The Corporate Economic Influence and Corporate Social Responsibility: Evidence from China

Abstract

This paper uses a panel threshold model to examine the relationship between corporate social responsibility (CSR, hereafter) and enterprises’ economic influence on regional gross domestic product (GDP, hereafter) and employment. We find that there is a threshold effect between CSR and enterprises’ economic influence on regional GDP and employment in different regimes of local GDP and unemployment rates. When local GDP is low, the relationship between CSR and enterprises’ economic influence on regional GDP is significantly negative; however, when local GDP is high, the relationship between the two factors is significantly positive. Meanwhile, firms employing more staff do less CSR when the local unemployment rate is higher. Furthermore, in terms of different government types, the relationship between CSR and corporate influence on regional GDP is negative in predatory and collusive governments, but not in market-leading governments, and the relationship between CSR and corporate influence on regional employment seems insignificant. The findings imply that local leading enterprises exert influence on the social responsibility rules stipulated by local governments by decreasing or increasing regional GDP and regional employment.

1. Introduction

Corporate social responsibility (CSR, hereafter) is of great concern to both governments and companies. Governments can restrain or promote CSR, solve the “market failure” and the externality problems of enterprises through policies, regulations, and tax incentives [1,2]. However, state-owned enterprises (SOEs, hereafter), large companies, and political-related corporations with more resources are more likely to influence the decision of the government by supporting the campaign, lobbying the government, and bribing officials [3]. Leading enterprises can exert influence on the performance of government officers and regional development plans by resorting to their financial power, especially in the Gross Domestic Product (GDP, hereafter). As a saying coined after the 2008 financial crisis goes, they are “too big to fail” [4], so the enterprise tycoons attempt to seize the governments, policymakers, and regulators, who can loosen CSR regulations, and cannot undertake the responsibilities of CSR supervision. According to some researchers [5,6], corporations with greater economic impact in local places can force the government to compromise by setting up factories at boundaries, which is known as the “Slope Effect” [7]. Especially when national capture of the surroundings is more prevalent, and CSR regulation is unclear or ambiguous, firms tend to shoulder less CSR because it is costly [1], and there is a negative relationship between local corruption and CSR [8].

In China, monopoly SOEs or regional leading enterprises have the right of voice on regional public policy. They can not only kidnap the government to acquire regional protection policy [9] but also force the different regional governments to “race to the bottom” [10] to pay attention to the environment [11] and employee rights and interests [12]. For example, the widespread haze and smog, continuous sewage water discharged by Henan Lotus Gourmet Power Group, and leakage accidents concealed and unreported by Zijin Mining Group all imply that the enterprises conspire with the local governments.

There is little evidence in the existing literature to suggest that local governments may reduce or relax CSR regulation on leading enterprises because there are shared interests between local government and leading enterprises. This kind of literature has two limitations. First, the interests of the government and leading enterprises are not always the same. Second, the influence of government capacity on enterprises is rarely considered in previous studies, and the implicit assumption is that the government has unlimited regulatory power. In fact, local governments are posed to resource constraint while executing their local strategies. In China, local governments have multiple goals, such as political promotion, financial pressure, and interests of regional governments. For local government officials, political promotion is the priority, and political promotion always serves GDP promotion and social stability. Therefore, government officials are not always on the sides of enterprises. This paper studies the relationship between CSR and corporate economic influence on local GDP and employment rates under different local GDP and unemployment conditions. We further investigate the effect of corporate economic influence on local GDP and employment rate on CSR under different government types.

This paper attempts to answer how the various economic impacts of firms lead to different CSR behaviors occurring across different types of governments. As our study shows, the greater the firm’s impact on total regional output and employment, the higher potential protection they can obtain from governments, and the stronger enterprises’ voices can be. Thus, leading enterprises are more likely to obtain protection from social responsibility regulation, which can help them shoulder less social responsibility. However, the marginal utility of regional governments’ dependency on economic influence is diminishing. When the economy reaches a certain level, enterprises cannot use total regional output as an excuse to negotiate with governments. On the contrary, local governments will ask leading enterprises to take on more social responsibility. Moreover, according to this study, the relationship between corporate GDP influence and CSR is different in the regression coefficients of subgroups of different government types. This shows that it is different in the regression coefficients of subgroups of different government types.

