China: An International Journal

RESEARCH BACKGROUND

The term corporate social responsibility (CSR) refers to companies' fulfilment of social responsibilities; their neglect of such responsibility can be termed corporate social irresponsibility (CSIR or CSI). Armstrong¹ defines CSIR as the act of disregarding the impact certain corporate decisions may have on stakeholders and focusing instead solely on individual interests, potentially sacrificing overall societal well-being. Clark and Grantham² extend this definition, asserting that CSI activities include illegal actions and behaviours that exploit negative externalities for personal gain, leading to [End Page 146] systemic unsustainability. Conversely, Lange and Washburn³ present CSIR as the opposite of CSR, defining it as "acting irresponsibly", while considering CSR as corporate behaviour with positive societal outcomes and CSIR as corporate behaviour with significant societal harm or damage.

The issue of CSIR has become particularly severe in emerging industries such as the internet. The internet industry in China is developing rapidly, including sectors such as e-commerce, live streaming, social media, instant messaging, online gaming, and navigation. While these industries do not pose environmental pollution problems as some industries may, they give rise to new types of CSIR incidents, such as personal privacy breaches, copyright infringement, false data and vulgar content. These commercial realities indicate that although companies have recently improved in fulfilling social responsibility, the frequent occurrence of CSIR in the internet industry remains significant and alarming. Moreover, CSIR may cause more significant losses to society than the social benefits brought about by CSR.⁴ Some studies argue that reducing CSIR is an effective way to enhance CSR and improve its current state.⁵ Therefore, studies on CSIR in the internet industry are more urgent and significant than other CSR studies.

Empirical studies of CSIR however have inherent limitations. It can be expected that companies will not willingly disclose their CSIR activities; consequently, only data and information that are exposed or revealed can be observed. Moreover, the goal of government and media exposure of corporate CSIR incidents is not only to penalise the implicated companies but also to serve as a deterrent for others, employing a "kill one, warn a hundred" strategy. In the internet era, constructing an efficient public relations system and early crisis warning mechanism for companies requires attention to the impact generated by exposure of CSIR incidents in other companies. Accordingly, this article addresses the following questions: (i) What impact does the exposure of CSIR incidents have on peer companies? (ii) How can punitive exposure be optimised to maximise the deterrent effect of CSIR exposure?

Existing CSIR studies have three limitations. First, most examine how CSIR is affected by internal and external factors concerning the involved companies and rarely discuss the impact of CSIR exposure on peer companies. Although some studies focus on the industry spillover effect of negative brand publicity and product-harm crises,⁶ there is a lack of theoretical discussion and empirical support for the direct impact of CSIR exposure between companies. Based on the deterrent effect, this study explores [End Page 147] the implications of CSIR exposure from a comparable corporate standpoint, thereby enriching the knowledge of CSIR. The prototype deterrent effect initially appeared in the military strategy field and has been the cornerstone of defence policies and strategies in various countries. It was later developed in criminal behavioural research and has been commonly used in information system studies.⁷ Recently, a deterrent effect has been introduced into business management to study corporate executive corruption,⁸ employee compliance with security policies⁹ and audit effectiveness.

Second, although most studies utilise data on CSIR exposure as research samples, few have examined the exposure mechanism. Studies on CSIR exposure can be divided into two main categories. The first category examines the monitoring effects of media supervision on corporate violations, executive compensation and financial performance. For example, media exposure can force companies to correct violations, effectively regulate executive compensation, curb financial restatements and affect short-term corporate financial performance.¹⁰ The second category investigates the negative impact of certain irresponsible behaviours from a company perspective,¹¹ including negative brand publicity,¹² product-harm crises¹³ and corporate crimes.¹⁴ However, both approaches provide only a partial understanding of CSIR, as CSIR exposure relates not only to the media and involved companies but also to factors such as governments, public opinion and peer companies. This [End Page 148] article establishes a CSIR exposure mechanism to cover these roles for empirical investigation, focusing on the emerging internet industry.

