TH-cam recently changed the way my content will be monetised. My channel now needs 1,000 subscribers. So it would be amazing if you show your support by both watching my videos and subscribing to my channel if you haven’t done so already. Monetising my videos allows me to invest back into the channel with some new equipment so this small gesture from you will be extremely huge for me. Many thanks for your support….CrunchEconometrix loves to teach, support my Channel with your subscription and sharing my videos with your cohorts.
The only person who included all the required steps, my god, I am so thankfull for that. I watched already 100+ of videos from some s....t heads, that are d...b enough to not even explain what are the prerequisites and why is someone doing specific tests, etc.
@@CrunchEconometrix I have a question, in your videos, you have not showed how to test for serial correlation, etc. Are those tests not required in case of Panel ARLD? I can not find any information on that. Everyone is using xtserial, but I am unsure wheather that can or should be applied to Panel ARLD with PMG? Could you maybe give some advice on that? I would highly appreciate that, and I think that would an important topic for a video too.
Thank you so much Prof. I managed to estimate my panel ardl model following all your videos. My thesis is due in the morning. You are the best econometrics teacher on TH-cam so far.
Hi Ali, videos on panel VAR-GMM are available on my Teachable paid platform cruncheconometrix.teachable.com. Enrollment fee of $200 allows you access to all videos published in the School. Thanks
@@kevinwanjala4223 Awesome!!! Kindly spread the word about my videos and YT Channel to your students, friends, and colleagues in Kenya...they will learn some useful skills and tips too!
Thank you for the good content, however I just have one inquiry regarding the unit root test, that is how de we select the number of lags in this test, from where did you come up with lags(1) for all variables. why it is not lags(0) or lags (2). Thanks in advance knowing that I support you for the amazing work you are producing.
Dear Dr. Ngozi, Thank you for the videos. I have two questions for you: 1. How can I reference the ARDL model you presented here in a paper? 2. In the model, ∂ stands for what? 3. Why ∅ is expected to be must be less than 0. Thank you in advance. José Lopes
Hi Jose, I ALWAYS do my best to give detailed explanations on the models specified. So, you many need to watch the clip again. Also, there are references listed at the end of the video and my ARDL papers are available for FREE downloads on my website for detailed understanding of the ARDL model cruncheconomerix.com.ng/shop. Regards,
Dr. Ngozy, I thank you very much for your promptly reply. I deeply appreciate it. I am adapting the ARDL specification model you used in your videos. What citation you recommend.
Hi again Dr. Ngozi, I am sorry. I think I did not express myself correctly. What I want to say is that I am writing a paper and I am using ARDL model. I adapted the model you presented in your videos and I need to reference it in my paper. Is it published in a particular paper or I should only make reference to the video itself. I hoped I explained myself sucessfully. I apologize. Thank you
@@joselopes4588 No problem, I just need clarity on the query. Kindly cite these my ARDL papers (most of my videos are based on methods I have used): (1) Adeleye, B. N. (2020). “Unbundling Interest Rate and Bank Credit Nexus on Income Inequality: Structural Break Analysis from Nigeria”. Journal of Financial Regulation and Compliance, DOI 10.1108/JFRC-04-2020-0035 dx.doi.org/10.1108/JFRC-04-2020-0035 (2) Adeleye, N., Osabuohien, E., and Asongu, S. (2020). “Agro-Industrialisation and Financial Intermediation in Nigeria”. African Journal of Economic and Management Studies, DOI 10.1108/AJEMS-02-2019-0078 dx.doi.org/10.1108/AJEMS-02-2019-0078 (3) Adeleye, N., Osabuohien, E., Bowale, E., Matthew, O., and Oduntan, E. (2018) “Financial reforms and credit growth in Nigeria: Empirical insights from ARDL and ECM techniques”, International Review of Applied Economics, 32(6), 807-820, DOI:10.1080/02692171.2017.1375466 dx.doi.org/10.1080/02692171.2017.1375466 ...thanks!
Thank you very much Prof. Please, without using panel ardl due to the presence of l(2) variables in a model, what estimator is best suited for a variable combination of l(0), l(1) and l(2) in panel data analysis. Thank you.
Hi there, I was just wondering on.. 1. Why do you use lag of 1 when running the unit root test? 2. May I know the difference between the IPS and LLC in a more detailed version?
@@ariefshamsul_ A bit late, but maybe that helps you still. The lags are for testin stationarity at differences. 0 lags would mean, testing stationarity at level of the variable as it is at the moment. Once you include lags(1), you are testing stationarity of that variable at its first difference, not at level! This is important, if you want to use Panel ARDL, you confirm in that way that at least at the 1st difference, your variables are stationary and you can use it.
Great explanation, ma'am professor, but I want to ask something That's the meaning of the ARDL model below of the notation for the sigma p and q? There is j=1 and j=0? Does j represent a unit or group sample from a cross-section or something else?
Is there a video that shows how to arrange the data in Microsoft Excel before we step to open the STATA application, I am confused when you explain how to separate CFA and NCFA Countries. There must be a specific format for data arrangement before running the STATA.
Thank you Dr Ngozi for the videos and very simple and easy to follow teaching. I’m trying to run a panel ARDL with N= 84 and T=12. He are my main questions: 1. Is ARDL model applicable in this sample? 2. When I tried to run an IPS Unit Root test for one of my variables (Polity IV), Stata keeps on saying that there are not enough time periods. However, other variables are not any issue. What could be the problem with my democracy indicator and/or how can I overcome the challenge?
