Endogeneity and endogenous independent variables

[@samanjamshidi1939]

The best explanation that I have encountered so far. Thanks.

[@MasumVatandas]

Perfect explanation, this is what I have been looking for hours! Thanks a lot

[@SimmelPimmel]

Very well explained!!! Thanks helping me pass my exams!!

[@mronkko]

You are welcome!

[@rihamsatti2611]

The best explanation so far. thank you

[@ananthkrishna1196]

Neat and clear explanation. Especially from the theoretical point of view, of addressing endogeneity by ruling out all possible causal variables from causal pathways. Excellent.

[@mronkko]

Thanks!

[@haoyanli8405]

very clear explanation. Thanks!

[@mronkko]

You area welcome!

[@aidanglaserschoff5925]

Hi, where did you learn to use path diagrams for regressions? I haven't seen this diagram style used before and would like to learn more about it. Fantastic explanation by the way. I hope more people can find your videos.

[@mronkko]

Not sure where I got the idea of using path diagrams for regression. There may be differences between disciplines, but for me it seems like an obvious thing to do. That being said, one of the things that I teach my students is that they should all learn SEM. Not because they should all apply it, but because it allows them to understand regression better.

[@arpansagar6453]

Hello thanks for the explanation, so for example if I difference sequential migrant stocks to estimate migrant flows and run a twoway fixed effects model where flows are the dependent variable and log of stocks is the independent variable, then is it a case of potential simultaneity ?

[@mronkko]

Depend on your theory and data. I would say that if you have stock of immigrants at the beginning of the year and you use it to explain the flow during the year, simultaneity is not a problem. However, you are almost certainly running into dynamic panel bias because the stock at the beginning of the year is determined by last year' s flow. You thus have a lagged dependent variable modeled indirectly and unit fixed effects. See my video on Arellano-Bond estimator on the problem that this causes.

[@laxmanpokhrel5153]

Remarkable...

[@mronkko]

Thanks.

[@JOHNSMITH-ve3rq]

Brilliant!!

[@mronkko]

Happy that you liked it!