Create smooth forecast that match level on annual frequency

[tvonbrasch]

Hi,

I have a dataseries with forecasts on annual frequency and a seasonally adjusted series on a monthly frequency. Then I would like to extend the seasonally adjusted series with a smoothed forecast on the monthly frequency, in such a way that the annual value for the forecasted smoothed series becomes equal to the annual value of the annual forecast series. How can one smooth a series in EViews, with such a restriction imposed?

To be more precise:

Code: Select all

wfcreate(page=a) a 2020 2030
   genr u_f= 4+nrnd

pagecreate(page=m) m 2020 2030
   copy a\u_f m\u_f

   smpl 2020 2023:6
      genr u_sa=4+0.5*nrnd

How can I extend the series u_sa with a smooth line from 2023:7 to 2030:12 while ensuring that: @meansby(u_sa, @year) = @meansby(u_f, @year) ?

Thomas

[EViews Gareth]

There isn't a technique built in (but if you find an econometric technique, let us know and we can add it in the future).

I guess you could always use your forecasting technique of choice and then just subtract/add the difference between the forecast and the annual average you want to hit.

[tvonbrasch]

Hello Gareth,

I've created a custom filter, which I've named "smoother," addressing the specific requirements I had. The rationale behind this filter is detailed below:

smoother.png (84.85 KiB) Viewed 3481 times

Please refer to the attached documentation for more information. Run the program example_1.prg to see how it can be used. You may also install it as an add-in using the smoother_install.prg file. It's important to note that the process is somewhat slow, taking a few seconds.

I would greatly appreciate it if someone from the EViews team could incorporate this into the built-in functionality or develop it as an add-in. Doing so would significantly enhance the speed of execution. Can you please add a filter of this type to EViews?

The smoother-filter utilizes the solveMCE add-in, as indicated in the forum: https://forums.eviews.com/viewtopic.php?f=10&t=21565. This example also serves as a reminder of the solveMCE routine's utility in solving forward-looking models and identifying optimal solutions, where instrumental variables minimize a specified loss function. These functionalities are essential and should ideally be incorporated as built-in features.

Best regards, Thomas