Hi Eddie, if had watch the 1st video, as advised, on "Basics of panel ARDL", you'd realise that panel ARDL is not applicable. May I know from where (location) you are reaching me?
Hello Professor, thank you for your help in this video. I have a question regarding the panel unit root test. I used cross-section Im Pesaran Shin test, which accounts for cross-sectional dependence, for my variables. One variable, which was population, still contained unit roots even for the first difference values. The second difference value did show to be stationary. So I was wondering if I am still able to incorporate this variable in the panel ARDL model? If not should I discard this variable or take other actions? Thank you again for your help!
@@CrunchEconometrix Hello again Professor, thank you for your help. I tried the 5. Optimal Lag Selection (In your next video) and it said that collinearity was detected for the lag variable of population. I am not sure why this is the case as it seemed that there is no high correlation over 0.8 with other independent variables when calculating the Correlation Matrix. Nonetheless, I have dropped this variable accordingly. Thank you again for your help!
Hi Soo, collinearity occurs with the lags used in the underlying algorithm. You will see that from the Stata error messages indicated at the top of the results output.
@@CrunchEconometrix Hello Professor, thank you for your help, may I ask one more question? So if collinearity occurs with the lag variable, is my best option to drop these variables from the model as well? Thank you again!
Hi professor, I have 25 countries and 7 variables , I used this command forval i= 1/25 { . 2. ardl FEIR GII PSCH WBLI GDP ATM GINIIndex if( Countrycode == `i'), maxlag(1 1 1 1 1 1 1) aic matrix list e(lags) . 3. di . 4. } I got this error message : 2. is not a valid command name
Hi Dear Crunch There is Multicollinearity problem in step3 between variables for my panel data. I want to estimate Trans-log production function. Dependent variable is output and independent variables are labor, energy, and capital (in LN form). Panel data is available for 20 years and 23 sections ( (N=20, i=23). what is your suggestion? use other approaches such as Ridge regression? Thanks
Hi Meidy, I deactivated the link due to unethical conduct and several attempts to hack into my Google Drive. All Stata dofiles and some datasets used in my videos are now available on my website upon payment. Here's the link cruncheconometrix.com.ng/shop
Hi, how can i applied the test of second generation on stata or eviews ? (Bai and Ng (2001) , Moon and Perron (2004),Phillips and Sul (2003) Pesaran (2003),Choi (2002) ).
Dear Professor, how did you create "ccy" for two different groups of countries CFA and NCFA. Did you convert your data to different subsamples or you just arrange them in one column? How did you perform the "Sum" command for CFA and NCFA? I have 16 banks, and they have three different groups A, B, C, according to their size. How can i create their groups or this "CCY"
when ever, I tried unit root command I got an error "Im-Pesaran-Shin test cannot have gaps in data". My data is strongly balance. Can you guide how to remove these gaps?
Thank you so much for the video. It was immensely helpful.. I just have a question: Can we use ARDL if some variables of the model are stationary at level form while others are stationary at first-difference? Thank you in advance.
Yes, Debosmita that's one of the core strengths of the ARDL model. See references at the end of the video (and my papers on the Community Tab) for better understanding. Thanks. Please may I know from where (location) you are reaching me?
@@CrunchEconometrix Thank you so much for the reply... I am from India. And also, thank you for the videos as an undergraduate student I benefited immensely from those.
Hello Prof, thank you for sharing the video God bless you more. I have a problem, I installed the pmg command in stata ( ssc install xtpmg, replace pmg), but when i try to start the pmg estimation, i get an error message (numerical derivatives are approximate, flat or discontinuous region encountered). Please, what does this mean? and what can i do?
Hi Kate, thanks for the encouraging feedback on my video. Deeply appreciated. That occurs when there's multicollinearity of the regressors. You may need to drop some and use at most 4.
I believe you are doing fine. At the moment I am doing dissertation but I am struggling to make a final decision on my model due to type of data I have. The data is panel but the number of time periods is larger than the number of observations/ companies (T>N). so I came to find out 2 most talk about model one is Time Series Cross Sectional and Another is Heterogenous Panel ARDL model. I am abit confused and all these models seems legit to be used. What is your Take? Regards Edgar Ndani
Edgar Ndani Hi Edgar, there is no direct answer to your query because many factors inform a methodology. Such as your research question and type of data. So, my answer will be that you study the assumptions of the approaches you have stated and see which best can address your research question and fits your data structure. More importantly, discuss with your Supervisor to harness more comments on the best approach to adopt. Hope these tips are helpful.
Thank you very much for the video. Very helpful! I was wondering why we do not run a hetroskedasticity test for panel data? And we have to run one, what is the command for it? Thanks
@@CrunchEconometrix Thanks professor. When I write the command provided in the do file for both lag length and cointegration, it does not work? Do you have any idea why is that? thanks
Thank you for such informative and useful videos. I have a very large panel dataset with monthly observations for values of a large number of firms. So I have a variable called yearmo: eg. (2005m1) and year: 2005 and month: 1 for each firm. So, my time values are repeated for each firm. When I try to do the xtset value yearmo, it says time values are repeated and cancels the command. How can I go about this?
