Dynamic Panel IV in Stata

Adding a lagged dependent variable in a panel regression is very valuable, but also adds a new source of endogeneity bias. Here's one way to deal with it using basic Stata commands.

Link to tutorial for basic IV/2SLS using cross-sectional data:

[adityasivakumar]

Mike, you were very helpful through your explanations of both theory and Stata applications. Thank you!

[LynnPinski]

Hi Mr. Jonas I have watched so many of your videos and now it is like you are one of my professors all through my Thesis writing. Thank you very much for such overly insightful videos, I am writing my thesis and using this method in analysing my panel. I am finding rather many papers comparing the estimation with GMM and DIV by Hayakawa 2009 and FOD. that aside, do you think with a panel of 54 observational units and 25 time points this dynamic panel IV estimation is consistent and efficient or just cosistent? Which R2 should i use in interpretation, "between"?

[yuchapa]

Thanks Prof. Jonas. I found this video valuable to my research

[dr.elliskakwaa-sekyi]

Wao, this is a very useful tutorial. I wish I had a longer version.

[ttwhyy]

Thanks for this tutorial. Does it matter that the R-squared is .(nothing)?

[vikrantspeaks]

why you are using reg command and not xtreg for this panel data estimation?

[muborizmirzoshoev]

Great video. Just one question. Is it better to use the first difference or fixed effects? What is the difference between the two methods?

[khalik]

Hello teacher,
First of all, I would like to thank you for your videos which help us a lot. Please I have a question for you: I would like to estimate a dynamic model by applying the GMM for a number of individuals which is equal to 16 and T=10, is this feasible? If not, is there another method that I can apply for this case? Thanks in advance.

[MulengaMukuka-io]

Hello Prof do you have the same video using Arellano-Bond

[jacklarx]

Wonderful and helpful video. Is it possible to use the Anderson-Hsio apporoach in a Granger-causality test with panel data? Can you suggest a code?

[votuankiet]

Thanks Prof. Jonas. It's great to know your lecture. I am working on my master thesis and a little bit stuck with these problems. I wonder if i could you dynamic panel model with FGLS in stata to deal with autoregression? Thank you Sir.

[trash_whisperer]

Thank you. This was really helpful. Can you use the dynamic panel data approach while still having fixed effects at the unit of analysis?

[noahspencer]

What would you do if X_{it} was endogenous? Is there a way to do Anderson-Hsiao or Arellano-Bond to create lagged instruments for both X_{it} and Y_{it-1}?

[JohnVandivier]

I've watched this three times over a period of months as I write my first academic paper using ivreg + an Anderson-Hsiao adjustment.

It just keeps getting better.

I did google around about Anderson-Hsiao and heard about a preferred Arellano–Bond estimator. Would love to hear something on this too from you.

[MariaVeronikCC]

Hi Mike! I need to ask you something...I need to specify a model and I have airports and cities, the thing is the Yvariable is the ln of number of employees and the exog/instrumental are the airport's size, if that is the case, wich one has to be the id. I mean...xtset airports year or xtset cities year? And another one...I want to do a MCO with this panel and fixed effec, the command could be "reg DEPENDIENTE110 VariableEnd i.AÑO i.aeropuerto, nocon"??. I hope you can help me, please.

[mustanggemini]

How the number of observation is calculated?

[avazyusibov]

Do you have Arellano and Bond estimation video. I could not find that one. Can you please share the link. If not, can you please film it. Thank you in advance.

[fatemehyavari]

Hi dear Mike
may I aske you about dynamic spatial durbin model in stata soft?

[moonsafar]

Thank you prof, could you please helpe with video talking about panel VARX ( VAR with exogenous variables, if thats possible.❤❤

[jaellelo]

how do you apply the first difference estimator if you only have two time periods?