That's fine sir, but i read from damodar gujarati that if trend is present then we have to regress it against trend variable for removing trend. and i did the same thing and solved the problem. Thank you so much for your reply.
Sometime it is difficult to make the variable stationary. In that case, I would advise, either convert variables into log or increase the data size and try. In future I have plan to upload video on ARMA. but normally we insert AR(1) to capture serial correlation problem
Thank you. I would like to invite you to join Hossain Academy Facebook Group at below link and join our group discussion. Thank you. Sayed Hossain from Hossain Academy. facebook.com/groups/hossainacademy/
There are three ADF equations to test whether a particular time series data is stationary or not. All these three must give the same decision to come to a decision whether a particular time series is stationary or not.
Dear Sir, I saw your videos for unit root test but I'm not getting the video which shows how to detrend the series if trend is present. Please revert back as soon as possible.
When I did "level" and "intercept" unit root was present but the coefficient was positive. For every other combination the coefficient was negative. Does it matter that we couldn't accept the result for level and intercept, as the P and T values showed it to have unit roots anyway?
Sir, In Eviews manual, there is no metion of obsoulte value The manual says " Notice here that the t statistic value is greater than the critical values so that we do not reject the null at conventional test sizes. " the t value shown in manual is -1.41 which is mentioned as GREATER than at 1% level -3.45 Are you correct or is the manual correct? Pl clarify
Sir, In one of my results on GDP, I am trying to find out stationarity. ADF Statistics = 5.4255 and critical values at 1%, 5% and 10% are -3.83, -3.02, -2.65 respectively and p-value is 1.00. Please tell me how to interpret such result. Thank You.
In my model, I have 5 variables, 3 of them are stationary at level but the other two are stationary at first difference. So is it okay or I have to make them stationary at level? If YES, then please tell me the technique.
what is the best criterion to choose the right model for example the first différence with 1: none 2: intercept 3: trend and intercept thank you in advance
Sir, for one of my series i went for adf test, in which the p- value for test statistics is 0.0110 and the trend co-efficient is significant as the p- value is 0.0082. This is the result of E-view Augmented Dickey-Fuller test statistic -3.934060 0.0110 @TREND(1) 0.213549 0.0082 Should I consider this series stationary? If yes then there is presence of deterministic tred, isn't it? N if yes then how do i make the series trend stationary?
Hello, what if ADF results are not consistent across 3 models(intercept, intercept and trend, none)? One of my variables is exhibiting significance at intercept and has unit root on other 2. Can i take it as no unit root or should i go to 2nd difference?
Dear Bolor, Thank you. I would like to invite you to join Hossain Academy Facebook at below link and post your question there. Actually I am in that group and may help you. Thank you once again, Sayed Hossain from Hossain Academy. facebook.com/groups/hossainacademy
nice video! i have a question.. if the data is stationary at 1st difference with "none" and "intercept", but when include "trend and intercept" it becomes not stationary. so, can we call it stationary or should we continue checking it at 2nd difference? thanks before.
but my proff said that the data would not be that reliable when it comes to the 2nd difference. so that the result not so accurate. what do you think sir? anyway, thanks so much for the reply and keep up the good work!
God bless you sir, I most confess to you that videos on running regression has been great. But sir, my question is, what would you do when you discover that a variable is not stationary?
Sayed Hossain I would like to invite you to join Hossain Academy Facebook for greater interaction about economics, finance and econometrics. Thank you Sayed Hossain from Hossain Academyfacebook.com/groups/hossainacademy/
Dear Sayed, You're absolutely helping me out with your videos, i've seen quite some already! Just one question: you happen to use only one variable for unit root testing. I would like to test it for multiple variables, (4,5,6,) but it does not work the same. Should I test it for all variables individually or as a group? Thank you
Sayed Hossain Alright thank you, and then what should one do if not all three (intercept, trend & intercept, none) follow the same line? If one (none) is significant while the others are not? Convert to first difference?
Sayed Hossain But what should I do then? If I cant take a decision? In the next step (Johansen test) I need to have all variables in first difference anyway..
thanks!!!!your video have help me a lot! but i still have 1 question. why the coefficient value of the variable must be negative sign ? if I get a positive sign?what its mean?
Prob*:1.000 (do not reject H0) Augmented Dickey-Fuller test statistic:-4.852017 Test critical values: 1% level -3.482035 5% level -2.884109 10% level -2.578884 (more than critical value=reject H0) GE(-1): 0.082220 (positive sign) this is the result i get, i should conclude it have unit root or not ?
you mean the prob*? but after i proceed to 1st differencing test, prob*=0.0000 ADF test statistic more than critical value coefficient is negative sign. so can i conclude that the variable is stationary?
ASALAM ul KUM Sir; I had taken log of all five variables and now all five variables are stationary at second difference (NONE). Is it possible to proceed to next test with variables stationary at second difference or need to transform the variables to make them stationary at first difference then can proceed to next test? I'll be using ARDL model to test the long run relationship then to VECM followed by Granger Causality test. I am really struck jazakallah if you can help me to proceed.
