[R] strucchange breakpoints (Bai and Perron, 1998, 2003)

[Achim Zeileis]

On Sun, 26 Feb 2012, buehlerman wrote:

> If I try the breakpoints() function (strucchange package) with a minimum
> segment size = the number of regressors, there appears the following error
> message:
> "minimum segment size must be greater than the number of regressors"
>
> According to the documentation:
> "breakpoints implements the algorithm described in Bai & Perron (2003) for
> simultaneous estimation of multiple breakpoints."
>
> Bai & Perron (2003) assumed: minimum segment size >= number of regressors
> (and not >).
> What is the reason for this deviation?

It was introduced to make the results numerically more stable. Typically, 
neither of the restrictions will be appropriate in practice and larger 
minimal segment sizes should be used. For example, for a regression with 
an intercept and a slope, I would typically use at least h = 10 (and not h 
= 2 or h = 3) so that the parameters estimators are meaningful on each 
resulting segment.

> --
> View this message in context: http://r.789695.n4.nabble.com/strucchange-breakpoints-Bai-and-Perron-1998-2003-tp4422361p4422361.html
> Sent from the R help mailing list archive at Nabble.com.
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>