Third, existing studies employ research methods which explore the "net effect" of a single influencing factor while ignoring the "joint effect" of multiple factors. Consequently, the interaction between different factors, including substitution, complementarity and suppression, cannot be explained. CSIR exposure results from the combined action of events, companies, governments and media. The complexity of causality cannot be entirely explained by the net effect of a single factor through traditional regression analysis or by the regulatory impact of up to three variables.¹⁵ However, fuzzy-set qualitative comparative analysis (fsQCA) is an effective method to explore the "joint effect" and "interactive relations" and has been widely applied in various fields of the management discipline.¹⁶

Based on these considerations, this article establishes the CSIR exposure mechanism based on deterrent effect theory to explore the "joint effects" on the deterrent power of CSIR exposure in the Chinese internet industry of seven factors: policy-oriented media involvement (PMI), media number (MN), event influence (EI), public opinion proportion (POP), punishment (PUN), corporate attitudes (CA) and cluster of events (COE). This study aims to identify the different paths through which exposure of unethical and illegal CSIR affects the Chinese internet industry, to guide companies in building high-performance public relations systems and to provide actionable suggestions for government decision-makers.

CORPORATE SOCIAL IRRESPONSIBIITY (CSIR) EXPOSURE

Corporate social irresponsibility infringes on the legitimate rights and interests of the public and diminishes overall societal welfare. Although CSIR is socially harmful, some companies evince such behaviour in their pursuit of significant returns with the view that the profits outweigh the costs.¹⁷ For example, some Chinese family-owned firms engage in philanthropic activities to divert public attention from environmentally unfriendly behaviour.¹⁸ Greve, Palmer and Pozner¹⁹ emphasise that such ill-intentioned acts as pollution should not be tolerated and punishment should be imposed on CSIR companies to prevent societal harm and restore market order. [End Page 149] Exposure acts as an external governance mechanism. It serves as an effective medium for social supervision, significantly reflecting society's oversight function.²⁰ This article initially analyses the exposure mechanism from a role perspective and clarifies the functions of different role factors concerning the deterrent effect. CSIR exposure involves two time points and three roles. The two time points refer to the moment of exposure and the outcome after one year. The three roles include the source of the exposure (that which brings the CSIR to light), the target of the exposure (that which is exposed, i.e. the company and the nature of the incident) and those affected by exposure.

When exposure occurs, the CSIR of the company is revealed to the media, which may include government investigation of the event and legal punishment of the company. There are two exposure sources: the government and the media. The government reviews and penalises illegal CSIR, whereas the media publicly discloses illegal or unethical CSIR.²¹ The government's role has three aspects: reviewing reported CSIR events such as infringements and the sale of counterfeit products; regularly checking and revealing related behaviour of companies, such as by annual reports and product quality inspections; and initiating investigations and interviews to standardise corporate behaviour, especially targeting common or newly defined CSIR with legal improvements. The government targets primarily illegal CSIR, whereas the media exposes both illegal and legal but unethical CSIR.

In the internet era, the influence of the media is reflected primarily in both mass and social media.²² Mass media exposure of CSIR events is more reliable, even if the influence of social media on public opinion should not be ignored. The authority of the media to expose CSIR events, the quantity of media coverage and attention to public opinion directly affect the level and effect of the exposure. Media exposure occurs in three ways: first, reporting certain scandals or misconduct of listed companies²³ to perform post-monitoring and disclosure functions, subsequently guiding government intervention and correction of corporate CSIR; second, serving as an effective information intermediary²⁴ to reduce information asymmetry and enhance the visibility of CSIR events, aiding investors and other stakeholders in protecting their interests; and third, triggering a reputation-related mechanism that shapes the reputations of managers and board members among shareholders and future employers, thereby [End Page 150] reducing the incentive to engage in CSIR and prompting companies to improve governance.²⁵

