Gauss-Markov violations: summary of issues

[@hansmoleman777]

Again, great work. I am going to recommend your channel to my fellow students.

[@SpartacanUsuals]

Hi, that's great - many thanks for your recommendation. If you have any suggestions for videos on econometrics please let me know. Best, Ben

[@sarimzahid3481]

This course is incredible - thank you

[@labyrinthChecksIn]

Hi Ben, your video lectures are nothing short of phenomenal! Thank you so much! I have a slight confusion with the use of the term "Endogenous Regressors". They have been used to describe the violation of Zero-Cond. Mean of Errors (09:30) and violation of Homoskedasticity in Erros (06:55) in this video. They have also been spoken about in the video "SERIAL CORRELATION TEST - THE BREUSH-GODFREY TEST" in the context of making "Auxillary Regressions robust to Endogenous Variables". My question is, which of these situations are due to Endogenous Regressors? Regards

[@SpartacanUsuals]

Hi, thanks for your message, and kind words. Ok. So endogenous regressors means that there is correlation between the error term and that regressor (explanatory variable) in the equation. In both these cases I am using endogenous regressors to refer to this. In the context of zero-conditional mean of errors I refer to endogenity as a violation of this assumption. In the Breush-Godfrey test, I am speaking about how to make this test robust to endogenous regressors (in other words missing information). Let me know if that helps. Best, Ben

[@labyrinthChecksIn]

Ben Lambert Right, I see, got it! Thanks for the prompt reply and the clarification!!

[@yanbowang4020]

Great work, hope you can keep update.

[@TheGoldenBeerg]

based god ben lambert

[@jasonwong5189]

Hi Ben, this series along with the graduate course are extremely good, as they are condensed but comprehensive, saving me much time and not missing any important points. I've been following them till now. In this video I do have a puzzle. The perfect collinearity, as what you define, should include the condition where age and age square are both in the model, which is quite commonly seen in the literature. Since age square=age^2, there should be perfect collinearity in the model, but it seems fine to almost all econometric papers; why is it the case then? Thanks very much.

[@probablybadvideos]

This would not be a perfect linear relationship, as y=x^2 is by definition a non linear relationship. As such there is no perfect collinearity, just a functional relationship between the two variables. For any finite data, there is always some function you can find which would relate two variables, so this is a meaningless problem. The issue is only with a perfect linear relationship

[@danx2932]

Thanks for the videos. They are very helpful. I have a question with regards to the difference between "homoscedastic error" and "zero conditional mean of error". "homoscedastic error" refers to var(u_i) does not change along with xi. "zero conditional mean of error" mean s E(u_i|x_i) = 0, meaning knowing x doesn't help predict u. If understanding above is correct, isnt Breusch-Pagan testing "zero conditional mean of error", instead of "homoscedastic error" ?

[@Shauracool123]

In 4th error here of conditional mean of error being zero. You say there would be endogeniety So can we check the covariance between error term and explainatory variables, and if cov(xi, ui)= not zero. Then that may suggest endogenous relationship between error and vairable