Forecasting
Posted: Fri May 05, 2017 8:02 am
Hi Gareth
I am finding eviews documentation confusing and limited (based on my forecasting)
Using other econometric packages if I can do a 1.static,2. recursive or 3.rolling forecast using an out of sample
eviews seems to only do a static 1. forecast ie it uses the estimated equation coefficents and then forecasts n steps ahead to fill in the out of sample period. or does it do 2. recursively ie estimate the equation calcualate a single step forecast and then inlcude that estimate in the equation and re-estimate and do the next step forecast.(in other words the estimation sample gets larger as each step is forecast). or does it do 3. A rolling window so that it estimates one step ahead, then drops an observation at the start and adds the new estimated observation to the sample re-estimates and continues.
I would greatly appreciate some clarity on this as I have had to resort to writing code to do a rolling forecast or go to the other econometric packages for forecasting. It may be my confusion but I must say the docs are not clear.
I am finding eviews documentation confusing and limited (based on my forecasting)
Using other econometric packages if I can do a 1.static,2. recursive or 3.rolling forecast using an out of sample
eviews seems to only do a static 1. forecast ie it uses the estimated equation coefficents and then forecasts n steps ahead to fill in the out of sample period. or does it do 2. recursively ie estimate the equation calcualate a single step forecast and then inlcude that estimate in the equation and re-estimate and do the next step forecast.(in other words the estimation sample gets larger as each step is forecast). or does it do 3. A rolling window so that it estimates one step ahead, then drops an observation at the start and adds the new estimated observation to the sample re-estimates and continues.
I would greatly appreciate some clarity on this as I have had to resort to writing code to do a rolling forecast or go to the other econometric packages for forecasting. It may be my confusion but I must say the docs are not clear.