Panel unit root tests based on sample variance
2013 (English)Report (Other academic)
In this paper, we propose a novel way to test for unit root in a panel setting.The new tests are based on the observation that the trajectory of the cross sectional samplevariance behaves dierently for stationary than for non-stationary processes. Three dierenttest statistics are considered and their limiting distributions are derived. Interestingly, oneof the statistics has a non-standard limiting distribution which can be described in terms offunctionals of a Gaussian process. A small scale simulation study indicates that our proposedtests have good power properties, quite close to the test of Levin, Lin and Chu (2002)(LLC).However, the empirical size of one of our tests is better than LLC when T is small and N islarge, and this suggest a good property for unit root tests in micro panels. In addition, the studyalso suggests that our tests are robust to cross section dependence for a particular covariancestructure.
Place, publisher, year, edition, pages
Department of Statistics, Uppsala University , 2013. , 24 p.
Working paper / Department of Statistics, Uppsala University, 2013-3
IdentifiersURN: urn:nbn:se:uu:diva-199582OAI: oai:DiVA.org:uu-199582DiVA: diva2:620150