Thank you so much for this nice and professional tutorial. If i am to ask; is it possible to use the generated residual to interpret the convergence of series ?
@ChekwubeMadichie3 жыл бұрын
Yes, the one-period lag of the residual appearing in the ECM model is expected to be negative and significant for convergence to be assured.
@saifsaidalaviuae4 жыл бұрын
THANK YOU
@ChekwubeMadichie4 жыл бұрын
Welcome sir
@prince-uba3 жыл бұрын
If the order of integration of variables where X is I(o), Y is I(1) & Z is I(2), what is the Appropriate Model for estimating the relationship among the variables.
@ChekwubeMadichie3 жыл бұрын
Use the TY model like I have explained over the phone call.
@muhammadfaizan37222 жыл бұрын
What test i need to perform after VECM?
@vishalvarsani71003 жыл бұрын
If all your variables are i1 variables shouldn’t you use the differences when making your equation ?? When finding the et?
@ChekwubeMadichie3 жыл бұрын
If all your variables are I(1), and they are cointegrated, then you can estimate your model in their level (to serve as long-run) and apply ECM. If they are not cointegrated, then you can use the differences but can only refer to the shortrun.
@kieranchard67533 жыл бұрын
thanks for the video! what do I do if my Durbin Watson from the regression is very close to 0?
@ChekwubeMadichie3 жыл бұрын
When your DW is very close to 0, it shows that there exists first-order autocorrelation. The remedy is either to add a period lag of the dependent variable or impose the ar(1) in your model and re-estimate it.
@abir39563 жыл бұрын
I should use FMOLS methode but unit root test show that my variables are not all in same order what can i doo please !!
@ChekwubeMadichie3 жыл бұрын
You should consider using the ARDL model if your variables are a combination of I(0) and I(1) with an I(1) dependent variable. But if the dependent variable is I(0), you may have to use the Bootstrap ARDL model.
@abakarannour51953 жыл бұрын
Salut J'ai une question sur le modèle vecm au niveau de l'interprétation CointeQ1 si on a négatives et positif.
@usmansaleem12533 жыл бұрын
Sir please make clear what r conditions to estimate Vecm And when we estimate Ecm I feel for Johnson's we estimate Vecm not Ecm And in ARDL we estimate Ecm Please clear? Am I right
@ChekwubeMadichie3 жыл бұрын
Johansen cointegration is for system of equations and it goes with VECM, while Engle-Granger residual-based cointegration is for a single equation and it goes with ECM.
@samfisher12502 жыл бұрын
hello i just read somewhere that the ect should be significant and negative. why does the ect is not significant in your example? and what are the solutions i need to do to make it significant?
@ashveenaashveena5702 жыл бұрын
you can take lags of different variables in the estimation window, it may solve the problem.
@user-tk8dp4bc2q4 жыл бұрын
Please can you answer... My all result signifigant but ecm(-1) is not negative.. In these case there is long term or not?
@ChekwubeMadichie4 жыл бұрын
Dear Athraa, kindly note that a positive ECM(-1) doesn't speak well of your model. It shows evidence of divergence from the longrun equilibrium. This may be as a result of specification problems or data issues or error during estimation. Kindly ensure that there is no variable that is originally I(2).
@osmanguldur25344 жыл бұрын
ECM(-1) is not significant so we cant trust this test? I mean we cant be interpreted this results?
@ChekwubeMadichie4 жыл бұрын
The ecm(-1) should be significant to ensure there is a strong feedback effect of short run deviation in the longrun. No meaningful interpretation can be generated about the convergence process and the speed of adjustment, when the ecm(-1) is not significant. However, the video is meant to provide the useful steps on how to perform the test in Eviews.
@ikwujesongeorge49043 жыл бұрын
increase the volume and graphics of your videos
@egbearafat72723 жыл бұрын
His sound and audio is quite good
@adrenalinerush3694 жыл бұрын
What if the residuals of the OLS only becomes stationary after first difference?
@ChekwubeMadichie4 жыл бұрын
If the residuals turns out to be stationary at first difference, that is I(1), then there is no cointegration. According to the EG residual-based cointegration approach, cointegration exists if the linear combination of I(1) variables is stationary or I(0). In other words, if variables X and Y are both I(1), then they are cointegrated if their residuals (i.e their linear combination) is I(0).
@Wendy-kr7dr4 жыл бұрын
Hello, may I ask if I have four variables, and my result shows that there are two asterisk sign at "none" and "at most 3", how can I explain it and can I still use VECM model. Thank you :)
@ChekwubeMadichie4 жыл бұрын
In your own case, there is only one cointegrating equation. Yeah you can run the VECM if your main dependent variable is the one in the identified cointegrating equation.
@Wendy-kr7dr4 жыл бұрын
@@ChekwubeMadichie Thank you so much! Could I ask where can I check my dependent variable is the one in the identified cointegrating equation?
@ChekwubeMadichie4 жыл бұрын
If you have four variables, that means four equations in the Johansen system specifications. Thus, the dependent variable in the first equation is obviously the main dependent variable and if there's only one cointegrating equation, then the main dependent variable is cointegrated with other variables.
@Wendy-kr7dr4 жыл бұрын
@@ChekwubeMadichie Thank you so much for your reply:) I'm wondering why my model just has one cointegrating equation instead of two. As I cannot reject the "at most 1" and "at most 2"
@ChekwubeMadichie4 жыл бұрын
What's exactly your interest in having two cointegrating equations?
@alphawhiskey39703 жыл бұрын
What is the software you are using?
@ChekwubeMadichie3 жыл бұрын
Eviews 10
@greatbus3 жыл бұрын
Can panel data do the ECM?
@ChekwubeMadichie3 жыл бұрын
Yes of course. Search for my video on panel unit root test and cointegration. That would give you an insight. I will post a video on panel ARDL and ECM.
@aniksaha99252 жыл бұрын
Hi, is it possible to analyse non performing loan (dependent) GDP and inflation rate (independent) by unit root test, panel cointegration model, VECM test using 9yrs data?
@ChekwubeMadichie2 жыл бұрын
No
@aniksaha99252 жыл бұрын
What can be done to troubleshoot the problem? What do you recommend? If i increase years of data, would that help?
@ChekwubeMadichie2 жыл бұрын
@@aniksaha9925 9 yrs data coverage is too small for such analysis. You should have at least 30 yrs.
@aniksaha99252 жыл бұрын
That's why Johansen panel cointegration didn't work. When i stretched to 13 years, it produced some result though
@ChekwubeMadichie2 жыл бұрын
@@aniksaha9925 That result is questionable
@sakhawatemon47762 жыл бұрын
I wish I could have the patience to deal with cockny accent in my dire times.