The target of the exposure (or exposure object) refers to the exposed roles, including the company involved and the target event. Any company could be involved; however, owing to the large number of listed companies, the CSIR of industry leaders and well-known companies often causes more significant harm than that of ordinary companies. Consequently, these companies are primary targets of social concern and research. Regarding CSIR events, most studies utilise the KLD index (Kinder, Lydenberg, Domini Social Performance Index) for evaluation.²⁶ According to the KLD Social Ratings Database, in line with China's national conditions, CSIR events are divided into six types or "issue areas": community, human rights, employee, environment, product and corporate governance. This study's research samples consist of news about the top 100 companies in the Chinese internet industry; these samples are representative and reliable. The characteristics of the internet industry and event categories covered by the KLD indicators are comprehensively considered, reflecting the samples' pertinence and practicality.

After exposure, the involved and their peer companies are affected. Most studies have focused on the negative impact of CSIR exposure on companies, including legal punishment, employee turnover, stock decline and brand depreciation.²⁷ Several studies have examined the spillover effect of CSIR exposure on peer companies, discovering both positive spillovers (where consumers appreciate the excellence of competing companies through comparison),²⁸ and negative ones (as consumers' suspicion of related companies is aroused).²⁹ However, these spillover effects are indirect influences and are described from the consumer perspective. This study focuses on the direct impact of CSIR exposure on peer companies. By determining whether these companies exhibit similar CSIR behaviour within a period after CSIR exposure, the deterrent effect of such exposure can be assessed. [End Page 151]

DETERRENCE THEORY

As viewed from the economic theory of criminal behaviour decision-making, the theory of deterrence is that by punishing the defendant, stipulating that the defendant pay the corresponding price, potential offenders can be deterred, making them realise that the cost of violations exceeds the benefits and forcing them into compliance.³⁰ According to deterrence theory, the exposure system can deter CSIR behaviour through several paths. First, exposure attracts the attention of local government or industry regulatory authorities. They may punish involved companies by law, issue regulatory documents and organise business leaders to talk to and conduct strict regulatory reviews of regions or industries prone to CSIR, thereby suppressing the motivation for it. Second, exposure alerts investors, creditors and other capital market stakeholders. For example, investors may doubt the CSR of peer companies in the same industry and speculate that these companies may have CSIR issues and that it would be prudent to sell their stock. Corporate executives may strive to improve corporate operations and suppress CSIR to cope with external pressures. Third, exposure attracts media and consumer attention to peer companies in a region or industry. Strengthening media supervision can facilitate the discovery of CSIR and reduce the public reputation of those involved. Fourth, exposure may allow corporate executives to witness the consequences of CSIR directly. They might consequently enhance their consciousness of self-observation and evaluation, focusing more on social evaluation. This generates guilt, shame and fear, thereby enhancing morality.

Deterrence theory uses three factors to measure the degree of deterrence: certainty, severity and celerity.³¹ Certainty is calculated based on the probability of enacting requisite penalties, severity is assessed based on the severity of the penalty result and celerity is evaluated based on the timeliness (speed) of punishment.³² In the context of the exposure of an incidence of CSIR, the model does not consider the celerity of exposure. Due to the development of network technology and the popularity of We-Media, the exposure of CSIR events is disseminated rapidly; however, there is no way to know about such incidents which have not been exposed. Therefore, this study defines two deterrent dimensions for CSIR exposure from the exposure source perspective: certainty and severity. While exposure celerity is used as the sole criterion for case selection, as only cases where exposure occurs in the same year as the event are chosen, this study also examines the impact of exposure target characteristics: corporate attitudes (CA) and cluster of events (COE). Figure 1 incorporates these into the logical framework. [End Page 152]

Characteristics of Exposure Targets

Corporate attitudes (CA)

