Page 1 of 1
tsls in panel estimation
Posted: Mon Oct 28, 2013 9:12 pm
by shuchang
In the panel tsls estimation, there is a choice for @cxinst, @perinst. What does it mean? Say if I have a Variable1, and I use Variable1(-1) after the @perinst. Then what am I estimating? The Eviews manual says that it’s the instrument that differs in each period. What would it mean in this context? The instrument is always Variable1(-1) for each period. It means in the first stage, the coefficients on the IV differ in each period? So if I have 1000 observations in the time dimension, then I will have 1000 coefficients for the IV at the first stage?
Re: tsls in panel estimation
Posted: Wed Oct 30, 2013 3:01 am
by shuchang
Hi Gareth, please help. I really need to understand this. For the project I am working on, the results are bad when using @inst, but good when using @perinst. I want to know what is making the difference. Until I understand it, I won't feel comfortable in presenting the 'good' results.
Re: tsls in panel estimation
Posted: Wed Oct 30, 2013 6:21 pm
by shuchang
Hi Gareth, I understand you are the expert on panel estimation in the Eviews team. Grateful if you can help on this one. Let me know if you need any further info from me. Many thanks in advance.