Abstract
The predominant use of shift-share analysis has been to analyze regional changes in employment over a given time period. This article focuses on how shift-share should, rather than should not, be used. In particular, the focus is the analysis and economic interpretation of shift-share results at the industry level. We attempt to codify some of the more recent shift-share modifications and mesh them with the earlier suggestions of Dunn. There is a brief review of the popular conventional shift-share model, noting it does not provide meaningful industry-level results due to the assumption of proportionality. A shift-share model whose industry- and region-level interpretations are consistent is illustrated. The modified version also resolves, somewhat at least the structural weighting dilemma of the conventional model. Furthermore, we believe this model, in conjunction with the Barif-Knight dynamic approach, provides more robust shift-share results by enriching the economic interpretation of industry results. Finally, given our pedagogical thrust, we provide the shift-share formulae for the two models in a Lotus spreadsheet format.
Get full access to this article
View all access options for this article.