@@CrunchEconometrix thank you so much for your offer, I have already solved the problem. But now I have another problem. My data is severely unbalanced so that when doing the unitroot fisher test which is the only one appropriate for such a problem, a lot of panel are not computed in the test and I.m sure the results are biased. Do you know how to go about unbalanced data?
Hi, I'm a master student currently working on my thesis using the program Stata to analyse my data. I have been looking at your videos and they are very helpful. I've been receiving the error: repeated time values in sample r(451); I have checked for duplication, dropped observations but still receiving the error message
mrs bozede...can I analysis betwen social variable an economic like democration index and growth, and how much data do we need to run panel ardl....thank you for advise
Yes you can Andres, but make sure that you have sufficient time series data of not less than 30years for each group in the panel....make sure you watch the listed prerequisite videos, read the references listed at the end of the video and subscribe to my YTC if you've not done so!
@@CrunchEconometrix From China, South Western University of finance and economics. By the way how to get the ECM in a panel data every time i'm trying your command it is requiring me to use a time series data, but my research is a panel of 32 countries
Greetings again, what would make STATA give you an error of “code follows on the same line as open brace” after running the command you suggested to check for the optimal lag lengths?
Dear Dr. Ngozi, I am performing a unit root test (syntax: xtunitroot ips var, lags(1) ) and i get the following output: Insufficient number of time periods to compute W-t-bar. Why is that?
Respected mam, many thanks for your videos and explanation. I am pleased with your guidance. I have followed your all videos on Panel ARDL. I have a question that is it necessary to run more than one unit root test for panel ARDL? My data is balanced panel data. I am eagerly waiting for your response. Thank you mam.
@@piyalikumar3497 My humble pleasure...kindly share my TH-cam Channel link with your students, friends and academic community on social media for awareness. They'll learn some useful tips and skills too 😊
Prof Thanks for the educative videos you have offered us. Through the videos i am now learning how to use stata. Please after running the ardl lag length command, it does not work. the feedback i get is " 'i' invalid name " . Please what should i do? Thanks
Hi Felix, good to hear that your interests in Stata has picked up. The "i" is my country identifier. You need to create yours. Watch my video on "Reshape wide to long" to do same.
Hi Felix, good to hear that your interests in Stata has picked up. The "i" is my country identifier. You need to create yours. Watch my video on "Reshape wide to long" to do same.
Hi Dr. Rashmi, I explained this in the video. Note that this video is on panel ARDL so choosing optimal lags is slightly different from that of time series. May I know from where (location) you are reaching me?
Thanks for replying. I am from India. Actually I want to ask if I have to test each of my dependent and independent variable for unit root .Should i specify optimal lag (using aic,sbc) in unit root test command ? Plz guide.
Please when i run the co integration test using the command xtpedroni gdpgr dcf exp tr, nopdols feedback i get is "unrecognized command: xtpedroni" what should i do? thank you
I honestly have no idea about the procedure. I'm yet to familiarize with it. You may need to check other online resources for procedures and Stata codes. May I know your name and from where (location) you are reaching me?
@@Prof.Imadeddin-Almosabbeh It's my humble pleasure, Prof...thanks for the positive feedback on my videos and for subscribing to my TH-cam Channel. Kindly share with your students, friends and academic community in Syria 🇸🇾 and Saudi Arabia! 💕 😊
Hello Doctor, I hope this message reaches you well! Can I run ARDL with 26 observations (time series annual data)?? the estimated results will be correct Many thanks
Hi Dr., thank you for this very informative video. I have a panel dataset including 151 countries spanning 28 years of annual data. Do u think using pmg/ mg/ ardl would be a bad choice of model ? And do u know an alternative that overrides this problem but can still give me a long run/ cointegration like relationship between the variables ? Thank you.
Hi Mirna, most definitely. Once N>T, you cannot use the PMG, MG and DFE estimators. Your choices are FE, RE and GMM even though the time-dimensions is quite large. So, given what you have, I'll say you have a large panel dataset. Thanks for the kind words on my videos, deeply appreciated. May I know from where (location) you are reaching me?
@@CrunchEconometrix thank you for ur reply, I live in Egypt. Just another quick question if I may, do u think a CSDL model would work, from what I understand it might override the small T problem. What do u think ?
I hope all is well with you. In my attempt of running a Pannel ARDL (Stata), I am facing the following problems, just know how to circumvent them. A positive Error Correction Term Unable to perform Unit test root (IPS-Pesaran and Shin ) for on variable. I am getting this answer ( Insufficient number of time periods to compute W-t-bar) While my data time varies (1960-2014). Trying xtunitroot llc lnpol Levin-Lin-Chiu (test requires strongly balanced data Please, Help.
Hi Kidaya, a positive ECT implies that the model is explosive. There is no reversion to long-run equilibrium. You need strongly balanced panel dataset to run panel unit root test. The only option is to source for variables with sufficient observations for all the cross-sections in the panel.
Nice videos, really helpful.. I am trying to conduct xtunitroot ips test however, STATA is giving me an invalid syntax error message, used the command "xtunitroot ips, var, lags(1)". What could be the problem?
Hi Alex, thanks for the positive feedback on my videos. Deeply appreciated. You made an error in the syntax. Remove "var,". May I know from where (location) you are reaching me?
