Mr. Lambert I know it has been a decade since you've posted these videos, but I've performed exceptionally well in my econometrics class and it's all because of your videos. THANK YOU! You are a blessing!
Hi Ben, this is a great video. I learned about the use of panel data and heterogeneity when I was doing my undergraduate degree 35 years ago. The lectures were dry and obviously I did not learn that well. Your explanation is the clearest by miles. It all makes sense! I think I will watch a few more videos of yours. Many thanks for the hard work.
i have my econometrics exam tomorrow and I know nothing since I didn't attend to classes and now im here studying only from your videos, if I can get over 40, I owe you a big thanks
@@FemkeHuisman hahaha i have my econometrics 2 final this Thursday and im sooooo scared because I haven't even started to study. the first comment that I wrote 5 months ago was for econometrics 1, and on Wednesday I have my linear algebra final which is the one im studying right now. good luck on your finals. hope you the best!!
Thank you for your videos, It is very useful to understand the concept behind each econometrics models. I wish I found you while I was doing my courses, however, still very useful for my thesis and teaching a practical approach to econometric. Thank you
Hi, in gretl in order to show that you are dealing with panel data you need to save the data in that particular format. Not sure about SAS I must admit. In Excel I wouldn't try it! Hope that helps, Ben
Great Video! I was just wondering if in such a model is there a method to check for a structural break (especially if the model is a time series cross section with many time periods)?
Thank you for the detailed film. However, there is a question I want to ask. Why are the dummy variables representing Vt not included in a fixed effect model or LSDV model in other vedios?
Hey Ben, could you please explain why do you call city dependent variables as unobservable? The ones mentioned on the video seem to be measurable (demographics, education?). Shouldn't variables used be more like: culture, customs (unobservable metrics)?
Hi Pat. I have a data question. Suppose I am building a model of state-level consumer expenditure on macroeconomic factors using least squares dummy variables regression. Also suppose that inflation data does not exist at state but at national level and the impact of inflation is something I want to capture. Can I take the national-level data and duplicate it for all states?
Good question. Every dummy refers to the time-only effect (similar for all cities). Allowing for such dummies will make it possible for house prices to be predicted differently depending on the time period.
If you have a dummy variable for each observation you will get perfect multicollinearity, with the regressors being linearly dependent with one another (you could form a linear equation with them all). This means that the OLS coefficients cannot be derived as for these to be calculated we need (X'X)^-1 (matrix notation), with X'X only being invertible if X is a full rank matrix - full rank is only present when all the columns are not linearly dependent (i.e. linearly independent) on one another - if full rank is not present we cannot calculate do OLS. Beta 0 represents the house price in year one and the other regressors represent the change in house price in the following years. Hope this helps (may not be perfectly answered).
@@TheMrMoled thank you for the explanation, does this mean that if there were 100 time periods and 100 cities and you wanted dummy variables for them all you would include only 98? Really hope you see this would be very helpful to answer :)
Dear Ben, I am in Advanced Econometrics class which is on Panel data econometric A-Z. I was looking at your videos of Panel Data Econometrics which are 25 in total, Can you kindly specify an arrangement of which video follows which one so that it is easier to follow? Kindly reply me in inbox with names of videos that come one after the other and I can watch them and learn. Your Intro to Econometrics was really great and I learned a great deal from them. waiting Regards, Ateeb
Dude, he has playlists of his videos in his channel. First check there. Also related videos are all his, you can see the time stamp on them. You can make a list yourself.
The example in this video is a good illustration. My understanding is that, observing the same individuals over time can lead to a systematic bias, as certain individuals will display behavioural characteristics specific to them over the period, (assuming the time frame isn't too long). While this may not lead to auto correlation for each observation across time( as the characteristic is fixed, i.e. time-invariant), it will lead to a biased outcome, as not accounting for(or not being aware of) these individual characteristics will lead to incorrect conclusions about the relationship between the covariates and the outcome. You don't have to worry about this in cross sectional data because the sample is assumed to be independent and identically distributed, and hence the average or these individual characteristics(the error term) will cancel out, assuming the sample is large enough. Hope that helps. I may be wrong though! I'm still learning myself.
Mr. Lambert I know it has been a decade since you've posted these videos, but I've performed exceptionally well in my econometrics class and it's all because of your videos. THANK YOU! You are a blessing!
Thank you Ben
Hi Ben, this is a great video. I learned about the use of panel data and heterogeneity when I was doing my undergraduate degree 35 years ago. The lectures were dry and obviously I did not learn that well. Your explanation is the clearest by miles. It all makes sense! I think I will watch a few more videos of yours. Many thanks for the hard work.
BEN YOU ARE A STAR! THANK YOU SO MUCH - I wish you were my teacher :(
@Ignacio Tyson maybe try to find a channel where people aren't trying to learn a university level concept? Wrong market for a shit scam attempt mate
Love your Econometrics videos it's so much clearer than at Uni thanks a lot!
i have my econometrics exam tomorrow and I know nothing since I didn't attend to classes and now im here studying only from your videos, if I can get over 40, I owe you a big thanks
And... what did you score?
F i don’t remember exactly but i think it was like 36. It’s not bad because average score was like 20
@@elifkloss2554 Ooohh nice! I have my econometrics final this Wednesday but another final on Monday so I'm a bit scared for mine...
