Stay Vigilant in Peace Time? Antitakeover Provisions and Corporate Social Responsibility

Abstract

We examine how antitakeover provisions (ATPs) impact corporate social responsibility (CSR) disclosure for Chinese listed firms. The empirical results show that the level of CSR disclosure is positively related to the level of ATPs, and this relationship is primarily observed in State-Owned Enterprises (SOEs). Further evidence indicates that this association is most prevalent in observations with lower efficiency or higher legitimacy pressures. Our paper provides new evidence on the mixed motivations for CSR disclosure in Chinese firms and offers insights into the moderating effects of ATPs in the context of weak corporate governance.

1. Introduction

Antitakeover provisions (ATPs) and corporate social responsibility (CSR) represent significant corporate phenomena that have sparked considerable debate in academic research. The existing literature agrees that ATPs can effectively reduce pressure from the corporate control market, but there is disagreement on how this reduction affects managerial incentives and the corresponding economic outcomes. Critics of ATPs argue that reduced pressure from the corporate control market leads to greater managerial entrenchment and exacerbates information asymmetry between management and shareholders [1]. Proponents, however, contend that decreased pressure mitigates managerial short-termism and fosters implicit incentives, thereby enhancing a firm’s long-term performance [2,3,4]. Empirical studies on shareholder value have produced mixed results [5]. The conflicting theories and results could potentially be reconciled by Sundaramurthy [6], who argued that the impacts of ATPs on shareholders’ value depend on the status of corporate governance. Specifically, under strong corporate governance, ATPs may enhance shareholder value, whereas under weak governance, they may diminish it [7]. Additionally, Cohen and Wang [8] underscore the importance of examining the specific contexts or types of firms that influence the relationship between ATPs and firm value in academic research.

Existing research shows a consensus that CSR involves addressing a broader range of stakeholders beyond just shareholders. However, there is disagreement regarding whether this focus enhances or diminishes shareholder value. Research based on agency theory argues that CSR is an externalization of the internal moral hazard between management and shareholders and diverts scarce resources from their optimal use, thereby decreasing shareholder value [9,10]. Conversely, the strategic view of CSR argues that CSR, as a strategic instrument, will increase shareholder value in the long run by satisfying stakeholders’ demands and contributing to the firm’s legitimacy [11,12]. Jeong and Kim [13] highlight that CSR decisions are influenced by the interplay between social legitimacy and economic efficiency, with a firm’s legitimacy management costs fluctuating based on changes in efficiency or legitimacy pressures. Legitimacy pressures arise from legitimacy benefits and/or illegitimacy penalties [13]. The former can be traced to the argument that “organizations that incorporate societally legitimated rationalized elements in their formal structures maximize their legitimacy and increase their resources and survival capabilities” [14]. The latter “can lead to financial loss and thus focus organizations to conform to existing institutionalized structures, norms, and values” [13]. Efficiency pressures come from economic efficiency and focus on corporate performance [13]. We do not intend to distinguish between the legitimacy and efficiency pressures driving CSR activities. We intend to demonstrate that CSR is affected by the change in the tension between legitimacy and efficiency pressures. Given that ATPs directly impact efficiency pressures, an important question arises regarding the potential effects of ATPs on CSR practices.

In contrast to the association between ATPs and shareholder value, the existing literature consistently indicates a positive relationship between ATPs and CSR [15,16]. However, the operational mechanisms involved are subject to debate. Surroca and Tribó [15] view CSR as an effective entrenchment strategy used by management to collude with non-shareholder stakeholders, thereby reinforcing their entrenchment position. Kacperczyk [16] takes a positive perspective, finding that an exogenous increase in ATPs leads to a higher level of CSR, which in turn increases long-term shareholder value. This mixed and inconsistent evidence suggests that the relationship between ATPs and CSR may depend on specific contextual factors. For instance, Surroca et al. [17] argue that when managerial entrenchment practices (MEPs) and CSR are adopted with the same rationale, they complement each other and enhance firm value. However, in China, a country characterized by coordinated market economies, discretionary increases in MEPs and CSR do not necessarily target shareholder value.

We argue that the unique characteristics of China’s mixed ownership structure complicate the straightforward application of coordinated market economy principles to explain CSR practices. In China, CSR activities are significantly influenced by social-political pressures, which serve as a form of legitimacy pressure [18]. Moreover, the motivations for engaging in CSR differ between state-owned enterprises (SOEs) and non-state-owned enterprises (non-SOEs). SOE managers, who are often bureaucrats, prioritize achieving political objectives over maximizing operational efficiency. Consequently, a reduction in efficiency pressure through ATPs may lead SOE managers to increase CSR activities to fulfill political mandates and advance their careers [19]. Conversely, non-SOEs engage in CSR to establish political connections with the ultimate goal of securing economic resources, such as lower capital costs and increased government subsidies, to enhance operational efficiency [20,21]. Therefore, a decrease in efficiency pressure due to ATPs may result in reduced CSR engagement among non-SOEs as the incremental value of CSR diminishes. Alternatively, CSR practices among non-SOEs may remain unchanged as they are compelled to meet minimum requirements. In conclusion, the relationship between ATPs and CSR in China is contingent upon ownership structure, with findings indicating a positive association predominantly observed in SOEs.

We further analyze whether the variance of efficiency or legitimacy pressure moderates the relationship between ATPs and CSR. Our results show that the positive relationship between ATPs and CSR mainly occurs in observations with lower efficiency pressure (as proxied by product market competition or total factor productivity) or higher legitimacy pressure (as proxied by media coverage), particularly driven by SOEs. Chen et al. [18] found that mandatory disclosure of CSR has positive externalities in that firms spend more on environmental expenditures, and the cities they are located in have less pollution, especially for observations of SOEs. That means pollution is a primary source of social-political pressure regarding CSR. Following their research, we examine whether the variance of local pollution moderates the relationship between ATPs and CSR. The overall results reveal that the link between ATPs and CSR is evident solely in locations with higher pollution levels, stemming from industrial wastewater, SO₂ emissions, and particulate matter emissions, thereby reinforcing the notion that social and political pressures are key drivers of CSR in China [18]. Furthermore, the moderating impacts are primarily observed in instances involving SOEs. Conversely, for non-SOEs, the relationship between ATPs and CSR remains consistently insignificant and is not influenced by pollution variations. This suggests that efficiency pressure is the primary catalyst for non-SOEs’ engagement in CSR, with political connections serving as a facilitating factor.

