Dear All,
I would estimate the following model:
Due to the presence of the lagged dependent variable, I intend to use a system-GMM estimator (Blundell-Bond).
My command would be:
where x2=x^2. Moreover, for the variable x I intend to use an external instrument, z. Clearly, I will use z^2 to instrument the squared values of x and z*k and z*h for the interaction terms.
I am confused on how I could enter the above instruments in my routine. Specifically, I do not want to use z (the squared value and the interaction terms) for all my regressors, but only for x x2 xh xk.
Is there any way to do that? Just to further clarify, I would like to mantain the internal instruments included above and only for x x2 xh xk, I would like to further add z, z^2, z*h and z*k.
Thanks for your suggestions.
Dario
I would estimate the following model:
Code:
y_it=alpha+gamma*y_it-1+beta1*x_it+beta2*(x_it)^2+beta3*x_it*h_it+beta4* x_it*k_it+eta_i+eta_t+error_term
My command would be:
Code:
xtabond2 y l.y x x2 xh xk, gmm(y x, lag(2 4) eq(diff)) twos cluster(id) gmm(y x, lag (3 3) eq(lev)) nocon
I am confused on how I could enter the above instruments in my routine. Specifically, I do not want to use z (the squared value and the interaction terms) for all my regressors, but only for x x2 xh xk.
Is there any way to do that? Just to further clarify, I would like to mantain the internal instruments included above and only for x x2 xh xk, I would like to further add z, z^2, z*h and z*k.
Thanks for your suggestions.
Dario
