关键词: 数据包络分析, 风险评估, 社会网络分析, 关联企业

Abstract: In market economic activities, enterprises do not exist independently. They usually become associated enterprises due to interpersonal, asset or transaction relationships. The occurrence of risks in one enterprise tends to have an impact on other associated enterprises that are more strongly connected with it. If the risk of an associated enterprise is evaluated solely on the basis of data from disclosed indicators, the enterprises may be considered “safe” in the field or as a whole. And it is difficult to truly and accurately assess the overall risk of an enterprise. This poses a significant obstacle to risk avoidance, sustainable and healthy business operations, as well as timely prevention and control of market risks by regulatory authorities. It is critical to characterize the impact of this risk on associated enterprises. Online public opinions reflect negative or positive information of an enterprise, which can reveal the company’s image and business situation condition to some degree. Moreover, public opinion risk has a greater detrimental effect on the associated enterprise. For this reason, it is necessary to take into account the associated impact of public opinion risk when conducting a comprehensive risk assessment of an enterprise. Data envelopment analysis (DEA), as an effective evaluation method, is also applicable to comprehensive enterprise risk assessment. However, few of the existing DEA evaluation method studies have explored the association relationships between decision-making units (DMU) and the impact of the association relationship on the performance of individual DMU. Therefore, it is of great theoretical value and practical significance to investigate how to assess comprehensive risk of associated enterprises based on DEA in network public opinion environment.
Based on the above issues, this paper considers the risk of corporate public opinion, and proposes a corporate risk assessment method considering individual association relationships which is based on social network analysis (SNA) and DEA. Our approach aims to quantify the impact of association relationships and evaluate the overall risk status of an enterprise more accurately. In our approach, the enterprise association network is firstly constructed. Then from individual attributes and network topology two aspects, we introduce the node global importance, node similarity, and node attributes, and propose a model that can quantitatively portray the public opinion risk under the influence of association relationships. After that, the quantitative association influence is integrated with the modified slacks-based measure (MSBM) model to develop a comprehensive risk assessment model for enterprises. Finally, the risk data of 345 real estate enterprises are used to verify the effectiveness of our proposed method.
We conduct an empirical analysis of 345 real estate firms in Hunan Province. The changes in the relative ranking of the overall risk of associated firms among all the sample firms before and after considering shareholder association are discussed comparatively. The findings are as follows: (1)In the enterprise association network, the overall risk rankings of all good public opinion enterprises that are directly associated with the public opinion risk enterprises have risen to a varying degree. (2)There are 10 firms of the 32 associated enterprises that have the changing level in the risk rankings reaching 50% or more. Furthermore, from the perspective of social networks, we rationally analyze the reasons for the changes in the risk rankings of these 10 enterprises based on the structural characteristics of the networks and the attributes of the enterprises. The results demonstrate that: (1)The close connection between individuals in the network and its impact cannot be ignored, and focusing only on the risk data of the enterprises themselves will lead to an “underestimation” of the risk status. (2)The methodology proposed in this paper can effectively quantify the associated impact of network public opinion risks among enterprises and reveal the associated effect of enterprises and its influence on the comprehensive risk of single enterprise in the context. Additionally, a risk ranking of enterprises that is closer to the actual situation and more explanatory can be obtained. (3)It also provides a new perspective for the relevant regulators to make risk prevention decisions, which can help them focus on monitoring the core enterprises and cluster hubs in the industry, and curb the adverse effects of enterprise risk spillover through association relationships in time. At the same time, it is conducive to urging enterprises to regulate their own business behaviors. On the other hand, our proposed method considers association relationship between individuals in the process of DEA-based assessment, which enriches DEA theoretical research to a certain extent.

Key words: data envelopment analysis, risk assessment, social network analysis, associated enterprise