Although our results are based on Chinese firms, it does not mean that they lack generalization, because the tension between efficiency and legitimacy pressures in CSR decisions is a global phenomenon [11,13]. Our study aligns with Sundaramurthy [6], who suggested that examining when, rather than whether, ATPs enhance or diminish shareholder value is more meaningful.

Our potential contribution mainly focuses on three aspects. First, we extend the research on the combination of ATPs and CSR to emerging markets. Prior studies by Surroca and Tribó [15] and Kacperczyk [16] focused on large companies in different regions and periods, leaving a gap in understanding the impact on small and medium enterprises, particularly in emerging markets. Given the idiosyncratic nature of ATPs’ effects on managerial behavior, which varies with firm lifecycle and corporate governance levels [22,23,24], our results offer insights into this area and address the concerns raised by Cohen and Wang [8].

Second, we provide further evidence on the impacts of social-political factors in China and their positive effects from the perspective of CSR. The negative impacts of social-political factors on listed firms in China have long been criticized, both in practice and in academic research [25]. Our research aims to explain how China and its companies have achieved rapid growth despite these challenges.

Third, we extend research on the tension between efficiency and legitimacy pressures by showing its variation due to ownership structure. Our results reinforce the notion that mandated CSR might be ineffective because the market-based signals provided by CSR are weak [26,27]. Consequently, changes in the balance between efficiency and legitimacy pressures may not significantly impact CSR decisions, particularly for non-SOEs in China. It also has broader implications for stakeholders and policymakers in effectively promoting the development of CSR.

The remainder of the paper is organized as follows. Section 2 provides the background. Section 3 discusses the prior literature and develops our predictions. Section 4 describes our sample and research design. Section 5 presents our empirical results, and Section 6 offers further analysis and robust tests. Section 7 concludes the paper.

2. Institutional Backgrounds

2.1. ATPs in China

Before 2002, the autonomy of a company’s corporate charter in China was relatively limited due to the country’s code law system. Significant changes occurred in 2002 with the issuance of the “Code of Corporate Governance for Listed Firms” by the China Securities Regulatory Commission (CSRC), which introduced cumulative voting for the first time. Subsequent amendments to the “China Company Law and Securities Law” in 2005, along with the issuance of “The Guidance for the Corporate Charter of Listed Company” in 2006, further expanded the autonomous rights of a company’s corporate charter. Consequently, an increasing number of companies began incorporating autonomous articles, such as ATPs, into their corporate charters. The ATPs used by listed firms in China include cumulative voting rights, classified boards or staggered boards, supermajority requirements, golden parachutes, checks on the qualifications of directors and restrictions on board nomination rights in terms of the threshold or the duration of shareholdings, provisions to increase either the threshold or the duration of shareholdings qualified to propose an interim shareholders’ meeting, and provisions to decrease the threshold of changes in shareholdings qualified to an announcement. According to our data, the percentage of companies with ATPs increases year by year. In 2009, there were 1535 companies with at least one ATP in their corporate charter, accounting for 89.35% of listed firms. By 2017, the number had risen to 3464, representing 97.52% of listed firms. However, there is great variation in specific provisions. The most frequently adopted provision is the supermajority requirement, appearing in almost all firm–year observations with ATPs. The second is cumulative voting rights, accounting for 96% of firm–year observations with ATPs. Restrictions on board nomination rights in terms of the threshold of shareholdings, checks on the qualifications of directors, restrictions on board nomination rights in terms of the duration of shareholdings, and staggered boards account for 22.77%, 20.29%, 7.39%, and 5.07%, respectively. Although the golden parachute provision is the least adopted, the number of firms adopting it has increased from 30 in 2009 to 125 in 2017.

The development level of ATPs in China coincides with that of hostile takeovers. Before the non-tradable reform occurring from 2004 to 2007, the majority shares of listed companies were non-tradable, making it impossible to exert a hostile takeover. The non-tradable reform includes a lock-up period of at least three years for those shares to become fully tradable. Consequently, hostile takeovers have become theoretically possible for most firms since 2007. However, the presence of a block shareholder holding voting rights exceeding 30% has deterred most hostile takeover attempts. For companies lacking controlling shareholders, ATPs play a crucial role as they provide a defense mechanism. In the absence of ATPs, management would have to rely on government intervention when faced with a hostile takeover, as exemplified by the Vanke case. While the economic significance of Vanke led to eventual government intervention, this has not been the case in other hostile takeover instances, such as the acquisition of CSG by the same entity. Consequently, ATPs are perceived as a more effective protective measure for management teams.

2.2. CSR in China

The Chinese government has undertaken several CSR initiatives in recent years, reflecting the view that CSR can contribute to “building a harmonious society”, a key goal outlined by the Chinese Communist Party at the 2006 National People’s Congress [18]. AsChen, Huang, and Wang [18] summarized, in addition to the mandate specified in the 2006 Chinese Company Law, the government has implemented specific CSR specifications to recognize firms with excellent CSR actions while punishing firms with poor CSR performance. Attention on CSR has intensified since 2013, when the Third Plenary Session of the 18th CPC Central Committee clarified that CSR is a critical focus in deepening reform. On 1 January 2016, the “National Standards of the People’s Republic of China: Guidance on Social Responsibility” came into effect.

In December 2008, the Shanghai Stock Exchange (SSE) and Shenzhen Stock Exchange (SZSE) implemented regulations requiring certain listed firms to disclose their CSR activities. This mandate applies to firms in SSE’s “Corporate Governance Sector”, those with overseas shares, financial industry firms, and those listed on the “Shenzhen 100 Index”. These companies are obligated to include a CSR report along with their annual reports, detailing efforts in areas such as the protection of shareholders’ and creditors’ interests, workers’ rights, suppliers’ and customers’ rights, environmental protection, sustainable development, and public relations and social welfare services.

Rating agencies in China provide scores or ratings on CSR in a less comprehensive and detailed manner. For example, the “Research Report on Social Responsibility of China” by the Chinese Academy of Social Science usually covers companies on the “Shanghai and Shenzhen 300 Index”, which are only a subset of companies providing CSR reports. Nevertheless, RKS is the most popular database on CSR in Chinese academic research [20,28]. The RKS database is provided by Rankins, who provides China-specific products [20]. RKS ratings are widely used in China-related studies [20,28]. Zhong et al. [29] demonstrate that RKS is more suitable for measuring CSR disclosure quality. Please see Zhong et al. [29] for detailed descriptions of the RKS index system (

Table A1. Definition of variables.

