Andy
The SUR and NLSUR functions in finmetrics do not implement a HC or HAC
covariance matrix estimator. If you want to do it, you have two options
1. program it yourself. This is relatively straightforward and you could use
the finmetrics function var.hac() to do the main kernel estimation. All you
need is to pass the sur moment equations to this function.
2. Write the sur model as a GMM model (as in Hayashi's textbook) and use the
GMM funciton to estimate it.
-----Original Message-----
From: s-news-owner@lists.biostat.wustl.edu
[mailto:s-news-owner@lists.biostat.wustl.edu] On Behalf Of Andy Cooper
Sent: Friday, May 09, 2008 9:03 AM
To: s-news@lists.biostat.wustl.edu
Subject: [S] SUR() and NLSUR()
Hi all,
I'm using Splus 7.0 with Finmetrics 2.0 runing in Windows XP Professional.
I've searched the s-news archives and have not found an answer to either
question.
1) Is there any way to apply a Newey-West HAC covariance matrix estimate to
a SUR (or a NLSUR) regression model? (N.B. This question was asked over in
the r-help list back in Oct. 2007, but it did not appear that methods were
available there)
2) The formula list for SUR() does not permit the inclusion of ar() terms
but does let me use a tslag(). E.g.:
y1 ~ x1 + tslag(y1,1)
y2 ~ x2 + tslag(y2,1)
y3 ~ x3 + tslag(y3,1)
But, just because I'm allowed to do it, doesn't mean I should. Am I
violating any of the SUR or NLSUR assumptions by doing this? Are the
statistical tests still valid?
Any advice you might have would be hugely appreciated!
cheers,
Andy
--------------------------------------------------------------------
This message was distributed by s-news@lists.biostat.wustl.edu. To
unsubscribe send e-mail to s-news-request@lists.biostat.wustl.edu with the
BODY of the message: unsubscribe s-news
|