This video demonstrates the use of EViews to estimate Panel ARDL using the PMG estimator. It's fun and sweet, you'll love it! :-) Link to the PDF file: drive.google.com/file/d/1WMnC...
Really informative many thanx. What i noticed is you used the 1st generation unit root test, as this is a panel data, we should test the CSD so that we can decide weather to use 1st or 2nd generation unit root test? Many thanx again
Thanks for this video! I noted that you didn't test for Pedroni cointegration before performing the panel ARDL estimation. 1. Is that something that can be skipped? 2. If there is no Pedroni cointegration, does it make sense to estimate the panel ARDL and look at the long and short run statistics? 3. Is it ok to perform the panel ARDL with only one dependent variable and one regressor? Thank you!
This particular video is the last in my panel VAR series. Earlier videos in the series show unit root and cointegration tests. Please review those first, since those must precede panel VEC or panel ARDL. Yes, you can estimate a bivariate model if that's your hypothesis.
Thanks for the great video! I am working with a dynamic panel model myself and I was thinking of applying panel ARDL, however, literature tells me that applying the ARDL-PMG estimator is biased for short T (12) and N is large (100). Do you have a suggestion of working around this bias by e.g. using a different estimator? I have been looking into the Arellano-Bond estimator however I am not sure if I can deduce short-run dynamics using this estimator? Kind Regards, Hein
should i use original data or differenced data in testing for ardl model or ardl bound testing as my all variables become stationary at I(1) while working in R software
Thank you so much for your understandable videos. But maybe something is wrong in terms of the coefficient of ECT. I think the system corrects about 11% (instead of 0.11%) of the deviation from long-run equilibrium in each period.
Hey Mr Obi! Thanks for the concise video! When I try to run the ARDL model for my data, it gives an error saying near singular matrix. Should I simplify the model, or is there a way I can tweak something to get around this? Thanks!
@@isnisyafhiraadha5602hey, I realised the problem was high multicollinearity among some variables that gave rise to the error in my case. The model worked after dropping one from each pair of problematic variables.
Good question. First, ensure there are no issues with your data and estimation. Coefficients are elasticities, meaning, in this case, that Y reacts differently to changes in X at different lags.
Thanks for this video. However, I ran a panel ardl with 1,0,0,0,0,0,0 lag structure and the short-run result came out with only the ECT coefficient without the coefficients of the other variables. What should I do? how do I report this?
Many thanks to you sir. Could you please show us how to use Hausman test to choose between PMG and MG as you mentioned in the previous video (Panel ARDL: The Concept)
Good day sir. For my thesis, I have to perform ARDL test. I want to know, if I have to perform the test on the first difference or level of my variables. my dependent variable is stationary at level, while dependent is at first difference. Is it possible for me to run ARDL on their levels.
Respected Sir. I have few questions. 1. Ur Eview version. 2. Basic difference between panel Ardl and ARDL, when we have to use panel ARDL. 3 . What does ROOT MSE tells in PANEL ardl model? 4. If ECT is negative but insignificant,how to interept it? 5. Why we use log(natural log) in ardl model.
Regular ARDL cointegration model uses time series data only. Panel ARDL uses panel data and for that reason, incorporates fixed effects in the modeling. In general, root MSE is the standard error of regression and is therefore a measure of variability around the regression line. If ECT is not significant, then there is no long-run Granger causality jointly from the regressors to the dependent variable. In general, log transformation improves the distributional properties of financial and economic data. Hope this helps.
@@PatObi sir, I have 2 questions. 1. If we find multicollinearity in a data(ARDL MODEL), is this a serious problem. How to remove it. If not removed ,what are it's consequences? 2. Same question for AUTOCORRELATION ?
