Thank you again Sir Ijaz for your great lecture. I just want to add something that if your data have "heteroscedasticity" then you can check the stationarity of the data, if all the variables are non-stationary at the level and stationary at first diff then you can go for Jhonson "co-integration" test. But if some of the variables are stationary at the level and some at first diff, then you can go for "panel ARDL" or GMM. Thanks again Sir Ijaz....
thanks alot from useful vidoe and explanantion. My question is does robust command for the model which hausman test suggested, overcomes the problem heteroskedacity. is the result is acceptable. if not how we can solve the problem of heteroskedacity. thanks
Hello sir , Greetings i would like to know of any other poolability test if the companies are more than 30 . or can i just create dummies ? i have 37 companies and three year data. So will it be appropriate to make so many dummies to check for poolability ? If not please suggest any other test for checking poolability. Thankyou :)
Well in my opinion, with that much low number of cross sections, do not go for dummies as it shrinks degree of freedom. U can go for the three comparisons I.e. pooled VS fixed, Pooled VS Random and Fixed VS Random
Hi Sir..my variables have no value in points...like audit committee meeting..it will be 4 or 5 but not in points between them e.g 4.5..rvpplot of it not like as you show in video..how i can solve normality problem and hetroscadicity.
Hello and thanks for the video. I am having some challenges with the testparm command. I have varlist required. I am using dummy variables [language (1 and 0) and border (1 and 0]. please, can you guide me on this?
th-cam.com/video/dn1ZIhAjmec/w-d-xo.html&lc=z13gh5mghxb3v5wyt04cer0xvvfscxdpu0k you can download it from this link and subscribe for more at my youtube channel as well at th-cam.com/channels/Mcl97AA1OOCKtZOz9vUxQA.html
Thank you, Sir I am subscribing surely. I will like to know how can we apply post estimation commands like residual normality, autocorrelation and heteroscedasticity in fixed effects or random effects model.
well Miss Sumera Arshad, you can run fixed effects or random effect with robust standard errors or cluster command or VCE robust to control for heteroscedasticity in these models while for autocoreelation u can use xtserial command but u have to install it first. commands are: findit xtserial install it and then execute it and if u r using facebook, be a member of hossain academy group. i and a lot more experienced ppl discusses such kinda questions and give solutions to the students.
Love and respect from China. As you examine the "stationarity in the variance" But Sir for examining "stationarity in mean", If we run the unit root test i.e. "Lin Lu Chi test", some variables are stationary at the level and some are at the 1st difference. What would you suggest? which model is suitable?
Thank you for this informative video. I would like to know if on running a normal regression, no heteroscedasticity is present but after running the panel fixed effect model, the 'xttest3' command ( Wald Test) shows there is hetero present, then what should be the course of action. Will FGLS be helpful if autocorrelation is also present?
thank you very much how about if my data have heteroskedasticity, both test has same result indicating there is heteroskedasticity how i can fix? thank you
Thank You.. Let me find my results.. FInding difficult in using stata due to observations and outliers.. When I drop the variable one by one all my data got vanished :-(.... But thanks once again
Thank you for the video, Sir I would like to know that how do I use PCA on panel data. I have data of "crime against women" which is cross sectional and time series. And I want to create an index using PCA. But how should I approach pca for panel data. Please explain. Or if possible please make a video.
well Mam in case u wana use PCA to panel data, first u have to see whether it's appropriate and necessary to use or not and that u can do in light of ur research topic and the variables u have included and most importantly , the theoretical relationship b/w them (Moderation or mediation). Moreover in case of applying PCA to panel data, u have two choices. either u will get an index variable (one with largest eigen value), after performing PCA, to use for further study or u will run PCA to all the components of ur variable and then use all components for analysis. in first case u may be throwing out information which isn't good as long as u have an excellent and strong logic for it. in the later case, No information is gained or lost. Nothing else in your analysis will change, and, indeed, the omnibus test of the joint significance of all the components of ur variable will come out the same regardless of whether you use the original variables of the components.
briefly i would say about applying PCA to panel data is that, If you can safely assume that the covariance among the items is constant over time, then the loadings, coefficients, and eigenvalues you get from applying PCA to the panel data will be fine. What you cannot rely on, however, would be standard errors of any of these quantities, nor standard errors of estimated component scores. This is because in panel data, the assumption of independent observations is violated and I am not aware of any software that does PCA and adjusts for nesting of observations within panels. If the covariance structure of the items do vary over time, then a PCA done on the entire data set will produce results that are difficult if not impossible, to interpret and use. And there is an additional consideration. Remember that PCA does not involve the outcome variable in your model, only the proxy measures. If the relationships between each of the measures and the outcome vary differently over time, using PCA will completely obscure that.
Thank you again Sir Ijaz for your great lecture.
I just want to add something that if your data have "heteroscedasticity" then you can check the stationarity of the data, if all the variables are non-stationary at the level and stationary at first diff then you can go for Jhonson "co-integration" test. But if some of the variables are stationary at the level and some at first diff, then you can go for "panel ARDL" or GMM.
Thanks again Sir Ijaz....
very nice informative video. keep making more
Sir is there a way to contact you for professional assistance? I tried to mail but i think you did not find the mail.
thanks alot from useful vidoe and explanantion. My question is does robust command for the model which hausman test suggested, overcomes the problem heteroskedacity. is the result is acceptable. if not how we can solve the problem of heteroskedacity. thanks
Can u plzz give the data set on panel
Hello... @ijaz ... do you give tuitions on this stuffF please.? Is there a way to contact you? Lef me know please.