Corporate attitudes (CA) in this study refer to the crisis management strategies of companies involved in CSIR after exposure. Coombs⁴³ divides company responses after crises into seven response strategies: attack accusers, deny, excuse, justify, ingratiate, correct and apologise, according to the standard of "exculpation-reconciliation". Griffin, Babin and Attaway⁴⁴ classify coping strategies into denial, reticence and apology. The addition of reticence considers a situation in which a company remains silent and does not respond to crises. This article combines both views to divide and measure CA. From an effectiveness perspective, positive CAs such as product recalls, taking responsibility and apologies can effectively restore consumers' willingness to purchase.⁴⁵ However, these strategies also acknowledge the authenticity of negative news. Counterarguments, denials and other defence strategies prevent public opinion from confirming the faults of the involved companies. Additionally, these strategies can resist negative information and increase loyal consumers' confidence, but they apply [End Page 155] only to defensible adverse events.⁴⁶ Therefore, the involved company's attitude is a response to the news of exposure. For other peer companies, this indicates whether the company involved can bear the exposure cost and whether the punishment after exposure has sufficient deterrence.

Cluster of events (COE)

Cluster of events (COE) refers to the number of companies involved in the exposed CSIR event. Exposure of clustered CSIR events is highly significant. Failure to promptly expose events and deter peer companies may result in imitation, a clear manifestation of the "contagion effect". In a highly competitive environment, companies may resort to plagiarism, cheap labour and false information to reduce costs, capture market share and improve corporate profits. When other companies in the industry regard these practices as default behaviour, they are encouraged to imitate them, pursuing the short-term performance gains from CSIR. However, as the number of CSIR companies in the same industry increases, a psychological paralysis of companies sets in, strengthening the belief that "the law cannot be enforced when everyone is an offender", which dilutes the sense of guilt due to lack of responsibility. Clustered CSIR affects corporate impressions, public trust in the industry and even national reputation.

In summary, CSIR exposure results from multirole and multilevel interactions. Based on the seven factors presented above, this study establishes a logical framework (Figure 1). The article uses fsQCA for empirical analysis due to the difficulty of investigating the interactive relationship between multiple conditions using traditional regression analysis.

Click for larger view
View full resolution

Figure 1.

Logical Framework of Corporate Social Irresponsibility (CSIR) Exposure

Notes: PMI denotes policy-oriented media involvement; MN, media number; EI, event influence; POP, proportion of public opinion; PUN, punishment; CA, corporate attitudes; and COE, cluster of events.

[End Page 156]

METHOD

Samples and Data

The authors chose to study exposed and deterred companies within the top 100 Chinese internet companies in 2019 for three reasons. First, the top 100 companies hold prominent positions in their respective industries and are well-known brands; they are therefore more likely to be subject to media scrutiny and public opinion. In addition, the news data for these companies are also more comprehensive and authentic. Second, the top 100 leading companies in the internet industry are representative and the deterrent effect of exposure for CSIR is more pronounced among them. Third, these leading companies have a higher proportion of intangible assets and are more concerned about brand image loss caused by CSIR exposure, making them more sensitive to the deterrence of such exposure. The "Top 100 Chinese Internet Companies" is an annual ranking list jointly released by the China Internet Association and the Information Center of the Ministry of Industry and Information Technology, which has been widely recognised for years. Table 1 presents the sample descriptions. [End Page 157]

Click for larger view
View full resolution

Table 1.

Characteristics of the Selected Chinese Internet Enterprises

The case samples used in this study are obtained from Zhiweidata (https://ef.zhiweidata.com/), a Chinese social information intelligence agency specialising in data analysis and mining big data in media. Zhiweidata has built a comprehensive ecosystem for discovering, tracking, mining and predicting social hotspot events, and it has established practical event impact evaluation standards. An event is included in the Zhiweidata event database if it meets one of the following criteria: (i) it achieves a high transmission volume within a short period, with a peak transmission speed of more than three news articles per hour; (ii) it maintains a certain level of transmission volume over a period, with an exposure duration of at least two days; or (iii) it generates widespread discussion on online social media platforms, making it one of the top 100 on the Weibo hot search list.

