In the early 1970s, Oaxaca and Blinder popularised a framework for decomposing differences between two groups attributed to observable and non-observable characteristics. A typical application of the model is the creation of a counterfactual that divides any observed gap between two exclusive sub-groups into components that are observed as characteristics of individuals and a component that contributes to the difference in the structure of outcome variables (Fortin, Lemieux and Firpo, 2011[1]). Since then, the Oaxaca-Blinder decomposition has been one of the most widely used models for understanding what may be attributed to observable and non-observable characteristics between two groups. A simplified version of their model decomposes intergroup differences in two parts. The decomposition aims to understand what part of the differences in the mean outcomes of each group: R=EYa-E(Yb) where Y are expected outcome variables for groups a and b.

The proposal on adjusting innovation indicators for the occupational structure or rural economies comes from discussions with the OECD Expert Advisory Committee for Rural Innovation. During the sessions, several rural academics identified structural problems associated with how innovation is measured in rural areas and why the bias associated may not be territorially homogenous. To address this, work by Dotzel (2017[6]) and Wojan (2021[7]) proposes an occupation-driven approach for analysing regional invention. The authors argue that patenting rates should be computed on the subset of workers that might plausibly contribute to patenting. To do this, the authors regress the aggregate number of patents produced in the commuting zone during the period 2000-05 on the share of the workforce employed in a selection of detailed census occupations. The authors’ commuting zone-level regression includes controls on the patent stock, human capital share (working-age population with a bachelor’s degree or higher), population density, a natural amenity score and the wage-rental ratio. They apply the analysis to a core set of occupations (from the U.S. Department of Labor Employment and Training Administration O*NET database) defined by the National Science Foundation’s classification of science, engineering and technical (SET) occupations, along with an iterative random selection of other occupations that may have a strong association with patenting. Ten thousand regressions are estimated with 19 non-SET occupations randomly included in each estimation. The inventive subset inclusion criteria for the non-SET occupations are those occupations-associated coefficients that are positive and significant in at least 75% of their regressions in the metro or non-metro analysis and are characterised as inventive. Of the 300 non-SET occupations included in the analysis, 11 are identified as inventive, that is consistently associated with positive, significant coefficients.

More general trends in non-patent-related innovations are explored using the EU Community Innovation Survey’s most recent iteration in 2014. Because of the lack of availability of data on a subnational level, all analysis is conducted on a national level, splitting countries with relative shares of rural populations. This limitation does not permit a full territorial analysis using self-reported measures of innovation; however, it provides a general outlook of what is observed on the aggregate level.

[6] Dotzel, K. (2017), “Three essays on human capital and innovation in the United States”, Chapter 3, Graduate School of the Ohio State University.