0 cumulative citations
View corpus contextFirms that disclose stronger AI capability generate more eco-innovation, and the payoff is larger in digitally advanced countries; national digital ecosystems amplify firms’ ability to convert AI investments into sustainable innovation.
Digital Economies and Artificial Intelligence: Unlocking Eco-Innovation in Asian Firms
openalex
correlational
medium evidence
7/10 relevance
Summary only
summary available; pdf_status=paywall
DOI
Source
PDF
Using a disclosure-based AI-capability index for 5,564 listed Asian firms (2017–2024), the paper finds that greater firm AI capability is associated with higher firm-level eco-innovation, and this association is stronger in countries with more advanced digital economies.
Citation observations
Cumulative provider counts captured on specific dates; providers are never combined.
0 cumulative citations
View corpus contextArtificial intelligence (AI) is increasingly transforming firms’ innovation processes and creating new opportunities for sustainable development; however, empirical evidence on how firm-level AI capability influences eco-innovation under different digital environments remains limited. Drawing on the Natural Resource-Based View and Dynamic Capability Theory, this study examines the effect of AI capability on firm-level eco-innovation and investigates whether national digital economy development strengthens this relationship. The analysis uses an unbalanced panel of 5564 listed firms from 15 Asian economies over the period 2017–2024. Firm fixed-effects models are employed as the primary estimation approach, while moderation analysis is complemented by System GMM, propensity score matching, and Heckman two-step estimation to address potential endogeneity and sample-selection concerns. AI capability is measured using a disclosure-based index derived from textual analysis of firms’ annual and sustainability reports. The results indicate that AI capability is positively associated with eco-innovation and that this relationship is significantly stronger in countries with more advanced digital economies. These findings suggest that supportive digital ecosystems enhance firms’ ability to convert AI capability into sustainable innovation outcomes. The study contributes to the literature by demonstrating that national digital economy development serves as an important contextual condition shaping the effectiveness of firm-level AI capability in promoting eco-innovation.
Summary
Main Finding
Firm-level AI capability — measured via a disclosure-based textual index — is positively associated with eco-innovation among listed firms in Asia, and this positive effect is significantly stronger in countries with more advanced digital economies. Supportive national digital ecosystems amplify firms’ ability to convert AI capability into sustainable innovation outcomes.
Key Points
- The study integrates Natural Resource-Based View and Dynamic Capability Theory to theorize how AI capability enables eco-innovation and how national digital economy conditions moderate this effect.
- Primary result: higher AI capability → more eco-innovation.
- Moderation: the AI → eco-innovation link is stronger in countries with more developed digital economies.
- Robustness: results hold when addressing endogeneity and selection concerns using System GMM, propensity score matching (PSM), and Heckman two-step estimation.
- Sample coverage: 5,564 listed firms across 15 Asian economies, 2017–2024 (unbalanced panel).
Data & Methods
- Sample: unbalanced panel of 5,564 publicly listed firms in 15 Asian countries, years 2017–2024.
- Key independent variable: firm AI capability measured by a disclosure-based index derived from textual analysis of annual and sustainability reports (documented mentions/indicators of AI-related capabilities).
- Dependent variable: firm-level eco-innovation (paper summary does not specify the exact operationalization; typical measures include eco-related patents, environmentally focused product/process innovations, or disclosure indicators).
- Primary estimation: firm fixed-effects panel models to control for time-invariant firm heterogeneity.
- Endogeneity and selection robustness:
- System GMM to address dynamic panel bias and potential endogeneity of regressors.
- Propensity score matching to reduce confounding from observables when comparing firms with different AI capability levels.
- Heckman two-step to correct for potential sample-selection bias.
- Moderation analysis: interaction between firm AI capability and a country-level digital economy development indicator to test how national digital conditions change the effectiveness of AI capability.
Implications for AI Economics
- The effectiveness of firm AI investments depends on national digital infrastructure and ecosystem maturity; AI capability does not operate in a vacuum.
- Policy: investments in digital economy development (broadband, digital platforms, data governance, digital skills) can increase the sustainability payoff of private AI investments and accelerate eco-innovation.
- Firm strategy: complement AI capability investments with engagement in digital ecosystems (data-sharing platforms, digital partnerships, upskilling) to maximize eco-innovation returns.
- Research: highlights the need for contextualized assessments of AI’s economic impacts — cross-country heterogeneity in digital development matters. Future work should refine measurement of eco-innovation, explore mechanisms (e.g., data access, platform effects), and test generalizability beyond listed firms and the Asian region.
Assessment
Paper Typecorrelational
Evidence Strengthmedium — Multiple complementary observational techniques (FE, System GMM, PSM, Heckman) lend credibility beyond simple correlations, and a large multi-country panel supports external variation, but causal claims remain limited by reliance on disclosure-based AI measures, possible time-varying confounders, reverse causality, and the usual weaknesses of GMM/PSM/Heckman when exclusion restrictions or instrument validity are not clearly established.
Methods Rigormedium — The authors apply a reasonable suite of econometric approaches (fixed effects, System GMM, PSM, Heckman) and test moderation effects, which is methodologically sound for observational panel data; however, key methodological details are missing or potentially problematic (operationalization of eco-innovation not specified, disclosure-based AI measures susceptible to reporting bias, validity of GMM instruments and PSM covariate balance not shown, Heckman requires credible exclusion restriction). These issues limit confidence in causal identification.
SampleUnbalanced panel of 5,564 publicly listed firms across 15 Asian economies observed from 2017–2024; firm-level AI capability measured via a disclosure-based textual index from annual and sustainability reports; dependent variable is firm-level eco-innovation (operationalization not specified in the summary).
Themesinnovation adoption
IdentificationPanel firm fixed-effects models to control for time-invariant heterogeneity; dynamic panel System GMM to address potential endogeneity and lagged dependent-variable bias; propensity score matching (PSM) to reduce confounding from observables when comparing firms with different AI-capability disclosure scores; Heckman two-step to correct for potential sample-selection bias; moderation tested via interaction between firm AI-capability index and country-level digital-economy indicator. No natural experiment, instrument, or exogenous shock reported.
GeneralizabilityListed firms only — excludes private firms and SMEs, which may differ in AI adoption and eco-innovation capacity, Asia-only sample — results may not generalize to Europe, North America, Africa, or Latin America, Disclosure-based AI measure may reflect reporting practices rather than underlying technical capability, which can vary across sectors and countries, Eco-innovation measurement unspecified — results may depend on whether outcome is patents, product/process changes, or disclosure indicators, Heterogeneity across industries and firm sizes may limit applicability to specific sectors (e.g., manufacturing vs services)
Claims (4)
| Claim | Direction | Outcome | Confidence & Evidence | Details |
|---|---|---|---|---|
| Higher firm-level AI capability is positively associated with eco-innovation among publicly listed firms in Asia. Innovation Output | positive | Firm-level eco-innovation |
Reading fidelity
high
Study strength
medium
|
n=5564
|
| The positive association between firm AI capability and eco-innovation is significantly stronger in countries with more developed digital economies. Innovation Output | positive | Firm-level eco-innovation |
Reading fidelity
high
Study strength
medium
|
n=5564
|
| The positive relationship between AI capability and eco-innovation remains when the analysis addresses potential endogeneity and selection concerns. Innovation Output | positive | Firm-level eco-innovation |
Reading fidelity
high
Study strength
medium
|
n=5564
|
| The study examines 5,564 publicly listed firms across 15 Asian economies over 2017–2024 using an unbalanced panel. Other | null_result | Study sample coverage |
Reading fidelity
high
Study strength
high
|
n=5564
|