Variable	Definition
SCORE	The rating score of CSR provided by the RKS database in year t.
ATP9	The sum of nine ATP provisions in year t − 1.
SIZE	The natural logarithm of ending total assets in year t − 1.
AGE	The natural logarithm of the monthly gaps between the fiscal end of year t − 1 and listing time.
ROA	Earnings before income tax, interest, depreciation, and amortization in year t − 1 deflated by average total assets in year t − 1.
MTB	The sum of the market value of equity and the book value of debt deflated by the book value of total assets, all measured as of the end of year t − 1.
LEV	The ratio of ending total liabilities in year t − 1 to ending total assets in year t − 1.
R&D	R&D expenditures in year t − 1 deflated by total revenues in year t − 1, multiplied by 1000 to match the magnitude of CSR.
SH1	The holding ratio of the largest shareholder in year t − 1 multiplied by 100.
INS	The percent of institutional holding in year t − 1 multiplied by 100.
OUT	The ratio of independent directors on the board of directors in year t − 1.
BOARD	The natural logarithm of the number of directors in year t − 1.
DUAL	An indicator variable which is equal to 1 if the CEO is also the chairman, and 0 otherwise.
MAN	An indicator variable which is equal to 1 if the CSR report is in the mandatory regime in year t, and 0 otherwise.

) and the sub-indicator system of RKS ratings (

Table A2. Results using firm-fixed effects and instrument variables. This table presents the results after controlling firm-fixed effects and the use of instrument variables. The dependent variable is SCORE. The instrument variable is the mean ATP at the industry level (IND_APT9) and the city level (DIS_ATP9). Variable definitions are in Appendix A, Table A1. *** and * indicate statistical significance at the 1% and 10% levels, respectively. The values of t are in parentheses.

	(1) Firm-Fixed Effect		(2) Instrument Variables
	Coef.	t	Coef.	t
ATP9	0.303 *	(1.84)	13.001 ***	(2.71)
Controls	Yes		Yes
Year	Yes		Yes
Industry	No		Yes
Firm	Yes		No
N	5696		5696
adj. R2	0.1970		0.4470

).

3. Literature Review and Hypothesis Development

3.1. ATPs and CSR Initiative

The agency cost view of CSR suggests that management would not always act in the interest of shareholders. Instead, they might engage in CSR at the expense of shareholders to pursue personal visibility and reputation [9,10]. According to this view, the probability and level of CSR engagement are positively related to a firm’s agency costs, manifested as managerial entrenchment. Meanwhile, ATPs protect management from the governance effects of the control market, entrenching incumbent managers and damaging shareholder interests [1]. Empirical research has provided evidence from various perspectives. For instance, prior studies find that firms with ATPs pay more to both management and employees [30] and encounter significant decreases in management shareholdings [31] and innovative long-term investments [32], increases in agency cost [33,34,35,36], and decreases in operating performance and firm value [8,37]. Therefore, the shared theoretical basis of agency cost implies that ATPs would increase the level of CSR actions via the mechanism of management entrenchment, as ATPs lower managerial employment risk and efficiency pressures.

However, recent studies on developed markets argue that CSR is a strategic instrument that enhances long-term value. Although CSR might occupy scarce resources, it can improve a firm’s reputation, brand, and social trust in the long run [12,38], gaining consensus from the institutional environment and improving organizational legitimacy [11,39], attracting customers and employees, and ultimately enhancing profitability and value [40,41]. This strategic view of CSR coincides with the long-term benefit hypothesis of ATPs, suggesting that the protection of managerial positions enables more efficient investment decisions [42] and encourages long-term investments due to decreased managerial myopia [43,44]. For example, ATPs decrease the efficiency pressure that managers face from control markets, which are induced by sluggish share prices, and prevent them from making short-term decisions [3]. Furthermore, ATPs can be viewed as a postponement of compensation for long-term implicit incentives [2]. As a result, firms with ATPs have been shown to engage in less earnings management [45,46], devote more to innovative activities [47], decrease tunneling by block shareholders [48,49,50], and eventually increase in long-term firm value [32,46,47]. Therefore, the combination of strategic CSR and the decrease in managerial myopia induced by ATPs indicate that the level of CSR is positively related to the extent of ATPs.

Based on the above-mentioned combination, we developed our first hypothesis as follows:

Hypothesis 1 (H1). 

Other things being equal, the level of CSR is positively related to the extent of ATPs.

There are also scenarios where CSR and ATPs exhibit contrasting or inconsequential relationships. For instance, according to the ‘quiet life’ hypothesis proposed by Bertrand and Mullainathan [30], managers tend to avoid challenging decisions when shielded from external pressures. Consequently, they are less likely to actively pursue CSR initiatives compared to restructuring a business. If the ‘quiet life’ hypothesis predominantly influences the correlation between CSR and ATPs, then the degree of CSR activities would not escalate with increased ATPs.

Moreover, even if a positive association is identified between CSR and ATPs, the underlying mechanism remains unresolved. To address this issue, further investigation is required to analyze a specific form of legitimacy pressure, namely social-political pressures, particularly within the context of China.

3.2. Social-Political Pressures in China

It is well-known that central and local governments are the most important stakeholders for listed firms in China, not only because many firms in China are ultimately controlled by central or local governments (SOEs) but also because the central government provides direct guidance on the disclosure and performance of CSR. Therefore, a primary motivation for CSR engagement in China is to comply with these requirements and gain regulatory legitimacy [20]. Compared with other stakeholders in the capital market, social-political pressures from governments are more significant [51]. For example, family-owned firms seeking political connections and those in regions with higher government intervention are more inclined to voluntarily disclose CSR information to enhance regulatory legitimacy [52]. If efficiency pressure decreases due to ATPs, Chinese firms may engage in more CSR activities because legitimacy pressure becomes more urgent [13].

