Hi all,
I'm trying to estimate a production function using the gmm command. If I use a dynamic panel approach, it works relatively fine.
where ly is logged output, lk is logged capital, ll is logged labor, and lly is lagged logged output and so forth.
If I want to approach this using a proxy approach instead, it is simple enough for the linear case --
where I obtained lphi from the following:
If I wanted to axe the linear {rho} above and instead nest a non-parametric polynomial or kernel estimator inside the gmm, is there a clean way to do this within the line of code? Essentially looking to put a polynomial in there that for each guessed value of parameters will try to minimize the gmm criterion. Thanks for your time.
I'm trying to avoid
and
because I'm going to want to mess around with the law of motion for TFP later on and will really need to be able to get my hands into the nuts and bolts.
I'm trying to estimate a production function using the gmm command. If I use a dynamic panel approach, it works relatively fine.
Code:
gmm (ly - ( {rho}*lly + {beta0} + {betak}*(lk - {rho}*llk) + {betal}*(ll - {rho}*lll) ) ), instruments(lk lll le lle logwag loglwag) nolog
If I want to approach this using a proxy approach instead, it is simple enough for the linear case --
Code:
gmm (ly - ({beta0} + {betak}*lk + {betal}*ll + {rho}*(lphi - {betak}*llk - {betal}*lll))), instruments(lk llk lll le lle logwag loglwag) nolog
Code:
npregress kernel ly lk ll le, predict(phi) noderiv gen lphi = l.phi
I'm trying to avoid
Code:
acfest
Code:
levpet
