Constraints under the Halo: The Constraining Effect of Corporate Reputation on Corporate Social Responsibility Behavior

Abstract

This study explored the multidimensional influence of high reputation on corporate social responsibility behaviors from the “report disclosure”, “report whitewashing”, and “actual performance” perspectives. The study found that a high reputation may trigger high expectations and strict supervision among stakeholders, which may cause those companies with high reputations to disclose high-quality CSR reports and improve the quality of their CSR performance. The results also indicated that, under the influence of a reputation-constraining mechanism, the degree of whitewashing in the CSR reports of high-reputation firms is significantly lower than that of others. This study focused on facilitating a better understanding of the influence of organizational reputation on organizational behavior and providing theoretical support and practical insights regarding the improvement of the effectiveness of corporate social responsibility governance.

1. Introduction

Corporate reputation is a critical source of competitive advantage for businesses [1]. Existing research often considers a good reputation as a goal pursued by organizations and emphasizes analyzing the motivations behind companies seeking a high reputation and identifying ways to enhance their reputation levels [2,3]. The recognition of a company’s concept, product, or service by government entities, consumers, and stakeholders can send a signal to external investors, commonly known as the ‘halo effect’. However, does a high-level corporate reputation always bring benefits to companies? In early 2020, as the COVID-19 pandemic struck Wuhan, Huawei promptly donated CNY 30 million to the city on 26 January. However, it unexpectedly faced criticism from many netizens, and even moral pressure. Some netizens suggested that, as a national enterprise, Huawei should have contributed more during the pandemic situation. In this incident, the “halo” of a high reputation brought significant pressure to Huawei due to public opinion. Given this context, it is evident that a high-level corporate reputation can both benefit a company and burden it [4,5].

Upon reviewing the existing literature, scholars have noted the potential hidden burdens that the halo of a high reputation may impose on companies. However, most of these studies primarily focus on the impact that external stakeholders have on companies under the halo of a high reputation, such as declining market shares and fluctuating stock prices [4], among other factors. To date, only a few scholars have thoroughly explored the influence of the burden of a high reputation on organizational decision-making behaviors, such as preferences in merger decisions and in strategic alliance decisions [6,7]. However, research on the influence of a high reputation on the preferences for corporate social responsibility (CSR) behaviors remains scarce. CSR entails companies balancing the interests of various stakeholders, including shareholders, employees, consumers, communities, and the environment, while pursuing profit maximization [8,9]. CSR behaviors comprehensively reflect a company’s efforts to address the expectations of different stakeholders [10]. Using CSR behaviors as a starting point to examine the constraining effect of a high reputation on corporate behavior, it is possible to effectively supplement theoretical research on the influence of organizational reputation on non-economic behaviors.

In China, the governance of corporate social responsibility (CSR) poses significant challenges. CSR practices involve various dimensions, which prevents comprehensive oversight by governmental bodies. While China’s formal governance framework provides minimum standards for CSR, it does not impose mandatory requirements beyond these standards. As a result, the effectiveness of formal institutions in influencing CSR activities is limited. In light of this, policymakers and scholars are increasingly focusing on exploring external mechanisms, such as market-based incentives, public pressure, and voluntary industry guidelines, to enhance CSR practices and optimize governance frameworks.

Compared to existing research, the marginal contributions and possible innovations of this study are as follows. ① Existing studies mostly view a high reputation as an incentive for companies to actively fulfill their social responsibilities, exploring how corporate social responsibility (CSR) influences corporate reputation [11,12]. However, little attention has been paid to how companies adjust their CSR behaviors in future operations after receiving high reputation evaluations from stakeholders. This study focuses on the constraining effect of reputation mechanisms on CSR behaviors, thereby enriching the theoretical research on the constraining effects of organizational reputation on organizational behavior to some extent. ② When researching CSR-related issues, some scholars choose to use the quality of CSR reporting disclosure as a substitute variable for the quality of CSR performance [13,14], overlooking the possibility of whitewashing behaviors in CSR reports. This study examines the performance of companies in three dimensions, namely CSR report disclosure, actual CSR performance, and CSR report whitewashing, revealing the multidimensional impacts of a high reputation on CSR behaviors.

2. Theoretical Analysis and Research Hypotheses

2.1. Review of Related Studies

Corporate reputation has both resource and signaling attributes. It is not only an important intangible asset enabling enterprises to enhance their competitive advantage, but also reflects stakeholders’ comprehensive evaluation of the enterprise and their expectations for its future behavior [15,16,17]. In early studies on organizational reputation, scholars thoroughly explored the ways in which enterprises acquire and maintain a good reputation. Fombrun and Shanley found that the public’s evaluation of a firm is actually the result of a combination of factors, such as the firm’s financial performance, market performance, media coverage, and other non-economic behaviors [18]. Petkova et al. classified corporate reputation into overall reputation and local reputation and found that investment in human capital and social capital can effectively enhance the overall reputation of a firm, whereas investment in product quality and customer relationship management mainly enhances the reputations of enterprises in specific regions [19].

Scholars have also delved into the relationship between CSR, corporate reputation, and firm performance. Aguilera-Caracuel and Guerrero-Villegas found that the fulfillment of CSR by multinational corporations (MNCs) in developing countries can satisfy the expectations of some stakeholders and thus enhance the reputations of firms [11]. Hur et al. showed that the socially responsible behavior of corporations contributes to their reputation and brand credibility [20]. Lange et al. argued that corporate reputation mediates the effect of CSR on employee satisfaction and investor loyalty [21].

With the in-depth study of reputation theory, scholars have gradually realized that corporate reputation may be a “double-edged sword”: although it can help enterprises to gain a competitive advantage, it may also become a burden [22]. Teng et al. investigated the impact of environmental, social, and governance (ESG) practices on the risk taking of enterprises in high-tech industries. They found that although ESG practices can reduce risks, excessive disclosure or improper resource allocation may also increase the financial burden on enterprises [23]. In addition, Huang et al. showed that government-driven corporate social responsibility (CSR) initiatives may negatively affect firms’ profitability under certain conditions, especially in environments with slack resources and competitive markets, where CSR practices may become a burden to firms [24].

In recent years, scholars’ research focus has shifted from the impact of corporate reputation on the behavioral choices of external stakeholders to its impact on the behavioral decisions of the enterprise itself. Brooks et al. explored the pressure brought by a high reputation “halo” to venture capital firms and its impact on the company’s investment decisions in an ambiguous situation. Their study found that some firms with a high reputation “halo” tend to make high-risk investments in response to investors’ and firms’ own expectations of high returns on investment [25]. Haleblian et al. suggested that high-reputation firms need to cater to high investor expectations regarding their financial performance, which results in firms having to engage in more M&A activities and significantly more cross-industry mergers and acquisitions than non-high-reputation firms [6]. However, the above studies mainly focus on the impact of a high reputation on firms’ economic behavioral preferences, and there are relatively few studies on the impact of a high reputation on corporate social responsibility behavior. This study takes CSR behavior as an entry point to comprehensively examine the multidimensional impact of a high reputation on CSR behavior and its internal mechanism.

In the field of corporate social responsibility (CSR), when analyzing and evaluating CSR behaviors, researchers often focus on only one aspect of an enterprise’s performance, in terms of its “words” or “deeds”. For example, some scholars measure the fulfillment of corporate social responsibility through the disclosure quality of CSR reports. However, the “disclosure” and “fulfillment” of CSR correspond to “words” and “deeds”, respectively [25]. China’s CSR disclosure is generally concentrated between March and October of the following year, i.e., “action” first and “words” later, which does not rule out the possibility of “overstatement” [26]. Some studies have shown that China’s social culture of “reporting the good news but not the bad news” affects the information content of CSR reports, resulting in the tendency of some enterprises to use positive but uncertain language in their reports to influence the judgment of stakeholders, despite their poor performance in terms of actual social responsibility [27]. Further research by Wu et al. suggests that media coverage plays an important mediating role between corporate CSR disclosure and stakeholder responses. Positive and neutral reports can promote corporate innovation, while negative reports may adversely affect the corporate image and market reaction [28]. Based on the above analysis, this study argues that it is necessary to examine CSR behaviors from both the “words” and “deeds” perspectives. Therefore, it investigates the impact of a high reputation “halo” on “social responsibility report disclosure”, “CSR fulfillment”, and “social responsibility report whitewashing” (Figure 1), and explores the internal mechanisms of such influences.

2.2. Impact of High Reputation on Quality of CSR Reporting Disclosure

A reputation constraint mechanism means that a good reputation restricts corporate behavior by attracting higher stakeholder expectations and broader stakeholder attention [4,6]. Among these, the high expectations of stakeholders and the easily induced expectation gap are important factors influencing CSR reporting disclosure decisions.

High levels of corporate reputation lead stakeholders to have higher expectations for the company’s future behavior. In situations of information asymmetry, stakeholders often judge a company’s future behavior based on its past performance [29]. A strong reputation implies that the company has consistently met the diverse demands of multiple stakeholders at a high standard over an extended period [4,30]. This leads stakeholders to believe that high-reputation companies will significantly outperform others in their future behavior. According to the expectancy violation effect, the higher the expectations held by evaluators for the behavior subject, the greater the psychological gap when the behavior subject fails to meet the evaluators’ expectations [31]. Compared to other companies, high-reputation companies that fail to effectively meet stakeholders’ expectations must endure greater gaps in stakeholder expectations and more negative evaluations. Therefore, companies enveloped in a high-reputation “halo” are more strongly motivated to actively and qualitatively respond to stakeholder expectations in their future business development [6]. Existing research confirms that good companies prioritize reputation management more than poor companies [1]. The higher their reputation levels, the more companies care about whether they can provide stakeholders with high-quality expectation feedback [32].