SOEs and non-SOEs exhibit distinct motivations for engaging in CSR, primarily influenced by their levels of political resource allocation. SOEs, inherently possessing political legitimacy [53], may have reduced incentives to pursue CSR solely for legitimization purposes. However, SOEs are the mainstay of the Chinese economy and are expected to bear more socio-political responsibilities from the perspective of both the government and the public [54,55,56]. With the promotion of CSR in China, CSR performance has become an important part of SOE assessment, influencing managerial promotion to bureaucratic positions. In 2008, the State-owned Assets Supervision and Administration Commission of the State Council (SASAC) released “Guidelines to the State-owned Enterprises Directly under the Central Government” to give the impetus to the SOEs directly under the central government (CSOEs) to earnestly fulfill corporate social responsibilities, so as to realize the coordinated and sustainable development of enterprises, society, and the environment in all respects. In 2010, CSOEs with CSR as part of their performance evaluation system make up over 60 percent of the total [19]). ATPs decrease the demand for economic benefits to boost share prices due to reduced pressures from the control market, releasing more resources for CSR activities. Consequently, SOE managers have more discretionary resources to meet the CSR requirements imposed by regulatory institutions and the public, driven by the implicit benefits of personal promotion [57].

On the other hand, although the explicit goal of CSR engagement by non-SOEs is to establish political connections or gain political resources [58,59,60], the ultimate goal is economic return. Governments in China can intervene with loan resources, government subsidies, and tax exemptions to relieve financial constraints faced by non-SOEs [61,62]. ATPs decrease the efficiency pressures faced by all firms, making the pursuit of economic returns from CSR less urgent for non-SOEs. Thus, non-SOEs may have less incentive to seek political rent via CSR engagement.

Alternatively, non-SOEs may engage in CSR initiatives to fulfill minimum legitimacy management requirements [63]. The dynamic between legitimacy and efficiency pressures is altered by government interventions. However, the voluntary increase in CSR activities may not send positive signals under a mandated regulatory framework [27], potentially dampening non-SOEs’ motivations for corporate charters compared to the ATPs used by listed CSR endeavors.

Based on the above analysis, we expect that the impact of ATPs on CSR mainly exists in SOEs rather than non-SOEs. Hence, our second hypothesis is:

Hypothesis 2 (H2). 

Other things being equal, the relationship between CSR and ATPs is more profound in SOEs in comparison with non-SOEs..

4. Research Design

4.1. Research Design

According to Attig et al. [11], we used the following Model (1) to examine the impacts of ATPs on CSR.

SCORE = α₀ + α₁ATP9 + Controls +ε

(1)

where SCORE is the rating score of CSR provided by the RKS database, which is to proxy for the disclosure and performance of CSR In China [64,65]. While Zhong et al. [29] demonstrated that RKS is more suitable for measuring CSR disclosure quality, Marquis and Qian [20] used RKS to measure CSR reporting substantiveness. The results are robust when we use ESG ratings provided by Bloomberg and the Sino-Securities Index Information Service as alternative proxies for CSR. We used ATP9 to proxy for the extent of protection provided by ATPs [49,66], which is the sum of nine ATPs, as listed in Section 2.1. If our hypothesis H1 is true, then we expect the estimated coefficient of α₁ to be significantly positive.

We also included a set of control variables in our regression according to Attig et al. [11], with some modifications due to the institutional backgrounds of China. In particular, we tried to control the potential impacts of firm-specific fundamentals including (1) financial attributes such as firm size (SIZE), financial leverage (LEV), firm age (AGE), profitability (return on total assets, ROA), market-to-book ratio (MTB), and innovation efforts (R&D); (2) characteristics of ownership structure such as shareholding by the largest shareholder (SH1) and institutional holding (INS); (3) corporate governance mechanisms such as the percentage of independent directors (OUT), the size of the board of directors (BOARD), and the duality of the Chairman and CEO (DUAL); and (4) the regime of CSR reports (MAN). To mitigate the potential problems induced by endogeneity, we used lagged control variables except MAN. We also included year-fixed effects and industry-fixed effects in our regressions. The detailed definitions of variables are provided in Appendix A, Table A1.

To test Hypothesis 2, we divided our sample into two subsamples according to the nature of the ultimate controller. In particular, if the firm is ultimately controlled by central or local governments, then it is classified as an SOE; otherwise, it is classified as a non-SOE. We reran Model (1) with the two subsamples, respectively. According to Hypothesis 2, we conjecture that the relationship between the extent of ATPs and the level of CSR mainly occurs in the subsample of SOEs.

4.2. Sample Selection and Distribution

Our initial sample comprised all firms covered in the RKS database for the period from 2010 to 2019. We restricted our sample to those firms with information on ATPs, and we excluded firms in the financial industries and those with missing control variables from the sample. Our final sample included 5696 firm–year observations.

Table 1. Sample selection. The final sample comprises 5696 firm–year observations for firms in China between 2010 and 2019.

	Observations
Observations with CSR score during the period from 2010 to 2019	6937
Less:
Observations without ATP data	817
Observations in financial industries or with missing control variables	424
Final sample	5696

reports the process of sample selection. Our rating score of CSR and information on the CSR regime were obtained from the RKS database, data on ATPs were manually collected, and other variables came from the CSMAR database. To exclude the potential impacts of extreme values, all continuous variables are winsorized at 1 and 99 percentiles.

Table 2. Sample distribution by year and industry. The table shows the sample by year and industry.

Panel A: Distribution by Year
Year	Obs.	Percent
2010	340	5.97%
2011	400	7.02%
2012	462	8.11%
2013	542	9.52%
2014	562	9.87%
2015	569	9.99%
2016	594	10.43%
2017	611	10.73%
2018	774	13.59%
2019	842	14.78%
Total	5696	100.00%
Panel B: Distribution by industry
Industry	Obs.	Percent
Agriculture (A)	69	1.21%
Mining (B)	276	4.85%
Manufacturing (C)	3234	56.78%
Utilities (D)	318	5.58%
Construction (E)	186	3.27%
Wholesale and retail (F)	311	5.46%
Transportation (G)	302	5.30%
Accommodation and food services (H)	12	0.21%
Information technology (I)	285	5.00%
Real estate (K)	392	6.88%
Service (L)	58	1.02%
Scientific Research and Technology Service (M)	21	0.37%
Water conservancy, environment, and public facilities management (N)	47	0.83%
Education (P)	1	0.02%
Health and social work (Q)	21	0.37%
Culture, sports and entertainment (R)	92	1.62%
Others(S)	71	1.65%
Total	5696	100.00%

displays our sample distribution by year and industry. The results in Panel A indicate that firms with CSR scores increase year by year, which is consistent with the development of Chinese capital markets and the increased attention on CSR. The industrial distribution is similar to Chen et al. [18], with over half of our observations coming from the manufacturing industry.

5. Empirical Results

5.1. Descriptive Statistics

Table 3. Descriptive statistics. This table presents descriptive statistics for the variables used in our analysis. The sample firms are in China between 2010 and 2019. Variable definitions are in Appendix A, Table A1.