Following established studies, this article adopts the following screening criteria to determine the initial sample. First, the names, abbreviations and product brands of China's top 100 internet companies were used as search keywords to collect all CSIR exposure events from January 2015 to February 2020. Second, based on the sentimental scores for the events from Zhiweidata, only those with negative scores were included, while rumours that were clarified were excluded. Third, because clustered events may involve multiple companies, the same event listed under different company names is considered a duplicate and is registered only once, thus duplicate events are excluded. After applying these steps and removing samples with missing data, 296 CSIR events were identified and obtained. Of these, 20 incidents were exposed more than one year after they occurred and 276 incidents were exposed within a year. The authors focus only on the 276 events with exposure celerity, i.e. that were exposed rapidly; 198 CSIR events occurred between January 2015 and February 2019. For [End Page 158] the fsQCA analysis, these incidents were used as samples for the antecedent variable. By contrast, the case database from January 2016 to February 2020 was used to derive the outcome variable (the number of companies exposed to similar events within a year) for the statistics. The condition variables, such as media data including PMI, MN, EI and POP, were obtained from Zhiweidata. Other case data, such as COE, PUN and CA, were derived from text analysis of related news.

Basing its categorisation on the KLD Social Ratings Database standard, this article categorises CSIR events according to the characteristics of internet companies, identifying eight types of CSIR events related to them: products, employees, unfair competition, corporate governance, false information, vulgar information, personal privacy issues and criminal use (Table 2).

Click for larger view
View full resolution

Table 2.

Types of CSIR (Corporate Social Irresponsibility) Events Associated with Internet Companies

Conditions

PMI (policy-oriented media involvement)

Policy-oriented media comprises 18 authoritative media companies covering the entire country. According to the industry ranking, the list of these outlets is: People's Daily, Xinhua News Agency, Qiushi, People's Liberation Army Daily, Guangming Daily, Economic Daily, China Daily, China National Radio, CCTV, China Central International Radio, Science and Technology Daily, China Discipline Inspection and Supervision News, Workers' Daily, China Youth Daily, China Women's Daily, Farmers' Daily, Legal Daily and China News Service. Subordinate media included in each policy-oriented media company, such as TV channels, network channels, Weibo company accounts and WeChat public CCTV accounts, are counted separately. The indicator PMI refers to the proportion of policy-oriented media relative to the total exposure media for each event and it is represented as a continuous variable ranging from 0 to 100 per cent. The Fiss⁵² calibration method is adopted, with percentage values of 95 per cent, 50 per cent and five per cent set as the thresholds for full membership, intersection and full non-membership, respectively.

MN (media number)

Zhiweidata has established a media authority index based on a list of media outlets issued by the National Cyber Office and news generated by these media outlets is considered authoritative so that it is allowed to be republished by other websites and by senior media professionals in their op-eds. Key media items are listed based on this index. MN refers to the number of crucial media reporting the exposure of a CSIR event, and the data present as continuous variables. According to the Fiss⁵³ calibration method, values of 95 per cent, 50 per cent and five per cent are set as the thresholds for full membership, intersection and full non-membership, respectively. [End Page 160]

EI (event influence)

In this study, EI refers to the communication effect of a single event on the internet, measured by the number of times the event was mentioned. The values are automatically identified and calculated using the Zhiweidata event database. The number of reports on self-media (predominantly Weibo and WeChat) and online media, as well as the EI index ranging from 0 to 100, are obtained through normalisation operation. The Fiss⁵⁴ calibration method is adopted, setting values of 95 per cent, 50 per cent and five per cent as the thresholds for full membership, intersection and full non-membership, respectively.

POP (proportion of public opinion)

Proportion of public opinion refers to the average value of public opinion on the exposed CSIR event relative to the total public opinion of all exposed events within 24 hours. The value is automatically identified and calculated using the Zhiweidata event database, reflecting the communication effect of occurrences on self-media (predominantly Weibo and WeChat) and online media per unit of time. This is statistically analysed according to communication nodes, information sources, number of comments, intensity of speaking time and so forth. The data are continuous variables ranging from 0 to 100 per cent. The Fiss⁵⁵ calibration method is adopted, with 95 per cent, 50 per cent and five per cent values considered as the threshold values for full membership, intersection and full non-membership, respectively.