Different types of stakeholders have varying expectations of companies, and high-reputation companies need to respond comprehensively to the diverse demands of different types of stakeholders [33]. Stakeholders directly involved with the company focus on its profitability and value growth, while entities such as governments, communities, and the general public expect the company to take responsibility for community development, environmental protection, and social issues [21,34]. If a company only addresses the demands of a specific stakeholder category while ignoring others, the interaction among stakeholders is likely to diminish the evaluations of those stakeholders who have already been satisfied. Existing research indicates that actively pursuing social responsibility can help companies to simultaneously gain recognition from different types of stakeholders, as compared to economic activities such as diversification [35,36]. As a crucial medium for communication between companies and stakeholders, corporate social responsibility reports serve as an essential means for companies with high reputation ratings to showcase their social responsibility performance and address the diverse demands of their stakeholders. The high-quality disclosure of social responsibility reports helps companies to establish a reputation for being socially responsible, enhances stakeholders’ identification with the company [37], and increases stakeholders’ loyalty to the company [25]. Given the significant role of social responsibility reports in shaping the company image and meeting stakeholders’ high expectations, this study posits that high-reputation companies are more likely to prioritize the use of corporate social responsibility reports as communication tools. Therefore, compared to companies without high reputation ratings, those with high reputation ratings are expected to have higher-quality social responsibility report disclosures. They are anticipated to respond to stakeholders by disclosing higher-quality social responsibility reports. Based on this, the following hypothesis is proposed in this study.

H1. 

Keeping other conditions constant, the CSR report disclosure quality of companies with high reputation ratings is higher than that of companies without high reputation ratings.

2.3. Impact of High Reputation on CSR Performance Quality

A high reputation may positively influence the quality of CSR performance for several reasons.

First, the ethical values and behavioral inertia of high-reputation companies may drive them to actively and effectively fulfill their social responsibilities [38]. As “good citizens”, high-reputation companies likely possess stronger ethical values [39] and pay more attention to social responsibilities beyond economic and legal obligations, such as community, environmental, and employee responsibilities. Therefore, they are likely to demonstrate their values and high moral standards to the outside world by fulfilling their social responsibilities with high quality [40]. Furthermore, a good reputation is the result of long-term investment and accumulation by companies [41]. Companies currently holding high reputations may have already formed habits of actively fulfilling social responsibilities in previous periods. Under the flywheel effect, companies will continue to perpetuate and enhance their behavioral inertia [42,43]. Behavioral inertia will have a sustained impact on the company’s future social responsibility actions.

Second, changes in the level of corporate reputation can affect the relationship between the company and its stakeholders [21]. After receiving high reputation ratings, the increased attention from stakeholders imposes a greater “concern burden” on the company, compelling it to improve the quality of its social responsibility performance. The media play a crucial role in exerting supervisory pressure on high-reputation companies. Before receiving high reputation ratings, only a few media outlets may pay attention to the company’s behavior. However, once a company is recognized as having a high reputation, it easily becomes the focus of media attention [4]. According to the evaluation criteria of news value, the prominence of the subject is positively correlated with the news value. The more familiar the subject, the higher the significance value of the news, making it easier for the negative behavior of good companies to attract public attention [44]. Therefore, compared to companies that have not received high reputation ratings, the media will pay closer attention to high-reputation companies, especially to their negative behavior [4]. In this context, if high-reputation companies evade their legal obligations related to social responsibility or passively fulfill their social responsibilities, their probability of exposure will be higher than that of companies that have not received high reputation ratings. Psychological studies have shown that when both positive and negative information is available, people tend to pay more attention to negative information [45]. This leads some stakeholders to firmly believe that they have sufficient reason to negatively evaluate high-reputation companies [46], thereby causing companies that have received high reputation ratings to suffer higher reputation and economic losses.

Finally, before and after a company receives a high reputation, stakeholders’ evaluation criteria for corporate social responsibility behavior may change. Before a company receives high reputation ratings, stakeholders typically use the company’s past performance or the industry’s average performance as the evaluation criteria. However, once a company successfully receives high reputation ratings, the stakeholders will instead use the social responsibility performance of similarly reputable companies as the reference standard. Therefore, to avoid the expectation gap caused by changes in evaluation criteria, companies that have received high reputation ratings have a stronger motivation to improve the quality of their social responsibility performance. Their performance in social responsibility will be significantly better than that of companies that have not received high reputation ratings. Based on the above analysis, this study proposes the following.

H2. 

Keeping other factors constant, the CSR performance of companies receiving high reputation ratings is higher than that of companies not receiving high reputation ratings.

2.4. Impact of High Reputation on Degree of CSR Report Whitewashing

While a high reputation has an impact on the disclosure quality and performance quality of CSR reports, it may also have an impact on their information content. The quality of CSR reports pertains to the overall structure, diversity of content, and technicality of information disclosure [47,48], while the content of CSR reports concerns the amount of information that they convey regarding the actual CSR performance. CSR reports serve as vital communication tools for companies to disclose their CSR information to the public [49]. In order to enhance stakeholder approval, companies may resort to tactics such as linguistic manipulation and data adjustment to whitewash their CSR reports [27]. Such whitewashing behavior diminishes the information content of CSR reports, making it challenging for them to accurately reflect a company’s actual CSR performance. Studies by Hemingway and Maclagan [50] and Guo [51] have shown that CSR reports can exhibit a “masking effect”, where reports with low information content may serve as a means for some companies to divert shareholder attention and conceal operational issues.

Although whitewashing behavior in corporate social responsibility (CSR) reports is difficult to detect, high-reputation companies become the focus of media and competitors’ attention. Thus, high-reputation companies need to handle whitewashing issues with caution. After receiving high reputation evaluations, companies’ core competencies are enhanced, prompting competitors to reassess their competitive strength and actively uncover negative information related to high-reputation companies. Considering the newsworthiness of high-reputation companies, the media are inclined to actively track and continuously report negative information about them [4]. Therefore, whitewashing behavior by high-reputation companies is more likely to be exposed. Once exposed, such companies will be perceived by their stakeholders as lacking credibility [52]. Moreover, the stakeholders will reassess the companies’ past behaviors based on the latest revelations [21]. Under the continued tracking and selective disclosure of the media, stakeholders’ negative evaluations of companies are likely to be continually magnified [5,53]. The losses brought to high-reputation companies by whitewashing in CSR reports will far exceed those of non-high-reputation companies. Therefore, we believe that when facing the choice of whether to whitewash their CSR reports, high-reputation companies will behave more cautiously. They will pay more attention to the information content of the CSR reports and reduce or avoid the use of whitewashing behavior. Based on this, we propose the following hypothesis.

H3. 

Under unchanged conditions, companies that receive high reputation evaluations will engage in less whitewashing behavior in their CSR reports compared to companies that have not received high reputation evaluations.

3. Research Design

3.1. Sample Selection and Data Sources

This study aims to investigate the impact of the high reputation “halo” on corporate social responsibility behavior. Therefore, it is necessary to select samples of high-reputation companies. Drawing from existing research, this study adopts whether a company is listed in Fortune magazine’s Most Admired Companies list (MAC) in China from 2010 to 2019 as a measure of whether the company has received high reputation evaluations [16]. The study also selects companies listed on the Shanghai and Shenzhen stock exchanges that are included in the Fortune China 500 list (referred to as Fortune 500 companies) as the overall sample for analysis. The study excludes the following: ① samples with significant data deficiencies that cannot be manually supplemented; ② ST- and PT-type listed companies; ③ financial and insurance industry listed companies; ④ companies with only one observation value. In the end, a total of 1372 company-year observation samples are obtained. To mitigate the influence of extreme values, this study conducts winsorization at the 1% and 99% levels for all continuous variables. The company reputation data used in this study are sourced from the Chinese website of Fortune magazine, and the data on the quality of social responsibility report disclosure are sourced from the Rankins CSR Ratings (RKS). Financial data and corporate governance-related data are sourced from the WIND database and the CSMAR database. The data used are cross-referenced to reduce statistical errors.

3.2. Variable Definitions

3.2.1. Independent Variable

Corporate Reputation ($\mathrm{Reputation}$). If a company is listed in the Fortune magazine MAC list published in year $\mathrm{t}$, it indicates that the company has a high reputation in year $\mathrm{t}$, with a ${\mathrm{Reputation}}_{\mathrm{t}}$ value of 1. If the company is not listed in the MAC list, the value is 0.

This study chooses whether a company is listed in Fortune magazine’s Most Admired Companies list (MAC) in China as the criterion to determine whether the company has a high reputation. This choice is based on two considerations. First, the evaluation of corporate reputation should primarily be based on the perceptions of stakeholders [54], and the evaluation results need to have a broad influence. Fortune magazine conducts MAC evaluations in several economically active countries, such as the United States, the United Kingdom, Germany, and China. The MAC list has extensive dissemination and influence, making it one of the most widely used methods of obtaining corporate reputation data. Inclusion on the MAC list indicates that a company has a high level of reputation [55]. Second, the MAC list is widely used in academic research. Many scholars at home and abroad have used MAC list data to measure whether companies have a high reputation [16,56,57]. In the robustness test section, this study also constructs a substitution variable for corporate reputation based on the annual newspaper coverage data of companies.