Variable	Obs.	Mean	Std.	Min	25%	Median	75%	Max
SCORE	5696	39.796	12.177	19.128	31.162	37.222	46.193	75.886
ATP9	5696	2.697	0.920	1	2	2	3	8
SIZE	5696	23.297	1.451	20.485	22.215	23.179	24.201	27.293
AGE	5696	2.585	0.532	1.098	2.302	2.708	2.995	3.332
ROA	5696	0.057	0.062	−0.209	0.029	0.051	0.082	0.252
MTB	5696	1.908	1.527	0.543	1.098	1.468	2.132	30.674
LEV	5696	0.497	0.198	0.074	0.351	0.510	0.647	0.905
R&D	5696	0.033	0.034	0.000	0.009	0.033	0.040	0.215
SH1	5696	36.853	15.846	7.822	24.052	35.701	49.012	74.984
INS	5696	55.442	22.295	1.560	41.063	58.537	71.391	95.455
OUT	5696	37.722	5.632	33.330	33.330	36.360	42.860	57.140
BOAR	5696	2.184	0.210	1.609	2.079	2.197	2.302	2.708
DUAL	5696	0.166	0.372	0	0	0	0	1
MAN	5696	0.559	0.496	0	0	1	1	1

reports the descriptive statistics of the main variables. The mean value of SCORE is 39.796, while the median of SCORE is 37.222. This is consistent with the results of Xiao et al. [67], who found that only a small number of firms have higher levels of CSR activities, with the CSR performance of the other firms being relatively low. The mean, median, and minimum values of ATP9 are 2.697, 2, and 1, respectively, indicating that over half of the firms have two ATPs. The mean value of ROA is 5.7%. Although it is a little higher than the average ROA of all listed firms, it is consistent with the fact that firms disclosing CSR reports are those with good governance and good performance. The average of MAN is 55.9%, indicating that over half of the observations belong to the mandatory regime. The mean and median of SH1 are 36.853 and 35.701, respectively, indicating that the ownership concentration is still higher in China and the phenomenon of a single majority shareholder is significant.

5.2. Preliminary Results

Table 4. The relationship between ATPs and CSR. This table presents the results of the relationship between ATPs and CSR using OLS regression. The dependent variable is SCORE, calculated as CSR scores provided by the RKS database. Variable definitions are in Appendix A, Table A1. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively. The values of t are in parentheses.

	(1) Full Sample		(2) SOEs		(3) Non-SOEs
	Coef.	t	Coef.	t	Coef.	t
ATP9	0.453 ***	(3.09)	0.852 ***	(4.14)	0.072	(0.35)
SIZE	4.412 ***	(31.22)	4.223 ***	(21.98)	3.870 ***	(16.84)
AGE	−1.619 ***	(−5.82)	0.043	(0.09)	−2.897 ***	(−7.48)
ROA	−5.110 **	(−2.08)	−2.212	(−0.58)	−4.452	(−1.41)
MTB	0.263 **	(2.39)	0.081	(0.39)	0.321 ***	(2.65)
LEV	−6.524 ***	(−7.04)	−7.004 ***	(−5.59)	−2.311 *	(−1.67)
R&D	3.217	(0.69)	2.999	(0.42)	28.480 ***	(4.35)
SH1	0.024 **	(2.17)	0.002	(0.11)	−0.002	(−0.14)
INS	0.018 **	(1.99)	0.088 ***	(5.59)	0.001	(0.07)
OUT	0.031	(1.15)	0.037	(1.07)	−0.019	(−0.44)
BOARD	4.232 ***	(5.62)	3.128 ***	(3.11)	5.582 ***	(4.82)
DUAL	−0.619 *	(−1.69)	−0.485	(−0.77)	−0.318	(−0.73)
MAN	−0.681 **	(−2.29)	−0.319	(−0.80)	−1.341 ***	(−3.02)
Constant	−77.483 ***	(−21.84)	−75.904 ***	(−14.52)	−63.204 ***	(−10.95)
Year	Yes		Yes		Yes
Industry	Yes		Yes		Yes
N	5696		3437		2139
Adj-R2	0.3457		0.3663		0.3185
F	80.177		56.181		28.008

reports the preliminary results of Model (1). In the full sample regression, the estimated coefficient of ATP9 is 0.453, which is significantly positive at the level of 1%. The results suggest that the level of CSR increases with the extent of ATPs, consistent with Hypothesis 1.

For the subsample of SOEs, the estimated coefficient of ATP9 is 0.852, which is significantly positive at the level of 1%. For the subsample of non-SOEs, the estimated coefficient of ATP9 is 0.072, which is insignificantly different from zero. The results indicate that the relationship between CSR and ATPs mainly occurs in the sample of SOEs, which is consistent with Hypothesis 2. As ATPs lower efficiency pressure, SOEs engage in more CSR activities to manage their legitimacy while non-SOEs remain unchanged.

5.3. Moderating Roles of Efficiency or Legitimacy Pressure

In this section, we provide further evidence on the mechanisms through which ATPs affect CSR. If it is the tension between efficiency and legitimacy pressures that drives the relationship between ATPs and CSR, then that relationship would vary with the level of efficiency or legitimacy pressures. Specifically, when efficiency pressures are lower, more resources will be released by ATPs and the relationship between ATPs and CSR will be stronger. When legitimacy pressures are greater, firms will have greater incentives to allocate more resources to CSR. Thus, the relationship between ATPs and CSR would also be stronger. We hypothesize that this holds true for both the whole sample and the sub-sample of SOEs, as SOEs are expected to bear more social-political burdens. If changes in the tension between efficiency and legitimacy pressures have insignificant impacts on CSR, then the relationship between ATPs and CSR would remain constant regardless of the level of these pressures. We conjecture that this is true for non-SOEs, as they are generally expected to meet only the minimum requirements of CSR.

We use product market competition and total factor productivity (TFP) as proxies for efficiency pressures, and media coverage as a proxy for legitimacy pressures [13].

