Shift-Share Analysis as a Linear Model

Abstract

Shift-share analysis continues to be popular among geographers, regional scientists, and planners despite widespread criticism of the method. In this paper, it is argued that insufficient attention has been paid to model-based approaches to shift—share analysis. It is shown that conventional shift—share and stochastic shift—share yield identical conclusions. Stochastic shift—share is easily extended dynamically and along the lines suggested by Arcelus. Thus, in the stochastic models several of the most persistent criticisms of the technique are addressed by allowing for testing of hypotheses, while preserving the practicality of the conventional accounting approach. It is suggested that stochastic shift—share should be used whenever practical.