(Stata13): How to Perform Johansen Cointegration Test

[@damilolaafolabi3679]

Thank you, my Mama :) I have learned so much from watching your videos. God bless you Ma.

[@CrunchEconometrix]

You are so welcome, Dami!

[@chidiogombelede9088]

May God bless you & everything you do! I’ll come back here when my MSc result is out to let you know the feedback. Thank you MA!

[@CrunchEconometrix]

U're welcome, Chidiogo. I really appreciate your positive and encouraging feedback... wishing you all the best! 🥰🙏

[@elidetenga2662]

Thanks alot, from Tanzania. This video is clear and quick to catch

[@CrunchEconometrix]

Thanks Elide, for the encouraging feedback. Deeply appreciated! Much love from Nigeria!

[@syedamarjanarazzak5130]

Thank you so much Ma'am! I can't express my gratitude to you in words. I go through your videos whenever I am working with stata. May the Almighty bless you.

[@CrunchEconometrix]

It's my pleasure, Syeda...deeply appreciated!

[@alejoforero89]

Awesome refresher. Thanks from Colombia.

[@CrunchEconometrix]

Glad it was helpful, Alejo!

[@okwordesmonda.8819]

you are a career saver. thank you very much

[@CrunchEconometrix]

Hahahaha, thanks Okwor for the positive feedback on my videos. Deeply appreciated! 💕 Kindly share my Channel link with your students and academic networks. May I know where you are reaching me from?

[@1nadjmi1]

this is clear quick and neat presentation. you save a lot of time, thanks!

[@CrunchEconometrix]

Thanks for the positive feedback, Nadjmi...deeply appreciated. May I know the country from where you are reaching me?

[@CrunchEconometrix]

Thanks Nadjmi! Please tell others by sharing my Channel link...

[@1nadjmi1]

@@CrunchEconometrix currently in UK but i live in Singapore

[@CrunchEconometrix]

@@1nadjmi1 Gr8! Keep watching...keep sharing too!

[@jamshidyolchi2497]

The first comment on KZbin. thumps up. you are doing a great job. tanx

[@CrunchEconometrix]

Thanks for the encouraging feedback, Jamshid!

[@elvisabudu5390]

Thank you for such an elaborate tutorial.

[@CrunchEconometrix]

Thanks Elvis, I am also glad that you find the content very helpful. May I know from where (location) you are reaching me?

[@elvisabudu5390]

Ghana

[@CrunchEconometrix]

@@elvisabudu5390 Awesome! Kindly spread the word about my KZbin Channel to your friends and academic community in Ghana 🇬🇭 for awareness. They'll learn some useful tips and hints too. Thanks 😊

[@elvisabudu5390]

Sure. I wish you well ma’am.

[@tawandapasipanodya8227]

Thanks, from Zimbabwe

[@CrunchEconometrix]

U're welcome, Tawanda. Give my love to the academic community in Zimbabwe. Kindly share my videos with them...grateful!

[@peterdavidkulyakwave6102]

Thankx xo much my journey has become shortened indeed. Madam receive more blesses🙏🙏

[@CrunchEconometrix]

U're welcome, Pete! Please tell others about my KZbin Channel...☺️

[@carlosramosvp]

Thanks very much, from Brazil!

[@CrunchEconometrix]

U're welcome, Carlos...kindly share my videos with your friends and and academic community :)

[@natro_09]

Awesome lecture👌

[@CrunchEconometrix]

Thanks for the encouraging feedback, Ernest. Deeply appreciated!

[@rupanandawidanage947]

Thank you very much for the presentation. It is excellent

[@CrunchEconometrix]

You are welcome Sir!

[@acewardell]

Thanks so much. You're the best

[@CrunchEconometrix]

U're welcome, Ace. Glad to know you find this video helpful. Please may I know from where (location) you are reaching me?

[@Thenotgivingmyname]

Thank you! PhD saver!

[@CrunchEconometrix]

U're very welcome! May I know from where (location) you are reaching me?