PUN (punishment)

The authors determine the nature and severity of punishment inflicted on a company by the government through text analysis of news on each event. Punishment ranges from no penalty to product delisting or suspension of business, and is divided into four levels: "no penalty" (–1), "criticism" (0), "fine" (1) and "product delisting or suspension of business" (2). Criticism refers to the government's public expression of disapproval concerning specific actions of businesses, along with its request for those businesses to rectify such actions. Based on the Fiss⁵⁶ calibration method, 95 per cent, 50 per cent and five per cent values are used as the thresholds for full membership, intersection and full non-membership, respectively. [End Page 161]

CA (corporate attitudes)

As mentioned above, according to the division of corporate crisis public relations strategies classified by Coombs⁵⁷ and Griffin, Babin and Attaway,⁵⁸ corporate response attitudes are divided into seven strategies based on the "denial–apology" spectrum: "counterattack" (1), "denial" (2), "silence" (3), "defence" (4), "catering" (5), "correction" (6) and "apology" (7). Based on the Fiss calibration method⁵⁹ at 95 per cent, 50 per cent and five per cent threshold values, this article considers seven as the full membership threshold, four as the intersection point and one as the threshold for full non-membership, respectively.

COE (cluster of events)

COE refers to the number of companies exposed in a CSIR event. If the exposed CSIR event involves more than one company, it is a collective event. Collective events often have more severe consequences and attract greater social attention. In this study, COE is treated as a dichotomous condition, where a value of 1 indicates two or more top 100 internet companies are involved and a value of 0 indicates that only one company is involved.

Click for larger view
View full resolution

Table 3.

Calibration of Results and Conditions

RESULTS

Necessity Analysis of a Single Condition

Similar to mainstream QCA studies, this article first examines whether a single condition (including its non-set) constitutes a necessary condition for high deterrence (~N). In [End Page 162] set theory, the necessity analysis of a single condition involves determining whether the result set is a subset of a certain condition set. In fsQCA, when a result occurs, a condition always exists that is necessary for such an occurrence.⁶⁰ Consistency is a crucial criterion for measuring the necessity of a condition; when the consistency level exceeds 0.9, the condition can be deemed necessary for the result.⁶¹ Table 4 presents the test results of the necessary conditions for high deterrence of unethical CSIR (~UN) and illegal CSIR (~IN) analysed using fsQCA 3.0. The results show that the consistency of all conditions does not exceed 0.9. Therefore, among the seven conditions, no single condition is a necessary condition for high deterrence of unethical CSIR (~UN) or illegal CSIR (~IN), indicating that each causal condition provides a weaker explanation for the deterrence of CSIR exposure.

Click for larger view
View full resolution

Table 4.

Necessary Conditions and Sufficiency Test of Condition Variables

Fuzzy-set Qualitative Comparative Analysis

The samples are grouped according to the degree of government participation. Those who violate the law and are investigated or punished by the government are grouped as the illegal group. By contrast, those who do not infringe on the law but breach moral standards and are exposed by the media are classified as the unethical group. The illegal behaviour of enterprises is punished by the government according to the law and simultaneously exposed by the media. Consequently, peer enterprises receive [End Page 163] deterrence from both legal and media exposure. Moral violations are not punishable by the law because they are not violations per se but breach socially recognised moral standards. However, moral violations when exposed by the media tend to result not only in a negative reputation, but also a loss of intangible assets and a decline in stock prices. Peer enterprises are deterred only through media exposure. Since the deterrent effects differ between the unethical and illegal groups, it is meaningful to discuss fsQCA results in groups. Of the 198 samples, 107 are from the unethical group and 91 are from the illegal group.

