Misleading Inferences from Panel Unit Root Tests: A Comment