This research sheds light on the literature in three important ways. First, focusing on the economic influence on corporate social responsibility behaviors, it extends the study from one single direction of enterprises’ property rights and political relation to the bilateral interaction of companies and governments. Second, the paper divides the government into predatory governments, collusive governments, and market-oriented governments based on the response of government capabilities and policy context. This lays an objective foundation for government capture, government behaviors, and the relationship between governments and corporations. Third, this paper effectively links the study of government and CSR behavior by presenting the differences and motivations for CSR behavior based on the conflicting interests and changing goals between leading firms and local governments. This paper dynamically presents the impact of economic influence on CSR behavior differences under different government actions. This paper illustrates how changes in local government types propagate to local CSR decisions. Finally, recent literature shows the impact of local corruption on CSR by highlighting the role of local factors on CSR [8]. The contribution of this paper to the relevant study is that it underlines the negative effect of local corruption on CSR. More importantly, a strong negative effect of the corporate GDP influence on local government on CSR is introduced in this paper.

2. Literature Review and Research Hypothesis

2.1. Corporate Social Responsibility as a Strategy Expense Highly Subject to Changes in Local Government Development

Corporations react to economic uncertainty [13]. Strategic firms respond to changes in economic growth by altering their CSR, regarding the conjecture that higher slack resources [14]. In China, local governments control the key competitive resources of companies, so Chinese companies must turn to the government for help. Sound CSR performance (CSP, hereafter) is an effective means of seeking political asylum or meeting the political demands of enterprises [15,16]. A good CSP can maintain sufficient flexibility in maintaining policy barriers to a firm’s competitive advantage [17]. Some scholars believe that corporate social responsibility and corporate donation are legal ways to capture the government, and CSR has an obvious rent-seeking tendency [18,19,20].

Consequently, the critical economic timeline of local government may represent the ideal time to make investments in targeted CSR projects so that companies differentiate themselves from their competitors, especially for enterprises with targeted CSP as a central part of their strategy. According to the literature [21], there are “thresholds” in the impact of the Genuine Progress Indicator (GPI) on changes in resource consumption and environmental pollution [22]. Regional economic growth is a prerequisite for environmental rules. When the region’s economy is relatively sluggish, economic development has become the top priority for local governments. Local enterprises can force local governments to loosen environmental regulations through border migration [5,6]. As the economy develops, the marginal effect of regional economic development on “performance enhancement” diminishes. On the contrary, the social environment and livelihood projects have become the next highlight of “political performance promotion”. Environmental performance is one factor in deciding whether an official can be promoted to mayor in China [23], so local officials shift their attention from economic growth to livelihood and public goods. It can be said that the degree and manner of local government regulation vary with the external social environment and their own political needs. Therefore, the level of economic development should be considered an important threshold variable for the impact of business economics on CSR behavior.

The assessment of the regional unemployment rate and economic growth rate and the pursuit of a loose financial environment force local governments to stop the liquidation of SOEs or large-scale enterprises [24]. During the introduction and implementation of pollution projects, local governments always find it hard to seek a balance between the regional regulation maximizing public interests and the regulation maximizing private interests. The huge differences in regional economic development and financial resources may also be magnified in terms of the supply quantity and regulation intensity of CSR behavior regulation. As revealed by [25], the Kuznets curve of the environment is more reflected as a result of changes in the intensity of real environmental regulation in China. Therefore, the influence of threshold variables, such as government hiring, management capacity, and self-financing capacity, needs to be further considered in the selection of thresholds for corporate economic influence and CSR behavior. Therefore, this paper proposes the following research hypothesis:

H1A. 

With regional GDP as a threshold variable, the economic impact of enterprises on GDP (hereafter referred to as GDPINF) is negatively correlated with social responsibility when the level of regional economic development is below the first threshold. When the level of regional economic development is above the first threshold value, $GDPINF$ is positively related to social responsibility.

H1B. 

With regional management as a threshold variable, when the difference between the regional unemployment rate and the average unemployment rate in the province is lower than the first threshold value, the corporate economic influence of employment (EMINF, hereafter) is positively related to social responsibility. When the difference between the regional unemployment rate and the average unemployment rate in the province is higher than the first threshold value, $EMINF$ is positively related to social responsibility.