[@Thenotgivingmyname]

@@CrunchEconometrix I am studying in the UK :)

[@CrunchEconometrix]

@@Thenotgivingmyname Awesome! Please spread the word about my videos to your students and academic community in the UK 🇬🇧...thanks! 💕 😊

[@clivemairura6191]

Good presentation

[@CrunchEconometrix]

Thanks Clive. Commendation is humbly taken...kindly help to share my YT Channel link with your students, colleagues and academic community...gracias!

[@davesamuel5864]

What a lecture! Thanks

[@CrunchEconometrix]

Thanks for the encouraging feedback, Dave. Deeply appreciated! Kindly share with your academic network.

[@guimausse3604]

Thank you for the lesson. I would like to understand how do you proceed the Johansen cointegration test if the variables show multicollinearity even though the series are I(1).

[@CrunchEconometrix]

Drop one of the collinear variables.

[@ibrahimmassaquoi3725]

Thanks for the video. Why do we reject the null hypothesis when maximum rank 2 shows that we have 2 cointegrating equations?

[@CrunchEconometrix]

Hi Ibrahim, I explained the hypothesis being tested in the video. Kindly watch again. Thanks.

[@nurlanrehimli2576]

Thank you sooooooooo muchhhhhh

[@CrunchEconometrix]

Glad you find this helpful, Nurlan!

[@dgscholar]

Thank you ma'am

[@CrunchEconometrix]

Thanks for the encouraging feedback, deeply appreciated ❤️

[@moviesanddramakorea]

Good day Ma. Thanks for your lecture. It has been of great use to me. I want to ask a question, can I use Johansen for Panel cointegration test?

[@CrunchEconometrix]

Hi Kezia, thanks for the encouraging feedback. Deeply appreciated ☺️ Not at all. Kindly watch my video on panel ARDL...well explained to guide you. Thanks.

[@chai4647]

Hi, This is very helpful ! I was wondering if how I can find the constant term for the normalized cointegration equation constant term as most thesis papers have this and I cant seem find it

[@CrunchEconometrix]

Hi Chai, I showed that in my VECM videos. Can't recall which. So, you may need my VECM clips. Thanks.

[@mimiemohamad]

Hi Prof, I have panel data for 6 countries n 25 years. I have done panel cointegration and found out the data is co-integrated. I am confused whether I should have Johansson Cointegration test as well. I tried to search your video for panel Johansson test, it is available for EViews but not for stata. Actually, what should I do next after found out my variables have long run relationship. Thank you very much for your kind advice.

[@CrunchEconometrix]

Hi Mimie, JCT is NOT applicable for panel data study. Kindly watch my videos on panel ARDL. Thanks.

[@walterandremonjane2312]

Hi, thank you for the lessons, it’s been helping me a lot. I would like to know how to make a conclusion and proceed in the Johansen test if we cannot reject the null hypothesis in two or more equations.

[@CrunchEconometrix]

Walter, this video is well explained. Please watch it again and take your decision.

[@frankosei-kusi9404]

thank you for the video, the automatic lag here is 2 can it be changed?

[@CrunchEconometrix]

Yes, but give a reason in your work as to why you are changing it.

[@MuskanSingh-ro8qt]

If a series is non stationary at levels but stationary at First difference, so johansen test is conducted on that original series or difference series please explain, if i am conducting Johansen test on original series VAR model is recommended but while using it on difference series there is long term relationship and VECM is recommended which one of the model is used for forecasting purpose?

[@CrunchEconometrix]

Hi Muskan, either can be used for forecasting.

[@MuskanSingh-ro8qt]

Thank you so much for your reply.

[@MuskanSingh-ro8qt]

I have one more doubt regarding johnsen cointegration test as per AIC and HQ criteria lag 4 is selected and as per SC criteria lag 1 is selected and if i perform Johansen test using K=2 it would recommend me to go for VAR(1) and as per K=4 it would recommend some kind pf cointegration among variable and then use VECM. Can you please help me out with this problem?

[@MuskanSingh-ro8qt]

Can i select the VAR model with 1 lag as per SIC criteria and go ahead with my problem of forecasting?

[@CrunchEconometrix]

I explained the rule of thumb for selecting the appropriate information criteria. Use the one with the lowest statistic.