@@CrunchEconometrix the var, is like general.. for instance I have Manufacturing Value Added (ManVA) and when I type the command "xtunitroot ips, ManVa, lags(1)" it brings the error message. I am located in Malawi
My problem is that my database of 14 African countries from 1980 to 2017 does not work in the STAT software. I used the following command: forvall i = 1/14 { ardl dep var indep var if (country_code == "i"), maxlag (p, q, ....) matrix list e (lags) di } the message was "no comments" Help me?
Hi Fizza, kindly note that dofiles are no longer free but available on my website upon payment of a token after which you are allowed a one-time download. Here's the link cruncheconometrix.com.ng/shop/
![เขามัทรี - เอ็กซ์ ศุภกฤต - จอนนี่มิวสิค [ Official MV ]](http://i.ytimg.com/vi/AS9ufKuQIVs/mqdefault.jpg)
TH-cam recently changed the way my content will be monetised. My channel now needs 1,000 subscribers. So it would be amazing if you show your support by both watching my videos and subscribing to my channel if you haven’t done so already. Monetising my videos allows me to invest back into the channel with some new equipment so this small gesture from you will be extremely huge for me. Many thanks for your support….CrunchEconometrix loves to teach, support my Channel with your subscription and sharing my videos with your cohorts.
The only person who included all the required steps, my god, I am so thankfull for that.
I watched already 100+ of videos from some s....t heads, that are d...b enough to not even explain what are the prerequisites and why is someone doing specific tests, etc.
Thanks, Nikolai for your encouraging feedback... deeply appreciated! 🙏
@@CrunchEconometrix I have a question, in your videos, you have not showed how to test for serial correlation, etc. Are those tests not required in case of Panel ARLD?
I can not find any information on that. Everyone is using xtserial, but I am unsure wheather that can or should be applied to Panel ARLD with PMG?
Could you maybe give some advice on that? I would highly appreciate that, and I think that would an important topic for a video too.
Thank you so much Prof. I managed to estimate my panel ardl model following all your videos. My thesis is due in the morning.
You are the best econometrics teacher on TH-cam so far.
Glad to hear, Keleb 🙌. Well done! 🍾🥂
Well explained. Thank you Prof.
Glad it was helpful, Dennis!
Merci Docteur !!! Please we need tutoriels to estimate panel Var bye GMM system
Hi Ali, videos on panel VAR-GMM are available on my Teachable paid platform cruncheconometrix.teachable.com. Enrollment fee of $200 allows you access to all videos published in the School. Thanks
Thank you very much. you have really come in handy in the completion of my Thesis
You made my day wth this positive feedback, Kevin...congratulations! May I know from where you are reaching me?
I am from Kenya. An Postgraduate economic student in Egerton university. The PMG code worked really well
@@kevinwanjala4223 Awesome!!! Kindly spread the word about my videos and YT Channel to your students, friends, and colleagues in Kenya...they will learn some useful skills and tips too!
Thank you for the good content, however I just have one inquiry regarding the unit root test, that is how de we select the number of lags in this test, from where did you come up with lags(1) for all variables. why it is not lags(0) or lags (2). Thanks in advance knowing that I support you for the amazing work you are producing.
Hi Abdelhadi, I chose the lag arbitrarily...and yes, you can run the test with 0 lag or 2 as you desire.
Dear Dr. Ngozi,
Thank you for the videos. I have two questions for you:
1. How can I reference the ARDL model you presented here in a paper?
2. In the model, ∂ stands for what?
3. Why ∅ is expected to be must be less than 0.
Thank you in advance.
José Lopes
Hi Jose, I ALWAYS do my best to give detailed explanations on the models specified. So, you many need to watch the clip again. Also, there are references listed at the end of the video and my ARDL papers are available for FREE downloads on my website for detailed understanding of the ARDL model cruncheconomerix.com.ng/shop. Regards,
Dr. Ngozy,
I thank you very much for your promptly reply. I deeply appreciate it. I am adapting the ARDL specification model you used in your videos. What citation you recommend.
Citation? I don't understand what you mean.
Hi again Dr. Ngozi,
I am sorry. I think I did not express myself correctly.
What I want to say is that I am writing a paper and I am using ARDL model. I adapted the model you presented in your videos and I need to reference it in my paper. Is it published in a particular paper or I should only make reference to the video itself.
I hoped I explained myself sucessfully.
I apologize.
Thank you
@@joselopes4588 No problem, I just need clarity on the query. Kindly cite these my ARDL papers (most of my videos are based on methods I have used):
(1) Adeleye, B. N. (2020). “Unbundling Interest Rate and Bank Credit Nexus on Income Inequality: Structural Break Analysis from Nigeria”. Journal of Financial Regulation and Compliance, DOI 10.1108/JFRC-04-2020-0035 dx.doi.org/10.1108/JFRC-04-2020-0035
(2) Adeleye, N., Osabuohien, E., and Asongu, S. (2020). “Agro-Industrialisation and Financial Intermediation in Nigeria”. African Journal of Economic and Management Studies, DOI 10.1108/AJEMS-02-2019-0078 dx.doi.org/10.1108/AJEMS-02-2019-0078
(3) Adeleye, N., Osabuohien, E., Bowale, E., Matthew, O., and Oduntan, E. (2018) “Financial reforms and credit growth in Nigeria: Empirical insights from ARDL and ECM techniques”, International Review of Applied Economics, 32(6), 807-820, DOI:10.1080/02692171.2017.1375466 dx.doi.org/10.1080/02692171.2017.1375466
...thanks!