2.2. Differential Influences of Local Government Types on CSP Strategic Choice

Corporate economic power refers to the ability of an enterprise to influence government rules and regulations without resorting to private payments to public officials. This power is usually a result of factors, such as firm size and country ownership [26]. The more powerful an enterprise is, the more influence it has on government rules and regulations, so that it can more easily access resources and effectively transform them into desired products or services. However, local government officials’ pursuits of promotion and interest impact the economic influence of enterprises [23,27]. Therefore, showing the characteristics of the threshold effect, the impact of corporate economic influence on the government’s social responsibility regulation and the role of corporate social responsibility behaviors are not only affected by the status of the enterprise itself but also affected by regional economic development, the ability of government self-financing, and regional employment and other governance capacity.

The Chinese government has already established a highly replicable and goal-oriented political governance system from the central government to the local government. An effective mechanism for selecting officials has been established at all levels of government. After meeting the “political red line” targets, such as “family planning” with the one-vote veto, personnel assessment system for GDP growth, political championships [28], and the pursuit of loosening financial revenue, local governments in China tend to lend “helping hands” for local economic intervention in their competition for economic growth [29]. It has been generally accepted by the academic community that the Chinese government uses the “helping hands’” model of the economy [30,31]. Under such a multitasking target assessment system (TRS, hereafter), local governments need to increase their investments in infrastructure projects, education projects, science and technology projects, and livelihood and welfare spending within their limited budgets. Abundant government spending makes the local financial system seem trapped in various problems. Therefore, “fees in lieu of levies”, extra-budgetary income, and extra-system income have become important legal channels for local government to generate revenue and become important evidence of local governments’ “grasping hands”.

In fact, the transitional local governments are always struggling to choose between “grabbing hands” and “helping hands”. Heterogeneous government resources in different regions, differences in government resources, and abilities in different development stages in the same region make governments behave differently. In Olsen’s national classification system, per capita income and human right protection are two important categories. From a comparative institutional perspective, countries can be divided into predatory governments, collusive governments, and democratized governments. However, [32] thinks that the game model of “government and people” is an important indicator of the type of country. However, he believes that the capture of enterprises only exists in the “collusive government”. Therefore, based on the characteristics of China’s economic and social development, this paper categorizes the government into predatory government, collusive government, and market-oriented government according to the panel threshold of the level of unemployment rate governance and per GDP. Specifically, (1) the market-oriented government refers to the local governments that are committed to protecting individual property rights through contracts rather than arbitrarily depriving or violating individual property rights in the TRS assessment system, when local economies expect higher levels of output. (2) Collusive government may be captured by the local elites and interest groups and set up policies that are beneficial to interest groups in terms of market access and development, or use “helping hands” to directly assist enterprises or industries, and even limit the development of other competitive enterprises or industries through “grabbing hands.” (3) The “predatory local government” refers to the local government, which is loosely organized. A group of bureaucrats, similar to bandits, plunder their economic subjects through administrative power and predominate their economic subjects under their jurisdiction.

Due to social and historical reasons, as well as the constraints of regional natural resources and the environment, the regional governments in economically underdeveloped regions will abandon economic competition and choose club convergence [33]. The highlights of the official promotion are as follows: the economic assistance provided by the central government to the backward areas, the appointment by the higher authorities, and the promotion determined by the working experience rather than by GDP in backward regions. Therefore, officials in backward regions regard family planning and social stability as the red line of political promotion, and thus, they are not sensitive to other indicators, such as economy, resources, and environment. The difference between economic appeals and political appeals intensifies the rules and regulations of local governments on corporate social responsibility. Consequently, in predatory government, firms confront such a tough time that they must make targeted investments that help themselves differentiate from their competitors, especially for firms that have taken CSP as a key part. Firms neglect some areas associated with CSP during economic downturns, resulting in increased concerns about community and employee relations, product safety/quality, and the environment [13], as they should be more attractive as producers, attract highly skilled workers to top companies on the best employer list, or reduce or internalize the company’s social issues [34]. This is a virtuous circle for the firms and the predatory government, especially for companies with high employment influence.

It is proposed that:

H2A. 

In predatory governments, corporate economic influence on GDP ($GDPINF$) is significantly negatively related to corporate social responsibility, and $GDPINF$ is negatively related to various dimensions of corporate social responsibility.

H2B. 

In predatory government, the corporate economic influence on employment ($EMINF$) is positively related to the social responsibility, and $EMINF$ is positively related to government, employee, environment, and partner responsibilities.

