Serial Correlation summary

[@allenbellas4340]

Thank you. My university's classes are all going online and your videos are a great resource for my econometrics students.

[@fruitylerlups530]

Would a functional misspecification essentially be a form of omitted variable? As in say happiness = c + dogs + e vs. happiness = c + dogs - dogs^2 + e ??

[@paulaspinola1110]

Can you have any of these 3 situations (omitted variable, functional misspecification or measurement error) and still get unbiased estimators?

[@willfowls2400]

great explanation

[@AI-ew1rj]

What do you mean that underestimating our independent variables/Omitting variables will be "carried over to other periods"?

[@DimitriBianco]

If the errors are serial correlated then an over-estimation in one period will result in an overestimation in the next period. This can be caused by an auto-regressive effect from the independent variable (i.e. people slowly adjust to changes in their environment) OR it can be the signal/pattern of a variable that is missing (omitted variable).

[@Dupamine]

​@@DimitriBianco where did 'periods' come from ? Are you talking about time series ? Because I am thinking of general linear regression which is applied once only

[@yfygs]

Thank you for the lecture! One thing I dont get is what does "other estimator" mean? Estimator given by MLE or other method? Can you give an example?

[@DimitriBianco]

GLS and fGLS can correct for serial correlation and are better than OLS given they can be BLUE. Some examples of these are the cochrane-orcutt and praise-winston estimations.

[@shreyastambat1813]

Autocorrelation vs Serial correlation

[@zoozolplexOne]

cool !!!

[@JustinGoedhart]

hello ben

[@JackIsNotInTheBox]

Can you please not write in cursive? Hard to read