An embedded and charset-unspecified text was scrubbed... Name: not available URL: <https://stat.ethz.ch/pipermail/r-sig-finance/attachments/20090317/0a1a19a8/attachment.pl>
Unit Root Tests: Empirical Results vs Theory
2 messages · Adams, Zeno, Matthieu Stigler
Adams, Zeno a ?crit :
I am currently analyzing a data set using panel cointegration. Like with any other cointegration method the variables have to be non-stationary. For my specific case (German commercial real estate market) I find the variables "total stock of office space" and "vacancy rates" to be non-stationary using standard unit-root tests.
From theoretic considerations however it is clear that the vacancy rate has to be between 0 and 1 and also that the office space is unlikely to wander off to any arbitrary values.
how would you deal with such a situation? Do you think it is possible to have "technical non-stationarity" in the sense that shocks in the past have a permanent impact on the series but that the variable exhibits a random walk behavior only within certain bounds? If that would be the case then it would be necessary to treat the vacancy rate as a non-stationary variables since ignoring this would lead to spurious regression.
The interesting concept that a variable is non-stationary within bounds and stationary outside (and so globally stationary) can be tested, using non-linear unit root tests. There are two of such tests (unit root vs stationary threshold autoregressive process) in the dev version of package tsDyn: BBCTest() and KapShinTest() (http://code.google.com/p/tsdyn/wiki/ThresholdCointegration) They may be nevertheless not so useful as if you find that the variable is indeed globally stationary, you may nevertheless not know what do do with it!! It is not clear and few discussed in the literature whether the fact that the variable is globally stationary does not lead to spurious regression. Note also that the concept of integrated (I(0) and I(1)) time series is not well defined in the case of non-linear time series (see Gonzalo and Pittarakis threshold effect in cointegrating relationships 2006). Mat
Thank you for your contributions EBS European Business School gemeinnuetzige GmbH - Sitz der Gesellschaft: Wiesbaden, Amtsgericht Wiesbaden HRB 19951 - Umsatzsteuer-ID DE 113891213 Geschaeftsfuehrer: Prof. Dr. Christopher Jahns, Rektor/CEO; Dr. Reimar Palte, Kanzler/CFO; Sabine Fuchs, CMO; Verwaltungsrat: Dr. Hellmut K. Albrecht, Vorsitzender [[alternative HTML version deleted]]
_______________________________________________ R-SIG-Finance at stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-sig-finance -- Subscriber-posting only. -- If you want to post, subscribe first.