Cointegration - an introduction

[@taniaobono978]

I normally dont comment on videos, however this was very clear and helpful!! Thank you very much

[@mikeysz1972]

i usually do not comment on comments on videos, however i agree!!

[@TheGodSaw]

Dude you make the absolute best Econometrics videos it is insane. Last year we had an extremely hard theory of multiple Regression course and your graduate playlist helped a TON. I recommended your videos to all my friends. Keep up the good work!

[@chariezwane3981]

Thank you! This topic made no sense until I gave this a try.

[@jakobforslin6301]

Best teacher out there, thank you for all the clarity you bring

[@richardwatson3484]

Great explanation - for a newcomer to econometrics this is is gold

[@SpartacanUsuals]

Hi Prathana, If a variable has no unit roots it is always 'cointegrated' in a sense with other I(0) variables. Hope that helps! Ben

[@sebastiankuhnert3639]

Great video!!! - thank you!! Made reading some articles a lot easyier. I read "Some Properties of TIME SERIES DATA..." by Granger (1981), where he defines: X_t = I(d) : X_t = a(B)e_t, where (e_t) ~ WN(0, sigma^2), B is the Lagoperator, a(B) = (1-B)^{-d}*a'(B), where a'(B) has no poles and roots in z=0. I don't understand the concept of the introduced "linear filter" a(B). Is it just a linear function?

[@superstarem]

totally awesome thankyou. im looking at options for my doctorate to test for causality between FDI, Exports and GDP so cointegration and causality models are my jam lately but this has been very useful.

[@홍성의-i2y]

Usually, it is banned for us to do regression when both Y and X are I(1), because it causes spurious regression. However, co-integration is devised so that we can discern such cases when regression is allowed. In cases they are both I(1), Y_t - \beta * X_t being I(0) means that they share the same pattern (up to constant multiplication) like the downward-dent case in 4:55. Then it is reasonable to think that Y_t and X_t share some sort of correlation, and thereby justifies the use of regression.

[@nackyding]

Goddman! Thank you. Thank you, thank you, thank you! Your series has been god send for me. Thank you again!

[@louismarcelmpundu8576]

how can I say thank you for this helpful video? Thank you to make plain as a day what cointegration really mean in simple words!

[@carlsousa]

Great explanation, I always found econometrics hard to understand and you make it super simple, maybe I always had bad econometrics professors. Thanks a lot.

[@johnsteedman7937]

I have the book but still found this useful. An extra column for the denominator might make things crystal clear even though I can see that the book does explain

[@ciaranbarrett5254]

Best explanation I have ever heard!

[@shihabuddintareq5151]

A simple but significant explanation

[@bang_goo]

Very simple and clear. It helps me a lot. Thank you so much!

[@subarkahsubarkah969]

You are the best, Ben!! I learn a lot from you. Thanks.

[@a.moizmaner2504]

Years later still benefiting GBU!

[@pelephantzoo]

Your videos are awesome! Keep it up! You're helping a lot of people :)

[@xphilster]

Your videos are still so useful, thank you Ben!

[@katerinamilaberska]

Perfect video, now I understand what a cointegration is! :)

[@MuhammadAsim-fy1qy]

Very very clear I must appreciate sir. Thank you so much

[@TheDominock]

Thank you very much, you are glorious! Could you please provide me with a title of a journal article/name of the authors where authors explain the case of using non-stationary variables of order 1 being regressed on each other? I am having difficulties in finding such a journal.

[@bartas8891]

Reall good explanations. Thank you for sharing your knowledge !

[@ssrouji4507]

Thank you Ben, excellent !

[@VainCape]

another way to put it: there exists a linear combination of yt, xt that is I(0)

[@JugaadTech]

Great explanation, Just could not understand I(0) or I(1) part, If someone can point me in right direction for this, that'll be great

[@fatimazahramoussaid350]

hello, does beta can be interpreted as the speed of ajustement? is it what we called the ECT( eroor correction term)?

[@abioduntaiwo8443]

Hi ben, please what is the weakness of the ARDL method of co-integration.

[@BansheeX]

Thank u Ben Lambert

[@theochhn7514]

GREAT explanation! It is very clear!

[@wenchaowu6204]

Great video. Very intuitive.

[@pranavkishorbaviskar5688]

Crisp and clear thanks sir

[@daanw]

Bitcoin stock-to-flow and price?

[@modiallo968]

Yessir 😎😎😎

[@eavjones]

I'm actually in the life sciences, not economics, but I analyse data from time-lapse experiments. I am looking at a relationship between an X_t and a Y_t in my time series. Do you think I could apply co-integration/Dickey-Fuller to this? I actually have 3 different time-lapse experiments (with about 10 times points per experiment). Can I just analyse all the data together? By the way, your movies are amazing. You make difficult statistical knowledge very accessible.

