Chinese listed firms with stronger AI-enabled process-performance scores report higher abnormal earnings and profitability, implying AI-linked operational upgrades are tied to firm value growth; the result is based on a large panel and a new interpretable index but does not establish causality.

Main Finding

The paper develops and implements a data-driven evaluation framework that links AI-driven process performance to enterprise value. Using a Process Performance Index (PI) built from firm-level indicators and integrated into a Feltham–Ohlson valuation structure, the study shows that higher AI-enabled process capability (PI) is positively associated with abnormal earnings and firm profitability — evidence that AI-driven operational improvements contribute to sustained enterprise value growth.

Key Points

• Framework design:
  • Integrates enterprise valuation (Feltham–Ohlson model) with a multi-level performance evaluation pipeline.
  • Weights indicators using a hybrid Analytic Hierarchy Process (AHP) combined with the Entropy Weight Method (EWM) to balance expert judgment and data-driven objectivity.
  • Aggregates multi-dimensional indicators via Fuzzy Comprehensive Evaluation (FCE) to handle measurement fuzziness and nonlinearity.
• Process Performance Index (PI):
  • Captures AI-enabled operational capability across three dimensions: resource allocation efficiency, coordination effectiveness, and production/service performance.
  • Constructed from firm-level observable indicators and weighted/aggregated through the AHP–EWM and FCE steps for interpretability.
• Empirical findings:
  • Using panel firm-level data, PI is positively correlated with abnormal earnings (as measured within the Feltham–Ohlson framework) and conventional profitability metrics.
  • Mechanisms associated with higher PI include increased digital-technology investment intensity, accumulation of knowledge-based human capital, and improved data governance, alongside enhanced production and service outcomes.
• Practical claim: The combined AHP–EWM–FCE within a valuation model provides an interpretable quantitative mechanism to evaluate how AI adoption and intelligent transformations translate into firm value.

Data & Methods

• Data:
  • Panel of 3,515 Chinese A-share listed firms, spanning 2014–2022.
  • Total of 20,076 firm-year observations.
• Index construction and weighting:
  • Candidate indicators mapped to the three PI dimensions.
  • AHP captures expert/structured preference information; EWM supplies data-driven weights based on indicator dispersion. The hybrid AHP–EWM blends both to reduce bias.
  • Fuzzy Comprehensive Evaluation aggregates weighted indicators into the multi-dimensional PI, accommodating vagueness and nonlinear aggregation across metrics.
• Valuation linkage and empirical strategy:
  • PI is embedded into the Feltham–Ohlson valuation framework to derive implications for abnormal earnings and value creation.
  • Empirical analysis uses the constructed PI as a key explanatory variable for abnormal earnings and profitability across the panel dataset (details on regression specifications and controls are reported in the paper).

Implications for AI Economics

• Measurement and valuation:
  • Offers a replicable, interpretable approach to quantify AI-enabled operational capability and map it to firm value — useful for researchers building micro-level models of AI’s economic impact.
  • The hybrid weighting + fuzzy aggregation approach balances subjectivity and data-driven rigor, addressing common measurement challenges in AI adoption studies.
• Managerial and investor use:
  • Firms can use the PI framework to benchmark AI investments, monitor operational transformation, and communicate value-creation channels to investors.
  • Investors and analysts can incorporate PI-like measures into firm valuation and due diligence to better price AI-related intangible improvements.
• Policy and industrial strategy:
  • Evidence that AI-related investments in digital tech, human capital, and data governance are associated with higher firm value supports policy measures promoting these investments (e.g., skills development, data infrastructure).
  • The framework informs evaluation of Industry 5.0 and data-driven industrial policies by linking operational upgrades to measurable enterprise value outcomes.
• Methodological contribution:
  • Demonstrates how combining multi-criteria decision methods (AHP–EWM) and fuzzy aggregation with a structural valuation model creates an interpretable bridge from process metrics to economic outcomes — a template for future microeconomic assessments of technology adoption.
• Caveats and future work:
  • Results are based on listed Chinese firms (2014–2022); external validity to other countries or non-listed firms should be tested.
  • The approach depends on indicator selection and the quality of underlying firm-level data; further robustness checks and extensions (e.g., causal identification of AI investment effects) would strengthen policy inference.