Estimating population change with a two-category shift-share model

Shift-share analysis is an accounting procedure that identifies three separate effects for regional employment change. But the analysis is ordinarily restricted to only one category: usually, industry employment. This paper presents a new shift-share model that simultaneously addresses both occupation employment and industry employment. Community-level occupation-in-industry employment effects (both mix and competitive) occurring in the 1980s are then used to estimate population change in the 1990s. Separate estimates are given for two data sets comprised of different-sized non-metropolitan U.S. communities – small towns and micropolitan centers. This expanded two-category model is shown to generate estimates that are clearly superior to those of the traditional one-category model.The first version of this paper was presented at the annual meeting of the Association of Pacific Coast Geographers in September, 2001. The senior author has benefited from various conversations with Kingsley Haynes regarding shift-share analysis. The authors also thank three referees and the editor for their comments on an earlier version of the paper.