shift-share analysis, shift-share technique This is a frequently used technique in the analysis of changing occupational distributions and employment growth. It begins from an observed change at the aggregate level and then attempts to decompose this into three distinct components in order to understand the cause of the change.

For example, suppose that we observed an increase in the number of professional employees in Britain from 200,000 in 1911 to 1,200,000 in 1991, and wanted to know how much of this increase of 1,000,000 professionals was due to the changing industrial structure (differences in the relative growth and decline of different industries) in the past 80 years, and how much was related to technical changes in industry which favour the growth of some kinds of occupations over others. We can first ask how many professional employees there would have been in 1991, if their proportion within each industrial sector had remained constant, while the relative size of industrial sectors had changed in the way which it actually did between 1911 and 1991. This could be called the ‘industry-shift effect’. Effectively we are asking what share of the change in the numbers of professional employees is due to changes or shifts in the industrial structure—hence the term ‘shift-share analysis’. Let us say that this calculation of the industry-shift effect indicated that there would have been 600,000 professional employees in 1991, simply due to changes in the industrial structure; that is, that the number of professionals tripled because the industries most likely to employ them grew in size, relative to other industries. Next we can play a second counterfactual game, by reversing the original proposition and asking how many professionals there would have been if their proportion in each industry had changed the way it actually did, but the relative size of each industrial sector had been constant between 1911 and 1991. This could be called the ‘occupation-shift effect’; that is, the change resulting from the changing use of professionals in industry. Let us say that this calculation produces the answer that there would have been 300,000 more professionals in the labour-force due to changes in their use by industry. The third component—the ‘interaction-shift effect’—is a residual term and represents the simultaneous effects of industry and occupation shifts or changes. Since industry-shift plus occupation-shift plus interaction (must) equal observed changes, in our example the interaction effect is 1,000,000 (observed change) minus 600,000 (industry-shift) minus 300,000 (occupation-shift), which equals 100,000. That is, the joint effects of industrial and occupational change boosted the number of professionals by a further 100,000. In this case, therefore, shift-share analysis tells us that the main reason we now have more professionals is because of changes in the industrial structure. In other words, in the industries which have increased in size, professional employees are more concentrated (industry-shift). However, the number of professional employees has also increased (though to a lesser degree), because all industrial sectors now tend to use more professionals than in the past (occupation-shift).

This technique could, of course, be applied to other data. For example, we could ask how much of the employment change in a particular region is due to national factors (a national-shift effect); how much due to the different industry mix between the region and the nation (an industry-mix shift effect); and how much is due to different growth-rates between the same industries at the regional compared with the national picture (regional-shift effect).