3.2.2. Dependent Variable

Disclosure Quality of Corporate Social Responsibility (CSR) Reports ($\mathrm{CSR}\_\mathrm{Dis}$). Drawing from existing research, this study measures the disclosure quality of corporate social responsibility (CSR) reports using the Rankins CSR Ratings (RKS) database. It evaluates the overall macrocosm, content, and technique (MTC) of CSR reports. The RKS is currently one of the most authoritative databases used to rate the quality of CSR reports in China.

Corporate Social Responsibility (CSR) Performance ($\mathrm{CSR}\_\mathrm{Perf}$). Drawing from Clarkson [58] and Zhang [59], this study assesses CSR performance in terms of five dimensions: shareholders, employees, suppliers, creditors, and government. Responsibility to employees is measured using the level of employee benefits. Responsibility to suppliers is assessed using the accounts payable turnover ratio and cash-to-accounts payable ratio. Responsibility to shareholders is evaluated based on the earnings per share and net assets per share. Responsibility to creditors is measured as the reciprocal of the current ratio and the debt-to-asset ratio. Responsibility to the government is assessed using the net tax amount, tax rate on assets, and number of jobs provided. Principal component analysis (PCA) is employed to combine these indicators and analyze the CSR performance. In the PCA process, factors with eigenvalues greater than 1 and cumulative variance contributions exceeding 80% are extracted. The weighted sums of the extracted principal components, based on their variance contributions, are then computed to derive $\mathrm{CSR}\_\mathrm{Perf}$ scores for each sample company, serving as a measure of the CSR performance.

The Degree of CSR Report Whitewashing ($\mathrm{MCSR}\_\mathrm{Dis}$). Following the research of Core [60], Brick [61], and Jiang [62], who measured managers’ or executives’ excess compensation using residuals of actual observations and estimated values, this study employs residuals between the actual CSR report disclosure quality and the expected CSR report disclosure quality determined by various influencing factors to measure the degree of CSR report whitewashing. The expected CSR report disclosure quality is estimated using Equation (1):

$$CSR\_Di{s}_{it}={\alpha }_0+{\alpha }_{1}CSR\_Per{f}_{it}+{\alpha }_{2}CSR\_Di{s}_{it-1}+{\alpha }_{3}Lnsiz{e}_{it}+{\alpha }_{4}RO{A}_{it}+{\alpha }_{5}Le{v}_{it}+\epsilon$$

(1)

In Equation (1), $\mathrm{CSR}\_{\mathrm{Dis}}_{\mathrm{it}}$ is the disclosure quality of the current corporate social responsibility report; $\mathrm{CSR}\_{\mathrm{Perf}}_{\mathrm{it}}$ is the current corporate social responsibility performance quality; $\mathrm{CSR}\_{\mathrm{Dis}}_{\mathrm{it}-1}$ is the disclosure quality of the $\mathrm{t}-1$ corporate social responsibility report; ${\mathrm{Lnsize}}_{\mathrm{it}}$ is the natural logarithm of the company size; ${\mathrm{ROA}}_{\mathrm{it}}$ is the rate of return on total assets; and ${\mathrm{Lev}}_{\mathrm{it}}$ is the asset liability ratio.

Specifically, first, the regression is conducted for the sample companies annually and by industry using Equation (1) to obtain the regression coefficients. Then, using Equation (1) and these regression coefficients, the expected quality of CSR report disclosure for each company is estimated. The difference between the actual CSR report disclosure quality and the expected CSR report disclosure quality is calculated as the degree of CSR report whitewashing, denoted as $\mathrm{MCSR}\_\mathrm{Dis}$.

3.2.3. Control Variables

Drawing from previous research, this study controls for the following variables: profitability (return on total assets, volatility of return on total assets, three-year average profit growth rate), debt-paying ability (asset–liability ratio), growth capability (operating revenue growth rate), market value (Tobin’s Q), stock price risk (risk assessment coefficient), equity concentration, equity restriction, institutional investor shareholding ratio, board size, board independence, managerial power, firm size, and year and industry effects [63]. The specific definitions of each variable are presented in

Table 1. Variable definitions and calculation methods.

Type	Symbol	Variable Name	Computing Method
independent variables	Reputation	Corporate reputation	If the enterprise is included on the MAC list, the variable’s value is 1; otherwise, the value is 0
dependent variables	CSR_Dis	Quality of CSR report disclosure	Quality evaluation of social responsibility reports of listed companies based on RKS database
	CSR_Perf	CSR performance quality	Using principal component analysis, the larger the value, the better the performance quality
	MCSR_Dis	Degree of CSR report whitewashing	See Equation (1) and relevant instructions for details
control variable	ROA	Return on total assets	2 × net profit/(total assets at the beginning of the period + total assets at the end of the period)
	V_ROA	Volatility of return on total assets	Standard deviation of ROA in past three years
	G_ROA	Three-year average profit growth rate	When ROAt-3 > 0, G_ROA = (ROAt/ROAt-3)^(1/3) − 1 When ROAt-3 < 0, G_ROA = (1 − ROAt/ROAt-3)^(1/3)
	Lev	Asset–liability ratio	Total liabilities/total assets
	Growth	Growth rate of operating revenue	(current operating income-previous operating income)/previous operating income
	Q	Tobin’s Q	Market price of enterprise/replacement cost of enterprise
	Beta	Risk assessment coefficient	Annual beta value of comprehensive market
	Top1	Equity concentration	Shares held by the largest shareholder/total shares
	Sbalance	Equity restriction	Shares held by the second to tenth largest shareholders/shares held by the first largest shareholder
	IInv	Shareholding ratio of institutional investors	Shares held by institutional investors/total shares
	Boardsize	Board size	Total number of directors
	IDrate	Board independence	Number of independent directors/total number of directors
	UnDual	Managerial power	Takes 1 when CEO and chairman are set separately; otherwise, it is 0
	LnSize	Firm size	Natural logarithm of total assets of company
	Ind	Industry	Dummy variable, classified by reference to 2012 CSRC industry classification standard
	Year	Year	Dummy variable

.

3.3. Model Design

To test H1, H2, and H3, this study constructs a regression model, as shown in Equation (2):

$$CSR\_Di{s}_{it}/CSR\_Per{f}_{it}/MCSR\_Di{s}_{it}={\alpha }_0+{\alpha }_{1}Reputatio{n}_{it-1}+{\alpha }_{2-15}Control{s}_{it}+{\alpha }_{16}Year+{\alpha }_{17}Ind+\epsilon$$

(2)

In Equation (2), ${\mathsf{\alpha }}_0$ is the intercept item, ${\mathsf{\alpha }}_1$~${\mathsf{\alpha }}_{17}$ is the regression coefficient of the corresponding variable, and $\mathsf{\epsilon }$ is a random error term. ${\mathrm{Controls}}_{\mathrm{it}}$ is the selected control variable, $\mathrm{Year}$ is the year, and $\mathrm{Ind}$ is the industry. ${\mathrm{Reputation}}_{\mathrm{it}-1}$ is the reputation level of the sample company in $\mathrm{t}-1$ year. In view of the possible interaction between corporate reputation and corporate social responsibility [12], this study uses the method of Haleblian [6] for reference when building the model, and it processes the dependent variable with a lag. It is obvious that the social responsibility behavior of enterprises in the $\mathrm{t}$ phase cannot affect the reputation of enterprises in the $\mathrm{t}-1$ phase. See Table 1 for the descriptions of the other variables in Equation (2).

4. Empirical Analysis and Discussion of Results

4.1. Descriptive Statistics and Variable Correlation Analysis

Table 2. Descriptive statistics results.

Variable	Mean	SD	Minimum	25th Quantile	Median	75th Quantile	Maximum
CSR_Dis	44.052	13.377	22.740	33.494	40.937	54.667	69.507
CSR_Perf	0.218	0.414	−0.462	−0.063	0.080	0.368	1.347
MCSR_Dis	−0.007	0.076	−0.292	−0.031	0	0.021	0.219
Reputation	0.072	0.258	0	0	0	0	1
Lev	0.595	0.164	0.054	0.490	0.619	0.715	0.952
Q	0.932	0.861	0.187	0.387	0.641	1.170	7.105
Growth	0.132	0.256	−0.341	−0.022	0.101	0.240	1.220
Top1	43.611	15.568	12.042	32.163	44.006	55.556	68.535
ROA	0.045	0.051	−0.166	0.014	0.036	0.065	0.215
G_ROA	−0.120	0.681	−3.676	−0.213	−0.047	0.063	1.916
V_ROA	0.001	0.004	0	0	0	0.001	0.060
Lnsize	24.294	1.046	20.683	23.491	24.213	25.082	26.097
IInv	5.307	4.413	0.117	2.120	4.210	7.280	24.569
Boardsize	9.701	2.080	5	9	9	11	15
IDrate	0.383	0.065	0.333	0.333	0.364	0.417	0.571
Beta	1.078	0.269	0.291	0.905	1.081	1.257	1.993
SOE	0.494	0.500	0	0	0	1	1
UnDual	0.884	0.320	1	0	1	1	1

presents the descriptive statistics for the main variables. The mean of $\mathrm{CSR}\_\mathrm{Dis}$ is 44.052, with a median of 40.937 and a skewness coefficient of 0.463. The mean of $\mathrm{MCSR}\_\mathrm{Dis}$ is −0.007, with a median of 0 and a skewness coefficient of −0.528. Both $\mathrm{CSR}\_\mathrm{Dis}$ and $\mathrm{MCSR}\_\mathrm{Dis}$ have relatively small skewness coefficients, making them suitable for fixed-effects regression based on the mean. The mean of $\mathrm{CSR}\_\mathrm{Perf}$ is 0.218, with a median of 0.08 and a skewness coefficient of 1.332. Given the high skewness coefficient of $\mathrm{CSR}\_\mathrm{Perf}$, using a fixed-effects regression model based on the mean may result in biased regression results. Therefore, in the robustness test section, we will further use quantile regression methods to test H2. The mean of $\mathrm{Reputation}$ is 0.072, with a standard deviation of 0.258 and a coefficient of variation of 3.583. $\mathrm{Reputation}$ shows significant differences across different samples and is suitable as an explanatory variable for regression analysis.

