0 cumulative citations
View corpus contextAI adoption strengthens resilience among China's listed manufacturers, chiefly by accelerating innovation and improving talent incentives; the gains vary substantially by firm ownership, size and region.
The Application of Artificial Intelligence and the Resilience of Manufacturing Enterprises: Mechanisms of Action and Heterogeneity Boundaries
openalex
quasi_experimental
medium evidence
7/10 relevance
Full text
usable extracted full text
DOI
Source
PDF
Using a text-mined AI adoption index for China A-share manufacturing firms (2015–2024), the paper finds that higher AI application is associated with stronger corporate resilience—mainly transmitted through technological innovation, talent incentives, and reduced management costs—with heterogeneous effects by ownership, size, region, and industry.
Citation observations
Cumulative provider counts captured on specific dates; providers are never combined.
Against the backdrop of increasing external environmental uncertainties and accelerated intelligent transformation in the manufacturing sector, enhancing the resilience of manufacturing enterprises has become a critical issue for ensuring the security of industrial and supply chains and promoting high-quality development in manufacturing. This study uses China A-share listed manufacturing companies from 2015 to 2024 as research samples, constructs an indicator of corporate AI application levels based on annual report text mining methods, and measures corporate resilience across three dimensions: resistance capacity, recovery capacity, and innovation capacity, empirically examining the impact of AI application on manufacturing enterprise resilience and its underlying mechanisms. The findings reveal that AI application significantly enhances the resilience of manufacturing enterprises, with conclusions remaining robust even after substituting core variables, adjusting model specifications, controlling for cluster standard errors at the firm level, incorporating provincial fixed effects, and employing instrumental variable methods. Mechanism tests indicate that AI primarily strengthens corporate resilience by improving talent incentives, fostering technological innovation, and reducing management costs, with technological innovation serving as the primary transmission pathway; while internal control is partially influenced by AI, its mediating effect fails the Bootstrap robustness test. Further heterogeneity analysis demonstrates that the impact of AI on corporate resilience varies across ownership structures, firm size, regional differences, and industry attributes. This study enriches research on factors influencing corporate resilience in the digital economy context and provides empirical evidence for manufacturing enterprises to rationally advance AI adoption and enhance risk resistance and recovery capabilities.
Summary
Main Finding
AI adoption significantly increases the resilience of manufacturing firms in China. This positive effect is robust to alternative variable definitions, model specifications, firm-clustered standard errors, provincial fixed effects, and instrumental-variable approaches. Mechanism analysis shows AI mainly operates by strengthening talent incentives, boosting technological innovation (the primary channel), and reducing management costs; internal control is partially affected by AI but its mediating effect is not robust in the Bootstrap test. The effect varies by ownership, firm size, region, and industry characteristics.
Key Points
- Sample and period: China A‑share listed manufacturing firms, 2015–2024.
- Core variables:
- AI application: constructed from annual report text mining (firm-level measure).
- Corporate resilience: measured across three dimensions — resistance capacity, recovery capacity, and innovation capacity.
- Main empirical result: higher firm-level AI application correlates with higher resilience across the three dimensions.
- Mechanisms tested:
- Talent incentives: AI improves HR processes and incentives, helping attract/retain and activate digital/high-skilled talent — positive mediator.
- Technological innovation: AI accelerates R&D, knowledge recombination, and innovation spillovers — identified as the primary transmission pathway.
- Management/administrative costs: AI lowers search, coordination and information frictions — contributes to resilience.
- Internal control: AI can improve internal control quality, but the mediating effect failed the Bootstrap robustness check.
- Robustness checks: alternative core variables, adjusted specifications, clustering at firm level, provincial fixed effects, instrumental variables.
- Heterogeneity: AI’s resilience-enhancing effect differs by property rights (ownership), firm size, geographic region, and industry attributes.
Data & Methods
- Data:
- Universe: Listed manufacturing firms on China A‑share market (2015–2024).
- AI measure: Text-mining of firms’ annual reports to construct an AI application index at the firm level.
- Resilience measure: Composite/multi-dimensional measure covering resistance, recovery, and innovation capacities (constructed at the firm level).
- Empirical strategy:
- Benchmark panel regressions relating firm-level AI application to resilience outcomes.
- Mechanism/mediation analysis with mediators for internal control, talent incentives, technological innovation, and management costs.
- Robustness/identification: alternative variable definitions, model specifications, firm-clustered SEs, provincial fixed effects, instrumental variable estimations.
- Bootstrap tests used to assess the robustness of indirect (mediated) effects.
Implications for AI Economics
- For firms: AI can be a meaningful productivity and resilience technology, but benefits require complementary investments — especially in skills/talent, R&D capabilities, and organizational processes to capture the innovation-led channel.
- For policy:
- Promote AI diffusion in manufacturing with targeted support for complementary assets (training, R&D subsidies, data infrastructure), since technological innovation is the dominant pathway to resilience.
