Fixed rolling window forecast with Eviews 8.1

[Stunti_Eviews]

Dear all,

I am currently coding a fixed rolling window forecast for the simple linear equation: ls return c ep fcf. The data ranges from 2000 to 2014 whereas I want to forecast the years 2011 to 2014 with the prior years. Therefore, the window size must remain 15 over time, which means that per forecasted figure the oldest observation must drop out (e.g. the forecast for 2012 includes observations from 2001 to 2011 whereby 2011 is already a forecasted value).

By giving it a try, Eviews showed the following error message:
- Data does not uniquely define observation in panels in "%FIRST = @OTOD(@DTOO("2000")+1)".

Can anyone help? Furthermore, I am not sure whether my code accounts for a fixed or expanding window. Could any one tell me how I can check that and if necessary how to change it to a fixed rolling window forecast. Please find the workfile and program attached.

Many thanks for your efforts!
Bests,
Stunti

[EViews Gareth]

You're setting the start and end points with the @otod function. Unfortunately that function doesn't work in panel workfiles. You'll have to use some other arithmetic to figure out the start and end dates of each iteration of the loop.

Having said that, I'm not sure I follow what you're doing at all. How can the window be 15? You only have 15 time periods total.