Structural Modeling and the Kalman Filter

[Carole Birrell <cbirrell@uow.edu.au>]

Summary of Responses 

Thank you to everyone who responded to the following query so quickly.  A
summary of the responses are below the original message.

Could anyone  tell me if  SPLUS (or R) will handle structural time series
models. The general approach of first putting the structural model into state
space form and then using the Kalman filter is what I am hoping to do. 
I can't seem to find any mention of it in the help files or the manuals.
If not, could anyone recommend any other software (I know about STAMP but
do not have it as this point). 
Carole.

SPLUS
In SPLUS, there is a module called ‘FinMetrics’ which will handle general
state space models using the Kalman filter.  S+ FinMetrics includes the Ssfpack
software, which is the same code used underneath STAMP.

References for this module are:
a)    Eric Zivot’s homepage has a survey paper.  http://faculty.washington.edu/ezivot
 
b)    The following book handles structural time series models (STSM) with
state-space representation (page 511)  see http://faculty.washington.edu/ezivot/modelingFinancialTimeSeries.htm
 
R
In R, there is a module called ‘dse’ which does state space modeling and
has both a Kalman filter and Kalman smoother.  (from ‘dse’ in CRAN.)  User’s
guide in directory dse1/inst/doc.  
NB. The code does not handle time-varying models.  It can handle non-innovations
form of state space models.

Also in R there is a package called ‘sem’ which handles structural equations.
Try StructTS() in the Help library [library (help=ts)] in the current R 1.8.1.

Other Software :  OX or RATS

Many thanks to:  Don McKenzie, Prof Brian Ripley, Eric Zivot, Erin Hodges,
Henrik Nielsen, Laurent Ferrara, Jeffrey Wang , Paul Gilbert, Fernando Tusell,
Mark Leeds, Achim Zeiles, Sven Knudsen.