Estimating a GMM regression
Posted: Mon Apr 10, 2017 3:00 am
Hi,
I’ve got an assignment for an econometric course, and for this course we are asked to repeat an econometric procedure done in a SSIC ranked article.
The paper I have chosen is attached just in case you want to get a better idea, but I will try describe my issue as best as possible.
The equation estimated is;
g(Y/L)i,t = B0 + B1g(co2/L)i,t + B2g(FDI/L)i,t + B3g(K/L)i,t + e i,t
Where g stands for the growth rate of a variable.
This equation here is estimated using a dynamic panel specification with one-period lagged levels of the dependent variables. The equation is estimated using Arellano-Bond (1991) GMM estimator. Another point to note is that the author takes the first differences when applying GMM.
The results that correspond to this equation is;
GDP per capita (growth) = 0.218 + 0.375GDP(-1) + 0.442FDI - 0.289C02 + 0.263capital
AR2test = 0.089
What I am interested to know is only for specification 1 (results above). Mainly Arellano-Bond result.
I have attached the file that I used in eviews, that tries to capture the same effects described in specification 1. The end result of this is called ‘eq01’ in my eviews workfile. I would appreciate it if you would be able to tell me if I have managed to estimate this correctly by taking a look at my workfile. Some additional information that might be useful is that in step 5 out of 6, in the ‘Dynamic Panel Data Model Wizard’, In the transform box I put explanatory variables ‘GRFDI’, ‘GRCO2’, ‘GRGCF’, whilst in the no transform box I didn’t put anything.
I hope I have managed to describe the issue as easy to follow as possible. Thank you for your time in reading this.
I’ve got an assignment for an econometric course, and for this course we are asked to repeat an econometric procedure done in a SSIC ranked article.
The paper I have chosen is attached just in case you want to get a better idea, but I will try describe my issue as best as possible.
The equation estimated is;
g(Y/L)i,t = B0 + B1g(co2/L)i,t + B2g(FDI/L)i,t + B3g(K/L)i,t + e i,t
Where g stands for the growth rate of a variable.
This equation here is estimated using a dynamic panel specification with one-period lagged levels of the dependent variables. The equation is estimated using Arellano-Bond (1991) GMM estimator. Another point to note is that the author takes the first differences when applying GMM.
The results that correspond to this equation is;
GDP per capita (growth) = 0.218 + 0.375GDP(-1) + 0.442FDI - 0.289C02 + 0.263capital
AR2test = 0.089
What I am interested to know is only for specification 1 (results above). Mainly Arellano-Bond result.
I have attached the file that I used in eviews, that tries to capture the same effects described in specification 1. The end result of this is called ‘eq01’ in my eviews workfile. I would appreciate it if you would be able to tell me if I have managed to estimate this correctly by taking a look at my workfile. Some additional information that might be useful is that in step 5 out of 6, in the ‘Dynamic Panel Data Model Wizard’, In the transform box I put explanatory variables ‘GRFDI’, ‘GRCO2’, ‘GRGCF’, whilst in the no transform box I didn’t put anything.
I hope I have managed to describe the issue as easy to follow as possible. Thank you for your time in reading this.