Thank you very much Prof.
Please, without using panel ardl due to the presence of l(2) variables in a model, what estimator is best suited for a variable combination of l(0), l(1) and l(2) in panel data analysis. Thank you.
Hi Dennis, I have no idea at the moment. You may want to check out other online resources. Thanks.
Hi dear professor,
I would be grateful, if you share a video or step-by-step manual to estimate ridge regression approach in STATA.
Thanks
Hi Mohamad, thanks for the suggestion but I have no idea about that technique at the moment. Thanks 😊
Tk you
U're welcome, Sir 🥰
Hi there, I was just wondering on..
1. Why do you use lag of 1 when running the unit root test?
2. May I know the difference between the IPS and LLC in a more detailed version?
Hi Arief: (1) to control for serial correlation; (2) related articles on both tests will provide you with detailed information.
@@CrunchEconometrix Thank you. But may I know what will happen if you use a lag of 2 or maybe 3 when running the unit root test. Will it be a problem?
Ariel, using more lags strengthens the power of the test BUT you lose more observations by doing so.
@@ariefshamsul_
A bit late, but maybe that helps you still.
The lags are for testin stationarity at differences.
0 lags would mean, testing stationarity at level of the variable as it is at the moment.
Once you include lags(1), you are testing stationarity of that variable at its first difference, not at level!
This is important, if you want to use Panel ARDL, you confirm in that way that at least at the 1st difference, your variables are stationary and you can use it.
Great explanation, ma'am professor, but I want to ask something That's the meaning of the ARDL model below of the notation for the sigma p and q? There is j=1 and j=0? Does j represent a unit or group sample from a cross-section or something else?
Hi Joko, j=0, j=1 represent the time specifications.
Is there a video that shows how to arrange the data in Microsoft Excel before we step to open the STATA application, I am confused when you explain how to separate CFA and NCFA Countries. There must be a specific format for data arrangement before running the STATA.
Kindly watch my video on "Building a Panel Data."
Thank you Dr Ngozi for the videos and very simple and easy to follow teaching. I’m trying to run a panel ARDL with N= 84 and T=12.
He are my main questions:
1. Is ARDL model applicable in this sample?
2. When I tried to run an IPS Unit Root test for one of my variables (Polity IV), Stata keeps on saying that there are not enough time periods. However, other variables are not any issue. What could be the problem with my democracy indicator and/or how can I overcome the challenge?
Hi Eddie, if had watch the 1st video, as advised, on "Basics of panel ARDL", you'd realise that panel ARDL is not applicable. May I know from where (location) you are reaching me?
What if one of my explanatory variables has a correlation statistic of 0.7955 may i still include this in the model?
That's a high correlation. You may want to drop the variable.
Thanks so much Prof.
Please, is this procedure applicable to unbalanced data?
Yes, absolutely on the condition that there are no TOO many missing observations.
May I know what is the difference between;
1st: xtunitroot ips x
2nd: xtunitroot ips x, lags(1)?
When should I use the second option?
Thanks.
Hi Alice, can use either. You can also search online about the underlying assumptions of the test.
In code CFA the ccy== mentioned. this is for what it’s?
Shaws, ccy is the country classification.
Hello Professor, thank you for your help in this video. I have a question regarding the panel unit root test. I used cross-section Im Pesaran Shin test, which accounts for cross-sectional dependence, for my variables. One variable, which was population, still contained unit roots even for the first difference values. The second difference value did show to be stationary. So I was wondering if I am still able to incorporate this variable in the panel ARDL model? If not should I discard this variable or take other actions? Thank you again for your help!
Soo, I will drop that variable and replace with closer proxy that is stationary either at level or 1st difference.
@@CrunchEconometrix Hello again Professor, thank you for your help. I tried the 5. Optimal Lag Selection (In your next video) and it said that collinearity was detected for the lag variable of population. I am not sure why this is the case as it seemed that there is no high correlation over 0.8 with other independent variables when calculating the Correlation Matrix. Nonetheless, I have dropped this variable accordingly. Thank you again for your help!
Hi Soo, collinearity occurs with the lags used in the underlying algorithm. You will see that from the Stata error messages indicated at the top of the results output.
@@CrunchEconometrix Hello Professor, thank you for your help, may I ask one more question? So if collinearity occurs with the lag variable, is my best option to drop these variables from the model as well? Thank you again!
Drop it from the model and use a closer proxy, if necessary.
Hi professor, I have 25 countries and 7 variables , I used this command
forval i= 1/25 {
. 2. ardl FEIR GII PSCH WBLI GDP ATM GINIIndex if( Countrycode == `i'), maxlag(1 1 1 1 1 1 1) aic matrix list e(lags)
. 3. di
. 4. }
I got this error message : 2. is not a valid command name
Panel ARDL works best with few countries and few variables.
Hi Dear Crunch
There is Multicollinearity problem in step3 between variables for my panel data.
I want to estimate Trans-log production function. Dependent variable is output and independent variables are labor, energy, and capital (in LN form). Panel data is available for 20 years and 23 sections ( (N=20, i=23). what is your suggestion? use other approaches such as Ridge regression?
Thanks
Mohammad, N is TOO large. Reduce to less than 10....and increase T to > 30 years.
