Storing information criterion values

[sohrabahari]

Hi Eviews support,

I'm doing a project which involves recursive modelling and analysis of information criterion. I have a recursive modelling and forecasting pattern, but I want to be able to use the set up I have of individual model estimation to be able to produce values of AIC, SIC (or BIC), and R^2 for individual observations. Below is an example of the program I have written so far.

series a=cpi
series b=m1
series c1=dgdpm
series d1=t1y
series e=t30d
series f=vwrincd
series g=wti
series h=indout
series i=sp500(-1)


'run rolling regression
!window=60

!step=1

!length=@obsrange

series ser=1

%start=@otod(@ifirst(ser))
%end=@otod(@ilast(ser))

!j=0

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

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

equation eqcons{%first}.ls sp500 c
eqcons{%first}.forecast fsp500_eqcons{%first}
equation eqa{%first}.ls sp500 c a
eqa{%first}.forecast fsp500_eqa{%first}
equation eqb{%first}.ls sp500 c b
eqb{%first}.forecast fsp500_eqb{%first}
equation eqc{%first}.ls sp500 c c1
eqc{%first}.forecast fsp500_eqc{%first}
equation eqd{%first}.ls sp500 c d1
eqd{%first}.forecast fsp500_eqd{%first}
equation eqe{%first}.ls sp500 c e
eqe{%first}.forecast fsp500_eqe{%first}
equation eqf{%first}.ls sp500 c f
eqf{%first}.forecast fsp500_eqf{%first}
equation eqg{%first}.ls sp500 c g
eqg{%first}.forecast fsp500_eqg{%first}
equation eqh{%first}.ls sp500 c h
eqh{%first}.forecast fsp500_eqh{%first}
equation eqi{%first}.ls sp500 c i
eqi{%first}.forecast fsp500_eqi{%first}

next
exitloop

I understand the labelling conditions and how it selects the period for observation for estimation, but I don't know how to get information criterion values for individual observations. What I would like to know is if its possible to store all the AIC values, for example, the AIC values of model y=mx+c for all observations between 1982M12 to 2011M12 (December 1982 to December 2011).

Your help is much appreciated

[EViews Gareth]

Just grab it using equation.@aic.

There's an example of storing the R-squared, which works in the same way, in the Programming Guide http://forums.eviews.com/viewtopic.php?f=5&t=1638

[sohrabahari]

Hi,

I've tried some of the things you have suggested and variations on it, but I can't seem to get it to do what I want it to. It'll give me the rows of estimation for 10 equations but nothing after it. As I've shown you, I have a recursive modelling setup as it, with a 60 period window, but there are 350 observations in that period.

Is it possible to create a matrix such that the rows are the equations estimated and the columns are the period of estimation for that equation?

1982M12 1983M01 1983M02 . . .
y=ma+c (AIC for this estimation) (AIC) . .
y=mb+c (AIC) . .
. . .
. .
.

The labelling of colums is set up using {%first}

Kind Regards

[EViews Gareth]

You could create a table, and put things in the table.