EViews.com

Posted: **Mon Aug 30, 2021 10:24 pm**

After reviewing the forum discussions on this topic, such as:

http://forums.eviews.com/viewtopic.php? ... ion#p65681

I noticed there has been consistent and even growing interest in rolling panel regression functionality.

However, it would appear that EViews doesn't offer a native function for panel rolling regressions, but one could attempt a program as is discussed here:

http://forums.eviews.com/viewtopic.php?t=18592

However, in many of such threads, the complete rolling regression program hasn't been provided for the panel case.

I was hoping that we could adapt the basic rolling regression EViews Gareth begun:

http://forums.eviews.com/viewtopic.php?f=15&t=878

To mold it into a panel rolling regression, it might be nice to have comments to denote what would change based on cross section and periods included

Purely as an example, in my case:

Periods:53

Cross sections: 670

Total (unblanced) observations: 21558

I tried the following code with the equation.ls(cx=f) as suggested:

Code: Select all

'Configuration

%y = "y"

%const = c
%var1 = "x1"
%var2 = "x2"
%var3 = "x3"
%var4 = "x4"
%var5 = "x5"
%var6 = "x6"
%var7 = "x7"

'-------------------------------------------------------------------------------------
'run rolling regression

' set window size
!window = 21558
 
' set step size
!step = 3000
 
' get size of workfile
!length = @obsrange
 
' declare equation for estimation
equation eq1
 
'calculate number of rolls
!nrolls = @round((!length-!window)/!step)

'matrix to store coefficient estimates
matrix(8,!nrolls)  coefmat ' where 8 is the number of coefficients

'variable keeping track of how many rolls we've done
!j=0

' move sample !step obs at a time
for !i = 1  to  !length-!window+1-!step step !step
   !j=!j+1

   ' set sample to estimation period
   smpl @first+!i-1 @first+!i+!window-2

   ' estimate equation - where the equation is y=c(1) + c(2)*x1 + c(3)*x2
   eq1.ls(cx=f) {%y} {%const} {%x1} {%x2} {%x3} {%x4} {%x5} {%x6} {%x7}
   'store coefficients
   colplace(coefmat,eq1.@coefs,!j)
next

show coefmat

However, this just gave me a 8x3 matrix. I was expecting 8x50-ish, to cover most of the periods observed. At least for me, it's difficult to figure out how to tweak the syntax to output betas across my desired time horizon.

Posted: **Tue Aug 31, 2021 8:52 am**

Without thinking about it too much, really the only thing that needs to change in a panel case is the calculation of the number of rolls. Might be easier for you to just calculate it by hand and hard code it.

EViews.com

Verbose syntax/comment program for rolling panel regression

Verbose syntax/comment program for rolling panel regression

Re: Verbose syntax/comment program for rolling panel regression