Hi,
I have been grappling with -reg3- command which implements 3SLS for a system of equations. Trying to figure out if I can check if my instruments are strong enough, I came across two question posted earlier which have not been answered:
https://www.statalist.org/forums/for...ak-instruments
https://www.statalist.org/forums/for...ables-via-2sls
It would be odd if an important command like -reg3- missed the essential post-estimation tests. So, I suspect that there must be a reason that the results of the tests of instruments cannot be obtained after running -reg3-. What could the reason be? I know that 3SLS is supposed to work like 2SLS--in some ways-- and therefore, one can run 2SLS for each equation in isolation and get the post-estimation tests. But, in 2SLS choice of instruments is very different from the 3SLS' approach.
Thanks,
Navid
I have been grappling with -reg3- command which implements 3SLS for a system of equations. Trying to figure out if I can check if my instruments are strong enough, I came across two question posted earlier which have not been answered:
https://www.statalist.org/forums/for...ak-instruments
https://www.statalist.org/forums/for...ables-via-2sls
It would be odd if an important command like -reg3- missed the essential post-estimation tests. So, I suspect that there must be a reason that the results of the tests of instruments cannot be obtained after running -reg3-. What could the reason be? I know that 3SLS is supposed to work like 2SLS--in some ways-- and therefore, one can run 2SLS for each equation in isolation and get the post-estimation tests. But, in 2SLS choice of instruments is very different from the 3SLS' approach.
Thanks,
Navid
Comment