Q: thx for representation,my question:if the johansons test reject both and 1 cointegration but mention that:Trace test indicates 2 cointegrating eqn(s) at the 0.05 level still could we go for VECM modeling?
thanks for the excellent explanation, please I have a question if all our variables are stationary at l(1), just one variable is stationary at level as well as stationary at l (1) can I do Pedroni cointegration test for this regression?
First, if a variable is I(0), meaning stationary at level, you should NOT difference it further. Second, if others variables are I(1), then you specify and run ARDL bounds test. Please check out my 6-part videos on ARDL BOUNDS TEST FOR COINTEGRATION. Hope this helps.
I'm not sure. You could pose the question in an online econometrics forum. But as you know, to ensure that the property of CONSISTENCY is satisfied in the estimation of any regression parameter, your sample size must be large enough, more than 30 observations. How large? The larger the better.
i have another question, i have 4 variables, the 3 are I(1) and the 4th I(2). Can i use the fisrt diferrences pf the I(2), so as all to be I(1) and then check for cointegration and specify my VEC model? thanks
I don't believe you can do a VEC or ARDL, not if one of the series is I(2). I think manipulating the I(2) series is questionable unless there's theoretical support for it as an ORIGINAL series when I(1). My 2 cents :-)
thanks for the explanation. please I have a question: if we have more than 2 variables and only one of the coefficients significant, how will we interpret this? there is a long run equilibrium only between 2 variables? Also, what happens if we have more cointegration equations the signs of which do not match each other?
Thanks for this video. However, I need to be clarified in one aspect. In the normalized co-integration result, you said to get the t-statistics, you divide the coefficient by the standard error. Pls, using the t-statistics, what do I do to get the p-values for the normalized co-integration result? I guess all am asking is how to get the P-values of the normalized co-integration result. thanks
Thank you for a great vid. I have a question regarding the Johansen test, you said that you need to check that I(d) are the same for the variables. Say that all are I(1), is that enough to start the test or do I actually need to differentiate the variables one time Before I do the test?
Yes, you can perform the test if all are I(1). Note that differencing (caution! not "differentiating") the series and re-doing the stationarity test serves to identify the order of integration. On a different note, please don't forget to LIKE any of my videos that you find helpful and also, feel free to subscribe. Thanks.
hello i just wanna ask what if i exchange the variables? like i used oil instead of fx? it gave me a very different results.
Q: thx for representation,my question:if the johansons test reject both and 1 cointegration but mention that:Trace test indicates 2 cointegrating eqn(s) at the 0.05 level
still could we go for VECM modeling?
I'm not really sure.
thanks for the excellent explanation, please I have a question if all our variables are stationary at l(1), just one variable is stationary at level as well as stationary at l (1) can I do Pedroni cointegration test for this regression?
First, if a variable is I(0), meaning stationary at level, you should NOT difference it further. Second, if others variables are I(1), then you specify and run ARDL bounds test. Please check out my 6-part videos on ARDL BOUNDS TEST FOR COINTEGRATION. Hope this helps.
@@PatObi thank you
thank you for the explanation. please I have a question: what is the minimum size of the sample to make the cointegration test?
I'm not sure. You could pose the question in an online econometrics forum. But as you know, to ensure that the property of CONSISTENCY is satisfied in the estimation of any regression parameter, your sample size must be large enough, more than 30 observations. How large? The larger the better.
i have another question, i have 4 variables, the 3 are I(1) and the 4th I(2). Can i use the fisrt diferrences pf the I(2), so as all to be I(1) and then check for cointegration and specify my VEC model? thanks
I don't believe you can do a VEC or ARDL, not if one of the series is I(2). I think manipulating the I(2) series is questionable unless there's theoretical support for it as an ORIGINAL series when I(1). My 2 cents :-)
Hi, Could you pls help me learning bayer & hanck cointegration in stata or eviews. thanx
thanks for the explanation. please I have a question: if we have more than 2 variables and only one of the coefficients significant, how will we interpret this? there is a long run equilibrium only between 2 variables? Also, what happens if we have more cointegration equations the signs of which do not match each other?
Thanks for this video. However, I need to be clarified in one aspect. In the normalized co-integration result, you said to get the t-statistics, you divide the coefficient by the standard error. Pls, using the t-statistics, what do I do to get the p-values for the normalized co-integration result? I guess all am asking is how to get the P-values of the normalized co-integration result. thanks
P-values are not given. But with large samples, if your t > 2, you can assume it to be significant at the 5% level.
Thanks for your comprehensive lecture. If the number of co-integrating equation is more than one, can I still use VECM approach to analyse my data?
Not sure. But some studies do it. Please pose your Q at an online econometric forum.
Thank you for a great vid. I have a question regarding the Johansen test, you said that you need to check that I(d) are the same for the variables. Say that all are I(1), is that enough to start the test or do I actually need to differentiate the variables one time Before I do the test?
Yes, you can perform the test if all are I(1). Note that differencing (caution! not "differentiating") the series and re-doing the stationarity test serves to identify the order of integration. On a different note, please don't forget to LIKE any of my videos that you find helpful and also, feel free to subscribe. Thanks.
Hi, may I ask if my C(1) is insignificant, how can I adjust it? Thanks :)
Excelente presentación!
You sound like a rockstar
Great