Construction supply-chain failures cluster by position: brokers bear relational and contract breakdowns while peripheral suppliers absorb external shocks; targeting predictive analytics at central coordinators and simple traceability tools at peripheral nodes maximises network resilience.

Main Finding

Construction supply chain disruptions and recovery dynamics concentrate systematically by network position. Relationship and contract failures accumulate at high-centrality brokers, while external pressures disproportionately hit peripheral suppliers. Structural roles predict technology preferences: central coordinators favor predictive analytics; peripheral actors prioritize traceability systems. A network-theoretic framework that maps interactions from interview-derived nodes/edges reveals structural vulnerabilities and actionable leverage points for position-based interventions.

Key Points

• Research questions: (1) How do construction supply chain networks emerge? (2) Where are positional vulnerabilities concentrated? (3) How do network pressures steer technology adoption?
• Data collection: purposive + snowball semi-structured interviews covering all major supply chain roles (sample spans roles though not numerically specified).
• Qualitative → network: thematic coding translated reported interactions into nodes and edges and grouped operational challenges into six main categories (16 open codes).
• Network metrics used: degree, betweenness, eigenvector centralities to identify brokers, influencers, and structurally important actors.
• Quantitative descriptors reported: high-centrality brokers (degree centrality ≈ 0.818), network density ≈ 0.591 (moderate density).
• Findings on challenge concentration:
  • Relationship and contract issues concentrate at high-centrality brokers.
  • External pressures (e.g., weather, geopolitical shocks) disproportionately affect peripheral suppliers.
• Technology adoption patterns align with positional roles:
  • Central coordinators: demand predictive analytics and coordination tools.
  • Peripheral actors: prioritize traceability systems and simpler transparency tools.
• The paper presents a replicable framework for mapping structural vulnerabilities and designing position-based interventions.

Data & Methods

• Sampling: purposive and snowball sampling to capture representatives from owners, contractors, subcontractors, suppliers, logistics providers, and other supply chain roles.
• Interviews: semi-structured format to elicit interaction partners, failure modes, and technology preferences.
• Coding: thematic coding produced 16 open codes aggregated into six challenge categories, and converted reported interactions into a network adjacency structure.
• Network analysis: computed degree, betweenness, and eigenvector centralities to identify central/broker/influential actors; computed network density to characterize overall connectivity.
• Analytical approach: mixed-methods integration—qualitative themes explain where and why disruptions occur; SNA metrics pinpoint structural concentration of those themes and inform intervention targeting.

Implications for AI Economics

• Targeted AI deployment by network position:
  • Invest in predictive analytics at high-centrality coordination nodes to maximize system-wide resilience and upstream/downstream benefits.
  • Deploy lightweight traceability and provenance technologies at peripheral suppliers to reduce external-pressure spillovers and improve reliability.
• Heterogeneous ROI and adoption dynamics:
  • Adoption value of AI differs by position; economic models and diffusion studies must account for position-dependent returns and complementarities.
  • Procurement and contracting incentives should be redesigned (e.g., risk-sharing, subsidized adoption) to align peripheral actors’ incentives with network-level gains.
• Policy and market design:
  • Regulators and industry bodies can use position-based mapping to prioritize support (training, subsidies, standards) where marginal impact is highest.
  • Standards for interoperability and data sharing should focus on brokers/coordination nodes to unlock network effects.
• Modeling and welfare analysis:
  • Macro/meso models of industrial resilience and productivity should incorporate network structure (centrality, density) to capture systemic risk and benefits of AI.
  • Consider externalities from AI adoption (positive spillovers via brokers, negative displacement at some nodes) when assessing social welfare and policy.
• Research and evaluation priorities:
  • Evaluate cost-effectiveness of different AI interventions conditional on network position (predictive vs. traceability vs. coordination platforms).
  • Study dynamic network evolution, multi-project supply webs, and how adoption feedbacks reshape centrality and market structure over time.
• Implementation strategy:
  • Use the paper’s replicable mapping framework to run pilot deployments, measure differential impacts by node type, and iterate incentive and technical design.

Suggested next steps for AI economists: incorporate supply-chain network measures into adoption models, estimate position-specific adoption elasticities and welfare impacts, and design position-targeted subsidies or contracting innovations to accelerate resilient, equitable AI diffusion in construction and similar fragmented industries.