[@meisterthea]

If beta is a scalar value then surely it would just raise or lower X(t). Why would it create a constant spread with Y(t)?

[@itthipong]

Dear Mr.LamBert. Suppose that I have more than one independent variable say x1 and x2. What if I find that y and x1 are I(1) but x2 is I(0)? Can they be cointegrated despite their different integration orders? Am I allowed to estimate the ECM model between y x1 and x2? Could you please explain to me? Regards.

[@SpartacanUsuals]

Hi, good question. Yes, in theory there is no problem here, so long as y and x1 are cointegrated (in the presence of x2). However, I would be very careful about doing this sort of regression for fear of it demonstrating a spurious relationship between variables. Best, Ben

[@lory198]

hi Ben, your videos are really great! Just one question concerning the video: if there exists a b such that y_t-bx_t is stationnary. Why don't we say that there exists a and b such that ay_t-bx_t is stationnary (or said differently: why can we always assume that a=1?)

[@malikaallali2924]

Great video thank you so so much I need some information about cointegration thanks a lot that's great

[@sara5555555555]

Awsome vid! Just have one question regarding I(1). I get that it says that if you differentiate it once then it becomes stationary, am I right in assuming that I(1) in the vid is still "undifferentiated" and non stationary still? If they were both differentiated once then both would be stationary and we wouldn't have a problem, would we? Or am I wrong?

[@lrozenwater]

Yes, they show the levels of y_t and x_t, not the first-differenced variables

[@josefinaramos6534]

what happens if 2 series do not look like they cointegrate but when looked at in first differences you can see they do'?

[@notonlygeek]

Hi, trying to do french subtitles, at 1:44 he say " witch I(1) .... another" I don't succeed find missing word nor understand the meaning. Many tks for help.

[@muhammadirfanislami818]

Perhaps its nonstationer at level

[@dhruvkotecha8843]

Very helpful, thank you!

[@fatimazahramoussaid350]

in the first example where the two series aren't cointegrated because of the two random walks, can we interpret the random walk as breakpoints ??

[@fatimazahramoussaid350]

and the correct them by adding dummy variables to the model ?

[@fatherle]

the best expaination

[@fazoo1000]

Very Much help ful Video

[@joebedford4157]

Thanks Ben.

[@PrarthanaRaviKumar]

Thank you for these videos :) :) I was wondering, if a variable has no unit roots, does it mean it cant be cointegrated with any variable??

[@aishiaratrika]

If it doesn’t have unit root, this indicates that the time series variable is stationary. So it can't be cointegrated since conintegration involves two non-stationary processes.

[@josuecosta894]

god bless you!! you helped me a lot, thanks!!

[@chaozhang5864]

very helpful! thx Ben.....

[@EduardoGarcia-if2kv]

I am hooked!!!

[@yangliu5652]

Very useful! Thank a lot!

[@ivankissiov]

Thank you!!!

[@chrislam1341]

but what is the meaning of I(1) and I(0), i didnt really catch it..

[@SpartacanUsuals]

Hi Chris, I(1) means that you need to take the 1st difference of a series in order to make it stationary. I(0) means that the series is already stationary. Hope that helps! Ben

[@hassanaber392]

very helpful..thanx

[@KeddingtonKKB_Official]

What's I(1)?

[@Anna-zi6fy]

Hi Ben! Thanks for the video! What is I(1) here?

[@radufffp]

The date contains one (1) unit root. This means the data, in order to be stationary, has to be differentiated one (1) time.

[@sara5555555555]

But then we are supposed to assume that the series have been differetentiated yet? Otherwise the data would be stationary right?

[@flamingflamingo4021]

What's I(1) / I(0) here?

[@shaguftashabbar4982]

very helpful!

[@subhransusekhar289]

Why we call the non stationary series a I(1)?

[@mohdbahakim]

I(1) means they have a unit root meaning they are not stationary

[@mohdmaudehero7602]

what does the I(1) or I(0) in this video denote thank you

[@mohdmaudehero7602]

i get it

[@ben73010]

Thanks for this

[@zoozolplexOne]

cool !!!

[@ahmedtrabelsi3589]

i love you

[@sport8133]

velly velly noice!

[@thaitazzz]

what is formula of beta ?

[@noueruz-zaman7894]

I have these for my masters in finance program and in class I don't get anything, it seems like I am Chinese. .lol

[@홍성의-i2y]

en.wikipedia.org/wiki/Order_of_integration definition of I(d)

[@Sydney_Anuyah]

I normally dont comment on videos, however this was very clear and helpful!! Thank you very much