First, firms operating in markets with lower competition can enjoy higher profits and face lower efficiency pressures. To test the impact of product market competition, we calculated the Herfindahl–Hirschman Index (HHI) by year and industry, then divided our sample into two groups based on the sample median. One subsample consisted of observations with higher competition (lower HHI), and the other consisted of observations with lower competition (higher HHI). The impacts of product market competition are reported in Panel A of

Table 5. The moderating effects of efficiency and legitimacy pressures. This table presents the results of whether product market competition, total factor productivity, and media coverage moderate the relationship between ATPs and CSR using OLS regressions. “Lower” (“Higher”) represents lower (higher) product market competition, total factor productivity, and media coverage. The dependent variable is SCORE, calculated as CSR scores provided by the RKS database. Variable definitions are in Appendix A, Table A1. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively. The values of t are in parentheses.

Panel A: Moderating Roles of Product Market Competition
	Full Sample		SOEs		Non-SOEs
	(1) Lower	(2) Higher	(3) Lower	(4) Higher	(5) Lower	(6) Higher
ATP 9	0.628 ***	0.099	1.006 ***	0.524 *	0.109	0.104
	(3.05)	(0.48)	(3.64)	(1.74)	(0.37)	(0.37)
Controls	Y	Y	Y	Y	Y	Y
Year	Y	Y	Y	Y	Y	Y
Industry	Y	Y	Y	Y	Y	Y
N	2876	2797	1883	1547	955	1170
Adj-R2	0.4110	0.2869	0.4446	0.2731	0.3168	0.3600
F	55.210	34.087	42.847	19.153	13.288	20.928
Panel B: Moderating roles of TFP
	Full Sample		SOEs		Non-SOEs
	(1) Higher	(2) Lower	(3) Higher	(4) Lower	(5) Higher	(6) Lower
ATP 9	0.783 ***	0.031	1.448 ***	−0.041	0.294	−0.136
	(3.21)	(0.19)	(4.50)	(−0.17)	(0.77)	(−0.56)
Controls	Y	Y	Y	Y	Y	Y
Year	Y	Y	Y	Y	Y	Y
Industry	Y	Y	Y	Y	Y	Y
N	2846	2847	1991	1445	799	1339
Adj-R2	0.3586	0.2184	0.3631	0.2308	0.4074	0.2353
F	42.857	22.499	32.514	13.037	15.828	12.435
Panel C: Moderating roles of media coverage
	Full Sample		SOEs		Non-SOEs
	(1) Higher	(2) Lower	(3) Higher	(4) Lower	(5) Higher	(6) Lower
ATP 9	0.461 **	0.315	0.889 ***	0.340	0.125	0.251
	(2.13)	(1.64)	(2.94)	(1.26)	(0.41)	(0.90)
Controls	Y	Y	Y	Y	Y	Y
Year	Y	Y	Y	Y	Y	Y
Industry	Y	Y	Y	Y	Y	Y
N	2833	2834	1791	1623	958	1176
Adj-R2	0.3826	0.2498	0.4000	0.2665	0.3808	0.2535
F	48.434	25.823	35.093	17.367	17.347	11.785

.

For the full sample analysis, the coefficient of ATP9 is 0.628 (t = 3.05) in the subsample with lower competition and is insignificantly different from zero in the subsample with higher competition. The results for the SOE sample analysis are similar to those for the full sample, whereas the results for the non-SOE sample are different. For SOE observations, in the subsample with lower competition, the coefficient of ATP9 is 1.006, which is significantly positive at the 1% level. In the subsample with higher competition, the coefficient of ATP9 is 0.524, which is significantly positive at the 10% level. These results support the view that SOEs’ CSR activities are driven by socio-political pressure, a unique form of legitimacy pressure in China. For non-SOE observations, the estimated coefficients of ATP9 are insignificantly different from zero in both Columns (5) and (6), consistent with our conjecture that non-SOEs’ CSR is unaffected by the tension between efficiency and legitimacy pressures.

Second, we used the LP method to calculate TFP [68]. A higher value of TFP corresponds to lower efficiency pressure. We classified our observations into two groups according to the sample median. The results, as shown in Panel B of Table 5, are similar to those in Panel A. For example, in Columns (1) and (3), the estimated coefficients of ATP9 are 0.783 and 1.448, both of which are significantly positive at the 1% level. In the other columns, the estimated coefficients of ATP9 are insignificantly different from zero. These results suggest that when efficiency pressure decreases, SOEs engage in more CSR to meet their socio-political objectives, whereas non-SOEs simply enjoy the reduced pressure.

Lastly, we calculated the media coverage (MC). A higher value of MC corresponds to greater legitimacy pressures. We divided our sample into two groups according to the sample median. The regression results, as presented in Panel C of Table 3, are consistent with our conjectures. Specifically, the relationship between ATPs and CSR is more pronounced in observations with higher median coverage. Furthermore, this heterogeneity is driven by SOEs, as changes in media coverage have little impact on non-SOEs.

In sum, our results indicate that the positive association between ATPs and CSR concentrates in SOEs under lower efficiency or higher legitimacy pressures. It deepens our understanding of the tension between efficiency and legitimacy pressures as well as the differential consequences of ownership nature.

5.4. Further Analysis of Legitimacy Pressure

In this section, we provide further analysis of social-political pressures specific to CSR activities.

First, we examine whether socio-political pressures derived from environmental concerns moderate the relationship between CSR and ATPs. CSR concerns are more likely to relate to industry standards or the minimum performance levels expected by the public [69]. Furthermore, CSR concerns often stem from factors beyond a firm’s specific CSR efforts [70], thus relating more to socio-political pressures. For instance, China’s Eleventh Five-Year Plan explicitly links officials’ environmental performance to their promotions [68]. Chen et al. [18] find that firms located in more polluted cities face greater socio-political pressures due to higher environmental expenditures. Industrial sewage, emissions, and dust released at the city level serve as proxies for the extent of environmental regulation and performance [68,71]. Therefore, we used these indicators to measure the level of socio-political pressure at the city level. We divided our sample into two groups according to the sample median and reran our regressions. The results are reported in

Table 6. The moderating effects of city-level pollution. This table presents the results of whether city-level pollution moderates the relationship between ATPs and CSR using OLS regression. “Low” (“High”) represents lower (higher) pollution. The dependent variable is SCORE, calculated as CSR scores provided by the RKS database. Variable definitions are in Appendix A, Table A1. *** and * indicate statistical significance at the 1% and 10% levels, respectively. The values of t are in parentheses.