In collusive governments, on the premise that there are no major security incidents, local governments are willing to form a short-term equilibrium with the politically affiliated enterprises and the leading enterprises with strong financial influence because they can protect the interests of the government. However, once major security incidents happen, the local government may be forced by higher-level governments to strengthen the regulations on politically affiliated enterprises and local leading enterprises. The best example is Zijin Mining. It was punished after it caused leakage of wastewater containing sodium cyanide and toxic sewage containing metal and was found to hide the truth.

H3A. 

In a collusive government, $GDPINF$ is significantly negatively correlated with corporate social responsibility and negatively correlated with various dimensions of corporate social responsibility, except for partner and community.

H3B. 

In collusive government, $EMINF$ is negatively correlated with corporate social responsibility and various dimensions of social responsibility.

As a market-oriented government, due to a highly developed local economy, the promotion of government officials depends less on the local economy but more on the environment, social welfare, legal system, and so on. Therefore, local governments can restrain their own behaviors in a better way and seldom interfere with enterprises. Corporate behaviors are more affected by the company’s own conditions and market forces. Corporate CSP decisions are more driven by motives of marketing and strategies [35]. Companies should not only address the mechanism for communicating CSR to process stakeholders and customers but also focus on the green strategy and sustainability [35,36,37,38,39]. The following hypotheses have been proposed:

H4A. 

In market-leading governments, $GDPINF$ is not related to corporate social responsibility, and the economic influence on $GDPINF$ is negatively related to the government and public, and shareholders.

H4B. 

In the market-leading government, $EMINF$ is not related to corporate social responsibility and not related to social responsibility dimensions, such as environment, employee, and customers.

3. Data Collecting and Research Design

3.1. Declaration of Research Sample

This paper selects the financial data of the listed companies in Shanghai and Shenzhen from 2004 to 2016, and the panel data of the local economy of the regions where the listed companies are registered for study. To ensure the comparability of indicators, this study includes the level of regional economic development, financial revenue, and other indicators in prefecture-level city statistics. Among them, Beijing, Tianjin, Shanghai, and Chongqing, as well as 4 municipalities directly under the central government and 14 sub-provincial cities, are further studied. Each district is defined as a statistical unit for data collection. Data in this study mainly comes from the China Urban Development Yearbook, and the supplementary data from the China Statistics Network and Statistical Communique on National Economic and Social Development published by the regional statistics bureau. The Chinese Listed Companies Social Responsibility Evaluation Database (2004–2016) is based on the analysis of annual reports and social responsibilities reports (Sustainable Development Report) of the listed companies published by the Shanghai and Shenzhen Stock Exchange and then calculated using the method introduced by [40]. Corporate financial data of the listed companies from 2004 to 2016 are collected from the database of CSRMA; the missing data are supplemented by the Sina network. The following information has been excluded in this study: First, the banking, financial, and insurance listed companies are not included; second, GEM, Shenzhen B shares, and Shanghai B-share listed companies are not included; third, companies listed after 2004 and those with missing values are removed [41] to meet the measurement requirements of the panel threshold model. To overcome the effect of outliers, the main variables are Winsorized 1% contractions.

3.2. Panel Threshold Model of CSR Based on Government Economic Times

As mentioned above, the heterogeneous resources of governments have multiple effects on CSR regulation, but they are not simple linear relationships. Therefore, under different conditions of government heterogeneity, the extent and direction of the government regulation influence on CSR may change significantly; the threshold of CSR regulation is not unique. Considering the differences in resource endowments among different regions, in the government’s resources and government’s regulations in China, CSR will be affected by many “threshold” restrictions. Therefore, this study draws on threshold panel theory [41] and establishes the Panel Threshold Model of the impact of corporate economic influence and government heterogeneity on CSR. To avoid endogeneity problems among variables, this paper uses the lag of enterprise economic influence and government resource endowment for regression analysis.