[@peshalamadhuwanthi217]

Thank you for the lesson but i have a question . In my data set have combination of l(0) and I(1) . According to the bounds test it can use only stationary and level form. So which one is the most sutable for my cointegration test madam.

[@CrunchEconometrix]

Hi Peshala, this clip is about the JCT. So, kindly watch my video on the Bounds test again. Well explained.

[@peshalamadhuwanthi217]

@@CrunchEconometrix ok thank you madam..

[@shuaiyuan2164]

Dear Dr Ngozi, thank you so much for your video! I am a postgraduate student from the UK and feel your video is quite helpful! By the way, I have a stupid question and I would appreciate it very much if you could help me with it: After realising that a set of variables are cointegrated through the Johansen Conintegration Test, we will know that there should be a long-run relationship equilibrium among these variables. So how could we know the specific long-run relationship equilibrium (like Y= 0.5 + 0.3 X1 + 0.4 X2 + 0.1 X3)? (Is this long-run relationship equilibrium being achieved just by 'reg Y X1 X2 X3' ?) Thanks!

[@CrunchEconometrix]

Hi Shuai, thanks for the encouraging words and feedback on my videos. Deeply appreciated! I showed the long-run equation in this video. You may need to watch it again. Thanks.

[@lysianebyukusenge9056]

Thank you so much for the video but I am having a question ,my final result of cointegration test,in the table where it is marked Sample,for me it is excluding some periods(years)compared to what I have .What should I do?

[@CrunchEconometrix]

Hi Lysiane, I can't really tell what the issue is. Perhaps, you have some missing observations?

[@AB_mdia2112]

it is very good lecture, can you attach the ppt and the do file to all of your videos ?

[@CrunchEconometrix]

Thanks, AB Medi, for your encouraging feedback. Deeply appreciated. PPT slides are not available. Due to abuse and unethical conduct, Stata dofiles and some datasets used in my videos are no longer free but available on my website upon payment. Here's the link cruncheconometrix.com/view/datashop.php

[@CrunchEconometrix]

KZbin 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, help me stay online.

[@NusratJahan-eb6ox]

Thank you for the video. In my data set I have combinations of I(0) and I(1). So how can i perform jct in this case?

[@CrunchEconometrix]

Watch my video on Bounds cointegration test.

[@enu6957]

Thank you for this, Mrs Adeleye. Please what is the specific model used in carrying out a Johansen Cointegration test

[@CrunchEconometrix]

Hi Enuma, thanks for watching my videos...deeply appreciated. What do you imply by "specific model"?

[@enu6957]

Like with the VAR method you gave an example of a VAR model with three variables. I was wondering you could please proffer one for the Johansen technique?

[@CrunchEconometrix]

@@enu6957 But you just watched this video on JCT. Or s there something I'm not understanding from your query?

[@arriannea2525]

Thank you for the helpful video. I hope you can help me with this question, Is it possible to report p-value for trace and max test statistic using Stata?

[@CrunchEconometrix]

Hi Arrianne, thanks for the encouraging feedback. Deeply appreciated! Yes, you can report the pvalues.

[@arriannea2525]

@@CrunchEconometrix Thank you for your reply. Do you have a video tutorial for that? I found some online resources but using E-views and not Stata.

[@CrunchEconometrix]

Arrianne, Stata does not report pvalues. Sorry for the mix-up. Make do with the stats provided. That's all you need.

[@arriannea2525]

@@CrunchEconometrix Thank you for the clarification. I really appreciate it.

[@thetruth4712]

Thank you for the video. BTW, what is LL and parms refer to and how we interprete them? Regards.

[@CrunchEconometrix]

Toba, I didn't interpret them because they are not so important. If still interested, kindly seek other online resources. Thanks.

[@yonibork9156]

You didn't mention how to decide between using constant or trend in the test, or how many lags to use. Why does the test default to constant and 2 lags? When would you select "trend" or "rtrend" or "rconstant" instead of just "constant"? Also, I am using yearly data and think I should only use 1 lag instead of 2. And what if the trace statistic says there is 1 cointegrating equation, but the eigen says 0? Question from Vermont, USA. Thank you.

