Hello Everyone! Thanks a lot for your support and helping me reach +1,000 subscribers! ✅Please subscribe to my channel by clicking: ruclips.net/channel/UC5P21WGFO4WRUlAiGLcwymg ✅ Download the Dataset and/or Buy the DO File at: jdeconomicstore.com/b/cointegration-stata ✅Visit my website for all my content: www.jdeconomics.com/ Thanks a lot for all your support and positive Feedback! Good luck on your research and courses! JDEconomics
Great job on this explanation! I'd love to see more videos diving into panel data analysis and other co-integration techniques like ARDL, CRDF etc. Your clear explanations really make complex topics easier to understand. Looking forward to more content like this!
Brilliant way of walking the audience through theory and empirics! Can you please post a video on VECM in STATA using a multivariate model? it would be massively helpful! thanks so much :)
@@JDEconomics Juan I have been looking for your interpretation of GMM models in STATA couldn’t locate it if you haven’t done them yet pls do many thanks xx
Hi! Including a constant in a regression model assumes that the mean of the dependent variable is stable over time. If the series wiggles around a mean of zero, it suggests that the series is already mean stationary, and including a constant might not be appropriate because it would introduce bias into the model. However, if the series starts from a value other than zero (like 10), it could indicate that a constant term should be included to account for this initial level. So, in such cases, excluding the constant with the "noconstant" option might not be appropriate. Regards, JD
Doubt: So I am running a VECM model on 5 variables with 1 dependent variable. However, my variables are integrated of different order. What should be the way forward in this case??
Hi. Thanks for the explanation on ECM above. My worry is, if after testing for cointegration using the Engle and Granger test above and the variables were found not to be cointegrated, would the Error Correction model still be an appropriate technique to use? If not, what would be the proper technique of analysis to use?
Hi, it all depends on what you are working on. Normally you take differences of the variables and work with a var model. Or you can work with linear regression, and include some dynamics to ensure there is no autocorrelation and the blue conditions are satisfied. What methodology you use is upto you. I suggest you look in google scholar for the existing literature about your topic of interest. There is a lot of literature already which can help you. Good luck! jD
Thank you for your amazing videos. Pls, the "ssc install egranger" command didn't work on my stata13, version? Furthermore, is there any other way I could do the test without command?
Thank you sir. One confusion regarding whether or not to go for cointegration. Some are saying its appropriate if we have a mixture of I(0) and I(1) variables. Or should all variables be of the same order?
Hi!😃One question. Should we use: egranger x y, lags() or egranger x y, no lags. Which one to use? Because frequently happens to me that using granger cointegration test with specified lags can show no cointegration between series in comparaison to egrangers like in your case, with no lags . Thank you so much.
What does it mean when my long-run model is cointegrated, but in the short-run model my error correction term is statistically insignificant. I have a sample size of 2700 but both swilk and wntestq indicate the residuals in my error-correction model are autocorrelated and not normally distributed
Hi! In my website you can find the link to download the data to replicate the content. If you want to download any other data, you can try the website FRED (Federal reserve) or World data Bank or the International Monetary Fund. Regarda
Thank you for this part! I have one question about exploring the relationship between series. For a bivariate analysis, can I explore the short-run relationship by VAR model and explore the long-run relationship by ECM model displayed in this video? Thanks so much !
If you are working with a var models, you should use a VECM model. I will try to do a series of videos on it hopefully some time in the future. I have many tutorials to work on. But I will definitely try to cover VEC models. Regards, JD
Hi, Please, what can I do if my eganger test of error term isn't significant? I also noticed that because of the above, both the Logged-difference of the independent variable and the lagged error term aren't significant at 5% level of significance when I regressed the dependent variable against them.
Hello sir In the last table where you have shown the comparison between the long run and short run models there the p-values are different from the ones that you obtained earlier in the models. Why is it so?
Hi sir, I just want to ask if my variables are stationary at lever 1 but not cointegrated, what should I do next? Should I use a var model in the first difference?
@@nabilaa.vashti2602 If your variables are I(0), you can just estimate a normal linear regression. If your variables are I(1) and you confirm there is cointegration, you can estimate the short run model with the error correction term. If your variables are different order of integration, you can't use this model. Hope that helps to clarify your doubts, Regards, JD
"But do you have any hints to do it with panel data, please? I'm working on a project and I need to determine the short-term model with ECM in panel data."@@JDEconomics
Hello Everyone! Thanks a lot for your support and helping me reach +1,000 subscribers!
