[R] breakpoints

[Achim Zeileis]

On Tue, 28 Oct 2014, Erasmo Papagni wrote:

> Hello,
>
> Using the package strucchange to implement Bai-Perron methods I found 
> that sometimes intervals between 2 break dates are exactly equal to the 
> minimum that I set in the command "breakpoints". I think this is a 
> problem and a signal of possible different breaks with a trimming value 
> lower than the one I specified.

Yes, that's possible. You can easily explore, though, how the results 
change if you require a smaller minimal segment size. The idea is that the 
minimal segment size needs to be big enough to lead to reliable parameter 
estimates but small enough to be below the smalles true segment size. 
Then, if the minimal segment size is neither too small or too large, then 
the results shouldn't be affected (much) by the precise value of the 
minal segment size.

> This case arises in Bai-Perron 2003 when they apply their methodology to 
> the Phillips curve: the distance between the dates 1967 and 1975 is 8 
> years, and is equal to the minimum allowed.

Yes, however in help("PhillipsCurve", package = "strucchange") we also 
explore some additional specifications with a smaller value which still 
yields the same breakpoints.

However, I think that this particular data set and model are not bad but 
it's probably fair to say that they are not the most convincing case for 
structural change techniques either.

Best,
Z

> Thanks a lot
>
> Erasmo Papagni
>
> Departemnt of Economics
> University of Naples II
>
> ______________________________________________
> 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.
>