Download to read offline
Uploaded byOECD CFE
Economic complexity analysis: a policy framework - Neffke
Presentation by Frank Neffke, Team leader of the Science of Cities program, Complexity Science Hub, Austria at the 26th OECD Spatial Productivity Lab meeting "Smart moves: How data can help discover what regions could be great at" held online on 23 October 2025. More info: https://oe.cd/6dd
More Related Content
Similar to Economic complexity analysis: a policy framework - Neffke
Economic complexity analysis: a policy framework - Neffke
- 1. Economic Complexity Analysis Apolicy framework Meeting of the Spatial Productivity Lab October 23, 2025 Frank Neffke Complexity Science Hub
- 2. An economic complexityworld Three central assumptions 1. Many input factors or “capabilities” (Hausmann & Hidalgo, 2011) 2. Strong complementarities between these capabilities (Gomez- Lievano et al., 2016) 3. Complexity increases with the number of and complementarities between capabilities Metaphor: Scrabble • Capabilities are like letters • Products are like words
- 3. PRODUCTS Top Hat: ByNikodem Nijaki (Own work) [CC BY-SA COUNTRIES CAPABILITIES
- 4. The world aswe observe it COUNTRIES PRODUCTS
- 5. The world aswe observe it COUNTRIES PRODUCTS
- 6. The product space COUNTRIESPRODUCTS
- 7. metallurgy vegeta bles fishing garments coffee andcocoa products fruits 7
- 8. Emerging related diversificationliterature • Principle of relatedness: Hidalgo et al. (2018) • Products in export baskets: Hidalgo/Klinger/Barabasi/Hausmann 2007 – Science; Bahar/Hausmann/Hidalgo 2014 – JoIE • Industries in regions: Neffke/Henning/Boschma 2011 – EG, Delgado/Porter/Stern 2014 – ResPol; Zhu/He/Zhou 2017 – JoEG, … • Innovation in cities: Rigby 2015 – RS; Boschma/Balland/Kogler 2017 – ICC; … • Science fields in cities: Boschma/Heimeriks/Balland 2014 – ResPol; Guevara/Hartmann/Aristarán/Mendoza/Hidalgo 2016 – Scientometrics • Methodological papers • Li & Neffke (2024 - ResPol) • Beyond RCAs: Van Dam et al. (2023 – JoRS) • “Genotypic”approach to ECA: inferring the Ppa and Cca matrices (Diodato et al., 2024 – work in progress) • Policy frameworks (e.g., smart specialization) • Balland/Boschma/Crespo/Rigby 2018 – Regional studies • Rigby/Roesler/Kogler/Boschma/Balland 2022 – Regional studies • Green product spaces: e.g., Mealy and Teytelboym 2022 – Research Policy 8
- 9. The real resolutionof economic activity Metroverse
- 10. From product spaceto industry space Dun & Bradstreet World Base • Second approach • Dun and Bradstreet World Base • Large-scale global company registry • Entire economy, not just exported products • Number of establishments in dataset: • 2011 batch: 124,816,570 • 2016 batch: 121,019,655 • 2019 batch: 148,315,749 • 2020 batch: 170,000,618 • Variables: • Consistent industry codes: up to 6 NAICS (2019, 2020) or SIC (all batches) codes • Geocoded • Estimated number of employees • Total sales (not fully reliable) • Domestic and ultimate global owner Yang Li
- 11. City-industry level aggregates Citydefinitions taken from EU’s JRC • Establishments are merged through spatial joins to city polygons • Currently, we use the urban core definition, others can be implemented Middlesex County boundary Boston urban core boundary Boston functional urban area boundary
- 12. Industry space (co-occurrences inestablishments)
- 13. Understanding a city’sspecialization and its“adjacent possible” New York City Chicago
- 14. Shenzhen-Guangzhou Hong Kong Understandinga city’s specialization and its“adjacent possible”
- 15. Software: Stack Overflow Openresource for support in coding • Established 2008 • Alexa top 100 website at its peak. • 6M users • Locations (50k+) • 280+ programming languages • Content • Questions (23M) • Answers (33M) • 60k Tags
- 16. • In the2022 snapshot, out of ~6 million active users we can geolocate: • 4.0 million to a country. • 2.8 million to a FUA.
- 17.
- 18. Atlases of economiccomplexity Regions’ position in abstract activity spaces Metroverse (Vienna) Software task space (Data Science in Europe)
- 19. ECA and policy Ifit’s so easy to develop this, why hasn’t the region done so yet? • Possibility 1: • The region hasn’t found the time to enter the activity yet • Possibility 2: • The region has recently changed a lot and only now has come close by • Possibility 3: • There are (observable or unobservable) barriers to entry
- 20. ECA and policy Anomalydetection (Li and Neffke, 2024) • Sometimes, • opportunities are real opportunities (e.g., guided missiles in Seattle, home to Boeing) • In other cases, • they are an expression of a missing crucial capability (e.g., deep sea transportation in landlocked Arizona) • Use anomalies to learn about the missing capabilities, • complementing ECA with in-depth analysis of a region ECA as part of a Growth Diagnostics framework
- 21. Conclusion • Using ECAto inform regional development policy • Diversification trajectories are predictable • Study the “adjacent possible” • Three ways to apply ECA in development policy • Map a country’s options • Industry spaces: map capability base • Complexity: benchmark development against productive portfolio • Find “low-hanging fruit” of nearby, un(der)developed activities • Easier: weed out targets based on wishful thinking • Best combined with smart objectives: good jobs, green, digital, etc. • Anomaly detection • Understand what is hindering seemingly easy development trajectories • Integrate into wider diagnostics framework