Dear Statalists,
In the context of production function estimation, I am trying to estimate the following model:
yit = log(be + bm*mit + bl *lit + bk *kit + 2*bee *eit) + a0xt + a1xt*Sit + a2xt*(Sit)2 + a3xt*(Sit)3 + uit,
where i is an establishment, t is a year, and x is the location of establishment i. Variable yit is the log of the expenditure share of a flexible input on total establishment revenue, mit is log materials, lit is log employment, kit is log capital, eitis log energy, Sit is a measure of market share, and uit, is a mean zero error term uncorrelated with any of the other variables. The coefficients a0xt , a1xt, a2xt , and a3xt are location-year specific.
As seen from the above equation, the model I am trying to estimate has a non-linear part in which the coefficient of interest be, bm , bl , bk and bee are the same for all the establishments. This first part of the equation could be easily estimated using the command -nl-. The second part of the model involves a heterogenous intercepts and slopes structure as the coefficients of interest are location-year specific. However, all of them enter linearly. Therefore, this part of the model can be estimated using the command -reghdfe-.
Is there a command that readily estimates the type of equation I want to estimate? If not, what would be the best approach to tackle this problem? I thought to use a GMM procedure, but there might be an easier way to do it.
I appreciate any guidance,
Santiago
In the context of production function estimation, I am trying to estimate the following model:
yit = log(be + bm*mit + bl *lit + bk *kit + 2*bee *eit) + a0xt + a1xt*Sit + a2xt*(Sit)2 + a3xt*(Sit)3 + uit,
where i is an establishment, t is a year, and x is the location of establishment i. Variable yit is the log of the expenditure share of a flexible input on total establishment revenue, mit is log materials, lit is log employment, kit is log capital, eitis log energy, Sit is a measure of market share, and uit, is a mean zero error term uncorrelated with any of the other variables. The coefficients a0xt , a1xt, a2xt , and a3xt are location-year specific.
As seen from the above equation, the model I am trying to estimate has a non-linear part in which the coefficient of interest be, bm , bl , bk and bee are the same for all the establishments. This first part of the equation could be easily estimated using the command -nl-. The second part of the model involves a heterogenous intercepts and slopes structure as the coefficients of interest are location-year specific. However, all of them enter linearly. Therefore, this part of the model can be estimated using the command -reghdfe-.
Is there a command that readily estimates the type of equation I want to estimate? If not, what would be the best approach to tackle this problem? I thought to use a GMM procedure, but there might be an easier way to do it.
I appreciate any guidance,
Santiago