Table 3. Pearson correlation coefficient test and VIF test results.

Variable	VIF	CSR_Dis	CSR_Perf	MCSR_ Dis	Reputation	Lev	Q	Growth	Top1
CSR_Dis		1
CSR_Perf		0.349 ***	1
MCSR_Dis		0.294 ***	0.016	1
Reputation	1.11	0.221 ***	0.272 ***	0.037	1
Lev	1.84	0.007	−0.198 ***	−0.011	−0.050 *	1
Q	2.28	−0.096 ***	0.150 ***	0.028	0.046 *	−0.573 ***	1
Growth	1.22	−0.069 ***	0.038	−0.035	−0.024	0.028	0.133 ***	1
Top1	1.18	0.154 ***	0.185 ***	0.018	−0.106 ***	−0.031	−0.020	−0.058 **	1
ROA	2.44	−0.038	0.378 ***	−0.035	0.140 ***	−0.526 ***	0.587 ***	0.315 ***	−0.060 **
G_ROA	1.33	−0.003	0.121 ***	−0.016	0.034	−0.116 ***	0.131 ***	0.290 ***	−0.054 **
V_ROA	1.11	−0.055 **	0.021	−0.058 **	−0.021	−0.147 ***	0.219 ***	0.019	−0.003
Lnsize	1.59	0.463 ***	0.417 ***	0.021	0.145 ***	0.328 ***	−0.461 ***	−0.018	0.192 ***
IInv	1.19	−0.159 ***	−0.007	0.007	0.107 ***	0.005	0.134 ***	0.137 ***	−0.274 ***
Boardsize	1.13	0.118 ***	0.025	0.012	0.045 *	−0.039	0.016	0.030	−0.037
IDrate	1.19	0.112 ***	0.249 ***	0.016	−0.058 **	0.085 ***	−0.042	0.025	0.138 ***
Beta	1.17	−0.151 ***	−0.171 ***	−0.017	−0.176 ***	0.160 ***	−0.152 ***	−0.030	−0.003
SOE	1.03	0.000	0.086 ***	0.004	0.020	0.065 **	−0.002	0.014	0.006
UnDual	0.04	0.047 *	−0.017	0.007	−0.021	0.026	−0.039	0.003	0.120 ***
Variable	ROA	G_ROA	V_ROA	Lnsize	Iinv	Boardsize	IDrate	Beta	SOE
G_ROA	0.437 ***	1
V_ROA	0.025	−0.016	1
Lnsize	−0.200 ***	−0.029	−0.105 ***	1
IInv	0.200 ***	0.071 ***	−0.054 **	−0.113 ***	1
Boardsize	−0.005	0.024	0.067 **	0.075 ***	−0.051 *	1
IDrate	−0.016	0.000	−0.037	0.222 ***	−0.011	−0.280 ***	1
Beta	−0.166 ***	−0.047 *	0.121 ***	−0.122 ***	−0.112 ***	−0.036	−0.051 *	1
SOE	−0.013	−0.011	0.006	0.052 *	−0.041	−0.037	0.035	−0.100 ***	1
UnDual	−0.073 ***	0.023	0.041	0.063 **	−0.010	0.074 ***	−0.068 **	0.057 **	−0.040

Note: *** indicates statistical significance at the 1% level, ** at the 5% level, and * at the 10% level.

presents the correlation matrix and VIF test results for the main variables. The results indicate that $\mathrm{CSR}\_\mathrm{Dis}$ is significantly and positively correlated with $\mathrm{Reputation}$ at the 1% level, providing preliminary support for H1. $\mathrm{CSR}\_\mathrm{Perf}$ is significantly and positively correlated with $\mathrm{Reputation}$ at the 1% level, preliminarily supporting H2. The correlation coefficient between $\mathrm{MCSR}\_\mathrm{Dis}$ and $\mathrm{Reputation}$ is positive but not significant, indicating that the further testing of H3 is required. The maximum VIF value is 2.44, indicating that there is no significant multicollinearity issue between the control variables and independent variables, ensuring the validity of the subsequent regression results [64].

4.2. Empirical Analysis Results

4.2.1. Impact of High Reputation on Quality of Corporate Social Responsibility (CSR) Disclosure

Table 4. Regression results of the impact of a high reputation on the CSR report disclosure quality.

Variable	Model 1	Model 2	Model 3	Model 4	Model 5
	CSR_Disit	CSR_Disit	CSR_Disit	ΔCSR_Disit	CSR_Disit
	FE	FE	GMM	DID	Oaxaca–Blinder
Reputationit-1	9.108 ***	1.540 ***	5.996 ***	--	--
	(3.946)	(3.053)	(19.630)	--	--
ΔReputationit-1	--	--	--	1.724 *	--
	--	--	--	(1.776)	--
Levit	−7.969	−0.181	1.520 *	--	−0.233 *
	(−1.497)	(−0.109)	(1.672)	--	(−1.665)
Qit	0.837	0.966 ***	1.781 ***	--	−0.206
	(0.862)	(2.633)	(13.806)	--	(−1.319)
Growthit	0.723	0.660	1.013 ***	--	−0.004
	(0.534)	(0.851)	(4.199)	--	(−0.133)
Top1it	0.021	0.005	0.041 ***	--	0.262 *
	(0.449)	(0.457)	(4.780)	--	(1.690)
ROAit	−13.661	−4.504	−7.275 ***	--	0.209
	(−0.850)	(−0.750)	(−3.128)	--	(0.822)
G_ROAit	0.256	0.048	0.104	--	−0.029
	(0.492)	(0.174)	(0.998)	--	(−0.627)
V_ROAit	−14.098	−106.951 ***	45.911 ***	--	−0.033
	(−0.164)	(−3.051)	(5.207)	--	(−1.049)
Lnsizeit	4.761 ***	0.755 **	1.220 ***	--	−3.127 ***
	(5.564)	(2.432)	(6.032)	--	(−5.023)
IInvit	−0.326 ***	0.003	−0.213 ***	--	0.662 ***
	(−2.807)	(0.096)	(−7.450)	--	(2.728)
Boardsizeit	0.528 *	0.179 *	−0.151 ***	--	−0.208
	(1.681)	(1.659)	(−3.055)	--	(−1.538)
IDrateit	9.659	2.432	−5.124 ***	--	0.155 *
	(0.927)	(0.705)	(−2.775)	--	(1.702)
Betait	−4.218 **	0.516	0.074	--	−0.466 *
	(−2.123)	(0.700)	(0.337)	--	(−1.927)
SOEit	−0.307	0.120	0.078	--	0.023
	(−0.278)	(0.431)	(0.336)	--	(0.584)
UnDualit	1.003	0.840 **	2.860 ***	--	0.021
	(0.750)	(2.007)	(6.435)	--	(0.471)
CSR_Disit-1	--	0.877 ***	0.341 ***	--	--
	--	(58.123)	(26.668)	--	--
CSR_Disit-2	--	--	0.318 ***	--	--
	--	--	(25.902)	--	--
Year	Yes	Yes	Yes	Yes	−0.241
					(−1.329)
Ind	Yes	Yes	Yes	Yes	0.009
					(0.144)
ΔControls	--	--	--	Yes	--
Prediction0	--	--	--	--	43.107 ***
	--	--	--	--	(117.733)
Prediction1	--	--	--	--	54.690 ***
	--	--	--	--	(43.986)
Total Explained	--	--	--	--	−3.204 ***
	--	--	--	--	(−3.977)
Difference	--	--	--	--	−11.584 ***
	--	--	--	--	(−8.937)
N	1372	1089	830	1089	1372
Sargan	--	--	181.990	--	--
AR(1)	--	--	−4.858 ***	--	--
AR(3)	--	--	−0.269	--	--
Within R2	0.203	0.850		0.040	--
F|Wald χ2	6.18 ***	422.32 ***	680,912 ***	1.650 ***	--

Note: ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively. Values in parentheses are z-values or t-values adjusted for clustering at the firm level.

presents the empirical test results for H1. Model 1 displays the fixed-effects regression outcomes based on Equation (2). The coefficient of $\mathrm{Reputation}$ is significant and positive ($\mathsf{\beta }=9.108, p<1\%$), providing support for H1. However, considering the perspective that “corporate social responsibility may be influenced by corporate behavioral inertia” [41], this study employs several methods to disentangle the impact of corporate behavioral inertia on the quality of CSR disclosure in the current period, thereby obtaining the net effect of reputation constraints on the quality of CSR disclosure.

