An alternative approach for selecting TS vs. DS processes using the Nelson and Plosser time series

Abstract

The initial study of Nelson and Plosser (1982) has been established in the literature as a
point of reference and many research papers have worked on their data set trying to determine
whether each U.S. series is generated by a trend stationary or by a difference stationary process.
The objective of this paper is to re-examine the Nelson and Plosser data set using maximum
likelihood estimation and to comment on the results based on the existing testing procedures.