Page 1 of 1

Multiple endogenous variables in TSLS estimation

Posted: Tue Apr 07, 2020 1:48 am
by livebythecold
Hello,

Assume the following multivariate model

y = a + b_1x_1 + b_2x_2 + u
where x_1 and x_2 are both endogenous variables.

Suppose that lagged values of x_1 and lagged values of x_2 serve as instruments for x_1 and x_2 respectively.
How do I go about assigning instruments for each variables in the same tsls-estimation in EViews? I am using EViews11Univ.

Looking forward to receiving your input and I thanks in advance for any help.

Re: Multiple endogenous variables in TSLS estimation

Posted: Tue Apr 07, 2020 6:10 am
by startz
In tsls all instruments are used for all variables.

Re: Multiple endogenous variables in TSLS estimation

Posted: Thu Apr 09, 2020 12:48 am
by livebythecold
In tsls all instruments are used for all variables.
What estimation technique do I use in the case which I described?

Re: Multiple endogenous variables in TSLS estimation

Posted: Thu Apr 09, 2020 6:12 am
by startz
In tsls all instruments are used for all variables.
What estimation technique do I use in the case which I described?
I'm not sure you can in EViews. Why would you want to? TSLS is asymptotically efficient.