[@CrunchEconometrix]

Hi Yoni, you will agree with me that not everything can be put on the video. The idea is to spur the student to also go out and fish for more information. I also mentioned that you have the prerogative to use either the Trace or Max-Eigen statistic after resorting to their underlying assumptions. Thanks for watching my videos, deeply appreciated.

[@raihansiddika8463]

@@CrunchEconometrix Hi, I really appreciate your endeavor. But I feel that the above question (as raised by Yoni Bork) need to be addressed. These are the most critical issue and direct answers are difficult to find. Otherwise, researchers might be misguided. God bless you.

[@CrunchEconometrix]

@@raihansiddika8463 It is also advisable for you to to seek other sources for clearer understanding. My earlier comments stand.

[@theonnshimirimana6956]

Thank you for your presentation. So, if i may ask! How to perform Johansen cointegration test with series integrated of different orders?

[@CrunchEconometrix]

Hi Theon, no you cannot use the JCT in that situation. The Bounds cointegration test is the most applicable. Kindly watch my videos on the procedure. May I know from where (location) you are reaching me? Thanks.

[@theonnshimirimana6956]

@@CrunchEconometrix Thanks ! I am studying in China !

[@CrunchEconometrix]

@@theonnshimirimana6956 Awesome, Theon! Kindly share the link to my KZbin Channel with your colleagues...thanks!

[@theonnshimirimana6956]

@@CrunchEconometrix You're welcome!

[@samsonmayaka7896]

I received a r (130) "too many literals" error when I tried to run a vecrank on my variables(4 variables) how can I rectify this error?

[@CrunchEconometrix]

Click on the error code on guide to solving the problem.

[@selinozdamar3759]

Hello ! Thank you for the video. I would like to ask you how can I add dummy variables into VAR and VECM models?

[@CrunchEconometrix]

Hi Selin, thanks for the positive feedback. Deeply appreciated! Not sure how dummy variables fit into VAR and VECM. You may need to check other online resources. Kind regards.

[@lbarberia]

Excellent videos! Very helpful! Can you comment on how this test could be undertaken with a structural break?

[@CrunchEconometrix]

Hi Lorena, thanks for the positive feedback. Deeply appreciated. In case of structural break(s), several tests can be applied: Zivot-Andrew, Gregory-Hansen etc. The fundamentals of these tests differ. So, you may have to read up to know which is applicable to your work. However, you can watch my videos on Chow Test and Gregory-Hansen. You'll find them very helpful. ...may I know from where (location) you are reaching me?

[@lbarberia]

@@CrunchEconometrix I found the videos on your KZbin page. Thanks! a reaching you from Brazil!

[@banelemavimbela3883]

Dear Prof I am using annual data for 50 years and my issue is that I am not very sure how I assign the optimal lag for the VAR using the AIC. My variables are stationary at first difference of the logs but at 10% critical value instead of the 5% critical value. I guess this are the problems that make my VAR model to have some omissions on some of the variables and Granger is not giving the Chi square values to show the direction. Please help.

[@CrunchEconometrix]

Hi Banele, did you watch my video on Optimal Lag Selection?

[@macmillanjere4473]

Very good presentation, so clear. . How do I access part 1 ?

[@CrunchEconometrix]

Hi MacMillan, thanks for the positive feedback. Deeply appreciated! Check through the Playlists for "Cointegration Series I(0)". Please may I know from where (location) you are reaching me?

[@macmillanjere4473]

From China

[@varunmiglani110]

How do we get pvalues on vec rank command

[@CrunchEconometrix]

You can use t-stats instead.

[@joachimvanoppen5212]

Hi, This video and many other Stata video's really helped me in my project. But I have a problem with my time series data that there are missing observations, is there a specific way I can deal with this problem?

[@CrunchEconometrix]

Hi Joachim, thanks for the encouraging words and feedback. Deeply appreciated! If you have too many missing observations, I will advise you change the variables to closer proxies.

[@shanikarathnayake1072]

Thanks a lot for the video. How can I get the Johenson cointegration vector?

[@CrunchEconometrix]

Hi Shanika, unless you imply something else, that's what I explained in this video.