First, the regression process includes the first-order lag of the CSR disclosure quality as a control variable. If the difference in CSR disclosure quality between companies with and without a high reputation is merely the continuation of the difference between the two in the previous year, then the coefficient of $\mathrm{Reputation}$ should no longer be significant after absorbing this difference with the first-order lag. However, the regression results of Model 2 show that the coefficient of $\mathrm{Reputation}$ remains significant and positive at the 1% level. This indicates that, even after controlling for inertia factors, the CSR disclosure quality of high-reputation companies remains significantly higher than that of non-high-reputation companies. Furthermore, compared to Model 1, the coefficient in Model 2 decreases, suggesting that inertia factors indeed partially explain the difference in the CSR disclosure quality between the two types of samples. The inclusion of the first-order lag effectively absorbs the influence of inertia, enabling the research results to more accurately reveal the net effect of reputation on the CSR disclosure quality.

Introducing the first-order lag of the dependent variable as a control variable in the model may lead to new endogeneity issues. Following the procedure proposed by Wintoki [65] and guided by existing research [66], this study re-estimates the regression using the system GMM model, which can correct for unobserved individual heterogeneity, omitted variable bias, measurement errors, and potential endogeneity issues. Model 3 presents the regression results based on the system GMM model, where the coefficient of $\mathrm{Reputation}$ remains significant and positive ($\mathsf{\beta }=5.996, p<1\%$), supporting the findings of Model 2.

Next, we choose the first-order difference value of $\mathrm{CSR}\_\mathrm{Dis}$, denoted as $\mathsf{\Delta }\mathrm{CSR}\_\mathrm{Dis}$, as the dependent variable and the first-order difference value of $\mathrm{Reputation}$, denoted as $\mathsf{\Delta }\mathrm{Reputation}$, as the independent variable to construct the regression model, as shown in Equation (3). If a company is not on the MAC list in year $\mathrm{t}-1$ but enters the MAC list in year $\mathrm{t}$, then ${\mathsf{\Delta }\mathrm{Reputation}}_{\mathrm{it}}=1$; if a company is on the list in year $\mathrm{t}-1$ but exits the list in year $\mathrm{t}$, then ${\mathsf{\Delta }\mathrm{Reputation}}_{\mathrm{it}}=-1$; otherwise, ${\mathsf{\Delta }\mathrm{Reputation}}_{\mathrm{it}-1}$ takes the value of 0.

If a high reputation has no effect on the disclosure quality of the CSR reports of companies, there should be no significant difference in $\mathsf{\Delta }\mathrm{CSR}\_\mathrm{Dis}$ (the indicator of the annual increase in the CSR report disclosure quality) between high-reputation and non-high-reputation companies, regardless of whether the company enters or exits the high-reputation list or remains unchanged. However, if a high reputation causes companies to disclose their CSR reports to a higher standard, then, compared to samples where the reputation remains unchanged, companies entering the high reputation list should have higher $\mathsf{\Delta }\mathrm{CSR}\_\mathrm{Dis}$, while those exiting the high reputation list are expected to have lower $\mathsf{\Delta }\mathrm{CSR}\_\mathrm{Dis}$. The expected coefficient of $\mathsf{\Delta }\mathrm{CSR}\_\mathrm{Dis}$ should be significant and positive. The regression result in Model 4 ($\mathsf{\beta }=1.724, p<10\%$) satisfies this expectation.

$$\mathsf{\Delta }CSR\_Di{s}_{it}={\gamma }_0+{\gamma }_1\mathsf{\Delta }Reputatio{n}_{it-1}+{\gamma }_{2-15}\mathsf{\Delta }Control{s}_{it}+{\gamma }_{16}Year+{\gamma }_{17}Ind+\epsilon$$

(3)

Finally, this study employs the Oaxaca–Blinder model (hereafter referred to as the O-B model) to examine the explanatory power of reputation differences regarding the disparities in the quality of CSR reporting. Building upon the OLS regression results, the O-B model decomposes the differences in the conditional means of the dependent variable between two different sample groups [67,68,69]. The decomposition results of the O-B model are shown in Model 5, where $\mathrm{Prediction}0$ and $\mathrm{Prediction}1$ represent the mean CSR report disclosure quality for non-high-reputation and high-reputation companies, respectively. $\mathrm{Total} \mathrm{Explained}$ is the total explanation of the differences between the groups by all explanatory factors selected by the O-B model, and $\mathrm{Difference}$ is the difference in the conditional mean between the groups. It is observed that the intergroup difference in the CSR reporting quality between companies without high reputation ratings and those with high reputation ratings is approximately −11.584, which is significant at the 1% level ($\mathrm{Difference}=-11.584, p<1\%$). Among these, the explanatory variables in the model can account for 27.66% of the difference (i.e., the ratio of $\mathrm{Total} \mathrm{Explained}$ to $\mathrm{Difference}$), leaving 72.34% of the difference unexplained by the selected explanatory variables. Referring to Blau and Kahn’s [69] understanding of the O-B model, we attribute these unexplained differences to reputation disparities.

In summary, after controlling for factors such as organizational inertia, companies receiving high reputation ratings demonstrate significantly higher levels of corporate social responsibility (CSR) report disclosure quality compared to those without high reputation ratings. The reputation constraint mechanism can effectively account for the differences in the CSR report disclosure quality between these two types of companies, with the explanatory power potentially reaching up to 72.34%. Therefore, H1 is supported by the findings.

4.2.2. Impact of High Reputation on Corporate Social Responsibility Performance

Following the logic of H1, this study empirically tests H2, and the regression results are presented in

Table 5. Regression results of the impact of a high reputation on CSR Performance.

Variable	Model 1	Model 2	Model 3	Model 4	Model 5
	CSR_Perfit	CSR_Perfit	CSR_Perfit	ΔCSR_Perfit	CSR_Perfit
	FE	FE	GMM	DID	Oaxaca–Blinder
Reputationit-1	0.128 **	0.038 **	0.106 ***	--	--
	(2.125)	(1.988)	(12.689)	--	--
ΔReputationit-1	--	--	--	0.065 ***	--
	--	--	--	(3.294)	--
Levit	−0.395 ***	0.059 **	−0.068 ***	--	−0.006
	(−4.734)	(2.088)	(−2.627)	--	(−1.632)
Qit	0.015	−0.004	0.001	--	−0.007
	(1.048)	(−0.445)	(0.424)	--	(−1.327)
Growthit	−0.008	0.072 ***	0.100 ***	--	−0.003
	(−0.409)	(2.948)	(13.965)	--	(−1.005)
Top1it	−0.000	0.000	−0.000	--	0.014 **
	(−0.312)	(1.166)	(−1.069)	--	(2.562)
ROAit	2.604 ***	1.025 ***	1.762 ***	--	−0.086 ***
	(9.165)	(6.682)	(27.471)	--	(−4.434)
G_ROAit	−0.016 **	0.025 ***	−0.009 ***	--	0.001
	(−2.403)	(4.287)	(−3.716)	--	(0.814)
V_ROAit	2.260	−1.234	−6.416 ***	--	0.001
	(1.116)	(−0.801)	(−23.684)	--	(0.816)
Lnsizeit	0.096 ***	0.024 ***	0.103 ***	--	−0.124 ***
	(3.026)	(3.801)	(24.247)	--	(−5.226)
IInvit	0.002	−0.000	−0.004 ***	--	−0.005
	(1.488)	(−0.621)	(−5.838)	--	(−1.375)
Boardsizeit	0.007	0.003 *	0.000	--	−0.001
	(1.349)	(1.847)	(0.141)	--	(−0.540)
IDrateit	0.085	0.057	0.237 ***	--	0.013 **
	(0.660)	(0.997)	(4.917)	--	(2.550)
Betait	0.041 **	−0.015	0.034 ***	--	0.002
	(2.071)	(−0.898)	(4.534)	--	(0.339)
SOEit	−0.036 **	0.011 *	−0.003	--	−0.001
	(−2.542)	(1.768)	(−0.496)	--	(−0.697)
UnDualit	−0.020	0.015	0.049 ***	--	−0.000
	(−0.921)	(1.286)	(5.071)	--	(−0.078)
CSR_Perfit-1	--	0.875 ***	0.666 ***	--	--
	--	(40.757)	(76.653)	--	--
Year	Yes	Yes	Yes	Yes	0.004
					(1.279)
Ind	Yes	Yes	Yes	Yes	−0.013 **
					(−2.012)
ΔControls	--	--	--	Yes	--
Prediction0	--	--	--	--	0.183 ***
	--	--	--	--	(16.671)
Prediction1	--	--	--	--	0.617 ***
	--	--	--	--	(13.670)
Difference	--	--	--	--	−0.434 ***
	--	--	--	--	(−9.334)
Total Explained	--	--	--	--	−0.212 ***
	--	--	--	--	(−7.569)
N	1340	1136	1129	1113	1340
Sargan	--	--	186.237	--	--
AR(1)	--	--	−4.272 ***	--	--
AR(2)	--	--	0.702	--	--
Within R2	0.447	0.889		0.364	--
F|Wald χ2	13.77 ***	362.23 ***	65,413 ***	10.23 ***	--

Note: *** indicates statistical significance at the 1% level, ** at the 5% level, and * at the 10% level.