✅Please subscribe to my channel by clicking:
ruclips.net/channel/UC5P21WGFO4WRUlAiGLcwymg
✅ Download the Dataset and/or Buy the DO File at:
jdeconomicstore.com/b/cointegration-stata
✅Visit my website for all my content:
www.jdeconomics.com/
Thanks a lot for all your support and positive Feedback!
Good luck on your research and courses!
JDEconomics
Great job on this explanation! I'd love to see more videos diving into panel data analysis and other co-integration techniques like ARDL, CRDF etc. Your clear explanations really make complex topics easier to understand. Looking forward to more content like this!
Thanks a lot! Please feel free to subscribe to my channel and share it with your friends/social network. I wish you good luck in your studies/work!
Brilliant way of walking the audience through theory and empirics!
Can you please post a video on VECM in STATA using a multivariate model? it would be massively helpful! thanks so much :)
Thanks for your feedback! Will consider it for coming videos. Best regards, JDEc
@@JDEconomics I join the request for material on VECM. Thank you for this one -- well-explained theoretical part.
Thank you very much for the description of the relevant technics which i am using in my study.
I am glad it helped you! Good luck!
Very beneficial and excellent work. Thank you so much JD Economics!
Thanks a lot for your kind feedback! Good luck
Thank you, @JDEconomics this video was really helpful for my understanding
That’s great. Thanks for your feedback!
Above all Juan I love the economic interpretation you give to your models and results this helps a lot xx
Thanks for your positive feedback! Kind regards, JD
@@JDEconomics Juan I have been looking for your interpretation of GMM models in STATA couldn’t locate it if you haven’t done them yet pls do many thanks xx
Thank you, great lecturer
Thanks a lot! Please feel free to subscribe to my channel and share it with your friends/social network. I wish you good luck in your studies/work!
A video on Threshold Regression will be a big plus. Thank you for the effort.
Thanks for your feedback! JD
well explained with a good speech style. thank you
Thanks! Best Regards, JD
Hi, what is the purpose of the noconstant bit after dfuller test and when is it appropriate to use?
Hi! Including a constant in a regression model assumes that the mean of the dependent variable is stable over time. If the series wiggles around a mean of zero, it suggests that the series is already mean stationary, and including a constant might not be appropriate because it would introduce bias into the model. However, if the series starts from a value other than zero (like 10), it could indicate that a constant term should be included to account for this initial level. So, in such cases, excluding the constant with the "noconstant" option might not be appropriate.
Regards,
JD
Thank you, JD ,very helpful for my thesis presentation
Thanks! I am glad it was helpful! Regards, JD
EXCELLENT
Thanks a lot! Please feel free to subscribe to my channel and share it with your friends/social network. I wish you good luck in your studies/work!
Doubt: So I am running a VECM model on 5 variables with 1 dependent variable. However, my variables are integrated of different order. What should be the way forward in this case??
I love your videos so much and they have been very helpful in polishing us. I would be happy if you could do a video on Monte Carlo Simulation. Thanks
Thanks a lot for your positive feedback. I will consider your video suggestion and add it to the list. Regards! JDEC.
Tk you very much, I'm from IVORY COAST in West AFRICA🙏🙏🙏
You’re welcome! Great Ivory Coast!
thank you very much JD you're really a life saver
Hi! Thanks for your message. I am glad you liked the video. Feel free to share it with your friends. Best Regards, JD
Hello. this is a fantastic tutorial, thank you. Question - what happens if any of the diagnostic tests are failed?
No problem. Make sure to check my website, www.jdeconomics.com
If a test fails, will depend on the type of test. You can always work it out.
Regards
Waiting for garch video 🌻, have you ever did panel data garch? I would love !
Just in case you chase me everywhere, I will say I haven't done any panel data or anything else. hahahah. I will work on the Garch this weekend.
@@JDEconomics ok thank you !
Thanks, do a video on svar with stata
Thanks for your feedback! I will take it into account for future tutorials. Regards, JD
Hi. Thanks for the explanation on ECM above. My worry is, if after testing for cointegration using the Engle and Granger test above and the variables were found not to be cointegrated, would the Error Correction model still be an appropriate technique to use? If not, what would be the proper technique of analysis to use?
Hi, it all depends on what you are working on. Normally you take differences of the variables and work with a var model. Or you can work with linear regression, and include some dynamics to ensure there is no autocorrelation and the blue conditions are satisfied. What methodology you use is upto you. I suggest you look in google scholar for the existing literature about your topic of interest. There is a lot of literature already which can help you. Good luck! jD
Thanks J D！I am looking forward to learning a panel cointegration test in Eviews from your channel：）
Thanks!
