If VAR suggest 2 or 1, which means that I should have K = 1 or 0 for the ca.jo test. But ca.jo test requires K= at least 2. What does this mean if I get a result like this? What should I do?
Sir there is contradict result between r=0 and r=1, one accept null on the other hand one reject the null then on which basis we conclude that there is cointegration between variable
Sorry! I run an ADF test and could not reject the null hypotheses. But there was no evidence of a unit root in first difference, therefore I rejected the null hypotheses.
what do you think about using the ndiffs function on both time series to make sure not only that each serie is not stationary, but also that the two series have the same order of non stationarity ?
I get the following error while doing cointegration test: Warning message: In chol.default(SKK, pivot = TRUE) : the matrix is either rank-deficient or indefinite do anybody knows how to solve the problem
Thank you! I have a question : I have tested 10 variables, there the first one is the one I want to regress later. The Johansen test tells me that there are 2 cointegrated variables. How do I find out which variables these are?
Hi Katherin, prehaps this reply is a bit too late and you probably figured it out by yourself. Anyway, in my understanding the Johansen test gives you 2 cointegrating relations and not 2 cointegrated variables. It tells you that among the variables you have, there exsit 2 cointegrating relationship. To see those, you can type summary(ca.jo(...)) or as the example given here summary(cointest), and if you do what to see the coefficients, you can type cointest@V . Hope it helps you or anyone else seeing this.
To convert non stationy series in stationary, we use diff(), as I have read thatv the opposite function is cumsum(). When applying cumsum() to my stationary data, data is no stationary anymore. Does this make sense?
Hi, I think that no make sense, because when the series are not stacionaries they cant be cointegrated, it is a rule. It´s because cointegrated series share a common trend, stacionaries seires doesn't have a trend, so you cant use a Johansen model. In that case its better to use a simple VAR.
@@dumarvargas1467 Hi , your comment helped me to make a decision about choosing Johansen Cointegration or simple VAR. Thank you for sharing this knowledge in 3 years before, but it's helping me now. Can you tell me please how I can check whether my data is stationary or non-stationary data ?
I'm watching you from Turkey, and thanks for your help. Your explanation is really clear.
Great video.. thank you very much for explaining how to test co integration.
Thanks. Quick comment on video: zooming in an out is distracting; better to leave the R screen fixed in the video.
Thank you very much, your videos are very helpful!
Most welcome!
If VAR suggest 2 or 1, which means that I should have K = 1 or 0 for the ca.jo test. But ca.jo test requires K= at least 2. What does this mean if I get a result like this? What should I do?
What if it says that there are insufficient degrees of freedom?
Sir there is contradict result between r=0 and r=1, one accept null on the other hand one reject the null then on which basis we conclude that there is cointegration between variable
Thank You soooo much!!! Your videos saved me :)
Thanks a lot...
Sorry! I run an ADF test and could not reject the null hypotheses. But there was no evidence of a unit root in first difference, therefore I rejected the null hypotheses.
Since your series are I(1), you can run Johansen Cointegration.
Thank you
what do you think about using the ndiffs function on both time series to make sure not only that each serie is not stationary, but also that the two series have the same order of non stationarity ?
Thanks a lot!
is there a way to test with 1 lag? in R, you have to use minimum 2 lags, but I want to use 1 lag
can someone answer this? cause i have to try it in lag 1. my optimal lag is 1
@@rizka_khr Same
hi, thank you for video
if i tested model and for r=0 i can not reject the null hypothesis but for r
thanks friend
Thank you for the video, but pliz why did you take the number of lag.max =10 ???
you can take any as per your assumption
Sir please make me clear this concept. Hope soon
I get the following error while doing cointegration test:
Warning message:
In chol.default(SKK, pivot = TRUE) :
the matrix is either rank-deficient or indefinite
do anybody knows how to solve the problem
Thank you! I have a question : I have tested 10 variables, there the first one is the one I want to regress later. The Johansen test tells me that there are 2 cointegrated variables. How do I find out which variables these are?
Hi Katherin, prehaps this reply is a bit too late and you probably figured it out by yourself. Anyway, in my understanding the Johansen test gives you 2 cointegrating relations and not 2 cointegrated variables. It tells you that among the variables you have, there exsit 2 cointegrating relationship. To see those, you can type summary(ca.jo(...)) or as the example given here summary(cointest), and if you do what to see the coefficients, you can type cointest@V . Hope it helps you or anyone else seeing this.
What should I do if my data is stationary? How can I convert it to non-stationary?
Thanks a lot for you videos :)
To convert non stationy series in stationary, we use diff(), as I have read thatv the opposite function is cumsum(). When applying cumsum() to my stationary data, data is no stationary anymore. Does this make sense?
Hi, I think that no make sense, because when the series are not stacionaries they cant be cointegrated, it is a rule. It´s because cointegrated series share a common trend, stacionaries seires doesn't have a trend, so you cant use a Johansen model. In that case its better to use a simple VAR.
@@dumarvargas1467 Hi , your comment helped me to make a decision about choosing Johansen Cointegration or simple VAR. Thank you for sharing this knowledge in 3 years before, but it's helping me now. Can you tell me please how I can check whether my data is stationary or non-stationary data ?
Why don’t you upload the bloody script?