Holt-Winters forecasting

For technical questions regarding estimation of single equations, systems, VARs, Factor analysis and State Space Models in EViews. General econometric questions and advice should go in the Econometric Discussions forum.

Moderators: EViews Gareth, EViews Moderator

labenroth
Posts: 4
Joined: Mon Nov 09, 2009 11:03 am

Holt-Winters forecasting

Postby labenroth » Wed Nov 11, 2009 4:58 pm

Question from a newbie:
How do I get the Holt-WInters smoothing option to forecast m-periods beyond the end of my data range?

Thanks in advance...

trubador
Did you use forum search?
Posts: 1520
Joined: Thu Nov 20, 2008 12:04 pm

Re: Holt-Winters forecasting

Postby trubador » Thu Nov 12, 2009 2:52 am

Open your series and select Proc/Exponential Smoothing. A self-explanatory dialog box will be opened. You do not have to specify a forecast horizon in addition to estimation sample, since EViews will calculate forecasts starting from the first observation after the end of the estimation sample.

labenroth
Posts: 4
Joined: Mon Nov 09, 2009 11:03 am

Re: Holt-Winters forecasting

Postby labenroth » Thu Nov 12, 2009 10:34 am

trubador-

THanks for responding. I don't seem to be doing it right then:

1) I open a series called spa containing 43 quarterly observations, 1999Q1 to 2009Q3.
2) I open Proc>Exponential Smoothing>
3) From the dialogue box I select: Smoothing Method=Holt-WInters Multiplicative; Smoothing Parameters=E,E,E; Smoothed Series="spasm"; Estimation Sample=1999Q1 2009Q3; Cycle for Seasonal=4
4) I open the resulting series "spasm" and find that there are 43 quarterly observations, 1999Q1 to 2009Q3. There are no forecast values for the next 4 quarters (2009Q4, 2010Q1, 2010Q2, 2010Q3) beyond the end of the estimation sample.

Why doesn't resulting series "spasm" should contain forecast values for the four quarters 2009Q4, 2010Q1, 2010Q2, 2010Q3 beyond the end of the estimation sample?

EViews Gareth
Fe ddaethom, fe welon, fe amcangyfrifon
Posts: 13600
Joined: Tue Sep 16, 2008 5:38 pm

Re: Holt-Winters forecasting

Postby EViews Gareth » Thu Nov 12, 2009 10:52 am

Does your workfile contain observations for those dates?

If you don't have those dates in your workfile range, then they will not be forecasted.

labenroth
Posts: 4
Joined: Mon Nov 09, 2009 11:03 am

Re: Holt-Winters forecasting

Postby labenroth » Thu Nov 12, 2009 1:13 pm

Forgive me if I misunderstand:

I have actual (observed)quarterly data for the interval 1999Q1 to 2009Q3. I know the data exhibits seasonality just by inspection. My goal is to: 1) quantify the seasonality, and 2) forecast the data for the next 4 quarters into the future 2009Q4, 2010Q1, 2010Q2, 2010Q3.

I obviously have no actuals for these future periods. From my reading of H-W methods, I should be able to project this forecast for m-periods ahead, beyond the range for which I have observed data (actuals).

startz
Non-normality and collinearity are NOT problems!
Posts: 3797
Joined: Wed Sep 17, 2008 2:25 pm

Re: Holt-Winters forecasting

Postby startz » Thu Nov 12, 2009 1:16 pm

Make sure the workfile range goes through 2010Q3, even though your data doesn't.

labenroth
Posts: 4
Joined: Mon Nov 09, 2009 11:03 am

Re: Holt-Winters forecasting

Postby labenroth » Thu Nov 12, 2009 1:21 pm

Ahhh....(sound of forehead smack)

Thank you all for the help!


Return to “Estimation”

Who is online

Users browsing this forum: No registered users and 2 guests