. Model 1 in Table 5 is a fixed-effects regression based on Equation (2), and the results show that the coefficient of $\mathrm{Reputation}$ is significant and positive ($\mathsf{\beta }=0.128, p<5\%$). This indicates that companies receiving high reputation ratings exhibit significantly higher levels of CSR performance compared to those without high reputation ratings. Model 2 presents the regression results with the inclusion of the $\mathrm{CSR}\_{\mathrm{Perf}}_{\mathrm{it}}$ control variable. In Model 2, the coefficient of $\mathrm{Reputation}$ remains significant and positive ($\mathsf{\beta }=0.038, p<5\%$). Compared to Model 1, the goodness of fit of Model 2 is improved, and the coefficient of $\mathrm{Reputation}$ decreases significantly. This suggests that the $\mathrm{CSR}\_{\mathrm{Perf}}_{\mathrm{it}}$ variable effectively absorbs the influence of the previous CSR performance on the current CSR performance, indicating the significant constraining effect of a high reputation on CSR performance. Model 3, based on the system GMM model, supports Model 2. Model 4 examines the impact of reputation changes on changes in CSR performance. Model 4 shows that the coefficient of $\mathsf{\Delta }\mathrm{Reputation}$ is significant and positive ($\mathsf{\beta }=0.065, p<5\%$). Compared to samples with unchanged reputations, the increase in CSR performance ($\mathsf{\Delta }\mathrm{CSR}\_\mathrm{Perf}$) after entering the high reputation list is higher. Model 5 presents the decomposition results based on the Oaxaca–Blinder model, indicating that the selected explanatory variables can explain 48.85% of the CSR performance difference between companies receiving high reputation ratings and those without high reputation ratings, with approximately 51% of the CSR performance difference attributed to differences in reputation levels.

In summary, the empirical results of Model 1 in Table 5 confirm that companies receiving high reputation ratings exhibit significantly higher levels of CSR performance compared to those without high reputation ratings. Models 2 to 5 further demonstrate that the reputation constraint mechanism is a significant factor contributing to this difference. After controlling for other factors, the reputation constraint mechanism explains 51% of the difference in CSR performance between the two types of companies. Therefore, H2 is supported by the data.

4.2.3. Impact of High Reputation on CSR Report Whitewashing

Before investigating the influence of a high reputation on the level of CSR report whitewashing, we first explore the existence of information content in CSR reports based on Equation (1). The empirical results are presented in Model 1 of

Table 6. Regression results of the impact of a high reputation on CSR report whitewashing.

Variable	Model 1	Model 2	Model 3	Model 4	Model 5
	CSR_Disit	MCSR_Disit	MCSR_Disit	MCSR_Disit	ΔMCSR_Disit
	FE	FE	FE	GMM	DID
Reputationit-1	--	−0.030 ***	−0.028 **	−0.005 **	--
	--	(−3.141)	(−2.333)	(−2.342)	--
CSR_Perfit	7.522 ***	--	--	--	--
	(3.096)	--	--	--	--
ΔReputationit-1	--	--	--	--	−0.018 *
	--	--	--	--	(−1.887)
CSR_Disit	--	0.005 ***	0.006 ***	0.003 ***	--
	--	(8.737)	(7.307)	(34.644)	--
Levit	−6.380	−0.030	0.042	−0.020 ***	--
	(−1.139)	(−0.588)	(0.703)	(−2.891)	--
Qit	0.595	0.008	0.000	−0.010 ***	--
	(0.592)	(1.201)	(0.064)	(−8.684)	--
Growthit	0.849	−0.007	−0.000	0.036 ***	--
	(0.641)	(−0.500)	(−0.029)	(10.863)	--
Top1it	0.005	−0.000	−0.000	−0.001 ***	--
	(0.110)	(−0.563)	(−0.580)	(−8.707)	--
ROAit	−30.153 *	−0.162	−0.200	0.035 **	--
	(−1.792)	(−1.403)	(−1.482)	(2.301)	--
G_ROAit	0.345	0.002	0.003	−0.005 ***	--
	(0.659)	(0.409)	(0.366)	(−4.810)	--
V_ROAit	−53.101	−1.591 *	−2.215 ***	−1.427 ***	--
	(−0.582)	(−1.885)	(−3.102)	(−23.538)	--
Lnsizeit	3.923 ***	−0.012	−0.011	−0.008 ***	--
	(3.735)	(−0.834)	(−0.650)	(−6.391)	--
IInvit	−0.291 **	0.003 ***	0.002 *	−0.001 ***	--
	(−2.437)	(2.687)	(1.907)	(−5.204)	--
Boardsizeit	0.483	−0.003	−0.001	−0.003 ***	--
	(1.517)	(−0.925)	(−0.383)	(−7.365)	--
IDrateit	2.920	0.105	0.045	−0.189 ***	--
	(0.285)	(1.355)	(0.506)	(−7.919)	--
Betait	−4.821 **	0.011	0.008	0.012 ***	--
	(−2.437)	(0.802)	(0.590)	(5.371)	--
SOEit	−0.648	0.001	0.002	−0.003	--
	(−0.574)	(0.092)	(0.224)	(−1.610)	--
UnDualit	1.336	0.007	0.019	0.004	--
	(1.010)	(0.507)	(0.969)	(1.112)	--
MCSR_Disit-1	--	--	−0.246 ***	−0.079 ***	--
	--	--	(−7.103)	(−11.400)	--
N	1340	1265	984	1008	1008
Year	Yes	Yes	Yes	Yes	Yes
Ind	Yes	Yes	Yes	Yes	Yes
N	1340	1265	984	1008	1008
Sargan	--	--	--	199.656	--
AR(1)	--	--	--	−5.746 ***	--
AR(3)	--	--	--	0.607	--
Within R2	0.191	0.130	0.210	--	0.185
F	6.004 ***	6.597 ***	9.180 ***	14,060 ***	4.321 ***

Note: *** indicates statistical significance at the 1% level, ** at the 5% level, and * at the 10% level.

. Model 1 shows a significant and positive correlation at the 1% level on $\mathrm{CSR}\_\mathrm{Perf}$ and $\mathrm{CSR}\_\mathrm{Dis}$, with a coefficient of 7.522. Given the understanding of the practical significance represented by the model, as mentioned by Tarabashkina [70], CSR reports can, to a certain extent, reflect the actual performance in terms of social responsibility of listed companies. This suggests that CSR reports contain genuine information regarding corporate social responsibility performance.

Following the testing logic of H1, this study empirically examines the difference in corporate social responsibility (CSR) whitewashing between high-reputation and non-high-reputation companies. Model 2 in Table 6 represents the fixed-effects regression based on Equation (2). Model 2 shows that the coefficient of $\mathrm{Reputation}$ is significant and negative ($\mathsf{\beta }=-0.03, p<1\%$), indicating that the level of CSR whitewashing is lower for companies with a high reputation compared to those without a high reputation. Therefore, a high reputation has a significant inhibitory effect on CSR whitewashing in companies.

In Model 3, even after adding $\mathrm{MCSR}\_{\mathrm{Dis}}_{\mathrm{it}-1}$ as a control variable, the coefficient of $\mathrm{Reputation}$ remains significant and negative ($\mathsf{\beta }=-0.028, p<5\%$). After eliminating the influence of behavioral inertia, high-reputation companies still exhibit lower levels of corporate social responsibility whitewashing compared to non-high-reputation companies. Model 4, which includes the system GMM regression of $\mathrm{MCSR}\_{\mathrm{Dis}}_{\mathrm{it}-1}$, provides further support for Model 3.

Model 5 indicates that the coefficient of $\mathsf{\Delta }\mathrm{Reputation}$ is significant and negative at the 10% level ($\mathsf{\beta }=-0.018, p<10\%$). This result suggests that, after entering the high-reputation list, there is a significant decrease in the level of corporate social responsibility (CSR) report whitewashing.

In summary, as confirmed by Model 2 in Table 6, companies with high reputation ratings exhibit significantly lower levels of corporate social responsibility (CSR) report whitewashing compared to those without high reputation ratings. The regression results from Models 3 to 5 further support the constraining effect of the reputation mechanism on CSR report whitewashing behavior. Therefore, H3 is supported by the data.

4.3. Endogeneity Test

4.3.1. Endogeneity Test Based on Propensity Score Matching

To further reduce the impact of between-group differences in unobserved variables on the regression results, this study retests the three hypotheses using the propensity score matching method [71,72].

Table 7. PSM results based on 1:4 nearest neighbor matching and kernel matching.

Model	Variable	Matching Method	Treatment/Control Group	Treatment Group ATT	Control GroupATT	Diff	SD	t
1	CSR_Dis	1:4 nearest neighbor matching	97/1081	55.297	46.914	8.383	1.782	4.71 ***
		Kernel matching	97/1059	55.297	46.787	8.510	1.534	5.55 ***
2	CSR_Perf	1:4 nearest neighbor matching	97/1081	0.607	0.389	0.218	0.058	3.73 ***
		Kernel matching	97/1059	0.607	0.374	0.243	0.050	4.82 ***
3	MCSR_Dis	1:4 nearest neighbor matching	99/1055	0.003	0.006	0.003	0.009	−0.36
		Kernel matching	90/1028	−0.0003	0.003	−0.003	0.008	−0.35

Note: *** indicates statistical significance at the 1% level.

demonstrates that the between-group difference test results based on 1:4 nearest neighbor matching and kernel matching support H1 and H2, while H3 does not receive data support.