[@shanikarathnayake1072]

Hi, I mean the cointegration vector like (1, -1). For that, I guess it is necessary to run the VECM. Unlike stata, in eviews, once we test the Johenson cointegration we can observe the cointe vector. Thanks

[@himadri6096]

I cannot find Part 3 of this series - when the data are mixed I(0) and I(1). Link please.

[@CrunchEconometrix]

Hi Himadri, kindly watch the videos on the Bounds test. Please may I know from where (location) you are reaching me?

[@himadri6096]

@@CrunchEconometrix Thanks! I am watching your well organized and excellently explained videos from USA Tennessee. :)

[@CrunchEconometrix]

@@himadri6096 Wow! Awesome! Thanks again for the encouraging words. Please stay safe 🙏

[@alexandretnb]

Thank you so much CrunchEconometrix! You saved my day! I can't find the part 3 of cointegration (series with differetn order of I(d)), can anyone hlpe me with the link?

[@CrunchEconometrix]

Thanks, Alexandre for the encouraging feedback...appreciated! Please browse my Channel for videos on BOUNDS COINTEGRATION.

[@eddymwangi5787]

Why is your log of variables different? For instance log 1800.5 of gdp is supposed to be 3.255 and not 7.963

[@CrunchEconometrix]

Eddy, I'm trying to understand your query. So, if you think the log transformation as derived from Stata is wrong, why not use your data?

[@eddymwangi5787]

@@CrunchEconometrix why the bile? not a critic but a query too, for instance, I picked just one value of gdp in your data, log 1800. 5 is 3.255, but in your data is 7.963. kindly please make me understand because I follow your videos

[@CrunchEconometrix]

@Eddy Mwangi Please take no offence. Honestly, I try to understand why some of you whip emotions on a harmless statement. I said if you think the GDP transformation is wrong, why not use your data?

[@eddymwangi5787]

@@CrunchEconometrix Great, now got it. Thanks fellow Statistician. Anyway, I can say, now I know how to use STATA because of your videos. Keep up the good work

[@asanteka.2403]

Good video MMe . Please Can you do a video on Hansen Panel threshold regression ?

[@CrunchEconometrix]

Hi Asante, unfortunately I have no idea how it is performed. You may search other YT videos.

[@asanteka.2403]

@@CrunchEconometrix Thank you very much Madame.

[@oyedemijeremiah2623]

You are a great teacher....please is it possible we connect...i need a mentor.

[@CrunchEconometrix]

Hi Jeremiah, thanks for the kind remarks. Deeply appreciated! Kindly connect with me via my CrunchEconometrix Page. Please may I know from where (location) you are reaching me?

[@oyedemijeremiah2623]

@@CrunchEconometrix i am from Lagos University teaching hospital.

[@CrunchEconometrix]

@@oyedemijeremiah2623 Awesome! I'll appreciate it if you can share the link to my KZbin Channel with your friends and academic community. They will find the content helpful too. Thanks 😊

[@akhliddinismailov2412]

I have a question: If I have some variables stationary while others are not stationary in my model but non stationary variable stationary is solved by taking the first difference. That is I(0) and I(1) Should I do cointegration test or I can't do it?

[@CrunchEconometrix]

In that case, you perform the Bounds cointegration test.

[@akhliddinismailov2412]

@@CrunchEconometrix Thanks a lot. I have also very important question concerning the data and suitability of ARDL model to use for my data. The fact is that I have quarterly data for 12 commercial banks 9 variables for each of these banks for 10 years. When I create a pool of data in one data set STATA considers them all as if they are the data for single bank. Also another problem with data is that the data for one commercial banks for 10 years with 9 variables end and then the next bank data starts and so on. Can we use ARDL model in this case considering that I have 2 dependent variables and 7 independent variables which can not be a dependent variable according to my research question. As I remember from your example, each variable can be dependent as well as independent exploratory variable. I thank you very much in advance. You are saving so much time and efforts for me

[@CrunchEconometrix]

You need to know the differences between a time series data and a panel data. From your explanation, your data structure is the latter. Watch my panel data videos for better understanding.