This study will control the nature of the owners of the enterprise, political relations, and regional corruption to build the threshold regression models, as shown in Equations (1) and (2).

$$CS{R}_{it}={\mu }_i+{\alpha }_1^{\prime }GDPIN{F}_{it-1}·I(q_{it-1}\le {\gamma }_1)+{\alpha }_2^{\prime }GDPIN{F}_{it-1}·I(q_{it-1}>{\gamma }_1)+{\theta }_1^{\prime }{h}_{1it-1}+{\theta }_2^{\prime }{h}_2{}_{it-1}+{\epsilon }_{it},$$

(1)

$$CS{R}_{it}={\mu }_i+{\beta }_1^{\prime }EMIN{F}_{it-1}·I(q_{it-1}\le {\gamma }_1)+{\beta }_2^{\prime }EMIN{F}_{it-1}·I(q_{it-1}>{\gamma }_1)+{\theta }_1^{\prime }{h}_{1it-1}+{\theta }_2^{\prime }{h}_2{}_{it-1}+{\epsilon }_{it},$$

(2)

where $CS{R}_{it}$ index is the degree of CSR fulfillment, $PERGD{P}_{it-1}$ and $GOVE{R}_{it-1}$, on behalf of the region, were the previous economic development and government’s governance capacity, respectively, and they are the two panel variables in Equations (1) and (2). $GDPIN{F}_{it-1}$ is the ration of i enterprise’s prime operating revenue and the regional GDP at year t − 1, which refers to the economic influence variable affected by the threshold variable in the model. $EMIN{F}_{it-1}$ is the ration of i enterprise’s employer number and total number of employers within the whole region. The $h_{1it-1}$ $h_{2it}$ index is the variable that is not affected by the threshold variable. Among them, the $h_{1it-1}$ index is a lagged variable affecting CSR, $h_{1it-1}=(ANT{I}_{it-1},FCL{G}_{it-1},GOVE{R}_{it-1},PERGD{P}_{it-1})$. The $h_{2it}$ index is a controlled variable affecting CSR and $h_{2it-1}=(SIZ{E}_{i,t-1},RO{E}_{i,t-1},LE{V}_{i,t-1},AUFITO{R}_{i,t-1},UC{T}_{i,t-1},P{C}_{i,t-1})$. ${\mu }_i$ reflects the individual effect, and ${\epsilon }_{it}$ is a random disturbance term.

The variables are described in

Table 1. Economic Influence Variable and Variable Definition.

Variable Names	Variable Symbol	Variable Definition
Independent variable
GDP influence	GDPINF	GDPINF = prime operating revenue/regional GDP
Employment influence	EMINF	The regional employment rate = employers’ number of one enterprise/ total number of employers within the whole region
Dependent variable
Corporate social responsibility	CSR	CSR is based on analysis of annual reports and social responsibilities reports (Sustainable Development Report) of the listed companies published by Shanghai and Shenzhen Stock Exchange and then calculated with the method introduced by Wiseman. (1982).
Threshold variable
	PERGDP	Local per capita GDP
	GOVER	Government governance = The regional unemployment rate-province unemployment rate
Control variable
Firm size	SIZE	Ln (total assets)
Return on equity	ROE
Debt asset ratio	LEV
Auditor	AUDITOR	Dummy variable, whether the biggest four auditor firm, 0 means no; 1 means yes.
Per share earnings ratio	EPS
Ownership of enterprises	UCT	Ownership of enterprises, 1 means State-owned enterprises, 0 means no-state owned enterprises.
Political connection	PC	Whether the enterprise employs party and government officials, military leaders, or representatives of the CPPCC, 0 means no political connection;1 means political connection
The level of regional corruption 1	ANTI_1	Regional infrastructure expenditures/regional fix assets investment
Self-financing capacity	FCLG	Budget revenue/budget outlays
Economy developing level	PERGDP	Per capital gross regional product

in detail.

3.3. Government Types and CSRP Selection Based on Panel Threshold

In the first stage, Equation (1) is used to conduct threshold regressions based on per capita GDP, and Equation (2) is adopted to conduct threshold regressions based on the ability to govern unemployment in the region. Next, government behaviors are classified into predatory governments, collusive governments, and market-oriented governments. Finally, based on the known types of government, Equations (3) and (4) are adopted to discuss the relationship between the economic influence of enterprises and the various dimensions of CSR.

$$CS{R}_{it}={\beta }_0+({\theta }_{1}GDPIN{F}_{it-1}+{\theta }_{2}EMIN{F}_{it-1}+{\theta }_{3}contro{l}_{it-1})·gtype+{\epsilon }_{it},if gtype=1,2,3$$

(3)

$$CS{R}_{it}{}_j={\beta }_0+({\theta }_{1}GDPIN{F}_{it-1}+{\theta }_{2}EMIN{F}_{it-1}+{\theta }_{3}contro{l}_{it-1})·gtype+{\epsilon }_{it},if gtype=1,2,3$$

