restriction for coefficients

For econometric discussions not necessarily related to EViews.

Moderators: EViews Gareth, EViews Moderator

mehraban
Posts: 9
Joined: Tue Jul 07, 2020 9:53 am

restriction for coefficients

Postby mehraban » Fri Feb 26, 2021 1:57 pm

hello everybody,
can anybody tell me in the regression y=c(1)+c(2)*x + e , how can I test the hypothesis h0:c(1)=0 and h0: c(2)>=1 in eviews?
Should I test both together with wald test or I should separate them?
I want to tell that y converges to x. therefore I haveto tell c(1)=0 and c(2)<1.
many thanks in advance

startz
Non-normality and collinearity are NOT problems!
Posts: 3775
Joined: Wed Sep 17, 2008 2:25 pm

Re: restriction for coefficients

Postby startz » Fri Feb 26, 2021 3:16 pm

mehraban wrote:hello everybody,
can anybody tell me in the regression y=c(1)+c(2)*x + e , how can I test the hypothesis h0:c(1)=0 and h0: c(2)>=1 in eviews?
Should I test both together with wald test or I should separate them?
I want to tell that y converges to x. therefore I haveto tell c(1)=0 and c(2)<1.
many thanks in advance


This is actually a very hard problem. The first part of the hypothesis calls for a two-tailed test and the second calls for a one-tailed test. Putting those together would not be easy.

mehraban
Posts: 9
Joined: Tue Jul 07, 2020 9:53 am

Re: restriction for coefficients

Postby mehraban » Fri Feb 26, 2021 4:31 pm

startz wrote:
mehraban wrote:hello everybody,
can anybody tell me in the regression y=c(1)+c(2)*x + e , how can I test the hypothesis h0:c(1)=0 and h0: c(2)>=1 in eviews?
Should I test both together with wald test or I should separate them?
I want to tell that y converges to x. therefore I haveto tell c(1)=0 and c(2)<1.
many thanks in advance


This is actually a very hard problem. The first part of the hypothesis calls for a two-tailed test and the second calls for a one-tailed test. Putting those together would not be easy.





thank you for your answer. could you please guide me how can I test the second part i.e. c(2)<1 ?
for the first part I know I can use the Wald test separately. but the Wald test doesn't work when I restrict c(2)>=1 as the null hypothesis suggests!

startz
Non-normality and collinearity are NOT problems!
Posts: 3775
Joined: Wed Sep 17, 2008 2:25 pm

Re: restriction for coefficients

Postby startz » Fri Feb 26, 2021 4:40 pm

Doing the tests separately isn't really right I'm afraid.

If you want to test an inequality constraint, you ask EViews to do the test using the equal sign and then take one-half the reported p-value.

mehraban
Posts: 9
Joined: Tue Jul 07, 2020 9:53 am

Re: restriction for coefficients

Postby mehraban » Sat Feb 27, 2021 6:18 am

startz wrote:Doing the tests separately isn't really right I'm afraid.

If you want to test an inequality constraint, you ask EViews to do the test using the equal sign and then take one-half the reported p-value.



Wald test can not be used for one tailed constraint? i.e. for b>=1

startz
Non-normality and collinearity are NOT problems!
Posts: 3775
Joined: Wed Sep 17, 2008 2:25 pm

Re: restriction for coefficients

Postby startz » Sat Feb 27, 2021 7:46 am

mehraban wrote:
startz wrote:Doing the tests separately isn't really right I'm afraid.

If you want to test an inequality constraint, you ask EViews to do the test using the equal sign and then take one-half the reported p-value.



Wald test can not be used for one tailed constraint? i.e. for b>=1

I just explained how to do it.

mehraban
Posts: 9
Joined: Tue Jul 07, 2020 9:53 am

Re: restriction for coefficients

Postby mehraban » Sat Feb 27, 2021 8:43 am

startz wrote:
mehraban wrote:
startz wrote:Doing the tests separately isn't really right I'm afraid.

If you want to test an inequality constraint, you ask EViews to do the test using the equal sign and then take one-half the reported p-value.



Wald test can not be used for one tailed constraint? i.e. for b>=1

I just explained how to do it.




I see, but there is no p-value reported in the output page of the Wald test!

startz
Non-normality and collinearity are NOT problems!
Posts: 3775
Joined: Wed Sep 17, 2008 2:25 pm

Re: restriction for coefficients

Postby startz » Sat Feb 27, 2021 8:48 am

mehraban wrote:
startz wrote:
mehraban wrote:

Wald test can not be used for one tailed constraint? i.e. for b>=1

I just explained how to do it.




I see, but there is no p-value reported in the output page of the Wald test!


Yes there is. It's the column marked "probability."

mehraban
Posts: 9
Joined: Tue Jul 07, 2020 9:53 am

Re: restriction for coefficients

Postby mehraban » Mon Mar 01, 2021 4:09 pm

startz wrote:
mehraban wrote:
startz wrote:I just explained how to do it.




I see, but there is no p-value reported in the output page of the Wald test!


Yes there is. It's the column marked "probability."


Dear Startz
sorry to take your time again, but the reported probability is 0.0000
therefore, taking one-half of the probability, it is still zero. so H0:c(2)=1 is rejected and it is not equal to 1.that is right, but then how can I understand that c(2)>=1 or c(2)<1?
Last edited by mehraban on Mon Mar 01, 2021 4:14 pm, edited 1 time in total.

startz
Non-normality and collinearity are NOT problems!
Posts: 3775
Joined: Wed Sep 17, 2008 2:25 pm

Re: restriction for coefficients

Postby startz » Mon Mar 01, 2021 4:12 pm

mehraban wrote:
startz wrote:
mehraban wrote:


I see, but there is no p-value reported in the output page of the Wald test!


Yes there is. It's the column marked "probability."


Dear Startz
sorry to take your time again, but the reported probability is 0.0000
therefore, by divideding p by 2 , it is still zero. so it says b is not equal to 1. then how can I understand that b>=1 or b<1?


Well, given the p-value, if the estimate is below 1, you can be pretty sure b is not greater than1. If the estimate is above one, you can be pretty sure b is not below 1.

mehraban
Posts: 9
Joined: Tue Jul 07, 2020 9:53 am

Re: restriction for coefficients

Postby mehraban » Mon Mar 01, 2021 4:17 pm

startz wrote:
mehraban wrote:
startz wrote:
Yes there is. It's the column marked "probability."


Dear Startz
sorry to take your time again, but the reported probability is 0.0000
therefore, by divideding p by 2 , it is still zero. so it says b is not equal to 1. then how can I understand that b>=1 or b<1?


Well, given the p-value, if the estimate is below 1, you can be pretty sure b is not greater than1. If the estimate is above one, you can be pretty sure b is not below 1.



that is right, thank you very much for your help


Return to “Econometric Discussions”

Who is online

Users browsing this forum: Google [Bot] and 16 guests