Unbalanced SUR estimation

[lsym]

Hello,

I am trying to estimate a SUR system (seemingly unrelated regressions) using an unbalanced number of observations. I would like to know whether EViews uses all observations and adjusts for the missing data or it just truncates the larger series to estimate with a balanced sample. Thanks

[EViews Gareth]

It uses all the observations. The easiest way to see this is to simply try it, and see how many observations EViews reports :D

[marthaluff]

Hi Gareth, if I start with an unbalanced dataset, is it possible to do a balanced data SUR regression?
i.e. can Eviews truncate the larger dataset for me?

[EViews Gareth]

Put all series in a group and then use the @robs function in the estimation sample.

[marthaluff]

thanks for your reply. How do I use @robs exactly? When I create a system of 2 equations for the SUR, do I add it in then? Tried that but it doesn't work...

[marthaluff]

I figured it out - for anyone else who wants to know....

1. put all your variables into a group and name it g
2. Select 'Object, New, Sample', in the top box, enter your dates (or leave as @all). In the bottom box enter @robs(g)=g.@count. You will now have a sample called smpl1.
3. Estimate your equation as you would have otherwise done, but at the bottom where it says sample and would usually have dates, instead enter smpl1. Now the regression will only use datapoints where there is a observation for all of the variables in your group.

You have turned an unbalanced sample into a balanced sample.