Basic Rolling Regression

[EViews Gareth]

Each subsamples coefficients are stored in coefmat. You'll have to change the program to store the standard errors in the same way.

[lscx411]

Each subsamples coefficients are stored in coefmat. You'll have to change the program to store the standard errors in the same way.

Thanks Gareth! How to store subsamples R-squared in Eviews?

[S Mukhtar]

Dear all,
I am new to use Eviews and I have been trying to replicated the basic rolling regression. After following all the steps mentioned in the first post, I find the message "!step not defined" when I reach to the following point:

!nrolls = @round((!length-!window)/!step)

Although I do define !step=40 as done in the first post.

Sorry for naivety, please can anybody help me with this? Why do I get such message? I have tried many times.

Many thanks,
Mukhtar

[EViews Gareth]

Hard to say without seeing what you've done.

[The Yoyo]

For anyone still looking for a code that does a 1-period ahead recursive regression forecast (expanding window), I put together something from posts all over this forum. Most of the credits for these codes go to Esther and Gareth.
Try the code below. Note that I left out the coefficient matrix.

Code: Select all

'create some data create u 100 series y=nrnd series x1=nrnd series x2=nrnd series z=nrnd 'run rolling regression ' set window size !window = 20 ' set step size !step = 1 ' get size of workfile !length = @obsrange ' declare equation for estimation equation eq1 'calculate number of rolls !nrolls = @floor((!length-!window)/!step) 'series to store forecast estimates series fcast '*EDITED: catching start and end points %start = "@first" '@otod(@ifirst(ser)) %end = "@last" '@otod(@ilast(ser)) '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 for estimation period %first = @otod(@dtoo(%start)) %last = @otod(@dtoo(%start)+!i+!window-2) smpl {%first} {%last} ' estimate equation - where the equation is y =c(1) + c(2)*x1 +c(3)*x2 eq1.ls y c x1 x2 ' 1-period-ahead forecast %1pers = @otod(@dtoo(%start)+!i+!window-1) 'start point %1pere = @otod(@dtoo(%start)+!i+!window-1) 'end point '*EDITED if @dtoo(%end) < @dtoo(%1pere) then 'check whether the forecast end point is greater than the workfile end point exitloop endif ' set smpl for forecasting period smpl {%1pers} {%1pere} ' forecast with command *forecast* (see also *fit*) eq1.forecast(f=na) yf ' set sampl to obtain the 4th period observation smpl {%1pere} {%1pere} ' store forecasts fcast = yf next smpl @all show fcast.line d(noerr) yf

I have no real programming experience, so keep in mind that the code might not work 100% correctly. Any improvements are appreciated.

[EVG]

Hello,

Many thanks you for your help Gareth and Esther.

I am using EViews 7.2 and have been able to do a h-period ahead recursive regression forecast thanks to this post. However, I would also like to add exogenous variables to my model, and these need to be forecasted for each step with an ARMA structure (say AR(1)). I am struggling with the organization of the code, could anyone help me ?

[EViews Gareth]

I don't understand what you're trying to do

[gsourop]

Hallo everyone!

I am new in the forum and I am not familiar with the EVIEWS programming language, so I would really appreciate some help!!

My forecasting equation is: Exchange rate=c+c(1)(log_cpi_uk-log_cpi_us)

I want to run the out-of sample forecasting test with recursive window forecasting scheme. I use some commands seen in the forum such as:

' set window size
!window = 576

' set step size
!step = 90

' get size of workfile
!length = @obsrange

' declare equation for estimation
equation eq1

'calculate number of rolls
!nrolls = @floor((!length-!window)/!step)

'matrix to store coefficient estimates
matrix(2,!nrolls) coefmat ' where 2 is the number of coefficients

'series to store forecast estimates
series fcast

' set sample for estimation period
%start ="@first" ' @otod(@ifirst(ser))
%end="%last" '@otod(@ilast(ser))
smpl {%first} {%last}

'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 for estimation period
%first = @otod(@dtoo(%start)+!i-1)
%last = @otod(@dtoo(%start)+!i+!window-2)
smpl {%first} {%last}

' estimate equation - where the equation is y=c(1) + c(2)*x1
eq1.ls log_spot c (log_ipi_uk-log_ipi_us)

'store coefficients
colplace(coefmat,eq1.@coefs,!j)

' 1-period-ahead forecast
%1pers = @otod(@dtoo(%start)+!i+!window-1) 'start point
%1pere = @otod(@dtoo(%start)+!i+!window) 'end point

'*EDITED
if @dtoo(%end) < @dtoo(%1pere) then 'check whether the forecast end point is greater than the workfile end point
exitloop
endif

' set smpl for forecasting period
smpl {%1pers} {%1pere}

' forecast with command *forecast* (see also *fit*)
eq1.forecast(f=na) yf

' set sampl to obtain the 4th period observation
smpl {%1pere} {%1pere}

' store forecasts
fcast = yf
next

smpl @all
show coefmat
show fcast.line

d(noerr) yf


But when I run the program I get the message "Matrix or vector given non positive dimensions in the "Matrix (2,0) COEFMAT" .

[EViews Gareth]

Looks like !length is the same as !window. i.e. your rolling size is the same size as the number of observations you have. Which makes no sense.

[gsourop]

Thank you for your response. How can I deal with that? Which part of the programming shall I change? To make my self clear I will give you more details regarding my project: My sample consists of 576 have observations, starting from M1 1969 to M12 2013. The in-sample period is taken to be from the beginning to M12 1985 and the rest are the out-of-sample observations. I want to produce me first forecast on M1 1975 and then recursively estimate next coefficients. How do I change my code in order to adjust on the above?

Thank you very much for your time!

[EViews Gareth]

Do you want a rolling regression or an expanding one?

[gsourop]

A recursive regression (expanding)

[EViews Gareth]

Could you explain what that means?

[gsourop]

That I use a period of time to produce my first coefficient and then I add an observation and re-estimate my model..I do these iterations until the end of the sample period.

[EViews Gareth]

You'll need to completely re-write the sample parts then, since you're not doing rolling regression.