Page 1 of 1
iterative re-weighetd least squares method
Posted: Fri Sep 02, 2011 7:34 am
by tavera
Hi
has anyone already used the iteratively re-weighetd least squares method which seems to be an adequate robust method for estimating a Meta Regression model ?
I wonder if this can be implemented easily with winrats.
Many thanks in advance
Christophe Tavéra
Re: iterative re-weighetd least squares method
Posted: Tue Sep 06, 2011 10:26 am
by TomDoan
Is there a specific application that you had in mind?
Re: iterative re-weighetd least squares method
Posted: Thu Sep 08, 2011 2:42 am
by tavera
Yes, I am working on the paper "Rose effect and the euro : is the magic gone ?", T. Havranek, Review of World Economics, 146(2): 2010, 241-261.
On page 247, the author suggests estimating the equation (3) of the paper with an iterative re-weighetd least squares method.
I don't really understand what is this procedure.
Chistophe Tavera
Re: iterative re-weighetd least squares method
Posted: Thu Sep 08, 2011 9:45 am
by TomDoan
The standard RATS example ROBUST.RPF includes iterated weighted least squares for robust estimation, although, since RATS includes LAD directly through RREG, I would probably use that instead. (IWLS is often used to approximate LAD since it can be done with standard regressions, while LAD requires a specialized linear programming calculation).
Re: iterative re-weighetd least squares method
Posted: Thu Sep 08, 2011 11:50 am
by tavera
many thanks Tom
Christophe Tavéra