Please, why did you use lag (1)?
Hi Bari, one lag used to avoid losing too many degrees of freedom.
When command for ist difference it appears on screen “ factor variables and time series operators not allowed “ please tell by where lies the fault
Hi Saba, no idea what this means. Click on the Stata error code on guide to correct the problem. Thanks.
May I have to ask what does mean ccy in 4:32 mins.?
That's the country classification.
what did you mean when you use this comand
sum x1 x2 x3 if ccy== "cfa"
my question is what is ccy and cfa?
Hi Babure. ccy is the country classification id; CFA is for countries classified as such.
Hi I think I couldn't access the website link you put in the description box. Is there anyway I can get the do file? Thank you
Hi Meidy, I deactivated the link due to unethical conduct and several attempts to hack into my Google Drive. All Stata dofiles and some datasets used in my videos are now available on my website upon payment. Here's the link cruncheconometrix.com.ng/shop
Hi, how can i applied the test of second generation on stata or eviews ?
(Bai and Ng (2001) ,
Moon and Perron (2004),Phillips and Sul (2003) Pesaran (2003),Choi (2002) ).
Hi Ghada, you may need to consult other online resources on these procedures since I do not any videos at the moment. Thanks.
Dear Professor, how did you create "ccy" for two different groups of countries CFA and NCFA. Did you convert your data to different subsamples or you just arrange them in one column? How did you perform the "Sum" command for CFA and NCFA?
I have 16 banks, and they have three different groups A, B, C, according to their size. How can i create their groups or this "CCY"
I categorized them in a column. I told you to watch TIPS TO BUILDING PANEL DATA.
when ever, I tried unit root command I got an error "Im-Pesaran-Shin test cannot have gaps in data". My data is strongly balance. Can you guide how to remove these gaps?
I get the same error message too.
Thank you so much for the video. It was immensely helpful.. I just have a question: Can we use ARDL if some variables of the model are stationary at level form while others are stationary at first-difference? Thank you in advance.
Yes, Debosmita that's one of the core strengths of the ARDL model. See references at the end of the video (and my papers on the Community Tab) for better understanding. Thanks. Please may I know from where (location) you are reaching me?
@@CrunchEconometrix Thank you so much for the reply... I am from India. And also, thank you for the videos as an undergraduate student I benefited immensely from those.
How to set forval before running ARDL command on stata
Hi Veenu, kindly post this on Statalist.org on how to do this.
Hello Prof, thank you for sharing the video God bless you more. I have a
problem, I installed the pmg command in stata ( ssc install xtpmg,
replace pmg), but when i try to start the pmg estimation, i get an error
message (numerical derivatives are approximate,
flat or discontinuous region encountered). Please, what does this mean? and what can i do?
Hi Kate, thanks for the encouraging feedback on my video. Deeply appreciated. That occurs when there's multicollinearity of the regressors. You may need to drop some and use at most 4.
3 depvars and 1 indepvar? What sort of model is that?
3 models
is there a tutorial on ARDL panel done with Eview
No Lesley. You may want to check out other online resources.
I believe you are doing fine.
At the moment I am doing dissertation but I am struggling to make a final decision on my model due to type of data I have.
The data is panel but the number of time periods is larger than the number of observations/ companies (T>N). so I came to find out 2 most talk about model one is Time Series Cross Sectional and Another is Heterogenous Panel ARDL model. I am abit confused and all these models seems legit to be used.
What is your Take?
Regards
Edgar Ndani
Edgar Ndani Hi Edgar, there is no direct answer to your query because many factors inform a methodology. Such as your research question and type of data. So, my answer will be that you study the assumptions of the approaches you have stated and see which best can address your research question and fits your data structure. More importantly, discuss with your Supervisor to harness more comments on the best approach to adopt. Hope these tips are helpful.
Prof, I need theoretical proof for estimating a dynamic panel that have small N and large T with PMG please
Hi Masaud, please read the "Pesaran" papers listed at the end of the panel ARDL videos. Thanks.
Hi Masaud, please read the "Pesaran" papers listed at the end of the panel ARDL videos. Thanks.
Hi Masaud, please read the "Pesaran" papers listed at the end of the panel ARDL videos. Thanks.
Thank you very much for the video. Very helpful! I was wondering why we do not run a hetroskedasticity test for panel data? And we have to run one, what is the command for it? Thanks
Hi Attila, scan thru the empirical literature to see what obtains. Thanks
@@CrunchEconometrix Thanks professor. When I write the command provided in the do file for both lag length and cointegration, it does not work? Do you have any idea why is that? thanks
@@attilashaaran4799 What is the Stata error message?
@@CrunchEconometrix For lag it says ARDL is not recognised and for cointegration it says xtpedroni is not recognised
@@attilashaaran4799 It is because you are yet to install the commands. Use either "ssc install ardl" or "findit ardl". Do same for Pedroni
Thank you for such informative and useful videos. I have a very large panel dataset with monthly observations for values of a large number of firms. So I have a variable called yearmo: eg. (2005m1) and year: 2005 and month: 1 for each firm. So, my time values are repeated for each firm. When I try to do the xtset value yearmo, it says time values are repeated and cancels the command. How can I go about this?
I rarely do this. Send your data to cruncheconometrix@gmail.com and I'll guide you appropriately.
