## DOLS- fitted values

For questions regarding programming in the EViews programming language.

Moderators: EViews Gareth, EViews Jason, EViews Moderator

muin.chowdhury
Posts: 6
Joined: Thu Feb 23, 2017 4:56 pm

### DOLS- fitted values

I have a static model that was estimated using DOLS. (1) I produce residual and fitted values. Residual by using makeresids and fitted by using fit. (2) I use Actual/fitted/residual table (drop down box in the equation object).

Then I compare the values from two sources and I find..
1. Residuals match.
2. Fitted doesn't

Not sure what to make of it.

Thanks,

Muin

EViews Glenn
EViews Developer
Posts: 2555
Joined: Wed Oct 15, 2008 9:17 am

### Re: DOLS- fitted values

From the EViews documentation:

http://www.eviews.com/help/helpintro.ht ... on.html%23

Most of the relevant issues were discussed previously (e.g., construction of residuals and gradients), however you should also note that forecasts constructed using the Forecast... procedure and models created using Make Model procedure follow the Representations view in omitting DOLS short-run dynamics.

Put differently, the fitted in the Act/Fit/Residuals are derived by taking the difference between the residuals and the actuals and thus include the short-run dynamics. The forecast/fit omit the short run dynamics.

muin.chowdhury
Posts: 6
Joined: Thu Feb 23, 2017 4:56 pm

### Re: DOLS- fitted values

Thank you Glenn to clear up the reasons for difference.

In my case the fitted values from the DOLS is an input to another estimation problem. This secondary estimation is affected considerably between the two choices ...
(1) actual - residual
or
(2) in-sample forecast that can be derived by using Eviews' forecast/fit or Model's static solution for the estimation period.

I am trying to determine what is reasonable to do and how would I defend my choice.

Thanks

EViews Glenn
EViews Developer
Posts: 2555
Joined: Wed Oct 15, 2008 9:17 am

### Re: DOLS- fitted values

If you want truly want the fitteds, use the actual minus the residuals. Those will include the short-run dynamics which are relevant for analyzing the fit.

For out-of-sample forecasting, people are generally only concerned with the long-run forecasts from the cointegrating relationship.