Page 1 of 1

Midas output

Posted: Tue Apr 04, 2017 4:40 am
by Lobbi
Hello,
I have just installed eviews in order to run a Midas regression using Beta weighting.
I have a question regarding the Beta01, Beta02 and Beta03 in the output table. According to Ghysels, Sinko, and Valkanov (2006) the Beta weighting only depends on two weights denoted as w1 and w2.

Do the Beta01 and Beta02 mean the values for w1 and w2?
I would like to restrict the Beta weighting in the sense that w1 is fixed at 1 and w2 > 1 to ensure a decaying pattern. How can I do that in eviews?

Thank you for your comments.

Re: Midas output

Posted: Tue Apr 04, 2017 5:54 am
by EViews Gareth

Re: Midas output

Posted: Tue Apr 04, 2017 7:58 am
by Lobbi
The description is not 100% clear to me.

That means I must choose restriction "both" to set w1=1 and w3=0?
To set w3=0 I need to choose "endpoint" yes?

Re: Midas output

Posted: Tue Apr 04, 2017 8:50 am
by EViews Gareth
Yes.

Re: Midas output

Posted: Tue Apr 04, 2017 12:26 pm
by Lobbi
Thank you very much. There is another question I have regarding the specification in the estimation dialog.

In the estimation dialog window we have two boxes. I know that the dependent variable must be entered in the first specification box. In the second box regressors of higher frequency must be entered.
I have got regressors which are of the same frequency as the dependent variable and I only use variables of daily or monthly frequency. Where do I have to enter the regressors which are of the same frequency as the dependent variable?

Re: Midas output

Posted: Tue Apr 04, 2017 1:25 pm
by EViews Gareth
The same as the dependent.

Re: Midas output

Posted: Tue Apr 04, 2017 11:53 pm
by Lobbi
So, all regressors of the same frequency as the dependent variable in the first specification box?

Re: Midas output

Posted: Wed Apr 05, 2017 12:19 am
by EViews Gareth
Just as the dialog says.