@@CrunchEconometrix thank you so much for your offer, I have already solved the problem. But now I have another problem. My data is severely unbalanced so that when doing the unitroot fisher test which is the only one appropriate for such a problem, a lot of panel are not computed in the test and I.m sure the results are biased. Do you know how to go about unbalanced data?
@@nadahusseinshokry5886 Not at all...
Hi, I'm a master student currently working on my thesis using the program Stata to analyse my data. I have been looking at your videos and they are very helpful.
I've been receiving the error: repeated time values in sample
r(451);
I have checked for duplication, dropped observations but still receiving the error message
Hi Jamella, painstakingly look thru your data for double imputation of the time series and remove. This has happened to me. The error is there.
@@CrunchEconometrix I checked for duplications and there's none so I'm not sure
@@jamellafraser7135 Ok. To harvest more constructive suggestions post it on Statalist.org.
mrs bozede...can I analysis betwen social variable an economic like democration index and growth, and how much data do we need to run panel ardl....thank you for advise
Yes you can Andres, but make sure that you have sufficient time series data of not less than 30years for each group in the panel....make sure you watch the listed prerequisite videos, read the references listed at the end of the video and subscribe to my YTC if you've not done so!
Thanks a lot , i'm also working on CFA and NCFA countries hahahaha
U're welcome, Omalams. Please may I know from where (location) you are reaching me?
@@CrunchEconometrix From China, South Western University of finance and economics. By the way how to get the ECM in a panel data every time i'm trying your command it is requiring me to use a time series data, but my research is a panel of 32 countries
Greetings again, what would make STATA give you an error of “code follows on the same line as open brace” after running the command you suggested to check for the optimal lag lengths?
Means you have incorrectly specified the syntax. Check the braces makes sure you captured the one beneath. Otherwise, Stata won't run.
Hi Doctor, how can I choose the number of lags to run the unit root test? Can I make a different lag for each variable (either 1 or 2) ?
Hi EK, I showed how that is done.
Dear Dr. Ngozi,
I am performing a unit root test (syntax: xtunitroot ips var, lags(1) ) and i get the following output:
Insufficient number of time periods to compute W-t-bar.
Why is that?
It says exactly that. You have few time series to engage the test. How many years do you have?
I have 27 years (1990 - 2016).
Ok. 27years. No. of variables?
Respected mam, many thanks for your videos and explanation. I am pleased with your guidance. I have followed your all videos on Panel ARDL. I have a question that is it necessary to run more than one unit root test for panel ARDL? My data is balanced panel data. I am eagerly waiting for your response. Thank you mam.
I'll say yes to compare the results which makes your analysis more robust.
ok. thank you mam.
@@piyalikumar3497 My humble pleasure...kindly share my TH-cam Channel link with your students, friends and academic community on social media for awareness. They'll learn some useful tips and skills too 😊
CrunchEconometrix yeah sure. 😊
do we need to convert series at I(1) for running ARDL?
Alisha, browse through my Channel and watch my video on "This is how to specify ARDL model"... answers your query.
@@CrunchEconometrix thankyou for ur response but my question is do we need to create a difference series of I(1) variable??
Watch my video. I've just told you what you need.
Prof Thanks for the educative videos you have offered us. Through the videos i am now learning how to use stata. Please after running the ardl lag length command, it does not work. the feedback i get is " 'i' invalid name " . Please what should i do? Thanks
Hi Felix, good to hear that your interests in Stata has picked up. The "i" is my country identifier. You need to create yours. Watch my video on "Reshape wide to long" to do same.
Hi Felix, good to hear that your interests in Stata has picked up. The "i" is my country identifier. You need to create yours. Watch my video on "Reshape wide to long" to do same.
@@CrunchEconometrix thanks Prof. I am grateful
@@CrunchEconometrix thanks
tell me why my data having gap
Geethika, check your data source.
Can you tell why you have tested unit root for lag 1 ? How to choose lag to run unit root tests ?
Hi Dr. Rashmi, I explained this in the video. Note that this video is on panel ARDL so choosing optimal lags is slightly different from that of time series. May I know from where (location) you are reaching me?
Thanks for replying. I am from India. Actually I want to ask if I have to test each of my dependent and independent variable for unit root .Should i specify optimal lag (using aic,sbc) in unit root test command ? Plz guide.
@@dr.rashmiahuja2375 I have a dofile on how to obtain the optimal lags for panel ARDL on my website. Link is at the end of the video.
Thank you for your contribution. How can I obtain the do file of the panel ARDL estimation that you use in this video?
Thanks Gokhan, for the positive feedback. Dofile is available on my website...may I know from where (location) you are reaching me?
I have reached it from your website. Thank you so much again. @@CrunchEconometrix
My pleasure, Gokhan...and may I know from where (location) you are reaching me?
from Turkey @@CrunchEconometrix
@@gokhandemirtas Awesome...please share my YT Channel link with all Turks!!!
what's the command to test variable (xit) included trend and intercept??
I don't quite understand your question?
Please when i run the co integration test using the command xtpedroni gdpgr dcf exp tr, nopdols
feedback i get is "unrecognized command: xtpedroni" what should i do? thank you
Hi Felix, you have to install the syntax for it to run.
Pr, i need your help. What's the best model if N= 24 country and T= 3 years ( 3 inquest))??
