Chinese firms’ digital transformation reduces the cost of green debt: firms with higher text‑based CDT scores pay materially lower green bond spreads, a decline partly explained by more credible environmental disclosure and tighter internal controls.

Main Finding

Corporate digital transformation (CDT) is associated with significantly lower green bond credit spreads for Chinese A‑share issuers (2016–2024). The effect is robust and is partly mediated by improvements in environmental information disclosure quality and internal‑control quality, consistent with signaling and agency theories.

Key Points

• Sample: firm–bond observations of Chinese A‑share listed companies issuing green bonds, 2016–2024.
• Primary result: higher text‑based CDT scores correspond to lower green bond financing costs (credit spreads over matched risk‑free benchmarks).
• Identification and robustness: results hold under fixed‑effects regressions and when using alternative variable measures, entropy balancing, instrumental‑variable estimation, and other robustness checks.
• Mechanisms: CDT reduces green bond spreads partly by (1) improving environmental disclosure credibility and transparency, and (2) strengthening internal control quality (reducing greenwashing and use‑of‑proceeds risks).
• Mediation tests: Sobel and bootstrap tests support the partial mediation pathways.
• Theoretical framing: CDT operates through signaling (credible disclosure) and agency (better controls reduce monitoring and misuse risks).
• Contribution: extends CDT literature into sustainable finance, identifying firm‑level digitalization as a determinant of green bond pricing.

Data & Methods

• Data: Green bonds issued by Chinese A‑share listed firms (2016–2024); annual reports used to construct a text‑based CDT index.
• Dependent variable: green bond credit spread measured relative to matched risk‑free benchmarks.
• Primary econometric approach: fixed‑effects regression at firm–bond level (controls implied but not detailed here).
• Endogeneity and robustness strategies:
  • Alternative measurements of CDT and spreads.
  • Entropy balancing to improve covariate balance across treatment levels.
  • Instrumental‑variable estimation to address reverse causality/omitted variable bias.
  • Channel (mediation) analyses using disclosure quality and internal control measures.
  • Statistical mediation validation via Sobel and bootstrap tests.

Implications for AI Economics

• Digitalization (including AI adoption) can materially reduce the cost of capital for green financing by lowering information asymmetries and agency frictions. This links firm‑level tech adoption to the pricing of sustainability investments.
• Mechanisms suggest practical roles for AI tools: automated, consistent environmental disclosure (NLP, structured reporting), real‑time monitoring and compliance, and internal control automation — all of which can enhance credibility and reduce greenwashing risk.
• Policy and market design:
  • Regulators and bond underwriters could encourage or reward verifiable digital disclosure/monitoring practices to lower green financing costs and improve market integrity.
  • Standardized, machine‑readable green reporting and third‑party verification enabled by digital tech could strengthen signaling and reduce pricing frictions.
• Research directions for AI economics:
  • Disentangle effects of AI-specific adoption from broader digital transformation on financing costs.
  • Causal identification using exogenous shocks to digital/AI adoption (e.g., policy rollout, software subsidies) or firm‑level adoption timing (diff‑in‑diff).
  • Explore heterogeneity (firm size, industry, regulator stringency) and external validity beyond China.
  • Quantify macro impacts: how firm‑level digitalization scales to affect green investment flows and the pace/cost of the green transition.
• Caveats: generalizability beyond the Chinese context may be limited; residual endogeneity concerns persist despite IV and balancing methods; further work needed to isolate AI's unique contribution within CDT.