Industrial Policy from a Network Perspective: Targeting, Cascades, and Resilience, with Evidence from Turkiye’s Production Network

Modern economies are networks of interdependent sectors, yet conventional tools for industrial policy overlook the critical pathways, bottlenecks, and communities that de- termine how shocks propagate and productivity gains diffuse. This paper develops a replicable computational methodology—three graph-theoretic algorithms—to transform dense input-output tables into actionable policy diagnostics. The framework identifies critical upstream and downstream pathways, constructs cascading layers of distortion propagation, and quantifies network resilience through community detection and edge- betweenness centrality. Applying this toolkit to Turkey’s 2018 manufacturing sector reveals three principal findings: finance operates as a critical bottleneck where reg- ulated upstream inputs converge; the network exhibits only moderate resilience, with six between-community edges carrying disproportionate systemic risk; and two reinforc- ing cycles—{manufacturing → agriculture → construction → manufacturing} and {manufacturing → energy → construction → manufacturing}—amplify distortions. These results generate specific policy recommendations: prioritize financial sector reforms, coordinate regulation across energy-transport-finance pathways, and protect vulnerable between-community edges. The methodology enables evidence-based, network-aware indus- trial policy applicable to any input-output dataset.