Chinese listed firms that adopt AI take more business risk as AI bolsters firms' dynamic capabilities; the risk-taking boost partly translates into higher product-quality productivity and is largest among low‑ESG, low‑tech and less marketised firms.

Main Finding

AI adoption by firms significantly increases corporate risk-taking. This effect operates through strengthening firms’ dynamic capabilities (absorptive, adaptive, and innovative), and is stronger for firms with low ESG performance, in low-technology industries, and in less marketised regions. Increased risk-taking partly mediates the positive effect of AI adoption on "new-quality" productivity.

Key Points

• Sample: 30,725 firm–year observations from Chinese listed companies (2011–2021).
• Core result: Firms that adopt AI take on more risk than non-adopters.
• Mechanism: AI enhances dynamic capabilities — specifically absorptive, adaptive, and innovative capabilities — which in turn raise firms’ propensity to take risks.
• Heterogeneity: The positive AI → risk-taking relationship is larger for:
  • Firms with low ESG performance,
  • Firms in low-technology industries,
  • Firms located in regions with lower marketisation.
• Downstream outcome: Part of AI’s positive impact on new-quality productivity operates via increased risk-taking.

Data & Methods

• Data: Panel of Chinese listed firms, 2011–2021 (30,725 firm-year observations).
• Empirical approach (as summarized in the abstract):
  • Comparative analysis of AI adopters vs. non-adopters over time.
  • Tests of mechanisms linking AI to risk-taking via measures of dynamic capabilities (absorptive, adaptive, innovative).
  • Heterogeneity analysis across ESG status, industry technology intensity, and regional marketisation.
  • Mediation analysis showing risk-taking channels part of the AI → productivity effect.
• Robustness: The abstract implies a range of econometric checks (mechanism and heterogeneity tests), though specific identification strategies, variable constructions, and controls are not reported in the abstract.

Implications for AI Economics

• Theoretical:
  • Extends understanding of firm-level strategic response to AI by highlighting risk-taking as an important behavioral consequence and dynamic capabilities as the transmission channel.
• Firm strategy:
  • Managers can view AI not only as a productivity tool but as an enabler of greater strategic risk-taking (e.g., new products, markets, investments) through enhanced learning and adaptability.
  • Firms with weaker ESG or in less technologically advanced settings may gain disproportionate risk-taking benefits from AI — but this could raise governance or reputational concerns.
• Policy and regulation:
  • Encouraging AI diffusion could stimulate firm-level experimentation and productivity growth, but regulators should be aware of potentially higher corporate risk exposures and uneven regional/sectoral impacts.
  • ESG and governance frameworks may need adaptation to ensure responsible risk-taking alongside AI adoption.
• Research directions:
  • Establishing causal identification (e.g., exogenous variation in AI adoption) and exploring longer-run effects and welfare implications.
  • Cross-country comparisons to test external validity and the role of institutional context.
  • Deeper measurement of AI intensity and heterogeneity in AI technologies (e.g., automation vs. decision-support) and their distinct effects on risk behavior.