@@FemkeHuisman hahaha i have my econometrics 2 final this Thursday and im sooooo scared because I haven't even started to study. the first comment that I wrote 5 months ago was for econometrics 1, and on Wednesday I have my linear algebra final which is the one im studying right now. good luck on your finals. hope you the best!!
I wish I was a Brit so I can understand your explanation better. The note is very helpful, though. Best econometric lecture on youtube!
Thank you for your videos, It is very useful to understand the concept behind each econometrics models. I wish I found you while I was doing my courses, however, still very useful for my thesis and teaching a practical approach to econometric. Thank you
Best teacher I Ethiopian
no fucking clue what i just watched but to whoever needs this, goodluck because erm i was trying to find a video for call of duty
Hi, in gretl in order to show that you are dealing with panel data you need to save the data in that particular format. Not sure about SAS I must admit. In Excel I wouldn't try it! Hope that helps, Ben
Excellent video, very well explained. Thanks for uploading.
Great Video! I was just wondering if in such a model is there a method to check for a structural break (especially if the model is a time series cross section with many time periods)?
Thank you for the detailed film. However, there is a question I want to ask. Why are the dummy variables representing Vt not included in a fixed effect model or LSDV model in other vedios?
Hey Ben, could you please explain why do you call city dependent variables as unobservable? The ones mentioned on the video seem to be measurable (demographics, education?). Shouldn't variables used be more like: culture, customs (unobservable metrics)?
Wow!! Thats very clearly explained.
Hi Pat. I have a data question. Suppose I am building a model of state-level consumer expenditure on macroeconomic factors using least squares dummy variables regression. Also suppose that inflation data does not exist at state but at national level and the impact of inflation is something I want to capture. Can I take the national-level data and duplicate it for all states?
Excelent class! Thank you very much.
Thank you!!!! Love you explaination!
Very detailed explanations!
great videos, man! keep up! these videos helped me a lot.
Hi, if I want to examine the impact of a time invariant variable, like geography, how should I proceed? Assuming the data I have is panel
4:20 "which allows for house prices across different cities in the US to be changing over time". Why is this the case?
Good question. Every dummy refers to the time-only effect (similar for all cities). Allowing for such dummies will make it possible for house prices to be predicted differently depending on the time period.
la stua thanks a lot!
that's awesome! thank you for providing such a nice video~!
If t denotes time, why are we using the same t in the delta dummies? What does the subscript of t then refer to if not the individual time period?
should the t not be replaced with an i instead?
no, it should not. t is not inside vt anyway. The t is only to denote that the number does actually refer to time.
Can this also be used if you have group and individual level data?
Hello. Why are we using dummy variables for the first error term? Thanks!
Hi! can anyone kindly help me choose a topic for my econometrics coursework? we are required to use panel data? any ideas?
amazing! thank you Sir!
hey is an excellent video. i thank you as usual.
Hi, Ben, thanks for your excellent presentation. But I still don't understand why you take T-1 not T when adding dummy variable t?
If you have a dummy variable for each observation you will get perfect multicollinearity, with the regressors being linearly dependent with one another (you could form a linear equation with them all). This means that the OLS coefficients cannot be derived as for these to be calculated we need (X'X)^-1 (matrix notation), with X'X only being invertible if X is a full rank matrix - full rank is only present when all the columns are not linearly dependent (i.e. linearly independent) on one another - if full rank is not present we cannot calculate do OLS. Beta 0 represents the house price in year one and the other regressors represent the change in house price in the following years.
Hope this helps (may not be perfectly answered).
@@TheMrMoled thank you for the explanation, does this mean that if there were 100 time periods and 100 cities and you wanted dummy variables for them all you would include only 98? Really hope you see this would be very helpful to answer :)
Dear Ben,
I am in Advanced Econometrics class which is on Panel data econometric A-Z. I was looking at your videos of Panel Data Econometrics which are 25 in total, Can you kindly specify an arrangement of which video follows which one so that it is easier to follow? Kindly reply me in inbox with names of videos that come one after the other and I can watch them and learn. Your Intro to Econometrics was really great and I learned a great deal from them.
waiting
Regards,
Ateeb
Dude, he has playlists of his videos in his channel. First check there. Also related videos are all his, you can see the time stamp on them. You can make a list yourself.
@@cmfrtblynmb02 but for these videos, they are not in a playlist I think
@@rubenvaneupen7327 idk about back then but now they are at least gathered under "undergraduate course in econometrics part 2".
Thank You Sir
Thank you!
thanks mate!! much appreciated
I'm still a little bit confused why heterogeneity is such a problem in panel data as opposed to cross-section data. Why is this?
The example in this video is a good illustration. My understanding is that, observing the same individuals over time can lead to a systematic bias, as certain individuals will display behavioural characteristics specific to them over the period, (assuming the time frame isn't too long). While this may not lead to auto correlation for each observation across time( as the characteristic is fixed, i.e. time-invariant), it will lead to a biased outcome, as not accounting for(or not being aware of) these individual characteristics will lead to incorrect conclusions about the relationship between the covariates and the outcome.
You don't have to worry about this in cross sectional data because the sample is assumed to be independent and identically distributed, and hence the average or these individual characteristics(the error term) will cancel out, assuming the sample is large enough. Hope that helps. I may be wrong though! I'm still learning myself.
How do you account for the city dependent variable in a statistical software such as gretl SAS or stats..or even excel?
nice presentation
Thank you sooo much
Grüßen an BOEF bei professor schlag an der Goethe uni