Digital talent and industrial digitalization spur regional growth only up to a point: across Chinese provinces their combined impact follows an inverted-U, implying diminishing (and eventually negative) returns to excessive clustering—except in the Yangtze River Delta, where coordinated digitalization and talent agglomeration still drive growth.

Main Finding

• Digital talent agglomeration and industrial digitalization each affect provincial economic growth in China with an inverted‑U (nonlinear) relationship consistent with the Williamson hypothesis: beneficial at low–moderate levels but generating diseconomies past a threshold.
• The coordination (conjugate state) between digital talent agglomeration and industrial digitalization matters: most of China (nationally, and in eastern, central, and western regions) shows a deviation from a coordinated/conjugated state, while the Yangtze River Delta exhibits a positive, interactive, collaborative relationship between the two that strengthens regional growth.

Key Points

• Conceptual framing:
  • “Digital talent” = workforce able to deploy, integrate, and innovate on AI, IoT, big data, blockchain and related systems.
  • “Industrial digitalization” = the application of digital technologies across traditional industries to raise output and efficiency (distinct from digital industrialization).
• Three hypotheses tested:
• Both digital talent agglomeration and industrial digitalization have inverted‑U effects on regional economic growth.
• They promote growth primarily by raising the overall digital economy development level.
• Positive growth effects occur only when talent agglomeration and industrial digitalization are in a conjugated (coordinated/collaborative) state.
• Empirical conclusion: moderate, coordinated agglomeration is desirable; over‑concentration produces congestion, rising transaction costs, and negative spillovers.
• Regional heterogeneity: only some regions (notably the Yangtze River Delta) have achieved the conjugate, mutually reinforcing state that maximizes growth benefits.

Data & Methods

• Data:
  • Panel of 30 Chinese provinces/municipalities (Tibet, Hong Kong, Macao, Taiwan excluded), 2007–2022.
• Measurement:
  • Digital talent agglomeration and industrial digitalization indices constructed using location‑entropy methods.
    • Digital talent: location entropy of employment in IT/services and in electronic equipment manufacturing.
    • Industrial digitalization: estimated added value of digitalization in traditional industries derived from input–output relationships to digital core sectors.
  • Missing input–output years filled using the RAS (ratio adjusted squares) method; digital core industries mapped to national statistical classifications.
  • A composite digital economy development index (digital infrastructure, digital industries, external digital environment) is used as a mediating variable.
• Econometric approach:
  • Panel regressions of log provincial GDP on the digital talent index and industrial digitalization index including quadratic terms to capture inverted‑U effects.
  • Interaction terms and a conjugate‑effect framework are used to assess coordinated development between talent and industrial digitalization.
  • Control variables include at least unemployment rate and urbanization rate (and additional standard regional controls as employed in the study).
• Robustness:
  • The paper applies region‑level subsample analyses (e.g., Yangtze River Delta, eastern/central/western regions) to identify heterogeneity.

Implications for AI Economics

• Talent–technology co‑evolution is central: growth gains from AI and related technologies depend not only on technology diffusion but on synchronous development of appropriately skilled labor. Policies must treat talent supply as a strategic production factor.
• There are optimal agglomeration thresholds: unbounded concentration of AI/digital skilled labor or industry can create negative externalities. AI‑economics models should incorporate nonlinear (inverted‑U) effects and congestion/externality costs from over‑agglomeration.
• Coordination matters more than scale alone: the complementarities between digital talent and industrial digitalization determine whether technology adoption yields productivity gains. Evaluations of AI policies should measure coordination indicators (e.g., skill–industry matching, talent spillovers, infrastructure complementarity).
• Regional policy design:
  • Encourage moderate agglomeration and coordinated development across regions (training, mobility, R&D networks).
  • Promote regional integration strategies (labor mobility, cross‑regional platforms, shared digital infrastructure) to spread benefits and avoid damaging centralization.
• Measurement and research suggestions for AI economics:
  • Use location‑based indices and input–output derived measures to quantify digitalization and digital skill agglomeration in studies of AI diffusion.
  • Study thresholds and tipping points empirically (when gains turn to diseconomies), and model dynamic feedbacks between technology adoption and labor supply.
  • Investigate firm‑ and worker‑level microdata to identify mechanisms (wage premiums, hiring frictions, skill mismatches) and causal channels.
• Policy levers: targeted education/upskilling, incentives for distributed digital infrastructure, regional coordination mechanisms, and measures to reduce diseconomies (congestion, housing/transaction costs) in high‑agglomeration areas.