A structural network analysis of the Product Space (Hidalgo et al. (2007)), reproducing and extending the original paper's findings using Python and NetworkX.
The Product Space is a network of 774 economic products where two products are connected if they are systematically co-exported by the same countries — reflecting shared productive capabilities. The structure of this network shapes a country's ability to diversify its economy: countries can only realistically move into products that are close to their existing capabilities.
The notebook covers:
- Network description and sparsity
- Degree distribution (linear and log-log scale)
- Degree assortativity and core-periphery structure
- Community detection via Louvain algorithm (35 communities, modularity = 0.76)
- Degree and betweenness centrality analysis
- Configuration model test for clustering coefficient (Z-score = 160.20)
- Network visualization colored by community
The Product Space cannot be explained by a random null model. Its high clustering coefficient (0.43) sits 160 standard deviations above the configuration model baseline — confirming that the network reflects genuine productive capability structures, not statistical chance.
Python · NetworkX · Pandas · NumPy · Matplotlib
Coscia, M. & Hidalgo, C.A. — The Product Space (SITC classification)
Source: icon.colorado.edu
Hidalgo, C.A., Klinger, B., Barabási, A.L., & Hausmann, R. (2007). The Product Space Conditions the Development of Nations. Science, 317(5837), 482–487. https://doi.org/10.1126/science.1144581
Coscia, M. (2012). The Product Space Dataset (SITC classification). https://www.michelecoscia.com/?page_id=223
Kolic, B. (2025). UC3M Networks Workshop. https://github.com/blas-ko/uc3m_networks_workshop_2025