Sir please explain ARIMA Model in detail... Unit root test How data became stationary. All in detail.. in SPSS SOFTWARE PLEASE SIR PLEASE, please 🙏🙏🙏🙏🙏
A time series variable is said to be stationary if its statistical properties (especially, mean and variance) are constant over time. This means that the distribution of a stationary variable is the same at different points in time. A UNIT ROOT TEST will confirm if the variable is stationary or not. If it's nonstationary, you can take the 1st difference to see if it becomes stationary, i.e. I(1). If you have two or more I(1) variables and they are also COINTEGRATED, you can run ECM. But if the I(1) variables are NOT COINTEGRATED, you run OLS on their 1st differences, which would enable you to examine their SHORT-RUN relationship. Finally, if the unit root test shows the variables are stationary at level, then simply run OLS on the level variables; and this would allow you to examine their long-run relationship. My VAR playlist videos will help further. I will also be publishing a basic video on this concept soon. Hope this helps 🙂
Thank you for all of your resources Mr. Obi. Extremely helpful for my thesis
A top-quality and really didactic video. Thank you very much for sharing your precious knowledge with us.
thankyou Mr Obi, your explanation very clearly
Thank you so much for this!
Really informative many thanx. What i noticed is you used the 1st generation unit root test, as this is a panel data, we should test the CSD so that we can decide weather to use 1st or 2nd generation unit root test? Many thanx again
Excellent!
It is very useful. Thank you for sharing this video.
You're welcome!
Thanks for this video!
I noted that you didn't test for Pedroni cointegration before performing the panel ARDL estimation.
1. Is that something that can be skipped?
2. If there is no Pedroni cointegration, does it make sense to estimate the panel ARDL and look at the long and short run statistics?
3. Is it ok to perform the panel ARDL with only one dependent variable and one regressor?
Thank you!
This particular video is the last in my panel VAR series. Earlier videos in the series show unit root and cointegration tests. Please review those first, since those must precede panel VEC or panel ARDL. Yes, you can estimate a bivariate model if that's your hypothesis.
Thanks for the great video!
I am working with a dynamic panel model myself and I was thinking of applying panel ARDL, however, literature tells me that applying the ARDL-PMG estimator is biased for short T (12) and N is large (100). Do you have a suggestion of working around this bias by e.g. using a different estimator? I have been looking into the Arellano-Bond estimator however I am not sure if I can deduce short-run dynamics using this estimator?
Kind Regards,
Hein
Estimate panel GMM per Arellano-Bond, 1991 or Blundell and Bond, 1998. My full playlist: ruclips.net/p/PL6Y8SvWdPo08BIszhwcL2jydMgBXMCKwb
Thanx for your videos Dr. Obi. is it possible to obtain the CUSUM and CUSUMQ chart in case of panel data?
I'm not sure. But please ask others.
hi sir can we perform structural break in a panel data?
should i use original data or differenced data in testing for ardl model or ardl bound testing as my all variables become stationary at I(1) while working in R software
Thank you so much for your understandable videos. But maybe something is wrong in terms of the coefficient of ECT. I think the system corrects about 11% (instead of 0.11%) of the deviation from long-run equilibrium in each period.
informative video.
Thanks for the useful info sir, I have question for my thesis, I use 24 periods and 3 groups for the panel data, can it be use the PMG model?
I suppose you can, although as you know, sample size needs to be at least 30 to achieve sufficient degrees of freedom.
Hey Mr Obi! Thanks for the concise video!
When I try to run the ARDL model for my data, it gives an error saying near singular matrix. Should I simplify the model, or is there a way I can tweak something to get around this?
Thanks!
yeah. i got an error to like "near singular matrix" when i open as equation. Please help me Mr. Obi
@@isnisyafhiraadha5602hey, I realised the problem was high multicollinearity among some variables that gave rise to the error in my case. The model worked after dropping one from each pair of problematic variables.
How do we interpret the short run , if we have lags with difererent signs or significance for the same variable?
Good question. First, ensure there are no issues with your data and estimation. Coefficients are elasticities, meaning, in this case, that Y reacts differently to changes in X at different lags.
Thanks for this video. However, I ran a panel ardl with 1,0,0,0,0,0,0 lag structure and the short-run result came out with only the ECT coefficient without the coefficients of the other variables. What should I do? how do I report this?
