For best experience please turn on javascript and use a modern browser!
You are using a browser that is no longer supported by Microsoft. Please upgrade your browser. The site may not present itself correctly if you continue browsing.

2.1 Functional Structure in Production Networks

Carolina Mattsson, Frank Takes, Eelke Heemskerk, Cees Diks, Gert Buiten, Albert Faber and Peter Sloot

Keywords: Production networks; Inter-firm networks; Complexity economics; Economic statistics; Trade linkages; Functional networks; Bipartivity.

Author Carolina Mattsson on the article: 

Industrial clusters are some of the toughest areas of an economy to decarbonize and so are of particular concern for policymakers in regards to the energy transition. This paper comes out of a multi-institutional collaboration where we asked: what is the structure of production in industrial clusters? A better understanding of this question can help develop policy to accelerate the energy transition, and so we focused our scientific efforts on the local structure of production networks.

Production networks are well-studied at an aggregated level, considering trade linkages between industries as published in “input-output tables”. These are produced by many national statistics offices as part of national economic statistics. Work based on such data has taken us far, and we know that production networks are integral to economic dynamics over the short-, medium-, and longterm1–3. At the same time, there are compelling reasons to turn our focus towards dis- aggregated networks of trade relationships among companies4, 5. Customer-supplier ties are where mechanisms act, including price setting, supply constraints, and policies promoting decarbonization.

In some countries, notably Japan, there have been efforts to collect networks of customersupplier ties and study their structure6, 7. In most countries of the world, however, dis-aggregated networks are not available. Luckily for us, we don’t necessarily need precise empirical data to answer our question. Indeed, a qualitative understanding of a networks’ local connectivity structure is likely to be just as useful in the hands of a policymaker who already has substantive local knowledge.

Local connectivity structure is an especially useful way to categorize networks. Random networks, social networks, and two-mode networks are well-known network types with distinct local connectivity structures8–10. Recently, a fourth distinct network type has been identified in work on protein-protein interaction (PPI) networks. Specifically, squares feature more prominently than triangles in the local connectivity structure of PPI networks11, 12. In this work, we suggest and find that company-level production networks are best characterized as so-called functional networks. Because networks share important structural features with other networks of the same type, we can draw insights across domains: companies trade with complementary others and are especially similar to their close competitors, not their trading partners.

Our argument is broad, and we draw on several sources of evidence. First, we re-examine findings from studies out of Japan and note that empirical networks have relatively few trianglesand relatively many squares6. Then, we consider a regional and a national company-level production network reconstructed from Dutch national economic statistics13. We develop a clear hypothesis test whereby a network of person-person friendships14 has “social” structure (more triangles) while a network of protein-protein interactions11 has “functional” structure (more squares). Using this test, we find functional structure in both company-level production networks!

Our results have practical implications for the structure of production in industrial clusters and wider implications for network science and complexity economics. Policymakers should expect relatively little direct trade among large companies in a local area (i.e. disassortativity), but considerable indirect dependence via shared customers and suppliers (i.e. squares). In designing transformative policy, it may help to consider that it takes groups of companies (i.e. modules) to perform higher-level economic functions. More generally, we have demonstrated the usefulness of network categorization according to local connectivity structure for network science and complexity economics. 

References

1. Acemoglu, D., Carvalho, V. M., Ozdaglar, A. & Tahbaz-Salehi, A. The Network Origins of Aggregate Fluctuations. Econometrica 80, 1977–2016 (2012).
2. McNerney, J., Savoie, C., Caravelli, F. & Farmer, J. D. How production networks amplify economic growth. arXiv:1810.07774 [physics, q-fin] (2018).
3. Pichler, A. & Farmer, J. D. Simultaneous supply and demand constraints in input–output networks: the case of Covid-19 in Germany, Italy, and Spain. Econ. Syst. Res. 0, 1–21, DOI: 10.1080/09535314.2021.1926934 (2021).
4. Carvalho, V. M. & Tahbaz-Salehi, A. Production Networks: A Primer. University of Cambridge (2018).
5. Inoue, H. & Todo, Y. Firm-level propagation of shocks through supply-chain networks. Nat. Sustain. 2 (2019).
6. Ohnishi, T., Takayasu, H. & Takayasu, M. Network motifs in an inter-firm network. J. Econ. Interact. Coord. 5, 171–180, DOI: 10.1007/s11403-010-0066-6 (2010).
7. Fujiwara, Y. & Aoyama, H. Large-scale structure of a nation-wide production network. The Eur. Phys. J. B 77, 565–580, DOI: 10.1140/epjb/e2010-00275-2 (2010).
8. Newman, M. E. J., Strogatz, S. H. & Watts, D. J. Random graphs with arbitrary degree distributions and their applications. Phys. Rev. E 64, 026118 (2001).
9. Rivera, M. T., Soderstrom, S. B. & Uzzi, B. Dynamics of Dyads in Social Networks: Assortative, Relational, and Proximity Mechanisms. Annu. Rev. Sociol. 36, 91–115 (2010).
10. Borgatti, S. P. & Everett, M. G. Network analysis of 2-mode data. Soc. Networks 19, 243–269 (1997).
11. Kovács, I. A. et al. Network-based prediction of protein interactions. Nat. Commun. 10, 1240 (2019).
12. Kitsak, M. Latent Geometry for Complementarity-Driven Networks. arXiv:2003.06665 (2020).
13. Hooijmaaijers, S. & Buiten, G. A methodology for estimating the Dutch interfirm trade network, including a breakdown by commodity. Tech. Rep., Statistics Netherlands (2019).
14. Sapiezynski, P., Stopczynski, A., Lassen, D. D. & Lehmann, S. Interaction data from the Copenhagen Networks Study. Sci. Data 6, 315, DOI: 10.1038/s41597-019-0325-x (2019).

Figure 1. Toy example networks with seven nodes and eleven edges, each of a different type, shown in increasing order of spectral bipartivity. Functional networks have higher-than-random spectral bipartivity.

Mattsson, C. E. S., Takes, F. W., Heemskerk, E. M.,  Diks, C., Buiten, G., Faber, A. and Sloot, P. M. A. (2021) Functional Structure in Production Networks. Frontiers in Big Data 4, 666712. https://doi.org/10.3389/fdata.2021.666712