17 August 2026 in SSRN Electronic Journal
We analyze the three most recent Mexican input–output production networks at the national industry level (6-digit classification), focusing on the statistical characterization of degree, strength, and local entropy. Using these metrics, we develop a multi-centrality index to identify key industries and track their positional changes over time. The results show that the in-degree distribution follows a bimodal pattern, where agricultural industries exhibit lower mean values than the rest of the economy. Conversely, the out-degree distribution evolves toward a beta-like form with parameters 1.42 < a < 2.09 and 0.155 < b < 0.219. Edge weights display a dual structure: low and medium values follow a stretched exponential (0.199 < β < 0.233), while the tail follows a power law with exponents around 2.3. Strength distributions also show a dual pattern, combining stretched exponential or lognormal behavior for small and medium values and power-law tails with 2 < α < 2.5. Local entropy distributions are well fitted by compressed exponentials, with 6.6 < β < 7.5 for in-entropy and 3.3 < β < 3.6 for out-entropy. Finally, the top 50 industries ranked by the out-multi-centrality index remain stable across periods, while the in-multi-centrality index shows greater variability. Sectorally, construction and manufacturing dominate the in-direction in 2008, whereas public administration, educational services, health care and nonresidential construction occupy the highest ranks in 2013 and 2018. These findings reveal a structural shift in Mexico’s production network, highlighting the growing relevance of service-based activities and state-driven sectors in recent years.