A Review and Comparison of Shift-Share Identities

Abstract

Shift-share analysis is a means of decomposing change. Several alternative shift-share formulations appear in the literature. This research compares seven shift-share models through intra-model component correlations. The widely cited Esteban-Marquillas model performs less well under this test than do the classic model and two of its other derivatives. The Esteban-Marquillas model has other previously undetected weaknesses: 1) some components should logically sum to zero when all sectors and regions are added together, but do not; and 2) some components produce unrealistically high values for small volatile sectors.