Thanks
Well u cant say tution, but I do help out people
@Chandni Yaseen as far as the contact is concerned, u can contact me through email, through mobile phone, facebook, messenger, whichever suits u.
Please can you send me your details.. i only saw your messages now. I didnt get any notification earlier
@@mrmoodymusa7629 ijazkhan_ims@hotmail.com
rest details through email
plz upload a video on panel unit root test.
Hello sir..is there a way to contact you??
Hello sir , Greetings
i would like to know of any other poolability test if the companies are more than 30 . or can i just create dummies ? i have 37 companies and three year data. So will it be appropriate to make so many dummies to check for poolability ? If not please suggest any other test for checking poolability. Thankyou :)
Well in my opinion, with that much low number of cross sections, do not go for dummies as it shrinks degree of freedom. U can go for the three comparisons I.e. pooled VS fixed, Pooled VS Random and Fixed VS Random
How exactly did you check on heteroskedasticity in the panel data regression? You just estimated with robust errors thats it.
Hi Sir..my variables have no value in points...like audit committee meeting..it will be 4 or 5 but not in points between them e.g 4.5..rvpplot of it not like as you show in video..how i can solve normality problem and hetroscadicity.
Dear sir, could you please show us how to run these models in R. Thank you
We can also post videos on demand
Hello and thanks for the video. I am having some challenges with the testparm command. I have varlist required. I am using dummy variables [language (1 and 0) and border (1 and 0]. please, can you guide me on this?
would u make gmm estimation video and tanks
it is best on panel data. thank you so much. can you send me a link to download it
th-cam.com/video/dn1ZIhAjmec/w-d-xo.html&lc=z13gh5mghxb3v5wyt04cer0xvvfscxdpu0k
you can download it from this link and subscribe for more at my youtube channel as well at th-cam.com/channels/Mcl97AA1OOCKtZOz9vUxQA.html
Thank you, Sir I am subscribing surely. I will like to know how can we apply post estimation commands like residual normality, autocorrelation and heteroscedasticity in fixed effects or random effects model.
well Miss Sumera Arshad, you can run fixed effects or random effect with robust standard errors or cluster command or VCE robust to control for heteroscedasticity in these models while for autocoreelation u can use xtserial command but u have to install it first. commands are:
findit xtserial
install it and then execute it
and if u r using facebook, be a member of hossain academy group. i and a lot more experienced ppl discusses such kinda questions and give solutions to the students.
Love and respect from China.
As you examine the "stationarity in the variance"
But Sir for examining "stationarity in mean", If we run the unit root test i.e. "Lin Lu Chi test", some variables are stationary at the level and some are at the 1st difference. What would you suggest? which model is suitable?
Thank you for this informative video. I would like to know if on running a normal regression, no heteroscedasticity is present but after running the panel fixed effect model, the 'xttest3' command ( Wald Test) shows there is hetero present, then what should be the course of action. Will FGLS be helpful if autocorrelation is also present?
www.stata-journal.com/sjpdf.html?articlenum=st0128
Read the article and hopw this will b helpful
thank you so much, it helps me a lot. by the way would you please run a gravity model test? no one till now has done it.
I will try for that as well
Thank you for this video. Very helpful.
thank you very much how about if my data have heteroskedasticity, both test has same result indicating there is heteroskedasticity how i can fix? thank you
run robust analysis or VCE robust standard errors
hi, you drop some observation due to normality problems, so you have unbalanced panel data, is that ok?
I have demonstrated that just to check and Cure for Normality problems. That's another issue as what happens after deleting outliers.
Thank You.. Let me find my results.. FInding difficult in using stata due to observations and outliers.. When I drop the variable one by one all my data got vanished :-(.... But thanks once again
Thank you very much for this video.
so helpful...thanks
Thank you for the video, Sir I would like to know that how do I use PCA on panel data.
I have data of "crime against women" which is cross sectional and time series. And I want to create an index using PCA. But how should I approach pca for panel data. Please explain. Or if possible please make a video.
Well mam u r welcome but please clarify, what do mean by PCA. Principle component analysis or the one used in Finance?
Ijaz Khan I mean principle component analysis .
well Mam in case u wana use PCA to panel data, first u have to see whether it's appropriate and necessary to use or not and that u can do in light of ur research topic and the variables u have included and most importantly , the theoretical relationship b/w them (Moderation or mediation).
Moreover in case of applying PCA to panel data, u have two choices. either u will get an index variable (one with largest eigen value), after performing PCA, to use for further study or u will run PCA to all the components of ur variable and then use all components for analysis.
in first case u may be throwing out information which isn't good as long as u have an excellent and strong logic for it. in the later case, No information is gained or lost. Nothing else in your analysis will change, and, indeed, the omnibus test of the joint significance of all the components of ur variable will come out the same regardless of whether you use the original variables of the components.
briefly i would say about applying PCA to panel data is that, If you can safely assume that the covariance among the items is constant over time, then the loadings, coefficients, and eigenvalues you get from applying PCA to the panel data will be fine. What you cannot rely on, however, would be standard errors of any of these quantities, nor standard errors of estimated component scores. This is because in panel data, the assumption of independent observations is violated and I am not aware of any software that does PCA and adjusts for nesting of observations within panels. If the covariance structure of the items do vary over time, then a PCA done on the entire data set will produce results that are difficult if not impossible, to interpret and use. And there is an additional consideration. Remember that PCA does not involve the outcome variable in your model, only the proxy measures. If the relationships between each of the measures and the outcome vary differently over time, using PCA will completely obscure that.
Ijaz Khan thanks for explaining it very nicely. 😊
Thank you so much
what about GMM how can I run