Related Variety and Regional Development: A Critique

AbstractEvolutionary approaches in economic geography have contributed substantially to the growing body of knowledge of regional development processes and their underlying mechanisms. One key concept in the literature on evolutionary economic geography is that of related variety. Herein, regional industry structure is represented through the level of related variety of technologies, skills, or outputs. The related variety concept proposes that regional economic development is favored when an economy diversifies into products or technologies that are closely related to the stock of existing activities. In this article, we raise substantive questions regarding the internal logic of the concept of related variety, its spatial expressions, measurement specifics, empirical regularities and biases, and its possible short- and long-term effects on regional development. Based on this investigation, we make suggestions for improvements to future research.Key words: economic geographies of placesevolutionary economic geography (EEG)regional developmentregional specializationrelated varietyJEL codes: L23R11 AcknowledgmentsThis article was initially presented at the 2022 Global Conference on Economic Geography in Dublin. The authors would like to thank the audience, especially Tom Broekel, for supportive comments, as well as three reviewers and editor Jim Murphy, for constructive criticism. We are also indebted to Adams Aghimien, Max Buchholz, and He (Shawn) Shuang for superb research assistance.Notes1 Other work in EEG that focuses on topics, such as regional path development, path creation, or lock-in (Coenen et al. Citation2017; Hassink, Isaksen, and Trippl Citation2019; MacKinnon et al. Citation2019), is often based on qualitative studies and does not employ the related variety concept empirically.2 While this research often emphasizes the diversification aspect over specialization, this could also be viewed from a clustering perspective.3 Content, Frenken, and Jordaan (Citation2019) are aware that the mechanisms as to how regional knowledge spillovers are created are not fully clear and that evidence needs to be provided for these effects (Content and Frenken Citation2016). An exception is the study by Miguelez and Moreno (Citation2018), which directly tackles underlying mechanisms rather than assuming them.4 See, for instance, the studies by Saviotti and Frenken (Citation2008), Boschma and Iammarino (Citation2009), and Miguelez and Moreno (Citation2018).5 We suspect that the related variety index captures a wide array of different forms of relatedness. A particularly potent example of this comes from the findings of Storper et al. (Citation2015) in analyzing high-tech industries in California. At the six-digit NAICS level, the study reveals substantial differences in wages (up to 50 percent) for occupations between two comparable metropolitan regions. This implies that, at least for some industries, even six-digit industry codes may not be very homogeneous in what they are doing on the ground across places; at worst, they may be little more than chaotic descriptive aggregations.6 To make things even more complicated, Uhlbach, Balland, and Scherngell (Citation2022) find in a study about the impact of EU research funding on new regional specializations that positive effects seem largest when the level of relatedness is neither too high nor too low. Yet, as in many other studies, it essentially remains unclear what that precisely means.7 The same results in terms of related and unrelated variety scores can be found irrespective of overall employment of the regional economy, since employment shares rather than absolute numbers are decisive for the computation. However, as revealed further down, the industry structure in large and small cities is in reality rather different, producing some bias of high related variety values toward large cities.8 In fact, in the scenario with ten clusters, related and unrelated variety scores are the same.9 As a starting point, however, see Fitjar and Timmermans (Citation2017).10 Castaldi, Frenken, and Los (Citation2015) also find a positive correlation between unrelated and related variety in their study at the US state level but are neither concerned about this relationship nor investigate it further.11 This may also explain why recent studies use indicators, such as related variety density (consisting of a quotient of different related/unrelated variety indicators—e.g., Balland et al. Citation2019), that reduce the impact of scale on the overall relatedness measure computed. While this seems an improvement, there are also concerns with indicators, such as related variety density (Uhlbach, Balland, and Scherngell Citation2022) or a combination of relatedness and complexity variables (Deegan, Broekel, and Fitjar Citation2021), since these are even more difficult to make sense of in real geography or policy terms than conventional related variety.12 See Boschma, Eriksson, and Lindgren (Citation2009), Cainelli and Iacobucci (Citation2012), Ebersberger, Herstad, and Koller (Citation2014), and Tavassoli and Jienwatcharamongkhol (Citation2016).13 While their analysis draws from a small number of regions, other studies use a similar approach with much larger sample sizes, which improves reliability.14 An investigation by Spencer et al. (Citation2010) asks similar questions but with respect to the impact of industrial clusters on regional performance. In a study of Canadian city-regions in the early 2000s, they show that city-regions with a higher employment share in clusters have a higher average income, employment growth, and patenting intensity, although they find large variations. As with the related variety concept, however, the direction of causality remains opaque. Do clusters cause such development or are they attracted to high-income regions that have higher skill levels?