Use of nonparametric statistical tests in defining the number of periods to include in an intertemporal Data Envelopment Analysis

In any type of intertemporal efficiency analyses either locally or globally, units in different periods are compared against each other, and therefore it is assumed that no frontier shift exists within the periods of study. This assumption extends in window analysis to each window. In almost all studies in the past, the number of periods included in an intertemporal data envelopment analysis or the width of a window in a window analysis is determined without validating this assumption analytically. This is a problem which has not been fully addressed in the literature. This paper presents a new approach using nonparametric statistical tests to examine a frontier shift and determine the number of periods to include in an intertemporal analysis. The case of sawmills in Vancouver, Canada is used to demonstrate how to apply this new approach.