Mahjoubi, apply techniques that relate with N>T panels: RE, FE, GMM, etc
@@CrunchEconometrix exemple please of 3 years and 24 country please
Use the suggested techniques.
@@CrunchEconometrix GLM methode no?
You may need to read more on that.
In the ARDL command what is 'i'???
Veena, "I" is the country identifier.
@@CrunchEconometrix Thankyou, ma'am :)
Thanks alot CrunchEconometrix, could you PLZ tel us about NARDL Panel Data. if there is any stata code for this test? and how?
I honestly have no idea about the procedure. I'm yet to familiarize with it. You may need to check other online resources for procedures and Stata codes. May I know your name and from where (location) you are reaching me?
@@CrunchEconometrix I am Imadeddin Almosabbeh, associated prof. in economics, from Syria. Now, I am in saudi arabia.
@@Prof.Imadeddin-Almosabbeh It's my humble pleasure, Prof...thanks for the positive feedback on my videos and for subscribing to my TH-cam Channel. Kindly share with your students, friends and academic community in Syria 🇸🇾 and Saudi Arabia! 💕 😊
Hello Doctor,
I hope this message reaches you well!
Can I run ARDL with 26 observations (time series annual data)?? the estimated results will be correct
Many thanks
Hi Charifa, advisable to use more than 30 observations to account for lags.
Hi Dr., thank you for this very informative video. I have a panel dataset including 151 countries spanning 28 years of annual data. Do u think using pmg/ mg/ ardl would be a bad choice of model ? And do u know an alternative that overrides this problem but can still give me a long run/ cointegration like relationship between the variables ? Thank you.
Hi Mirna, most definitely. Once N>T, you cannot use the PMG, MG and DFE estimators. Your choices are FE, RE and GMM even though the time-dimensions is quite large. So, given what you have, I'll say you have a large panel dataset. Thanks for the kind words on my videos, deeply appreciated. May I know from where (location) you are reaching me?
@@CrunchEconometrix thank you for ur reply, I live in Egypt. Just another quick question if I may, do u think a CSDL model would work, from what I understand it might override the small T problem. What do u think ?
@@mirnah.soliman4459 What's CSDL?
@@CrunchEconometrix It is the Cross-sectionally augmented distributed lag model. I think its command on Stata is xtdcce2
@@mirnah.soliman4459 Oh ok, thanks.
I hope all is well with you. In my attempt of running a Pannel ARDL (Stata), I am facing the following problems, just know how to circumvent them.
A positive Error Correction Term
Unable to perform Unit test root (IPS-Pesaran and Shin ) for on variable. I am getting this answer ( Insufficient number of time periods to compute W-t-bar) While my data time varies (1960-2014). Trying xtunitroot llc lnpol Levin-Lin-Chiu (test requires strongly balanced data
Please, Help.
Hi Kidaya, a positive ECT implies that the model is explosive. There is no reversion to long-run equilibrium. You need strongly balanced panel dataset to run panel unit root test. The only option is to source for variables with sufficient observations for all the cross-sections in the panel.
@@CrunchEconometrix Thanks prof.
Nice videos, really helpful.. I am trying to conduct xtunitroot ips test however, STATA is giving me an invalid syntax error message, used the command "xtunitroot ips, var, lags(1)". What could be the problem?
Hi Alex, thanks for the positive feedback on my videos. Deeply appreciated. You made an error in the syntax. Remove "var,". May I know from where (location) you are reaching me?
@@CrunchEconometrix the var, is like general.. for instance I have Manufacturing Value Added (ManVA) and when I type the command "xtunitroot ips, ManVa, lags(1)" it brings the error message. I am located in Malawi
@@everythingnyimbo To be on the safer side, why not use my dofiles? They are available on my website.
why we choose number of lags 1, but we don't choose lags 0?? or 2??
Your query is unclear, Mahjoubi. What are you referring to?
@@CrunchEconometrix 8 variables and 24 country
How many years?
@@CrunchEconometrix 18
Use CCE, CCE-MG and AMG techniques. I don't have videos on these. You will need to search online for resources and guides.
Hello Mme, is it up to us to select the number of lags while doing the unitroot test?
Ashna Gangoo Yes
Ok Mme, thanks a lot for your reply
My problem is that my database of 14 African countries from 1980 to 2017 does not work in the STAT software. I used the following command:
forvall i = 1/14 {
ardl dep var indep var if (country_code == "i"), maxlag (p, q, ....)
matrix list e (lags)
di
}
the message was "no comments"
Help me?
Hi Rahma, the error message is a bit strange to me.
Do you have an order that can execute the ARDL panel for 14 African countries?
@@rahmanakbi1905 That is what I showed in the video.
Hello,
I watched this video and used all the commands but STATA 15 MP did not work for me then you have another command for 14 African countries
When the dependent variable is i(0). can we applied the ARDL model?
Adapting Pesaran et al (1999, 2001), yes you can.
non
Maam If time period covers from 2000 to 2018 and the number of cross section units are 3 (t=19, and N=3) should I go for Panel ARDL ?
Hi Fida, increase time to > 30.
@@CrunchEconometrix if not possible to increase t because of missing observations
The you have small sample bias which a Reviewer may take you up on.
I can't find do file please guide me
Hi Fizza, kindly note that dofiles are no longer free but available on my website upon payment of a token after which you are allowed a one-time download. Here's the link cruncheconometrix.com.ng/shop/