FsQCA yields three types of solutions: complex solutions (excluding "logical residues"), parsimonious solutions (including "logical residues" but without evaluating their rationality) and intermediate solutions (including "logical residues" that coincide with theoretical and practical knowledge). A key advantage of intermediate solutions is that they do not eliminate necessary conditions. Generally, the intermediate solution is superior to others.⁶² The core and peripheral conditions of the configuration are distinguished based on the parsimonious and intermediate solutions. If a causal condition appears in both the parsimonious and intermediate solutions, it is a core condition that significantly impacts the result; if it appears only in the intermediate solution, it is recorded as a peripheral condition, contributing to auxiliary outcomes.⁶³

FsQCA 3.0 is used to analyse the data of 198 internet company CSIR events. After comprehensively considering the following four best practices for QCA method application, a reasonable and natural threshold is determined. First, the truth table rows (configurations) with results of 0 and 1 should be covered and roughly balanced.⁶⁴ Second, 1 is selected as the frequency threshold for the case, including all truth table rows covered by the case included in the logical minimisation process. Third, the consistency threshold of the results exceeds 0.8.⁶⁵ The minimum proportional reduction in inconsistency (PRI) value should be ≥0.60 to reduce the potential for contradictory configuration. Fourth, possible simultaneous subset relations must be avoided; that is, a particular truth table row (configuration) serves as a sufficient configuration for both complete and partial mergers.⁶⁶ Based on the aforementioned four best practice standards, the consistency threshold of this study is determined to be 0.8, the PRI value is 0.7 and the frequency threshold is 1. The calculation results, listed in Table 5, identify four configurations with high deterrence: two for the illegal group and two for the unethical group. Following the presentation form of the results of Fiss,⁶⁷ filled circles indicate the presence of conditions, crossed circles indicate the absence of conditions and spaces indicate a fuzzy state (the condition may be present or absent). [End Page 164] Large circles indicate the core conditions (those that exist in both the parsimonious and intermediate solutions) and the small circles indicate supplementary conditions (those that exist only in the intermediate solution).

Click for larger view
View full resolution

Table 5.

Configurations for High Deterrence of Unethical and Illegal CSIR

The fsQCA results demonstrate that two configurations of unethical CSIR lead to high deterrence and the consistency of the single solution (configuration) exceeds 0.80. The overall solution's consistency reaches 0.83, indicating that these two configurations are sufficient for high deterrence. The overall solution's coverage exceeds 50 per cent, suggesting that these two configurations explain the main reasons for high deterrence. Furthermore, QCA indicates that two configurations for illegal CSIR lead to high deterrence: the consistency of individual solution (configuration) and overall solution exceeds 0.80, and the coverage is nearly 50 per cent, explaining the primary reasons for high deterrence in the illegal group.

Table 5 shows that two pathway configurations produce a high deterrent effect (~U1, ~U2) in the unethical group. ~U1 indicates that in the non-clustered (peripheral condition) CSIR events, regardless of the POP condition, having high PMI (core [End Page 165] condition), high MN (core condition), high EI (core condition) and active CA (core condition) produces a high deterrent effect. ~U2 indicates that for a CSIR event denied by the enterprise (peripheral condition), high PMI (core condition), high MN (peripheral condition), high EI (core condition) and high POP (core condition) can produce high deterrence.

Additionally, two pathway configurations produce a high deterrent effect (~I1 and ~I2) in the illegal group. When the events are clustered (core condition), regardless of the CA condition, the presence of all six remaining conditions—COE (core condition), EI (core condition), PUN (core condition), PMI (peripheral condition), MN (peripheral condition) and POP (peripheral condition)—can produce a high deterrent effect. When the events are non-clustered (core condition) with low POP (core condition), the presence of the other five conditions—MN (core condition), PUN (core condition), CA (core condition), PMI (peripheral condition) and EI (peripheral condition)—becomes important for producing high deterrent effects.

