One SE Bands on IRF

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

MRLCU
Posts: 3
Joined: Fri Aug 18, 2017 10:09 am

One SE Bands on IRF

Postby MRLCU » Fri Aug 18, 2017 10:37 am

Hi,

I am outputting impulse response functions for my 5 variable SVAR. I was wondering if there was anyway to customise the standard error bands that are given with your IRF. The default response standard errors is the +- 2 SD error bands. I was wondering if there was any way to change this to 1 SD error band for my Impulse responses.

Many thanks.

EViews Matt
EViews Developer
Posts: 157
Joined: Thu Apr 25, 2013 7:48 pm

Re: One SE Bands on IRF

Postby EViews Matt » Fri Aug 18, 2017 11:30 am

Hello,

You can do this via EViews commands. See http://forums.eviews.com/viewtopic.php?t=2883.

MRLCU
Posts: 3
Joined: Fri Aug 18, 2017 10:09 am

Re: One SE Bands on IRF

Postby MRLCU » Sat Aug 19, 2017 10:26 am

Hi,

Thank you for your response, I had a look at that but I couldn't quite understand the code.

My VAR command is as follows;

I'm running a 5 variable VAR and i've used the code below to output the IRF of a one SD shock in government spending to GDP;

var var1.ls 1 4 gov_spending gdp inflation tax_revenue interest_rate
var1.impulse(40, m, se=a) gdp @gov_spending @ gov_spending gdp inflation tax_revenue interest_rate

However, this outputs the IRF with the 'analytic' +-2 SD error band. I was wondering if you could tell me what command I need to display +-1 SD error band?

Thanks a lot.

dakila
Posts: 264
Joined: Tue Nov 24, 2015 4:57 pm

Re: One SE Bands on IRF

Postby dakila » Sun Aug 20, 2017 1:39 am

Use the sirf add-in (v2.0).

MRLCU
Posts: 3
Joined: Fri Aug 18, 2017 10:09 am

Re: One SE Bands on IRF

Postby MRLCU » Mon Aug 21, 2017 4:32 am

EViews Matt wrote:Hello,

You can do this via EViews commands. See http://forums.eviews.com/viewtopic.php?t=2883.


Hi I had a look at this and I did the command;

var1.impulse(40, m, se=a, smat=yourownimpulse) gdp @gov_spending @ gov_spending gdp inflation tax_revenue interest_rate

I have the yourownimpulse and yourownimpulse_se vectors saved.
I'm just unsure what I do now. I tried typing yourownimpulse+1*yourownimpulse_se into the command window but that didn't work.
Sorry I haven't used eviews before so could you kindly tell me where I need to type or what I need to do with the "yourownimpulse+1*yourownimpulse_se" command.

Many thanks.

dakila
Posts: 264
Joined: Tue Nov 24, 2015 4:57 pm

Re: One SE Bands on IRF

Postby dakila » Mon Aug 21, 2017 4:51 am

First install the sirf add-in. Then goto Add-ins/Download Add-ins. Then select the sirf add-in and install it.
After estimation IRF, goto Proc/Add-ins then select Scaled IRF. It will a dialog box. The rest is easy to fill.
You cant use command for the sirf add-in. It is only available for dialog interface.

EViews Matt
EViews Developer
Posts: 157
Joined: Thu Apr 25, 2013 7:48 pm

Re: One SE Bands on IRF

Postby EViews Matt » Tue Aug 22, 2017 10:22 am

"ourownimpulse+1*yourownimpulse_se" is just a mathematical expression for one of the bands, it's not a complete EViews command. You'd still need to create the graph from the IRF and SE data. That's where the sirf addin can save you a lot of work, but here's an example of recreating the 2 SE band graph for the response of the first variable to the first shock (top left graph). Obviously you can remove or replace the "2 *" in the code below to create bands at any scale.

Code: Select all

var1.impulse(40, m, imp=struct, se=a, smat=z) @imp 1 2 3 4 5
matrix tmp = @hcat(@hcat(z.@col(1), z.@col(1) + 2 * z_se.@col(1)), z.@col(1) - 2 * z_se.@col(1))
freeze(imp_graph) tmp.line
imp_graph.option linepat
imp_graph.elem(2) lcolor(red) lpat(dash1)
imp_graph.elem(3) lcolor(red) lpat(dash1)
show imp_graph


Return to “Estimation”

Who is online

Users browsing this forum: No registered users and 4 guests