Economic complexity of cities and its role for resilience

Abstract

The aim of the paper is to propose the construction of an index that captures the economic complexity of cities over the globe, as well as to explore whether it is a good predictor for a range of city-level economic outcomes. This index aspires to mitigate data scarcity for cities and to provide policy makers with the tools for monitoring the evolving role of cities in the global economy. Analytically, we implement the economic complexity methodology on data for the ownership, location and economic activities of the world's 3,000 largest firms and their subsidiaries to propose a new indicator that quantifies the network of the largest cities worldwide and the economic activities of their globalized firms. We first show that complex cities are the highly diversified cities that host non-ubiquitous economic activities of firms with global presence. Then, in a sample of EU cities, we show that complex cities tend to be more prosperous, have higher population, and are associated with more jobs, human capital, innovation, technology and transport infrastructure. Last, using OLS methodology and accounting for several other confounders, we show that a higher ECI, at the city level, enhances the resilience of cities to negative economic shocks, i.e., their ability to bounce back after a shock. Specifically, we find that the expected increase of the ratio of employment in 2012 over 2006 is 0.01 (mean: 0.992; standard deviation: 0.081) when the ECI increases by 1 unit (mean: 0.371; standard deviation: 1.094), i.e., a satisfactory pace of recovery, in terms of employment. The ability to diversify in the presence of a shock, the reallocation of factors of production to other sectors and the ability to extract rents associated with those diversified activities, uncovers the mechanics of the ECI index.

Fig 1. The city-activities of globalized firms network.

[a] The bipartite network of cities and economic sections (NACE Rev. 2). [b] The adjacency matrix using the economic activities at NACE Rev. 2 (2-digit). The figures have been generated using R 4.1.2.

Fig 2. Cities’ economic complexity (ECI), diversity and average ubiquity.

[a] ECI vs Diversity; [b] ECI vs Average ubiquity. The figures have been generated using StataSE 14.

Fig 3. Cities economic complexity index (ECI) across the largest cities.

Cities depicted in dark orange and big circle have a high ECI value (Data for 2010). The figure has been generated using Qlik Sense. ©OpenSTreetMap contributors.

Fig 4. Cities economic complexity index (ECI) in Europe.

Cities depicted in dark orange and big circle have a high ECI value (Data for 2010). The figure has been generated using Qlik Sense. ©OpenSTreetMap contributors.

Fig 5. The economic activities of London (GB)’s firms with global presence.

Paris ranks second in the economic complexity index of cities in 2010 (ECI: 5.394). The figure has been generated using Qlik Sense.

Fig 6. The economic activities of Paris (FR)’s firms with global presence.

Paris ranks second in the economic complexity index of cities in 2010 (ECI: 5.314). The figure has been generated using Qlik Sense.

Fig 7. The economic activities of Ibiza (ES)’s firms with global presence.

Ibiza ranks 580th in the economic complexity index of cities in 2010 (ECI: -0.407). The figure has been generated using Qlik Sense.

Fig 8. The economic activities of Kozani (GR)’s firms with global presence.

Kozani ranks 1100th in the economic complexity index of cities in 2010 (ECI: -0.763). The figure has been generated using Qlik Sense.

Fig 9. Cities economic complexity (ECI), population and development.

[a] ECI vs Population; [b] ECI vs GDP. The figures have been generated using StataSE 14.

Fig 10. Cities economic complexity (ECI), employment, education and innovation.

[a] ECI vs Total employment; [b] ECI vs Tertiary education; [c] ECI vs Patents. The figures have been generated using StataSE 14.

Fig 11. Cities economic complexity (ECI), digital and transport infrastructures.

[a] ECI vs Internet infrastructure; [b] ECI vs Transportation accessibility; [c] ECI vs Transportation performance. The figures have been generated using StataSE 14.

Fig 12. Correlation matrix.

The figure has been generated using StataSE 14.

Fig 13. Employment resilience and ECI.

Employment resilience measured using the index developed by ESPON [69]: 1 = resistant to the economic crisis; 2 = recovered from the economic crisis; 3 = not recovered but experienced economic upturn; 4 = not recovered and still experiencing economic downturn. The figure has been generated using StataSE 14.

Fig 14. GDP resilience and ECI.

GDP resilience measured using the index developed by ESPON [69]: 1 = resistant to the economic crisis; 2 = recovered from the economic crisis; 3 = not recovered but experienced economic upturn; 4 = not recovered and still experiencing economic downturn. The figure has been generated using StataSE 14.