Abstract
Risk of corporate social responsibility (CSR) refers to the possibility and consequences of the potential lack of CSR, which will cause damage and loss to society and enterprises. Hence, scientific evaluation of CSR risk is conducive to enterprises to take timely risk response measures and reduce potential risk factors. In order to fully reflect the CSR risk, a comprehensive risk evaluation index system is established from four dimensions of corporate stakeholders, law, environment and charity. In addition, hesitant fuzzy language is used to describe the CSR risk information, so as to reflect the uncertainty in the process of risk judgment. Considering the complexity of the CSR concept and the possible discreteness of risk results, the hesitant fuzzy language-cloud model is established to evaluate CSR risk level. Finally, utilizing China Huaneng Group as a case to apply the CSR risk evaluation model, it is found that the overall CSR risk level of Huaneng is relatively low. Through further comparative and hierarchical analysis, it is found that the CSR risk of community stakeholders is relatively high, and the CSR risk of legal, environmental and charitable is relatively balanced.
Introduction
Corporate Social Responsibility (CSR) has evolved from a discretionary public relations effort into a fundamental pillar of enterprise risk management. A failure to meet CSR expectations—whether through environmental negligence, labor disputes, or legal infractions—can result in severe financial loss, reputational damage, or even bankruptcy. However, quantifying CSR risk is inherently challenging because it deals with qualitative, "fuzzy" concepts that are difficult to capture with traditional precise metrics. Standard models often force experts to provide single-point scores, which discards the natural hesitation and uncertainty inherent in human judgment. This research addresses this gap by proposing a novel assessment framework that combines Hesitant Fuzzy Language (HFL) with Cloud Models to better capture the nuances of CSR risk.
Methodology: The Evaluation Index System
To systematically evaluate CSR risk, the researchers first established a comprehensive index system. This system translates broad CSR concepts into specific risk dimensions.
The framework identifies four primary dimensions for the criteria layer:
- Stakeholder Responsibility Risk: Focuses on the impact on shareholders, creditors, consumers, employees, managers, government, competitors, and the community.
- Legal Responsibility Risk: Evaluates compliance with national laws, international conventions, and industry-specific norms.
- Environmental Responsibility Risk: Assesses efforts in resource conservation and environmental protection.
- Charitable Responsibility Risk: Measures risks related to social donations, poverty alleviation, and public service.
The Hesitant Fuzzy Language-Cloud Model
The core of the proposed model is the integration of Hesitant Fuzzy Language (HFL) and the Cloud Model. HFL allows evaluators to use multiple linguistic terms (e.g., "between medium and high") to express their assessments, preserving the data contained in their uncertainty.
The Cloud Model then acts as a bridge between these qualitative terms and quantitative analysis. Each "cloud" is defined by three parameters:
- Expectation (Ex): The central value of the concept.
- Entropy (En): The measure of fuzziness or uncertainty.
- Super Entropy (He): The measure of randomness or dispersion.
By using the Analytic Hierarchy Process (AHP) to weight these indicators, the model aggregates expert opinions into a "Standard Cloud" that represents the overall risk level, ranging from "Very Low" to "Very High."
Case Study: China Huaneng Group
The model was validated using a longitudinal study of the China Huaneng Group, a major state-owned power enterprise, covering the period from 2010 to 2018. Data was sourced from sustainability reports, the China Stock Market & Accounting Research Database (CSMAR), and international standards like ISO9001 and GRI.
Key Results
The application of the model yielded several critical insights:
- Overall Risk Profile: The CSR risk for China Huaneng Group was generally "slightly low" (Level s4) throughout the study period, with 2011 being a notable outlier at "medium" risk (Level s3).
- Stability and Convergence: The probability of catastrophic CSR failure was found to be near zero, as the results showed no convergence with "High" or "Very High" risk clouds.
- Dimension Performance: The company demonstrated its strongest performance in Legal Responsibility (consistently "Low" risk). Environmental and Charitable risks were categorized as "Slightly Low."
- The Community Gap: A specific area of concern was "Community Stakeholder" risk, which consistently scored higher than other indicators, suggesting a need for better local social integration.
Discussion and Conclusion
The primary contribution of this research is the recognition that expert "hesitation" is a valuable data point. By visualizing risk as a "cloud" rather than a single point, the model provides a more realistic representation of the uncertainty and disagreement inherent in CSR assessments.
While the model offers a robust framework, future work should focus on developing real-time monitoring capabilities and expanding industry-wide databases for benchmarking. Furthermore, integrating the role of government supervision could help refine the model by accounting for the external "costs" of CSR non-fulfillment. Ultimately, this hesitant fuzzy language-cloud model provides a scientific basis for enterprises to identify, evaluate, and mitigate social responsibility risks more effectively.
Summary updated