(4)

In Equation (3), the $CS{R}_{it}$ index means the CSR score of enterprise i in year t. In Equation (4), the $CS{R}_{itj}$ index refers to the CSR score of enterprise i in year t at dimension j, which are $GOVERR$, $EMPLR$, $SHARER$, $ENVIRR$, $CUSTR$, $PARTER$, and $COMMR$. The data come from the ratings of the listed companies based on the annual reports from 2004–2016, social responsibility reports, and related websites. $IN{F}_{jit-1}$ $GDPIN{F}_{it-1}$ is the corporate economic influence j of enterprise $i$ at year $t-1$; $EMIN{F}_{it-1}$ is the corporate employment influence j of enterprise $i$ at year $t-1$; $control$ stands for controlled variables, including the size of enterprise ($SIZE$), return on equity ($ROE$), enterprise asset-liability ratio ($LEV$), the big four audit firms involvement ($AUDITOR$), ownership of enterprise ($UCT$), and political connection involvement ($PC$). Detailed variables are explained in Table 1.

4. Corporate Economic Influence, Government Heterogeneity Ability Threshold Test

4.1. Descriptive Statistical Analysis

As can be seen from the descriptive statistics of the variables in

Table 2. Descriptive Statistical Analysis.

	Statistical	CSR	GDPINF	EMINF	ANTI_1	PERPGD	GOVER	FCLG	ROE	SIZE	LEV	AUDITOR	EPS	PC
	Num	4748	4732	4748	4739	4735	4742	4716	4748	4748	4748	4748	3752	4748
Non-state-owned enterprises	Mean	0.23	0.03	0.01	0.27	6.91	−0.01	0.78	0.05	21.53	0.52	0.04	0.28	0.51
	Med	0.21	0.01	0.00	0.22	5.40	−0.21	0.77	0.07	21.49	0.51	0.00	0.17	1.00
	Sd	0.11	0.06	0.02	0.17	5.58	7.76	0.38	0.22	1.31	0.26	0.20	0.46	0.50
	Min	0.03	0.00	0.00	0.02	0.71	−3.00	0.19	−1.09	18.88	0.06	0.00	−1.08	0.00
	Max	0.73	0.51	0.23	0.80	30.72	273.63	2.43	0.67	25.75	1.40	1.00	2.25	1.00
State-owned enterprises	Num	8889	8879	8889	8814	8869	8877	8849	8885	8887	8887	8888	6730	8889
	Mean	0.26	0.05	0.02	0.26	7.18	−0.27	0.77	0.05	22.11	0.53	0.09	0.29	0.56
	Med	0.22	0.02	0.01	0.21	5.48	−0.18	0.74	0.06	21.98	0.54	0.00	0.20	1.00
	Sd	0.12	0.09	0.04	0.18	5.99	0.63	0.36	0.19	1.32	0.21	0.28	0.50	0.50
	Min	0.00	0.00	0.00	0.02	0.71	−3.00	0.19	−1.09	18.88	0.06	0.00	−1.08	0.00
	Max	0.80	0.51	0.23	0.80	30.72	8.70	2.43	0.67	25.75	1.40	1.00	2.25	1.00
Total	Num	13,637	13,611	13,637	13,553	13,604	13,619	13,565	13,633	13,635	13,635	13,636	10,482	13,637
	Mean	0.25	0.04	0.02	0.26	7.09	−0.18	0.77	0.05	21.91	0.53	0.07	0.29	0.54
	Med	0.22	0.01	0.00	0.21	5.45	−0.20	0.75	0.06	21.82	0.53	0.00	0.19	1.00
	Sd	0.12	0.09	0.03	0.18	5.85	4.61	0.37	0.20	1.35	0.23	0.26	0.49	0.50
	Min	0.00	0.00	0.00	0.02	0.71	−3.00	0.19	−1.09	18.88	0.06	0.00	−1.08	0.00
	Max	0.80	0.51	0.23	0.80	30.72	273.63	2.43	0.67	25.75	1.40	1.00	2.25	1.00

, the social responsibility index of SOEs is 0.26, slightly higher than the average social responsibility index of 0.23 of the non-SOEs. The economic influence of SOE $GDPINF$, employment influence $EMINF$, and enterprise scale $SIZE$ are significantly higher than non-SOEs.

The correlation analysis of the variables in

Table 3. Correlation Coefficient.