Panel A: Full Sample
	Industrial sewage		Industrial SO2 release		Industrial dust release
	(1) High	(2) Low	(3) High	(4) Low	(5) High	(6) Low
ATP 9	0.786 ***	−0.177	0.799 ***	−0.028	0.959 ***	−0.238
	(3.23)	(−0.88)	(4.24)	(−0.12)	(5.10)	(−1.02)
Controls	Yes	Yes	Yes	Yes	Yes	Yes
Year	Yes	Yes	Yes	Yes	Yes	Yes
Industry	Yes	Yes	Yes	Yes	Yes	Yes
N	2481	2479	3523	2172	3481	2214
adj. R2	0.3465	0.3897	0.3484	0.3632	0.3541	0.3644
F	36.537	43.763	50.564	34.470	51.195	35.294
Panel B: SOEs
	Industrial sewage		Industrial SO2 release		Industrial dust release
	(1) High	(2) Low	(3) High	(4) Low	(5) High	(6) Low
ATP 9	1.619 ***	−0.524 *	1.323 ***	0.506	1.580 ***	0.170
	(5.01)	(−1.82)	(4.83)	(1.60)	(5.76)	(0.54)
Controls	Yes	Yes	Yes	Yes	Yes	Yes
Year	Yes	Yes	Yes	Yes	Yes	Yes
Industry	Yes	Yes	Yes	Yes	Yes	Yes
N	1537	1552	2039	1398	2053	1384
adj. R2	0.3773	0.4042	0.3637	0.3892	0.3570	0.4127
F	27.590	31.067	33.359	25.724	32.651	28.001
Panel C: Non-SOEs
	Industrial sewage		Industrial SO2 release		Industrial dust release
	(1) High	(2) Low	(3) High	(4) Low	(5) High	(6) Low
ATP 9	−0.312	0.226	0.320	−0.368	0.348	−0.354
	(−0.86)	(0.75)	(1.29)	(−0.96)	(1.41)	(−0.90)
Controls	Yes	Yes	Yes	Yes	Yes	Yes
Year	Yes	Yes	Yes	Yes	Yes	Yes
Industry	Yes	Yes	Yes	Yes	Yes	Yes
N	907	855	1433	706	1374	765
adj. R2	0.3701	0.3659	0.3080	0.3609	0.3231	0.3333
F	15.787	14.689	18.223	12.058	18.710	11.610

.

The results indicate the following. First, the positive relationship between ATPs and CSR mainly occurs in observations with higher social-political pressures, i.e., firms located in cities with more severe pollution. Second, the moderating role of social-political pressures is driven by SOEs. Third, non-SOEs are insensitive to the variance in social-political pressures as their estimated coefficients of ATP9 are all insignificantly different from zero.

The above results have shown that for SOEs, social-political pressure is the main driver of the relationship between CSR and ATPs. Chen et al. [18] and Bradshaw et al. [57] found that local SOEs (LSOEs) face more social-political pressure than central SOEs (CSOEs). Therefore, we expect that the relationship between CSR and ATPs is more profound for LSOEs. The results shown in

Table 7. Further analysis of SOEs. This table presents the results of whether the level of government and the regime of CSR disclosure moderate the relationship between ATPs and CSR in SOEs using OLS regression. The dependent variable is SCORE, calculated as CSR scores provided by the RKS database. Variable definitions are in Appendix A, Table A1. *** indicate statistical significance at the 1%level. The values of t are in parentheses.

	(1)	(2)	(3)	(4)
	Local SOE	Central SOE	Mandatory CSR	Voluntary CSR
ATP 9	1.051 ***	−0.292	0.852 ***	0.352
	(4.34)	(−0.73)	(3.29)	(1.05)
Controls	Yes	Yes	Yes	Yes
Year	Yes	Yes	Yes	Yes
Industry	Yes	Yes	Yes	Yes
N	1508	1044	1748	890
Adj-R2	0.2446	0.5499	0.3981	0.3097
F	14.185	37.412	32.231	12.080

confirm our conjectures. In particular, in Column (1), the estimated coefficient of ATP9 is 1.051 which is significantly positive at the level of 1%; in Column (2), the estimated coefficient of ATP9 is −0.292, which is insignificantly different from zero.

Chen et al. [18] also demonstrate that firms with mandatory CSR disclosures have more social-political pressures. Therefore, we expect that the relationship between CSR and ATPs is stronger in observations with mandatory CSR. The results, shown in Columns (3) and (4), confirm our conjecture. For the sample with mandatory CSR, the estimated coefficient of ATP9 is 0.852 which is significantly positive at the level of 1%; for the sample with voluntary CSR, the estimated coefficient of ATP9 is 0.352, which is insignificantly different from zero.

The results in Table 4 indicate that the tension between efficiency and legitimacy pressures insignificantly affects CSR. A potential explanation is that non-SOEs are content with meeting the minimum requirements of legitimacy management, and a voluntary increase in CSR does not signal positive information to stakeholders. To provide further evidence, we examined two settings with varying legitimacy pressures. Specifically, we divided our sample based on the regime of CSR (mandatory or voluntary) and the presence of political connections (with PC or without PC). The results presented in

Table 8. Further analysis of non-SOEs. This table presents the results of whether the regime of CSR disclosure and the status of political connections moderate the relationship between ATPs and CSR in non-SOEs using OLS regression. The dependent variable is SCORE, calculated as CSR scores provided by the RKS database. Variable definitions are in Appendix A, Table A1. The values of t are in parentheses.

	(1)	(2)	(3)	(4)
	Mandatory CSR	Voluntary CSR	With PC	Without PC
ATP 9	−0.086	−0.016	−0.176	0.225
	(−0.29)	(−0.05)	(−0.57)	(0.79)
Controls	Yes	Yes	Yes	Yes
Year	Yes	Yes	Yes	Yes
Industry	Yes	Yes	Yes	Yes
N	860	1279	972	1167
Adj-R2	0.4613	0.2837	0.3353	0.3369
F	22.019	15.060	14.607	17.012

suggest that the estimated coefficients of ATP9 are all insignificantly different from zero. In other words, the insignificant relationship between ATPs and CSR for non-SOEs does not vary with legitimacy pressures.

