Hello, Im using strucchange package in R software in order to apply Bai and Peron (1998, 2003) structural break tests to a set of n=1671 observations with a constant term (no AR terms). For that purpose I have read several papers, for instance Validating Multiple Structural Change Models An Extended Case Study, in which its aim is to replicate the results from Bai and Perron (2003) in R software. But, as far as I had understood, the rule of thumb for selecting the number of breaks in data is the BIC (or RSS) criteria. However, that kind of procedure is often downward-biased. Bai and Perron (1998, 2003) argue in favour of a sequential procedure: a) First select a Sup F type test of no structural break (m=0) versus m=k breaks; b) Often, an investigator wishes not to pre-specify a particular number of breaks to make inference. To allow this BP have introduced two tests of the null hypothesis of no structural break against an unknown number of breaks given some upper bound M. These are called the double maximum tests; c) BP proposed a test for l versus l+1 breaks, labelled supFT (l+1/ l). The method amounts to the application of (l+1) tests of the null hypothesis of no structural change versus the alternative hypothesis of a single change. Questions: 1) I think it is possible to make a) and b) in the strucchange package. In a) one would simply make call of Fstats function whereas in b) one would call the efpFunctional. Am I right? How it really works in practice? 2) However, I think there is no code in the strucchange package for the test in c). Could you help me, please? Kind Regards, Joo Boavida "Confidencialidade: Esta mensagem (e eventuais ficheiros anexos) destinada exclusivamente s pessoas nela indicadas e tem natureza confidencial. Se receber esta mensagem por engano, por favor contacte o remetente e elimine a mensagem e ficheiros, sem tomar conhecimento do respectivo contedo e sem reproduzi-la ou divulg-la. Confidentiality Warning: This e-mail message (and any attached files) is confidential and is intended solely for the use of the individual or entity to whom it is addressed. lf you are not the intended recipient of this message please notify the sender and delete and destroy all copies immediately." [[alternative HTML version deleted]]
I’m using strucchange package in R software in order to apply Bai and Peron (1998, 2003) structural break tests. This sequential procedure includes: 1) Sup F test; 2) double maximum tests and 3) supFT (l+1/ l) tests. Could anyone help me to perform such tests in R, please? Kind Regards, João "Confidencialidade: Esta mensagem (e eventuais ficheiros anexos) é destinada exclusivamente às pessoas nela indicadas e tem natureza confidencial. Se receber esta mensagem por engano, por favor contacte o remetente e elimine a mensagem e ficheiros, sem tomar conhecimento do respectivo conteúdo e sem reproduzi-la ou divulgá-la. Confidentiality Warning: This e-mail message (and any attached files) is confidential and is intended solely for the use of the individual or entity to whom it is addressed. lf you are not the intended recipient of this message please notify the sender and delete and destroy all copies immediately." [[alternative HTML version deleted]]
On Mon, 19 Jul 2010, Jo?o Boavida wrote:> I???m using strucchange package in R software in order to apply Bai and > Peron (1998, 2003) structural break tests. This sequential procedure > includes: 1) Sup F test; 2) double maximum tests and 3) supFT (l+1/ l) > tests.Thanks for your interest, but you have sent this e-mail twice to me personally and twice to the list which is three e-mails too many. Either send one e-mail to me personally, or send an e-mail to the list with cc to me. See the posting guide for details.> Could anyone help me to perform such tests in R, please?The "strucchange" package implements the supF test for 0-vs-1 breaks (in BP terminology). The 0-vs-m and l-vs-l+1 tests (in BP terminology) are not implemented. For choosing the number of breaks, I typically employ a slightly different strategy: - perform a structural change test graphically (conveying information about the number and timing of potential breaks), - estimate breakpoints for m = 1, ..., M. - visualize associated RSS and BIC (or the modified BIC called LWZ), -> choose number of breaks. A paper showing this strategy in practice is: Achim Zeileis, Christian Kleiber, Walter Kr?mer, Kurt Hornik (2003). Testing and Dating of Structural Changes in Practice. Computational Statistics & Data Analysis, 44(1-2), 109-123. A preprint version is available on my webpage. Another paper which essentially uses the same idea but in a somewhat updated setting is Achim Zeileis, Ajay Shah, Ila Patnaik (2010). Testing, Monitoring, and Dating Structural Changes in Exchange Rate Regimes. Computational Statistics & Data Analysis, 54(6), 1696-1706. The preprint version is also on my webpage and all replication files are in the package "fxregime". hth, Z