	CSR	GDPINF	EMINF	ANTI	PERGDP	GOVER	FCLG	ROE	SIZE	LEV	AUDITOR	EPS	UCT	PC
CSR	1.000
GDPINF	0.204 ***	1.000
EMINF	0.210 ***	0.533 ***	1.000
ANTI	−0.006	0.123 ***	0.086 ***	1.000
PERGDP	0.226 ***	0.015 *	0.016 *	0.158 ***	1.000
GOVER	−0.027 ***	−0.003	−0.009	0.065 ***	−0.035 ***	1.000
FCLG	0.044 ***	0.112 ***	0.041 ***	0.261 ***	0.376 ***	−0.028 ***	1.000
ROE	0.099 ***	0.071 ***	0.026 ***	0.025 ***	0.059 ***	−0.023 ***	0.058 ***	1.000
SIZE	0.456 ***	0.474 ***	0.385 ***	−0.004	0.283 ***	0.003	0.075 ***	0.117 ***	1.000
LEV	−0.066 ***	0.148 ***	0.068 ***	−0.005	−0.004	0.008	−0.024 ***	−0.116 ***	0.138 ***	1.000
AUDITOR	0.142 ***	0.208 ***	0.190 ***	0.017 **	0.128 ***	0.079 ***	0.053 ***	0.058 ***	0.311 ***	−0.012	1.000
EPS	0.187 ***	0.191 ***	0.118 ***	0.016	0.080 ***	−0.020 **	0.042 ***	0.529 ***	0.304 ***	−0.181 ***	0.159 ***	1.000
UCT	0.106 ***	0.141 ***	0.116 ***	−0.021 **	0.022 ***	−0.027 ***	−0.020 **	−0.015 *	0.207 ***	0.023 ***	0.080 ***	0.004	1.000
PC	0.035 ***	0.061 ***	0.045 ***	−0.007	−0.027 ***	−0.007	−0.041 ***	−0.002	0.113 ***	−0.009	0.033 ***	0.035 ***	0.051 ***	1.000

Notes: ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively.

shows that the enterprises from the corporate GDP influence $GDPINF$ and employment influence $EMINF$ are significantly related to corporate social responsibility. The correlation coefficient between variables, the correlation coefficient only among $GDPINF$ and $EMINF$ is more than 0.5. Based on the correlation coefficient, this paper further extends Newy’s multicollinearity estimation. The estimation results show that the average VIF (variance inflation factor, hereafter) value is 1.32, and the single VIF value is 1.76, so there is no multi-collinearity among the variables.

4.2. Threshold Regression of the Level of Economic Development, Self-Financing Capacity, and Governance of Employment

4.2.1. Threshold Estimated Value

In order to accurately examine the threshold effect of government heterogeneity on the economic influence of enterprises and to avoid multiple collinearity in the threshold regression, this paper first uses Equations (1) and (2) as the threshold variables, including per capita GDP ($PERGD{P}_{it-1}$) and government employment governance capacity ($GOVE{R}_{it-1}$), fitting the inner relationship between Corporate Economic Influence ($GDPINF$, $EMINF$) and Corporate Social Responsibility ($CSR$). According to the model of [41], the stata15.0 calculation tool was used to perform repeated sampling 300 times by the Bootstrap method to estimate and test whether the model has the thresholds, single threshold, double threshold, and triple threshold, and the corresponding p-value and F statistics. The related test results are shown in

Table 4. Threshold Value.

Threshold Variable	Threshold Number	Threshold Value	95%		Marginal Values Generated by the Bootstrapping Method
			Confidence Interval	F Value	1%	5%	10%
PERGDP	single threshold	3.872	[3.685–4.154]	50.759 ***	33.966	26.637	23.046
	Double threshold	1.359	[1.281–15.660]	13.932 ***	2.020	7.535	12.404
		4.154	[3.578–4.419]
	Triple threshold	10.307	[8.366–13.885]	11.722 ***	29.605	52.091	60.100
GOVER	single threshold	0.210	[0.190–0.250]	30.098 ***	9.944	5.704	4.406
	Double threshold	−0.710	[−2.030–−0.270]	34.702 ***	5.777	14.031	17.960
		−0.380	[−0.210–0.020]
	Triple threshold	−0.190		3.455	5.544	0.469	2.244

Notes: 1. Based on the study of Hansen, repeated sampling was carried out 300 times using the bootstrap method to estimate both the p-value and critical value [41]. 2. *** denote statistical significance at the 1%.