Stability Test

The QCA results may vary due to the different calibration thresholds. Therefore, adjusting the calibration threshold is considered an effective stability testing method.⁶⁸ This article utilises 80 per cent, 50 per cent and 20 per cent, instead of 75 per cent, 50 per cent and 25 per cent, as the thresholds for full membership, intersection and full non-membership of result variables, respectively. Keeping the other conditions constant, the configuration analysis is performed again. The configurations remain robust, with only subtle changes in the consistency and coverage of each configuration and the overall solution. This demonstrates the validity of the study's conclusions.

DISCUSSION

Given the complex configuration of multiple governments, media and company factors, this article identifies four configuration paths that activate high-deterrence exposure. With regard to unethical CSIR exposure, PMI, MN and EI appear in each configuration, with both PMI and EI as core conditions, indicating that media play a key role. Moreover, policy-oriented media are the most critical exposure sources. Due to their official status, policy-oriented media can easily obtain relevant and authoritative information from the government and their reports are more likely to attract government attention. However, unlike ordinary media, policy-oriented media do not prioritise profit as their objective; therefore they forge limited public relations with companies and do not allow any opportunities to quibble or deny. Comparing ~U1 and ~U2, this article finds that when other events may attract public attention during unethical CSIR exposure (~U1), significant media exposure and a positive CA involving confession, apology and rectification are required for high deterrence. However, when [End Page 166] the involved enterprise has a negative attitude (~U2), being more attention-grabbing than any other news item becomes the determining factor for high deterrence, regardless of the scale of the event.

With regard to illegal CSIR exposure, media play an important role, as PMI, MN and EI appear in each configuration. Additionally, PUN is a core condition in both illegal group pathways, reflecting China's strict supervision of internet companies. Laws are continually updated and improved as the industry develops, and the government actively regulates corporate behaviour. ~I1 indicates the exposure path of clustered CSIR events with three common characteristics. First, severe punishment and exposure of incorrigible companies appear in clustered CSIR events. Second, such events are often caused by large-scale government rectification actions in the industry. Third, the companies involved are subject to extensive media exposure, resulting in high POP. Even if companies remain silent, defend themselves, or deny the accusations, they cannot mitigate exposure costs, ranging from negative word-of-mouth to removal of products from the stores. These measures serve as a significant warning to all companies in the industry. ~I2 indicates the exposure path of non-clustered CSIR events, particularly those that the public ignores. In non-clustered events, which often have minimal or non-severe consequences and losses, media attention and public opinion may be less prominent. The key to high deterrence lies in the attitudes of the companies involved. Confessions, apologies and rectifications by the involved companies demonstrate the effectiveness of government actions and media exposure.

Finally, the configurations are compared across the unethical and illegal groups. First, the presence of media conditions (PMI, MN and EI) plays a significant role in determining the exposure-deterrence effect, as all four configurations include the PMI, MN and EI conditions. This indicates that exposure certainty and severity are crucial for achieving a high deterrence effect and is a finding consistent with deterrence theory. Second, when comparing clustered and non-clustered CSIR exposure paths, CA emerges as the core causality condition in non-clustered events for both unethical and illegal groups, indicating that CA is a critical factor for producing high deterrence for non-clustered CSIR exposure. Third, when comparing illegal and unethical CSIR events, all unethical configurations presented in the results have PMI as the core condition, whereas all illegal configurations have PUN as the core condition and PMI as the peripheral condition. Therefore, in the absence of legal PUN, exposure by authoritative media serves as a substitute to cause deterrence. Finally, the existence of these four configurations demonstrates the diversity of pathways to achieving high-deterrence exposure. This highlights the advantages of QCA in explaining the configuration effect among various factors.⁶⁹ [End Page 167]

CONCLUSIONS, INSIGHT AND OUTLOOK

This study establishes the causal mechanisms inherent in the exposure of CSIR to determine its deterrent effects. From the perspective of peer companies, seven critical exposure conditions are analysed using configuration and QCA. This article uses the CSIR events from the top 100 Chinese internet companies from 2015 to 2020 as samples to validate these mechanisms. The results indicate that there are two pathways each for the exposure of unethical and illegal CSIR to achieve a strong deterrent effect.