Robustreg

[EViews Esther]

This thread is about the Robustreg add-in that perform robust regression computation by using the method of iteratively re-weighted least squares.

[Document typo] Page 1: replace n+p-1 with n-p-1

[Panny13]

Hi

I am trying to download the Robustreg Add-In but I can only download it in ".aipz" format which cannot be identified. Can you please give me some guidance?

Thank you in advance

[EViews Gareth]

You need EViews 7.1

[esquivelmm]

Hi. I downloaded and instaled the robustreg add-in. Where do I find documentation about how to use it?

Regards
Manfred.

[EViews Gareth]

From the Add-ins Menu, click on Manage Add-ins. Select the add-in you're interested in, then hit the "docs" button.

[esquivelmm]

Found it! I'm on it by now. Thanks!

[esquivelmm]

Hello, I am now running test regressions with the robustreg add-in. Single equation of this form y c x x(-1), that is y is my dependent, there is a constant and the explanatory variables are X and the first lag of X. I am using the default weight function (bisquare). When I run the add-in I get this message: "_SE01 is not found". What does this mean? I got some output but I am afraid the iterative processes might have stopped after the first iteration.

[EViews Esther]

To make a long story short, I would say that the current version of Robustreg addin cannot work with series with NA values (please note that the first element of "x(-1)" is NA).
This problem can be easily adjusted. Please reintall the Robustreg addin. Thank you.

[ROSSINL]

Hi,

Using the @expand function for including dummy variables does not work with this add-on.
The available space for specifying an equation is also too limited for me. Is there a way to solve these issues?

Thanks

[EViews Gareth]

There probably isn't any limit on the size of the expression you can use if you use the command line version, rather than the dialog.

[ROSSINL]

Thank you for the suggestion, in the command line it works.
However, the @expand function still does not work, I get the error "RSV is not defined". Bug in the program or currently not supported?

[EViews Gareth]

Just not supported.

[esquivelmm]

Hi, I am running the add-in but if I try to include several lags of an explanatory variable by using the standard command line x(-1 to -2) then I receive the message: "RSV is not identified". When I include the same lags of the explanatory but listed separately x(-1) x(-2) then it works fine.
I need to run a large number of different specifications (basically changing the lag structure) of several equatons, so it will really help if the add-in can manage the usual x(-1 to -j) list of explanatory variables

[EViews Esther]

As you mentioned, the list of variables should be specified separately: e.g. x(-1) x(-2).

[esquivelmm]

Hello, now I'm receiving the message “_SE01 is not defined” in some of my regressions. I cannot figure it out why since I am running regressions with the same data but aggregated differently. I have 1, 5, 10 and 15 minutes versions of the same data and I’ve got the problem in the 1 and 5 minutes versions. My regressions include up to 5 lags of the explanatory variables which initially was a problem that was solved (see my first post on this add-in)