In order to explore the effectiveness of H3 under matched samples, this study uses matched samples and Equation (2) to conduct a regression analysis of H3.

Table 8. H3 test results based on PSM matched samples.

Variable	Model 1	Model 2
	1:4 Nearest Neighbor Matching Sample	Kernel Matching Sample
Reputationit-1	−0.021 **	−0.016 **
	(−2.573)	(−2.087)
CSR_Disit	0.005 ***	0.005 ***
	(8.287)	(8.171)
Controls	Yes	Yes
N	1093	1052
Within R2	0.149	0.161
Ind	Yes	Yes
Year	Yes	Yes
F	5.967 ***	5.758 ***

Note: *** indicates statistical significance at the 1% level, ** at the 5% level.

presents the corresponding regression results, which support H3.

Table 9. Results of PSM matching quality test.

Type	Unmatched			1:4 Nearest Neighbor Matching			Kernel Matching
Variable	%bias	t	V(T)/V(C)	%bias	t	V(T)/V(C)	%bias	t	V(T)/V(C)
Growth	−10.3	−0.9	0.62 *	3.7	0.29	0.91	6.2	0.49	0.91
Top1	−37.6 ***	−3.99	1.52 *	−14.2	−0.95	1.25	−12	−0.82	1.4
ROA	53.3 ***	5.29	1.14	4	0.26	0.93	8.6	0.56	0.9
Lnsize	56.9 ***	5.46	0.99	−8	−0.58	1.14	−2.5	−0.18	1.1
IInv	37 ***	4.02	1.66 *	3.7	0.22	0.75	6.5	0.39	0.83
IDrate	−25.4 **	−2.18	0.54 *	2.7	0.22	0.93	3.5	0.28	0.91
Beta	−70.2 ***	−6.7	0.95	2	0.15	1.07	−1.3	−0.09	1.05
Meanbias	41.5			5.5			5.8
Medbias	37.6			3.7			6.2
B	132.8			17.2			15.3
R	0.9			0.9			1.48
%Var	57			0			0

Note: *** indicates statistical significance at the 1% level, ** at the 5% level, and * at the 10% level.

presents the results of the propensity score matching (PSM) quality test. In the matched sample, the differences between all control variables are not significant. Additionally, V(T)/V(C) is also not significant, indicating a high degree of consistency in the sample distribution. In the matching quality test, the B values are all less than the critical value of 25, indicating that the matching results effectively eliminate the differences in the control variables between different sample groups.

4.3.2. Two-Stage Least Squares (2SLS) Estimation Based on Instrumental Variables

In order to address potential endogeneity issues arising from mutual causality, this study employs instrumental variables in two-stage least squares regression. Given the strong correlation between news coverage in newspapers and corporate reputation [4], this study selects the average number of positive news reports in newspapers for companies in the same industry, excluding the focal company, as the instrumental variable for corporate reputation. The data are obtained from the CSMAR database’s Chinese Press News Analytics sub-database.

Table 10. Two-stage least squares regression results.

Variable	Model 1	Model 2	Model 3
	CSR_Disit	CSR_Perfit	MCSR_Disit
Reputationit-1	5.607 **	0.191 ***	−0.015 ***
	(1.979)	(2.737)	(−2.505)
Controls	Yes	Yes	Yes
N	1372	1340	1265
Within R2	0.185	0.171	0.051
Ind	Yes	Yes	Yes
Year	Yes	Yes	Yes
Wald χ2	752.31 ***	2718.74 ***	421.73 ***

Note: *** indicates statistical significance at the 1% level, ** at the 5% level.

presents the corresponding regression results. Models 1 and 2 show that, after controlling for potential endogeneity issues, the coefficient of ${\mathrm{Reputation}}_{\mathrm{it}}$ is significant and positive. Model 3 indicates that, after controlling for potential endogeneity issues, the coefficient of ${\mathrm{Reputation}}_{\mathrm{it}}$ is significant and negative ($\mathsf{\beta }=-0.015, p<1\%$). The two-stage least squares estimation supports the three research hypotheses of this study.

4.4. Robustness Test

As an important information intermediary [73,74,75], the news media disseminate a vast amount of information to the population. Compared to other means of information dissemination, traditional mainstream media exhibit stricter scrutiny in disseminating information to the public, making the information more persuasive. When the media provide more positive coverage of a company, it is easier to convey positive signals to the public, thus enhancing the company’s reputation. Conversely, when the media deliver more negative coverage, the public receives more negative information about the company, leading to reputational damage [74]. Based on this, this study chooses to use data on the number of positive and negative news reports about companies from newspapers to construct a substitution variable ($\mathrm{Media}\_\mathrm{P}$) for corporate reputation. The data are sourced from the CSMAR database’s Chinese Press News Analytics sub-database.

If there are many positive news reports in newspapers, it indicates that the media have a favorable evaluation of the company. Stakeholders can receive more positive information about the company, making it easier for them to form positive opinions about it. However, relying solely on the number of positive news reports in newspapers as a substitution variable may lead to biased results. This is because stakeholders do not only accept positive news but also form a comprehensive evaluation of the company by absorbing positive, neutral, and negative news related to the company. In light of this, this study constructs a substitution variable ($\mathrm{Media}\_\mathrm{P}$) to measure the level of corporate reputation as follows: first, we calculate the logarithm of the number of positive news reports about the company, denoted as $\mathrm{Pn}$; then, we calculate the ratio of positive news reports to non-positive news reports ($\mathrm{P}\_\mathrm{rate}$); finally, we standardize and sum $\mathrm{Pn}$ and $\mathrm{P}\_\mathrm{rate}$, denoted as $\mathrm{Media}\_\mathrm{P}$.

Table 11. Robustness test results based on newspaper reports.

Variable	Model 1	Model 2	Model 3
	CSR_Disit	CSR_Perfit	MCSR_Disit
Media_Pit-1	1.441 ***	0.046 ***	−0.007 *
	(−2.862)	(−3.1)	(−1.821)
CSR_Disit	--	--	0.005 ***
	--	--	(−8.759)
Controls	Yes	Yes	Yes
N	1372	1340	1265
Within R2	0.186	0.476	0.16
Ind	Yes	Yes	Yes
Year	Yes	Yes	Yes
F	5.345 ***	19.88 ***	6.573 ***

Note: *** indicates statistical significance at the 1% level, * at the 10% level.

lists the robustness test results using $\mathrm{Media}\_\mathrm{P}$ as a substitution variable for reputation. The regression results show that the effect of $\mathrm{Media}\_{\mathrm{P}}_{\mathrm{it}-1}$ on $\mathrm{CSR}\_{\mathrm{Dis}}_{\mathrm{it}}$ and $\mathrm{CSR}\_{\mathrm{Perf}}_{\mathrm{it}}$ is significant and positive at the level of 1%, and the influence coefficients are 1.441 and 0.046, respectively. $\mathrm{Media}\_{\mathrm{P}}_{\mathrm{it}-1}$ has a negative impact on $\mathrm{MCSR}\_{\mathrm{Dis}}_{\mathrm{it}}$, which is weakly significant at the level of 10%. The robustness test results obtained using newspaper news reports as a substitution variable for corporate reputation support H1, H2, and H3.

4.5. Further Study

The strategic choices and behavioral decisions of corporations regarding social responsibility issues are also influenced by various internal and external governance factors. At the internal governance level, the board of directors serves as a critical formal mechanism that directly impacts corporate social responsibility. In terms of external governance, government is recognized as a significant factor affecting corporate fulfillment of social responsibilities. Particularly within the Chinese context, governmental concern plays a pivotal role in corporate decision-making, motivating companies to actively engage in fulfilling their social responsibilities to align with governmental expectations. As an informal governance mechanism, what is the interplay between reputation constraints and these aforementioned governance mechanisms? To address this inquiry, this paper further investigates the interaction between the reputation constraint mechanism and the ethics committee’s governance structure within boards of directors, as well as its relationship with governmental emphasis on public environmental concerns.

4.5.1. The Moderation Effect of Board Ethics Committee in Governance

The establishment of an ethics committee within the board of directors signifies that business owners and senior managers are increasingly attentive to corporate ethical practices. The institutionalization of corporate ethics management through this internal committee can enhance the control and supervisory efficiency of the board over corporate social responsibility behaviors. It serves to enforce compliance with established systems and norms, thereby improving both the quality of social responsibility disclosures and actual performance while reducing instances of ‘whitewashing’ in social responsibility reports, which may diminish the effectiveness of external reputation governance mechanisms. Consequently, compared to corporations without ethics committees, those with such committees will experience a weaker restraining effect from reputation on CSR behaviors.

Table 12. The Regression Results of the moderation effect of the Board Ethics Committee.

