values of X(t) and Z(t). How can I solve this problem using reg3' Should I type a command like
"reg3 (eq1: y1(t) y1(t-1) y2(t-1) X(t)) (eq2:y2(t-1) y1(t-1) Z(t)), 2sls endog (y1(t-1) y2(t-1)) inst(l.X(t) l.Z(t))" ?
b) as I have monthly data, should exhist a monthly unit root test? And removing fixed errror and unit root by monthly difference (es. January 2001 - January 2000, February 2001 - February 2000) what kind of instruments I should have to use to perform GMM estimation? As I now that in a classic xtabond lagged dependent variable in first difference is instrumented with all lagged level of dependent variable and predetermined variables from t - 2, also d.y(t-12) [= y(t) - y(t-12)] shoul be instrumented with l.y(t) and l.X(t) from t - 2?
Thank you for your kindly reply,
Andrea Sisto
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/