Two Stage Least Squares - multiple endogenous explanatory variables

[@xiao6322]

I am a PhD specialized in Econometrics and I watch your lessons daily. I truly enjoy your lectures. Thank you

[@matheuscintrao4339]

You might just be the greatest econometrics teacher. Thanks!

[@itismetomnl]

Very nice video, but I have a question. Why do we need to include both instruments in both first stages? I am imagining a situation in which I have a clear IV for x1, which is not an IV for x2, and an IV for x2 that is not an IV for x1. Why in such a scenario do I need to use both instruments in both first stages?

[@farhimouad2860]

run out of expressions to thank you. i followed all your course till to this video and will ISA contnue my (more efficient and consistent) Learning with you. wish all the best and pry Allah for you

[@issasanou6346]

Hello! I really like this tutorial and I want to thank you. However, how this method can be implemented using stata code?

[@KavyaMenon-j1i]

Hi, I wanted to know if I have multiple endogenous variables, then do the IVs I use to estimate each of them, do they have to be the exact same? Is it okay if I use the same IVs but for different years?

[@piyushashah1]

Lovely videos. If you can introduce some specific examples like say health care cost, insurance, etc, instead of always using the Xs and Ys, it would make it a bit easier to visualize.

[@Kavafy]

"Use z1 as an instrument for z1" What does this mean?

[@tizianociamberlano797]

You are fantastic! Many thanks

[@salvadorhs6080]

Does someone know how to fix the endogeneity problem if the endogenous variables are dichotomic and there is an interaction between them?

[@CatherinePryde]

Thank you so much for your clear explanations! Does anybody know how write code syntaxe to do 2SLS with multiple endogenous variables in R?

[@lastua8562]

rstudio-pubs-static.s3.amazonaws.com/332491_1a839b62d1ed404dbef681d83f5d01c4.html

[@jennyji5732]

Do control variables count as explanatory variables? Cause I have read papers where all three independent variables are endogenous and 2sls is still used. I am not sure whether this satisfy the requirement you mentioned in the end "number of exogenous variables is bigger than that of endogenous ones".

[@lastua8562]

Could you specify what exactly you hint at with control variables?

[@Vangeeha]

thanks! it helps a lot

[@nguyenthanhtung1384]

Dear Prof. Ben Lambert, May I ask what if my model specification includes the interaction term of two endogenous variables, say X1 * X2 in this case? I would appriciate if you can suggest anything. Thank you very much.

[@lastua8562]

Good question. I would say if you include an interaction, you always include the isolated variables as well. Then when you add interactions, you will essentially use the same variables again, which does not increase the number of endogenous variables, and you could predict the interaction term with exogenous variables. Let us know if you found your answer!

[@sami-samim]

What if we have simultaneous equation model with multiple endogenous explanatory variables? can we still use 2sls?

[@lastua8562]

Do you mean VAR?

[@nicolasacosta5901]

There is a mistake. The conditions are (i) delta3 neq 0 or gamma4 neq 0 and (ii) delta4 neq 0 or gamma3 neq 0.

[@JJ-fb2lp]

I agree, otherwise I can just add one rubbish variable and use other one that explains both of the endogenous variable.

[@lastua8562]

Can you confirm this is true if you know the answer?

[@lastua8562]

I think that is also not accurate. If it were, one possible combination would be both delta3 and delta4 neq 0, while all gamma are zero. This would leave one x without instruments...

[@idilrahmat3879]

thankyou for then insight

[@jhbaik6245]

so confusing since you use "z" for both instruments and exogenous....not ideal and clear, definitely.