6. Robustness Tests

6.1. Alternative Proxies

To further ensure the robustness of our results, we used alternative proxies for ATPs and CSR performance. Our ATP9 is borrowed from the G-index proposed by Gompers et al. [66]. Bebchuk et al. [72] argue that the effects of some provisions are relatively small, so they selected the most effective six provisions to construct their E-index. Due to the legal restrictions in China, we could find only three of them to construct our E-index, namely ATP3. We replaced ATP9 with ATP3 in our Model (1) and reran the regression. The results are reported in Column (1) of

Table 9. Results using alternative proxies. This table presents the results of alternative proxies for CSR and ATPs. In Column (1), the dependent variable is SCORE and the key independent variable is ATP3, which is constructed according to Bebchuk et al. [72]. In Column (2), the dependent variable is RATE and the key independent variable is ATP9. In Columns (3) to (6), the dependent variables are the comprehensive and dimensional ratings of CSR from Bloomberg. Variable definitions are in Appendix A, Table A1. *** indicates statistical significance at the 1% level. The values of t or z are in parentheses.

			Ratings from Bloomberg
	(1) SCORE	(2) RATE	(3) ESG_Rate	(4) E_Rate	(5) S_Rate	(6) G_Rate
ATP 9		0.070 ***	0.601 ***	1.003 ***	0.332 ***	0.509 ***
		(2.63)	(7.01)	(6.80)	(3.54)	(3.80)
ATP 3	1.699 ***
	(5.05)
Controls	Yes	Yes	Yes	Yes	Yes	Yes
Year	Yes	Yes	Yes	Yes	Yes	Yes
Industry	Yes	Yes	Yes	Yes	Yes	Yes
N	5696	5696	5083	5056	5075	5083
adj. R2	0.3476	0.0962	0.5850	0.2782	0.2517	0.6366
F	80.822	2201.20	189.519	52.270	45.909	235.238

. As is shown, the coefficient of ATP3 is 1.699, which is significantly positive at the level of 1%. This means that our main results are robust with regards to the measurement of ATPs.

The RKS database also provides the rating grades of CSR, including A, BBB, BB, B, and CCC, and the first four grades would sometimes be suffixed by + or −. We set grade A as 3, grades BBB, BB, and B as 2, and grade C as 1. Then, we reran our Model (1) using ordered logit regression. Column (2) of Table 9 reports the results. The coefficient of ATP9 is 0.070, which is significantly positive at the level of 1%.

We also used ratings from Bloomberg as proxies for CSR. In addition to the comprehensive ratings, we also examined the impacts of ATPs on dimensions of CSR. The results, as shown in Columns (3) to (6), suggest that the relationship between ATPs and CSR is robust to alternative measures of CSR. Furthermore, that relationship applies to all dimensions of CSR.

6.2. Endogeneity Problems

Although we have lagged our independent variables to mitigate the potential problems of endogeneity, there are still potential problems induced by omitted variables and inverse causality. Specifically, the relationship between ATPs and CSR performance is driven by some common variables that are omitted, or it is CSR performance that determines ATPs.

To diagnose the severity of those problems, we conducted four tests including (1) controlling firm-fixed effects; (2) using instrumental variables in a two-stage regression; (3) using the hostile takeover of Wanke as an exogenous shock to perform group testing; and (4) placebo tests.

Firstly, we argue that the systematic firm-specific factors influencing the protection of ATPs and the level of CSR initiatives are unlikely to change in the short term, at least during our sample period. Therefore, firm-fixed effects should effectively control for the confounding effects induced by these factors [73]. As shown in Column (1) of Table A2, our baseline results remain robust.

Second, following Attig et al. [11], we used the mean ATP9 at the industry level (IND_APT9) and the city level (DIS_ATP9) as instrument variables. The regression results from the second stage, as reported in Column (2) of Table A2, are consistent with our baseline results.

Third, we used the hostile takeover of Vanke in 2015 for group testing. Although hostile takeovers are rare in China, the Vanke battle was highly significant and underscored the importance of ATPs. This event highlighted the weaknesses of ATPs for many firms, prompting adjustments to their corporate charters accordingly. Therefore, the Vanke battle can serve as an exogenous shock to ATPs but does not directly influence CSR. Therefore, comparing tests conducted before and after the battle allows us to clearly identify causality between ATPs and CSR. We divided our sample into two groups, data before 2015 and data after 2015, and reran our baseline regression. The results reported in

Table A3. Effects of Baoneng’s attempt to control Wanke via a hostile takeover. This table presents the impacts of Baoneng’s attempt to control Vanke via a hostile takeover on the relationship between ATPs and CSR using OLS regression. The dependent variable is SCORE. Variable definitions are in Appendix A, Table A1. *** indicates statistical significance at the 1% level. The values of t are in parentheses.

	(1)	(2)
	Before Year 2015	Since Year 2015
ATP 9	−0.009	0.759 ***
	(−0.04)	(3.88)
Controls	Yes	Yes
Year	Yes	Yes
Industry	Yes	Yes
N	2306	3390
adj. R2	0.3443	0.3040
F	36.595	45.836

are consistent with our expectations. Hence, we conclude that ATPs influence CSR, and any potential reverse causality does not significantly alter our main findings.

Lastly, we constructed a randomly assigned value of ATPs (Random_ATP9) variable to conduct our placebo tests. If true ATPs indeed affect CSR disclosure, then the randomly assigned ATPs should not show a similar effect. We repeated this random assignment 1000 times and plotted the distribution of the estimated coefficients of Random_ATP9. As shown in Figure 1, the estimated coefficients of Random_ATP9 are normally distributed around 0, excluding the influence of unobservable firm heterogeneity on our results.

7. Conclusions

With the protection provided by ATPs, will managers enjoy a quiet life or cater to the interests of more stakeholders? We find some interesting phenomena with Chinese data. Based on a sample of A-share companies for the period from 2010 to 2019, we find that ATPs increase the level of CSR and the relationship is mainly driven by SOEs. This positive association between ATPs and CSR is most pronounced in instances where organizations exhibit higher efficiency or face heightened legitimacy pressures, a trend predominantly observed in SOEs.

We have shown that the tension between efficiency and legitimacy pressures has little impact on the CSR of non-SOEs. We argue that non-SOEs might enjoy a quiet life with a minimum level of CSR. However, conclusive evidence supporting this assertion is limited by constraints in data availability and technological resources. Future research efforts could delve deeper into exploring this hypothesis. More comprehensive comparative analyses and longitudinal studies are necessary to validate and understand the interplay between efficiency and legitimacy pressures in shaping CSR practices among non-SOEs.

The existing literature posits that mandatory CSR initiatives may result in unintended outcomes [26,27]. Our results indicate that the effects of mandatory CSR vary with state ownership. The positive association between ATPs and CSR is driven mainly by SOEs under lower efficiency or higher legitimacy pressures. These findings will bring new insights and have broader implications for stakeholders and policymakers.