Thank you. I would like to invite you to join Hossain Academy Facebook Group at below link and join our group discussion. Thank you. Sayed Hossain from Hossain Academy.
Thank you. I would like to invite you to join Hossain Academy Facebook Group at below link and join our group discussion. Thank you. Sayed Hossain from Hossain Academy.
Thank you. I would like to invite you to join Hossain Academy Facebook Group at below link and join our group discussion. Thank you. Sayed Hossain from Hossain Academy. facebook.com/groups/hossainacademy/
Hi Sir, ADF test shows that the variable is nonstationary at level but PP shows that it is stationary. Of course both ADF and PP show that the variable is stationary after first difference. Which decision should be taken? Can running the correlogam and the LB test help to clear this uncertainty?
yoshua nugraha Dear Youshua, Please join Hossain Academy Facebook below for greater interaction with me regarding data analysis. Thank you Sayed Hossain from Hossain Academy facebook.com/groups/hossainacademy/
Fantastic video sir! God bless you
really thank you sir, you have helped me alot
That's fine sir, but i read from damodar gujarati that if trend is present then we have to regress it against trend variable for removing trend. and i did the same thing and solved the problem. Thank you so much for your reply.
Sometime it is difficult to make the variable stationary. In that case, I would advise, either convert variables into log or increase the data size and try. In future I have plan to upload video on ARMA. but normally we insert AR(1) to capture serial correlation problem
Fantastic video dear Hussein. Thank you.
Marc Jack You are welcome Jack. Thank you Sayed Hossain from Hossain Academy at www.sayedhossain.com
Thank you for the video. It was very enlightening.
michelangelolandgraf You are very much welcome. Thank you Sayed Hossain from Hossain Academy at www.sayedhossain.com
Thanks a lot sayed Hossain
Thank you. I would like to invite you to join Hossain Academy Facebook Group at below link and join our group discussion. Thank you. Sayed Hossain from Hossain Academy. facebook.com/groups/hossainacademy/
There are three ADF equations to test whether a particular time series data is stationary or not. All these three must give the same decision to come to a decision whether a particular time series is stationary or not.
You see all my videos properly. I guess you will understand finally and also read literature on it.
Hi Sayed, do you know if it is possible to carry out the Dickey Pantula test for multiplie unit roots on eviews?
Dear Sir,
I saw your videos for unit root test but I'm not getting the video which shows how to detrend the series if trend is present. Please revert back as soon as possible.
When I did "level" and "intercept" unit root was present but the coefficient was positive. For every other combination the coefficient was negative. Does it matter that we couldn't accept the result for level and intercept, as the P and T values showed it to have unit roots anyway?
Sir, In Eviews manual, there is no metion of obsoulte value The manual says " Notice here that the t statistic value is greater than the critical values so that we do not reject the null at conventional test sizes. " the t value shown in manual is -1.41 which is mentioned as GREATER than at 1% level -3.45 Are you correct or is the manual correct? Pl clarify
Dear sayed,
can you post anything to show us how to use DOLS on a money demand function?
Thank you....
Sir, In one of my results on GDP, I am trying to find out stationarity. ADF Statistics = 5.4255 and critical values at 1%, 5% and 10% are -3.83, -3.02, -2.65 respectively and p-value is 1.00. Please tell me how to interpret such result. Thank You.
Hi sir.
Its must to use intercept, trend and none, or i can use only one that i get stationary by using level first and second difference
In my model, I have 5 variables, 3 of them are stationary at level but the other two are stationary at first difference. So is it okay or I have to make them stationary at level? If YES, then please tell me the technique.
What do you do if the coefficient doesn't have a negative sign?
Hi, the question is what do i include my model? none?, drift?, or time-trend and drift?..
thank you Mr sayed, but still im confused, how to make cointegration and the propose of it
what is the best criterion to choose the right model for example the first différence with
1: none
2: intercept
3: trend and intercept
thank you in advance
Sir, for one of my series i went for adf test, in which the p- value for test statistics is 0.0110 and the trend co-efficient is significant as the p- value is 0.0082.
This is the result of E-view
Augmented Dickey-Fuller test statistic -3.934060 0.0110
@TREND(1) 0.213549 0.0082
Should I consider this series stationary? If yes then there is presence of deterministic tred, isn't it? N if yes then how do i make the series trend stationary?
You are welcome
Hello, what if ADF results are not consistent across 3 models(intercept, intercept and trend, none)? One of my variables is exhibiting significance at intercept and has unit root on other 2. Can i take it as no unit root or should i go to 2nd difference?
Dear Bolor, Thank you. I would like to invite you to join Hossain Academy Facebook at below link and post your question there. Actually I am in that group and may help you. Thank you once again, Sayed Hossain from Hossain Academy. facebook.com/groups/hossainacademy
Yes.
Better test all
May be but I have never used Dickey Pantula Test
nice video! i have a question.. if the data is stationary at 1st difference with "none" and "intercept", but when include "trend and intercept" it becomes not stationary. so, can we call it stationary or should we continue checking it at 2nd difference?
thanks before.
