Drop insignificant variables and reeestimate equation

[miksterz1]

Hello,

I am running a loop that estimates an equation for each of 29 series. Amongst the regressors are 12 seasonal dummies(@seas) since I am working with non-seasonally adjusted monthly data. Having estimates these equations, I would like to write some code that drops those seasonal dummies in each equation that are not statistically significant, and re estimates the equations accordingly.

Any guidance would be appreciated. At the moment I'm not sure what sort of commands I should be looking in to.

[EViews Gareth]

Use the built in stepwise estimation routine.

[miksterz1]

Thanks, Gareth. I've got it working but have one problem. I want to use @spec to save the specification of each estimation. However, when I use @spec after the stepwise function, it saves the specification that includes all the search regressors as opposed to the final selected regressors that I am interested in. Any way around this?

[EViews Gareth]

I'm in the middle of the Peruvian jungle and don't have EViews handy, but you could try @varlist. I that doesn't work, you'll have to hope Glenn has something clever :D

[EViews Glenn]

@varlist it is...

The @spec contains what you specified, the @varlist contains the information associated with the equation you ended up estimating.

As for the Peruvian jungle, we don't believe him. We're convinced it's Photoshop.

[miksterz1]

Thanks, that worked perfectly. Enjoy the jungle!