[@akhliddinismailov2412]

@@CrunchEconometrix Thank you . I will definitely watch them and return with questions)

[@akhliddinismailov2412]

Professor, I have a question which I couldn't solve. I have the quarterly data given as 01.07.09; 01.10.09; 01.01.10 so on. meaning that they are given for the first date of the next quarter. For example ROA for 01.07.09 is the ROA for the second quarter of the year of 2009. How can I convert them into 2009q2, 2009q3, 2009q4 etc in stata so that stata would accept each of these observations as quareterly data. When I show them as quarterly data in stata it shows them as 6479q4, 6502q4, 6525q4 so on for the data which normally belongs to 2009q2, 2009q3, 2009q4 so on. For some banks the same for example 6502q4 was given twice for two different observations which normally belong to two different quarters. Sorry for such a long explanation. I hope it is understandable

[@dipenmodi1807]

Please can you tell me how do we integrate of order 1 or 2 in Stata?

[@CrunchEconometrix]

Hi Dipen, by taking the first and second difference of a variable. For instance to generate these for "inflation", the Stata syntax is: gen dinfl=d.infl gen d2infl=d2.infl

[@dipenmodi1807]

@@CrunchEconometrix I see. Thank you! Works perfect.👍

[@CrunchEconometrix]

@@dipenmodi1807 I don't know what you mean by "find 50 years data that satisfies OLS". Just apply all the rules (stationarity, lag selection) guiding time series analysis and estimate your model. That's all.

[@MrStaron47]

can I ask you a question, if my johansen cointegration test gave me a cointegration results, can I use ECM instead of VECM?

[@CrunchEconometrix]

Hi Fry, know that JCT is to VECM as Bounds test is to ECM.

[@MuhammadAhmad-ei1ib]

i have a question is it necessary that for Johansen Cointegration Test all variables are integrated of same order at level/raw form/log form and must not have differenced ? Kindly guide me in this issue thanks

[@CrunchEconometrix]

Muhammad Ahmad Yes, all variables must be stationary at 1st difference b4 performing JCT. But when performing the test, u can use either the raw or log forms but not the 1st difference. Just the way I performed it. You can watch my video on how to perform the test.

[@MuhammadAhmad-ei1ib]

thanks ma'am

[@sabreenkhan3498]

I cant get the part one plz share the link

[@CrunchEconometrix]

Sabreen, I mentioned that at the beginning of this clip. Kindly search for it within my Channel. Thanks.

[@akhliddinismailov3766]

What about a test of cointegration between individual pairs of the model?

[@CrunchEconometrix]

Johansen test is on cointegration among the variables in the model.

[@seanh19954]

so if we have lags of 2, how do we create our ARDL model, if we get I(0) only after using 1st differences? do we create the ardl model using the 1st differences with max lag of 2?

[@CrunchEconometrix]

Sean, you are mixing this up. For one, this video is about JCT which is not correlated with ARDL modeling. Secondly, you don't obtain I(0)after taking 1st difference. Thirdly, I've simplified ARDL techniques, watch them. Suggestions: video tutorials are no substitutes for reading, you MUST read about the fundamentals from basic econometric textbooks to get the full import of what you are doing.

[@seanh19954]

@@CrunchEconometrix my apologies, i realised i confused the 2 together. Thanks for your help! Your channel is great!

[@CrunchEconometrix]

@@seanh19954 No worries Sean. Glad to hear that you find my videos useful. Gracias 🙏

[@arturox1997]

if i get a cointegration in levels but not on 1st differences should i do a vecm? (i only got 2 lagss)

[@CrunchEconometrix]

Hi Arturos, what do you imply by "cointegration in levels"?

[@arturox1997]

@@CrunchEconometrix umm the variables like they are ummm like years of education and pib per capital(this is just and example , not the variables I'm using), when I do the test with the variables in differences I don't get cointegration anymore, I'm doing my econometrics homework at my University in Peru

[@CrunchEconometrix]

@@arturox1997 Your query is not understandable. The guidelines to performing JCT are quite explicit. Watch the videos on cointegration AGAIN, jot down notes and follow my guide.