Building a Vector Error Correction Model in R

[@marianamonteiro2210]

The cointegration test suggests the cointegration rank equals 2. Here is why: The rule should be: if H0: r=0 (let's call it step (1)) is not rejected, then the cointegration (r) = 0. If in step (2) when H0: r

[@kvafsu225]

Gives a very clear explanation on how to carry out a VECM. Thank you very much. I wish you had shared the codes too.

[@mohammadirsad9285]

Sir, Thank you very much for this precious video. It clears many doubt of mine.

[@miercolo]

Thank you for the video. Could you explain how you should writte the VECM equation in terms of the significant variables and how you should distinguish short term and long term relations.

[@arnaudpradier4954]

Hello, many thanks for your video ! I was wondering how to get the VECM's R squared ?

[@hogrideeeeer]

How to access p values and r^2 though? The results only show coefficient and that's it...

[@antiilankedileli]

Thanks for the video, it is really educative. Yet, for the interpretation of the coefficients in the cointegrating equation, dont we need to reverse the signs of them ?

[@fritzmuller8761]

Is it not a problem that the ECT coefficient of the significant CPI (CPI --> EcT) is positive? 0.0079? It mesans that the curves do not approach but distance from each other

[@andreadebortoli6069]

Guys does anyone know how to change the color for the plot of the FEVD? Alternatively does anyone know how to plot the FEVD for only one of the variables?

[@getinenglish3472]

Great job explaining VECM. I've been trying to apply the structural breaks and the granger causality using what you called ''Model1VAR'' as the object but I keep getting an error. Would know why? Thanks and we’ll done again! Andre

[@mohammadsaifuddin1009]

Thanks for everything. Professor, what should I do if my residuals are not "Normally distributed"?

[@62294838]

The ECT should be negative and between Zero and 1. So even if it’s a significant it still shows mispecificafion.

[@subhankarsanyal8138]

what does the vertical ais of IRF plot interpreted as? ..? %

[@geetanjali3436]

Sir I have run VECM residuals diagnostic but my model found non normal and hetroskedastic residuals but it solution for it I already taking my variable as natural log form. What can I do for this problems Pls rpy

[@rafaeldemoraislima9080]

Thanks for the video. I'm not an economist so I've been struggling with the choice of the lags. Could you explain why did you choose the lag.max=7?

[@rizka_khr]

I also have same problem

[@joed3325]

Super useful practical guide. Thanks,

[@palashsrivastava7711]

Hi Justin. Thank you for the video.i had a query. If I use a dataframe of 7 variables to find a cointegration between them and the hypothesis for r

[@JustinEloriaga]

Hi, thanks for your comment! There are at least 5 cointegrating relationships in the data frame.

[@MrOitube]

Really good video! Congrats on the great content!

[@estaykylyshbek8347]

Hi, great videos but you may have made a mistake. The p-value of serial.test is smaller than 0.05 thus Null Hypothesis of no autocorrelation is rejected. GDP has autocorrelation since in the summary 3 out of 4 lags of GDP are significant. Please take a look at it and reply if I am wrong.

[@patrickonodje1428]

The background you chose is for your R makes it difficult to see your codes even in high resolution

[@thetruthsreality]

In the VEC model, the signs in the cointegration equation is negative (-3.327..) showing a negative relation to M1, while in the long run ECT results it is positive (+0.0079). Is it contradictory, and how can this be interpreted!

[@hogrideeeeer]

Did u find answer?

[@nickomanu]

Thank you very much for your work! If I found out serial correlation and heteroscedasticity in my model how can I control it or solve it?

[@sudhirpatil3922]

just to add that gdp has annual seasonal effect , that's the reason gdp-4 was super significant ..

[@jennykim3428]

How would you interpret a model with 3 ETC terms?

[@fvc1612]

One question, didn't the Johansen test show that there were 2 cointegrating relationships?

[@alexandrodisla6285]

If the serial test fails, the model isn’t good. A good model must at least succeed the serial test, arch test and even the cusum test. It’s possible for the residuals to not be normalized. Nevertheless they must be whitenoise with no arch effect.

[@Zolololol]

Great video, thanks!

[@mohammadirsad9285]

Sir, I am kindly requesting to you make a video on NARDL Model.

[@danielloza4710]

Does anyone know if to simulate a vec model, the variables must be stationary?

[@ThebigPYRO]

No, they just have to be integrated of the same order.

[@adilmasoodkhan4527]

Dear Sir, Thanks for your help. Also, could you plez make your R Studio console White in color so that the codes could be visible easily; although the blue color is soothing to the eyes. Thanks.

[@santamedinas8600]

Muy claro!

[@tjhnnnmbg]

Thank you for this interesting video. Could you make video in R for SSE. If you can pls make with economic examples

[@raulq.3519]

Justin, why did you finally picked 1 cointegrating relation instead of 2 since test value is significant when r=2 (7.89)

[@JustinEloriaga]

Hello! Thank you for your comment. I was using the 95 percent benchmark instead of the softer 90 percent one, for consistency across most literature. But you may also conclude 2 cointegrating relationships at the 90 percent bound.

[@adilmasoodkhan4527]

@@JustinEloriaga Sir, what would be the interpretation of getting 2 cointegrating vector then?

[@patrickbormann8103]

I'm wondering also, because with his argumentation he should have picked r = 0 because 87.77 is higher than all three critical values. What we want is to reject the null hypothesis, that there is no cointegrated relationship. After we passed this test, we checked for r

[@adilmasoodkhan4527]

Bez. in case of 2 cointegrating vectors we would be getting 2 ECT terms, then how will we interpret it. Thanks

[@greatbus]

@@JustinEloriaga based on your previous video (Johansen Cointegration Test in R), using the 95 percent benchmark you still conclude 2 cointegrating isn't it?

[@alexandrodisla6285]

The test suggest at least 2 cointegration relationship. Yet you did choose r equal 1. That doesn’t make any sense.

[@belkhir789]

Yes i think he miske, he must chose 2 cointegration

[@marianamonteiro2210]

Yes. The rule should be: if H0: r=0 (let's call it step (1)) is not rejected, then the cointegration (r) = 0. If in step (2) when H0: r

[@000Requiem]

No, that should be a requirement for a VAR model, not for a VEC.