The Hiroshima Economic Review Volume 39 Issue 2
2015-11-30 発行

Size Improvements In Unit Root Tests For Time Series With Serially Correlated Errors

This paper proposes a few nonparametric tests to reduce severe size distortions in the Phillips-Perron tests. It is shown that these lead to noticeable improvements by including terms that slowly converge to zero in asymptotic approximations.
Copyright (c) 2015 広島大学