Many thanks to you sir.
Could you please show us how to use Hausman test to choose between PMG and MG as you mentioned in the previous video (Panel ARDL: The Concept)
Ok, will do
@@PatObi many thanks to you sir 🙏
Good day sir. For my thesis, I have to perform ARDL test. I want to know, if I have to perform the test on the first difference or level of my variables. my dependent variable is stationary at level, while dependent is at first difference. Is it possible for me to run ARDL on their levels.
In my view, no, unless the stationary variable is included as a regressor. But please be sure to seek the opinion of others.
Good day. Thanks for the video. I ran a panel ARDL but my ECT is .99 not -.99 can I interpret this?
No.
any reason why you choose the none option for the trend specification selcection?
It didn't seem to be relevant in my case.
What if cointegrating equation (speed of adjustment/ error correction term) value is positive and insignificant ? kindly explain
There is no convergence in the series. The variables are said to be 'explosive,' incapable of attaining equilibrium.
In the results part, how long is the long run and how short is the short run?
Undefined. Conceptually, long-run means 'by and large' while short-run means 'in the near term.'
please wanted to find out if you have done a video on NARDL
Here's my NARDL video playlist: ruclips.net/p/PL6Y8SvWdPo0-LorWBVesMi51o4h3D56Nk
Does sample size matters in the value of optimal lag?
For the regression, yes. You need sufficient degrees of freedom.
What to do if COINTEQ01 -0,088, but p value0,064? How to interpret this result?
The coef of long run is significant at 10% level.
How if short run all probability not significant ?
How to choose between pooled mean group vs. mean group? Can you please explain Hausman test for Panel ARDL?
It's almost always better to use the PMG estimator.
Very useful content, thank you! Could please share the ppt as well?
Please click on the file link in the video description.
Respected Sir.
I have few questions.
1. Ur Eview version.
2. Basic difference between panel Ardl and ARDL, when we have to use panel ARDL.
3 . What does ROOT MSE tells in PANEL ardl model?
4. If ECT is negative but insignificant,how to interept it?
5. Why we use log(natural log) in ardl model.
Regular ARDL cointegration model uses time series data only. Panel ARDL uses panel data and for that reason, incorporates fixed effects in the modeling. In general, root MSE is the standard error of regression and is therefore a measure of variability around the regression line. If ECT is not significant, then there is no long-run Granger causality jointly from the regressors to the dependent variable. In general, log transformation improves the distributional properties of financial and economic data. Hope this helps.
@@PatObi Sir ,thanks alot .
@@PatObi sir, I have 2 questions.
1. If we find multicollinearity in a data(ARDL MODEL), is this a serious problem. How to remove it. If not removed ,what are it's consequences?
2. Same question for AUTOCORRELATION ?
Mr. Obi. please share your pdf handbook. thx for evertything
Sure, will be happy to.
@@PatObi Please let me know when your Handbook is published. Your Handbook is one of the best I have seen.
Sir please explain ARIMA Model in detail...
Unit root test
How data became stationary.
All in detail.. in SPSS SOFTWARE
PLEASE SIR PLEASE, please 🙏🙏🙏🙏🙏
A time series variable is said to be stationary if its statistical properties (especially, mean and variance) are constant over time. This means that the distribution of a stationary variable is the same at different points in time. A UNIT ROOT TEST will confirm if the variable is stationary or not. If it's nonstationary, you can take the 1st difference to see if it becomes stationary, i.e. I(1). If you have two or more I(1) variables and they are also COINTEGRATED, you can run ECM. But if the I(1) variables are NOT COINTEGRATED, you run OLS on their 1st differences, which would enable you to examine their SHORT-RUN relationship. Finally, if the unit root test shows the variables are stationary at level, then simply run OLS on the level variables; and this would allow you to examine their long-run relationship. My VAR playlist videos will help further. I will also be publishing a basic video on this concept soon. Hope this helps 🙂
Can you be my private tutor 😊
I'm flattered 🙂
Smart, thank you BRO
You're very welcome 🙏