Dear Respected Prof! Kindly make the videos on short-run and long-run causality, and the Panel VAR causality test.
Thank you for your amazing videos. Pls, the "ssc install egranger" command didn't work on my stata13, version? Furthermore, is there any other way I could do the test without command?
Hi, sorry for the late reply. You should probably check the stata 13 manual. I am not sure of it. Good luck! JD
Thanks
My pleasure! Make sure to check my website! Www.jdeconomics.com Good luck!
Thank you sir. One confusion regarding whether or not to go for cointegration. Some are saying its appropriate if we have a mixture of I(0) and I(1) variables. Or should all variables be of the same order?
Same
Order
Thank you sir 🙏🏻 please you make explanation vecm panel data.
Thankyou 🙏🏻
Thanks for your feedback! Will add it to the list.
Regards, JD
Thanks for your feedback! Will add it to the list.
Regards, JD
Is this video related to VECM or do you have a video about evryting VECM just like in the VAR?
Hi! This isn’t VEC. I haven’t posted a video about VEC yet. Regards
Thanks for the video. I need the Do file please.
Here is the link jdeconomicstore.com/collection/stata
Good luck!
Hi!😃One question. Should we use: egranger x y, lags() or egranger x y, no lags. Which one to use? Because frequently happens to me that using granger cointegration test with specified lags can show no cointegration between series in comparaison to egrangers like in your case, with no lags . Thank you so much.
Just egranger is fine. Cheers
What does it mean when my long-run model is cointegrated, but in the short-run model my error correction term is statistically insignificant. I have a sample size of 2700 but both swilk and wntestq indicate the residuals in my error-correction model are autocorrelated and not normally distributed
How can I reach the trade and cpi data you used in your video?
Hi! In my website you can find the link to download the data to replicate the content. If you want to download any other data, you can try the website FRED (Federal reserve) or World data Bank or the International Monetary Fund. Regarda
Please make a video on ARDL
Thanks for your suggestion!
Cool !!
Thanks!
what do you do if one was stationary upon the first check, do you still do cointegration?
They need to be of the same order of integration. If they aren’t, you need to try another model.
Thank you for this part! I have one question about exploring the relationship between series. For a bivariate analysis, can I explore the short-run relationship by VAR model and explore the long-run relationship by ECM model displayed in this video? Thanks so much !
If you are working with a var models, you should use a VECM model. I will try to do a series of videos on it hopefully some time in the future. I have many tutorials to work on. But I will definitely try to cover VEC models. Regards, JD
Hi,
Please, what can I do if my eganger test of error term isn't significant? I also noticed that because of the above, both the Logged-difference of the independent variable and the lagged error term aren't significant at 5% level of significance when I regressed the dependent variable against them.
Hello sir
In the last table where you have shown the comparison between the long run and short run models there the p-values are different from the ones that you obtained earlier in the models.
Why is it so?
Hey. I could have forgotten to update the pvalues in the slides. Just make sure you fill the table with the pvalues from the estimated models. Rrgards
How would you write out the short run model equation without a constant?
You would just remove the "c" from the equation. Lmk if you have questions, regards. JD
Hi sir, I just want to ask if my variables are stationary at lever 1 but not cointegrated, what should I do next? Should I use a var model in the first difference?
Depends on what type of model it is. Can be a var or linear regression. Cheers
@@JDEconomics thank you
@@JDEconomics thank you very much for the help
in min 19:14 I am not sure that is the interpretation of the coefficient? it is a first difference of a log-variable...
Hi Sir, if my variables are stationary at different levels, how is the command if I input ECT?
Hi, to use this type of models, you need the variables to be of the same order of integration. Regards, JD
@@JDEconomics Oh I see, so I can choose between I(0) or I(1) to regress it with ECT?
@@nabilaa.vashti2602 If your variables are I(0), you can just estimate a normal linear regression. If your variables are I(1) and you confirm there is cointegration, you can estimate the short run model with the error correction term. If your variables are different order of integration, you can't use this model. Hope that helps to clarify your doubts, Regards, JD
what if your residual is not significant?, how are we going to make residual significant?
In such case the variables are not cointegrated.
with panel data?
I don’t have videos about panel data. But I should
"But do you have any hints to do it with panel data, please? I'm working on a project and I need to determine the short-term model with ECM in panel data."@@JDEconomics