Adiatma Kasim All three must come to same decision to come to a single decision.
but my proff said that the data would not be that reliable when it comes to the 2nd difference. so that the result not so accurate. what do you think sir?
anyway, thanks so much for the reply and keep up the good work!
what if the result only shows two give the same result while the othe gives a different result. what does that imply?
God bless you sir, I most confess to you that videos on running regression has been great. But sir, my question is, what would you do when you discover that a variable is not stationary?
hosea handan It can always happen but you can solve it
Sayed Hossain I would like to invite you to join Hossain Academy Facebook for greater interaction about economics, finance and econometrics. Thank you Sayed Hossain from Hossain Academyfacebook.com/groups/hossainacademy/
Cool
Hi Sir, may i knw..did u have the video show the 2nd different model?
puong renn Lau No I do not have but the procedure is same like first difference.
It would be better if all three give the same result. Try to get same result from three.
Dear Sayed,
You're absolutely helping me out with your videos, i've seen quite some already!
Just one question: you happen to use only one variable for unit root testing. I would like to test it for multiple variables, (4,5,6,) but it does not work the same. Should I test it for all variables individually or as a group?
Thank you
You have to use each and every variable individually..
Sayed Hossain Alright thank you, and then what should one do if not all three (intercept, trend & intercept, none) follow the same line? If one (none) is significant while the others are not? Convert to first difference?
If all three give the same decision only then accept it. Otherwise do not take any decision.
Sayed Hossain But what should I do then? If I cant take a decision? In the next step (Johansen test) I need to have all variables in first difference anyway..
In that case, you run only VAR, not VECM model but make the data stationary before running it.
@NMAUTUBE the idea is to use absolute value of t-statistic
thanks!!!!your video have help me a lot!
but i still have 1 question.
why the coefficient value of the variable must be negative sign ?
if I get a positive sign?what its mean?
It should be negative but if not, meaning that there is no long run causality...
Prob*:1.000 (do not reject H0)
Augmented Dickey-Fuller test statistic:-4.852017
Test critical values:
1% level -3.482035
5% level -2.884109
10% level -2.578884 (more than critical value=reject H0)
GE(-1): 0.082220 (positive sign)
this is the result i get, i should conclude it have unit root or not ?
I have never seen it to be 1...There is problem somewhere either in data or data setting...
Test statistics is more than critical value...you can reject null hypothesis...what about probablity value?
you mean the prob*?
but after i proceed to 1st differencing test,
prob*=0.0000
ADF test statistic more than critical value
coefficient is negative sign.
so can i conclude that the variable is stationary?
ASALAM ul KUM Sir; I had taken log of all five variables and now all five variables are stationary at second difference (NONE). Is it possible to proceed to next test with variables stationary at second difference or need to transform the variables to make them stationary at first difference then can proceed to next test? I'll be using ARDL model to test the long run relationship then to VECM followed by Granger Causality test. I am really struck jazakallah if you can help me to proceed.
Thank you. I would like to invite you to join Hossain Academy Facebook Group at below link and join our group discussion. Thank you. Sayed Hossain from Hossain Academy.
I am also facing same situation..can I use ardl if some variables stationary at 2nd difference????
Pls reply ASAP
hello, sir
what if the variable is stationary with tested on level form, do we need to test it on 1st difference?
No
Sir, If the series is stationary at I(1) and also stationary at I(2) then what is the final conclusion? Is it I(1) or I(2)?
Thank you. I would like to invite you to join Hossain Academy Facebook Group at below link and join our group discussion. Thank you. Sayed Hossain from Hossain Academy.
Thank you. I would like to invite you to join Hossain Academy Facebook Group at below link and join our group discussion. Thank you. Sayed Hossain from Hossain Academy. facebook.com/groups/hossainacademy/
Hi Sir, ADF test shows that the variable is nonstationary at level but PP shows that it is stationary. Of course both ADF and PP show that the variable is stationary after first difference. Which decision should be taken?
Can running the correlogam and the LB test help to clear this uncertainty?
You can choose any one but if both gives the same decision, that is better always.
Sir, is it a common scenario of having ADF and PP showing different results and can running a correlogram and Q statistics be of any help?
Yes ...it is common scenario results may vary test to test...
Thank you Mr. Hossain.
Plan in future
Normally when three equation gives the same result, we accept it.
Mr Sayed, what is the meaning of p-value equal to 0.0000 after 1st difference?
yoshua nugraha it means that you can reject null hypothesis. What is your null?
yoshua nugraha Dear Youshua, Please join Hossain Academy Facebook below for greater interaction with me regarding data analysis. Thank you Sayed Hossain from Hossain Academy
facebook.com/groups/hossainacademy/
Sir, I'm a beginner in time series analysis. So, if you find my question silly, I am sorry for that. But please guide me properly.
Out of three options, at least two should be negative
Sir, what if my probability (p-value) for all three conditions is 0.00000?
+dada It means that reject null hypothesis. What is your null hypothesis?
In that case we cannot accept the result
I am not understanding your question
You are welcome
Better test all
You are welcome