Stronger digital economies improve cotton supply‑chain resilience in China: a 0.10 increase in a provincial digital‑economy index predicts a 0.0452 rise in resilience (about 0.6 standard deviations), effects partly mediated by industrial upgrading and strengthened by policy attention; findings are robust to IV and robustness checks but are limited to 16 provinces and aggregated indices.

Main Finding

A stronger digital economy is positively associated with greater resilience in China’s cotton agricultural value chain. In two-way fixed-effects models using provincial panel data (2011–2023), a 0.10-point increase in the digital-economy index is associated with a 0.0452-point increase in the resilience index — roughly 0.60 of the observed standard deviation of resilience. The positive relationship holds for multiple resilience dimensions (secure development; shock resistance; recovery & adaptation; renewal & transformation).

Key Points

• Effect size: 0.0452 increase in resilience per 0.10 increase in the digital-economy index; ~0.60 SD of resilience for that change.
• Multi-dimensional outcome: Positive associations observed across subcomponents — secure development, shock resistance, recovery and adaptation, and renewal and transformation.
• Mechanism: Mediation analysis suggests partial transmission through industrial-structure upgrading (i.e., digital development helps shift the industry structure in ways that improve resilience).
• Moderation: Policy-attention measures strengthen the positive association between the digital economy and resilience.
• Nonlinearity: Threshold analyses indicate a nonlinear (interval) pattern, but thresholds are closely spaced and middle-interval coefficients are only weakly significant — implying possible diminishing or varying marginal returns to digital development across levels.
• Robustness: Results are supported by robustness checks and an instrumental-variable specification that address endogeneity concerns.
• Cautionary note: Findings support digital investment but warn against inefficient or misallocated digital spending.

Data & Methods

• Data: Panel of 16 Chinese provinces covering 2011–2023.
• Variables: Constructed composite indices for (a) digital economy development and (b) cotton-value-chain resilience (and its sub-dimensions).
• Primary model: Two-way fixed-effects regression (province and year fixed effects).
• Supplementary analyses:
  • Mediation analysis to test mechanism via industrial-structure upgrading.
  • Moderation tests using policy-attention metrics.
  • Robustness checks (alternative specifications and variable constructions).
  • Instrumental-variable (IV) estimation to address endogeneity.
  • Threshold (nonlinear) analysis to detect heterogeneous effects across digital-economy levels.
• Interpretation: Estimates are presented for a 0.10-point change in the digital-economy index for ease of interpretation; significance varies across models and intervals.

Implications for AI Economics

• Digital capacity builds resilience: Digitalization (including AI-enabled tools) can strengthen agricultural value chains by improving risk detection, coordination, production efficiency, and recovery after shocks.
• Complementarity matters: Benefits flow partly through industrial upgrading — AI and other digital technologies are more effective when combined with structural shifts (e.g., value-added processing, logistics, firm capabilities).
• Policy coordination amplifies impact: Government attention and targeted policy measures increase the effect of digital development on resilience, suggesting public support and regulation shape the returns to AI investments.
• Nonlinear returns and targeting: The threshold results imply marginal returns to digital/AI investment may vary by development stage; policy and investment should avoid one-size-fits-all scaling and focus on where additional digital capacity produces the largest resilience gains.
• Causal inference & measurement: The study uses IV and robustness checks, illustrating the importance of careful identification when claiming causal effects of digital/AI adoption on economic outcomes.
• Sector specificity: Results are sector-specific (cotton in China); translating findings to other crops, countries, or global supply chains requires caution and further empirical work.
• Research directions for AI economics:
  • Micro-level causal studies of specific AI interventions (forecasting, precision ag, supply-chain optimization) on resilience and welfare.
  • Cost–benefit analysis of digital/AI investments across development stages and regions.
  • Heterogeneity analyses by farm size, firm type, and subnational infrastructure.
  • Comparative studies across commodities and countries to assess generalizability.
  • Investigation of potentially inefficient or misallocated digital investments and policies to improve targeting.

Short takeaway: Investing in digital capacity (including AI-enabled capabilities), combined with industrial upgrading and coordinated policy, can materially improve supply-chain resilience in strategically important agricultural sectors — but returns vary by context and poorly targeted digital spending can be inefficient.