Test the null hypotheses c(6)>c(5)

For technical questions regarding estimation of single equations, systems, VARs, Factor analysis and State Space Models in EViews. General econometric questions and advice should go in the Econometric Discussions forum.

Moderators: EViews Gareth, EViews Moderator

al.domenico
Posts: 3
Joined: Mon Nov 29, 2021 11:54 am

Test the null hypotheses c(6)>c(5)

Postby al.domenico » Tue Nov 30, 2021 3:21 am

Hello everybody, I have a test equation with 6 parameters and I would like to test the hypotheses c(6)>=c(5) against H1: c(6)<c(5).
I have performed a Wald test with c(6)=c(5) and I have a p-value equal to 0. How can I confirm that c(6) is larger or equal than c(5)?
Thank you for your answer.

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

Re: Test the null hypotheses c(6)>c(5)

Postby startz » Tue Nov 30, 2021 7:15 am

Basically, you just want to do a one-tailed test. See https://en.wikipedia.org/wiki/One-_and_two-tailed_tests for an explanation.

al.domenico
Posts: 3
Joined: Mon Nov 29, 2021 11:54 am

Re: Test the null hypotheses c(6)>c(5)

Postby al.domenico » Wed Dec 01, 2021 7:38 am

startz wrote:Basically, you just want to do a one-tailed test. See https://en.wikipedia.org/wiki/One-_and_two-tailed_tests for an explanation.

Yes, exactly, thank you. How could I perform it on EViews?

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

Re: Test the null hypotheses c(6)>c(5)

Postby startz » Wed Dec 01, 2021 7:47 am

EViews does not have one-tailed tests built in. Go through the steps for a two-tailed test and then compare the t-statistic to the critical value for a one-tailed test from a t-table.

al.domenico
Posts: 3
Joined: Mon Nov 29, 2021 11:54 am

Re: Test the null hypotheses c(6)>c(5)

Postby al.domenico » Wed Dec 01, 2021 8:10 am

startz wrote:EViews does not have one-tailed tests built in. Go through the steps for a two-tailed test and then compare the t-statistic to the critical value for a one-tailed test from a t-table.

Okok, thank you a lot!
So, do I have to pick the right tail to test if c(6)>c(5) testing on EViews c(6)-c(5)=0?

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

Re: Test the null hypotheses c(6)>c(5)

Postby startz » Wed Dec 01, 2021 8:28 am

The tails are symmetric. Of course, if the estimated c(6) is greater than the estimated c(5) then you can't reject the hypothesis no matter what the t-statistic is.

abd
Posts: 2
Joined: Thu Dec 02, 2021 11:07 am

Re: Test the null hypotheses c(6)>c(5)

Postby abd » Thu Dec 02, 2021 1:51 pm

I didn't know all of this, test wald is great, but I go directly to -> Prob(F-statistic)
H0: c(2)=c(3)=0
H1: c(2) != c(3)

if p>5% ---> accept H0
otherwise rejected

maybe false

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

Re: Test the null hypotheses c(6)>c(5)

Postby startz » Thu Dec 02, 2021 1:55 pm

abd wrote:I didn't know all of this, test wald is great, but I go directly to -> Prob(F-statistic)
H0: c(2)=c(3)=0
H1: c(2) != c(3)

if p>5% ---> accept H0
otherwise rejected

maybe false


I'm not sure if there is a question here. (Do note that H1 isn't quite right, as c(2)=c(3)=17 would lead to a rejection of the null.)

abd
Posts: 2
Joined: Thu Dec 02, 2021 11:07 am

Re: Test the null hypotheses c(6)>c(5)

Postby abd » Thu Dec 02, 2021 3:09 pm

startz wrote:
I'm not sure if there is a question here. (Do note that H1 isn't quite right, as c(2)=c(3)=17 would lead to a rejection of the null.)


noted. then H1: each one is != 0
Thank you

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

Re: Test the null hypotheses c(6)>c(5)

Postby startz » Thu Dec 02, 2021 3:11 pm

abd wrote:
startz wrote:
I'm not sure if there is a question here. (Do note that H1 isn't quite right, as c(2)=c(3)=17 would lead to a rejection of the null.)


noted. then H1: each one is != 0
Thank you


Actually, the alternative is that either one is not equal to 0.


Return to “Estimation”

Who is online

Users browsing this forum: Cscottmcwall-ea, cuthbertdavies, Majestic-12 [Bot] and 40 guests