Variable	Model 1	Model 2	Model 3
	CSR_Disit	CSR_Perfit	MCSR_Disit
Reputationit-1	9.798 ***	0.142 **	−0.034 ***
	(4.063)	(2.127)	(−3.371)
Reputationit-1 × Commitit	−8.473 ***	−0.116 *	0.033 ***
	(−3.114)	(−1.734)	(3.078)
Commit it	5.877 ***	−74,794.248	--
	(2.626)	(−0.006)	--
CSR_Disit	--	--	0.005 ***
	--	--	(8.739)
Controls	Yes	Yes	Yes
N	1372	1340	1265
Within R2	0.212	0.449	0.135
△R2	0.009 ***	0.097 ***	0.004 ***
Ind	Yes	Yes	Yes
Year	Yes	Yes	Yes
F	6.836 ***	14.04 ***	2.184 ***

Note: *** indicates statistical significance at the 1% level, ** at the 5% level, and * at the 10% level.

shows the regression results of the vicarious role of board ethics committee governance. Commit indicates whether the board of directors has set up an ethics committee. If the board of directors has set up ethics committee, environment committee, safety and health committee, sustainable development committee, social responsibility committee, etc., the value of Commit is 1; otherwise, it is 0. Reputation Commit is a cross between high reputation and the ethics committee of the board of directors. The remaining variable definitions are the same as previously. The relevant data on the establishment of ethics committees of corporate boards come from the CSMAR database.

Model 1 shows that the regression coefficient of Reputation*Commit is −8.473, which is significantly negative at 1% level, and is significantly opposite to the regression coefficient sign of Reputation. This suggests that reputation has a weaker impact on the quality of CSR report disclosure in companies with ethics committees than in those without. Model 2 shows that the regression coefficient of Reputation*Commit is −0.116, which is weakly significant at the level of 10%, which is opposite to the regression coefficient sign of corporate Reputation. The influence of reputation on the quality of CSR performance is weaker in companies with ethics committees than in those without them. The establishment of ethics committees in corporations means that they pay more attention to ethics, and the higher quality of fulfilling social responsibilities and the higher quality of disclosing social responsibility reports become the active behaviors of these corporations, and the constraint effect of reputation is correspondingly reduced. Model 3 shows that the regression coefficient of Reputation*Commit is 0.033, which is significantly positive at 1% level and significantly opposite to the regression coefficient sign of Reputation. The results also show that reputation has a weaker effect on the whitewashing of CSR reports in companies with ethics committees than in those without. Whether a company can honestly and credibly disclose its social responsibility report may itself be part of the ethics committee’s concern. When a corporation sets up an ethics committee, the ethics committee will also restrict and control the whitewashing behavior of the CSR report and take the initiative to reduce the whitewashing degree of the CSR report. In this case, the constraining effect of a high reputation “halo” will be affected. The regression results presented in Table 12 provide evidence that the board ethics committee system serves as a substitute for the reputation constraint mechanism.

4.5.2. The Moderation Effect of Government in Public Environmental Governance

Based on the theory of institutions, government governance behavior will exert a coercive isomorphism pressure on organizations. When the government demonstrates heightened concern for public environment, corporations may be motivated to actively fulfill their social responsibilities to conform to government needs, thereby reducing the motivation for “talking the talk but not walking the walk” behavior. Therefore, compared with corporations located in areas where the government shows less concern for public environment, the influence of reputation on promoting corporate social responsibility behaviors in areas where the government shows greater concern for public environment will be weaker.

Gov represents the degree of government’s emphasis on public environment. In order to effectively measure the degree of government’s emphasis on public environment, this paper selects the ratio of local government’s environmental protection expenditure to its general budget expenditure, the ratio of local government’s healthcare expenditure to its general budget expenditure, the ratio of local government’s social security and employment expenditure to its general budget expenditure, the ratio of industrial output value to local industrial water consumption, and the annual reduction in the annual urban registered unemployment rate as measurement indicators of government’s emphasis on public environment. By conducting principal component analysis on these indicators, and based on the principles of eigenvalue greater than 1 and cumulative variance contribution rate greater than 0.8, the first two principal components are extracted, and the weighted sum of the variance contribution rates of each principal component is used as a substitute variable for government’s emphasis on public environment. Reputation*Gov is the product of high reputation and government concern for the public environment. The definitions of the other variables are the same as before.

Table 13. The regression results of the moderation effect of government public environmental governance.

Variable	Model 1	Model 2
	CSR_Disit	CSR_Perfit
Reputationit-1	6.504 ***	0.138 *
	(2.950)	(1.927)
Govit-1	−0.720	−0.033
	(−0.347)	(−1.311)
Reputationit-1 × Govit	−9.092 **	−0.058 **
	(−1.984)	(−2.181)
Controls	Yes	Yes
N	1372	1340
Within R2	0.206	0.449
△R2	0.003 **	0.097 ***
Ind	Yes	Yes
Year	Yes	Yes
F	6.048 ***	12.37 ***

Note: *** indicates statistical significance at the 1% level, ** at the 5% level, and * at the 10% level.

shows the moderating effect of government public environmental governance on the regulation of corporate social responsibility. Model 1 shows that the regression coefficient of Reputation*Gov is −9.092, which is significantly negative at the 5% level and has the opposite sign with the regression coefficient of Reputation. As governmental attention to public environmental governance intensifies, the impact of Reputation on the quality of social responsibility report disclosures gradually diminishes. Table 13, Model 2 shows that the regression coefficient of Reputation*Gov is −0.058, which is significantly negative at the 5% level and has the opposite sign with the regression coefficient of Reputation. As the government’s attention to public environmental governance increases, the influence of Reputation on the quality of corporate social responsibility fulfillment gradually weakens. When the government’s attention to public environmental governance is higher, corporations have stronger motivation to conform to government needs, and they will improve the quality of social responsibility fulfillment and social report disclosure to show their good “social citizen” image to the government. At this time, the effectiveness of reputation as a constraint on corporate social responsibility misrepresentation behavior will also be affected to some extent, as the regression coefficient symbol of Reputation*Gov is significantly opposite to the regression coefficient symbol of Reputation. Since we cannot find a reasonable theory to support how the government’s attention to public environmental governance affects the moderating role of reputation on the constraint of corporate social responsibility misrepresentation behavior, we have not tested the moderating effect of the government’s attention to public environmental governance on the relationship between reputation and the degree of corporate social responsibility report misrepresentation.

5. Research Conclusions and Implications

Based on stakeholder theory and the reputation constraint mechanism, this study focuses on the social responsibility governance effect of corporate reputation, deeply analyzing the multidimensional impact of a high-level corporate reputation on corporate social responsibility (CSR) behavior. By comparing Fortune magazine’s Most Admired Companies and the Fortune 500 companies in China, which have significant differences in reputation levels, this study finds that a high-level corporate reputation brings benefits to companies, while also imposing “expectation burdens” and “concern burdens” from stakeholders, which constrain companies’ corporate social responsibility behavior. Empirical research shows that companies receiving high reputation ratings have significantly higher levels of social responsibility report disclosure quality and social responsibility performance than those without high reputation ratings. The explanatory power of reputation differences for differences in the quality and performance of social responsibility report disclosure is as high as 72.34% and 51%, respectively. Corporate social responsibility whitewashing behavior does exist, with the degree of whitewashing in social responsibility reports being significantly lower for companies with high reputation ratings compared to those without high reputation ratings. Compared to non-high-reputation companies, high-reputation companies tend to choose more sincere social responsibility reporting strategies, focusing on improving their social responsibility performance governance to enhance their social responsibility report disclosure quality rather than risking detection by excessively whitewashing their social responsibility reports.

This study has both theoretical and practical significance. On one hand, it confirms that the reputation mechanism is an informal institutional mechanism that is capable of influencing corporate social responsibility (CSR) behavior, thereby validating the governance effect of corporate reputation on CSR. The focus on reputation constraint effects helps to reduce stakeholders’ reliance on corporate social responsibility (CSR) reports, enabling them to overcome the dilemma of relying solely on “unverifiable” CSR reports to understand corporate social responsibility behavior. Specifically, the corporate reputation evaluation list or guide identifies companies with higher reputations. A good reputation usually indicates better management capabilities, higher ethical standards, and stronger risk management capabilities within a company. Additionally, a strong corporate reputation allows for recognition and support from society during the process of an enterprise’s development, enabling access to more resources and opportunities. These factors directly impact an enterprise’s ability to withstand risks, its market value, and its future development potential. This implies that a reliable corporate reputation evaluation list or guide can serve as valuable evidence for institutions or individual investors engaging in (or intending to engage in) socially responsible investing. Such resources aid stakeholders in making more informed, accurate, and long-term decisions through validated information. On the other hand, encouraging corporations to assume greater social responsibility is a matter of concern among governments in many countries, including China. Research indicates that different types of firms do not respond uniformly to formal government regulations, nor do they exhibit consistent feedback towards government-issued incentives or guiding policies. Our study confirms that “companies with a good reputation tend to perform well in subsequent social responsibility disclosures and practices”. Based on this, it is recommended that the government, in optimizing and refining the corporate social responsibility (CSR) regulatory framework, should implement a strategy of differentiated guidance and supervision, rather than a “one-size-fits-all” approach. Specifically, companies with higher levels of reputation might be granted greater autonomy in CSR guidance and regulation. This would enable the government to allocate its resources more rationally, improve the efficiency of CSR governance, and alleviate the challenges of CSR regulation.

Research Limitations: There is a complex relationship between the social culture and a corporation’s understanding and actions regarding social responsibility. This study uses Chinese firms as the sample for a series of empirical tests. In the future, it would be worthwhile to select companies influenced by different social cultures as samples to examine whether the explanatory power of reputation differences on the quality of CSR reporting and performance remains consistent. This could provide a decision-making reference for stakeholders across a wider range of countries, particularly governments and multinational investors.