Misleading Inferences from Panel Unit-Root Tests : A Comment

In an effort to resolve the "all or nothing" nature of panel unit root tests, Professors Boucher-Breuer, McNown and Wallace propose the SURADF test that can determine the mix of stationary and non-stationary series in a panel. We reveal a practical shortcoming of the test. Specifically, the results of the test appear to be highly sensitive to the selection of panel members. Since member selection is often driven by somewhat arbitrary rules such as missing data or membership in an organization, this sensitivity of the test is perhaps a significant shortcoming greatly limiting both its applicability and its generality