- Design heterogeneity‑aware programs: smaller firms, different ownership types, lagging regions, and certain industries may need tailored incentives or capacity-building to realize resilience gains.
- Support governance upgrades (data governance, internal controls) during digital transformation, because AI’s governance benefits are present but not automatically robust.
- For researchers/economists:
- The study highlights the need to model complementarities between AI and organizational/talent investments when assessing AI’s macro and micro effects.
- Future work should explore timing and nonlinearities (e.g., adoption lags, threshold effects), causal identification of mediating channels in more detail, and how specific AI subfields (e.g., generative models, industrial vision systems) differentially affect resilience and productivity.
- For industrial policy and supply chains: encouraging firm-level AI adoption—coupled with innovation and human capital policies—can strengthen industrial and supply-chain resilience, contributing to broader economic stability and high‑quality manufacturing development.
Assessment
Paper Typequasi_experimental
Evidence Strengthmedium — The paper uses firm-panel data across 2015–2024, applies multiple robustness checks and an IV strategy, and conducts mediation and heterogeneity analyses—supporting plausible causal claims. However, the excerpt does not specify the IV, its validity tests, or detailed empirical results; the text-mined AI measure may be endogenous or noisy (reporting bias), leaving residual concerns about identification and measurement.
Methods Rigormedium — The design demonstrates standard good-practice steps (panel controls, fixed effects, clustered SEs, IV, bootstrap mediation, heterogeneity checks). But key details are missing in the supplied text: the exact IV, its exclusion restriction, the construction/validation of the AI dictionary/measure, controls used, functional forms, and effect sizes; these omissions limit assessment of causal rigor.
SampleFirm-year panel of China A-share listed manufacturing companies, 2015–2024; AI application intensity constructed via annual report text mining; corporate resilience measured along three dimensions: resistance capacity, recovery capacity, and innovation capacity; additional firm-level controls and province fixed effects employed; heterogeneity examined by ownership, firm size, region, and industry.
Themesadoption innovation org_design governance
IdentificationPanel regressions on 2015–2024 firm-year data for China A-share listed manufacturing firms using a text-mined AI application index from annual reports; controls and fixed effects (firm-level and provincial) included; robustness checks with alternative variable definitions, clustered standard errors, and bootstrap mediation tests; causal inference strengthened by an instrumental-variable (IV) approach (IV not detailed in supplied text).
GeneralizabilityLimited to publicly listed manufacturing firms in China (omits SMEs and privately held firms)., China-specific institutional, regulatory, and policy context may limit applicability to other countries., Text-mined measure of AI adoption may not capture on-the-ground AI usage intensity or quality (reporting bias)., Findings pertain to manufacturing only and may not apply to services or other sectors., Study period (2015–2024) includes unique shocks (e.g., COVID-19, supply-chain disruptions) that may affect external validity.
Claims (7)
| Claim | Direction | Outcome | Confidence & Evidence | Details |
|---|---|---|---|---|
| AI application significantly enhances the resilience of Chinese manufacturing enterprises. Organizational Efficiency | positive | Corporate resilience, measured through resistance capacity, recovery capacity, and innovation capacity |
Reading fidelity
high
Study strength
medium
|
not reported
|
| The positive association between AI application and manufacturing-enterprise resilience remains robust after replacing core variables, changing model specifications, clustering standard errors at the firm level, adding provincial fixed effects, and using instrumental-variable methods. Organizational Efficiency | positive | Corporate resilience |
Reading fidelity
high
Study strength
medium
|
not reported
|
| Technological innovation is the primary transmission pathway through which AI application strengthens manufacturing-enterprise resilience. Innovation Output | positive | Corporate resilience mediated by technological innovation |
Reading fidelity
high
Study strength
medium
|
not reported
|
| AI application strengthens manufacturing-enterprise resilience through improved talent incentives and reduced management costs. Organizational Efficiency | positive | Corporate resilience through talent-incentive mechanisms and management-cost reductions |
Reading fidelity
high
Study strength
medium
|
not reported
|
| Although AI affects internal control, the mediating effect of internal control on the relationship between AI application and corporate resilience is not robust to the Bootstrap test. Organizational Efficiency | null_result | Internal-control mediation of the AI–corporate-resilience relationship |
Reading fidelity
high
Study strength
high
|
not reported
|
| The effect of AI application on manufacturing-enterprise resilience is heterogeneous across ownership structure, firm size, region, and industry attributes. Organizational Efficiency | mixed | Corporate resilience |
Reading fidelity
high
Study strength
medium
|
not reported
|
| The study measures corporate AI application using annual-report text mining and measures resilience across resistance, recovery, and innovation dimensions. Organizational Efficiency | positive | Resistance capacity, recovery capacity, and innovation capacity |
Reading fidelity
high
Study strength
high
|
not reported
|