.

The statistics of LR (Likelihood Ration Statistic, hereafter) in Table 4 show that under the single threshold, two threshold, and three threshold models of the level of regional economic development $PERGDP$ and the governance capacity of employment, the p values all pass the 1% hypothesis test. This indicates that the level of regional economic development has a triple threshold effect. According to the results of the threshold effect test, the level of regional economic development and local government employment governance capacity have non-linear threshold characteristics on the choice of corporate economic social responsibility behavior.

4.2.2. Threshold Regression Model

In order to further examine how the level of regional economic development influences the relationship between GDPINF and CSR, this study uses threshold panel regression analysis. The results are shown in

Table 5. The Result of Corporate GDP Influence Threshold Regression.

	Single	Double	Triple
GDPINF_1	−0.0613 **	−0.432 ***	−0.408 ***
	(−2.19)	(−3.77)	(−4.94)
GDPINF_2	0.0794 ***	−0.0420	0
	(3.90)	(−1.58)	(.)
GDPINF_3		0.0856 ***	0.0164
		(4.16)	(0.92)
GDPINF_4			0.141 ***
			(5.50)
ANTI	0.0014	0.0012	−0.0049
	(0.19)	(0.17)	(−0.66)
GOVER	−0.0003	−0.0002	−0.0005 **
	(−0.71)	(−0.68)	(−2.30)
FCLG	−0.0011	−0.0017	0.0011
	(−0.32)	(−0.46)	(0.29)
ROE	0.0010	0.0012	0.0021
	(0.17)	(0.22)	(0.38)
SIZE	0.0333 ***	0.0331 ***	0.0342 ***
	(29.63)	(29.40)	(31.79)
LEV	−0.0467 ***	−0.0467 ***	−0.0498 ***
	(−8.82)	(−8.82)	(−9.61)
AUDITOR	0.0145 ***	0.0145 ***	0.0182 ***
	(2.61)	(2.61)	(3.33)
EPS	0.0002	0.0004	−0.0054 **
	(0.06)	(0.14)	(−2.11)
UCT	0.0169 ***	0.0171 ***	0.0111 ***
	(5.78)	(5.84)	(3.85)
PC	0.0024	0.0025	0.0014
	(1.12)	(1.14)	(0.68)
Industry, year	yes	yes	yes
Constant	−0.440 ***	−0.435 ***	−0.453 ***
	(−18.19)	(−17.94)	(−19.54)
r2_w	0.155	0.156	0.154
r2_b	0.239	0.236	0.247
r2_o	0.197	0.198	0.196
N	9315	9315	9315

Notes: 1. Our model is Equation (1). 2. The t–values are in parentheses. Significance at the 1%, and 5% levels is denoted by ***, and **, respectively.

and Figure 1. When the local per capita GDP $PERGD{P}_{it-1}\le 3.872$, the coefficient of the marginal impact effect $GDPINF$ of enterprises on CSR is −0.0613 and is significant at the 5% level, which indicates that the greater the impact of enterprises on the local economy, the stronger the incentive of local governments to protect local enterprises and the stronger the voice of enterprises. Therefore, leading enterprises are more likely to be protected by the government in the regulation of social responsibility, thus shouldering less CSR. These results are consistent with the notions of [8,10]. However, when the local per capita GDP $PERGD{P}_{it-1}>3.872$, the marginal impact effect coefficient of $GDPINF$ on CSR is 0.079 and significant at the 1% level. As it shows, when the per capita GDP exceeds CNY 38,720, the local government’s reliance on leading enterprises is reduced, and regional economic growth will no longer be the government’s sole goal. The political promotion of local government officials emphasizes performance projects, such as environmental quality and people’s livelihood. Therefore, local governments may have higher social responsibility expectancy on leading enterprises, such as more donations, and lower pollution. As seen from the double threshold of the corporate GDP influence, when the per capita GDP is higher than the second threshold, that is $PERGD{P}_{it-1}>4.154$, the marginal impact effect coefficient of $GDPINF$ the on corporate social responsibility is improved to 0.0856, significant at the 1% level. This shows that the local economy is highly developed, the economic dependence of local governments on leading enterprises further diminishes, and the positive impact of corporate economic influence on corporate social responsibility is even more significant. Meanwhile, the Chinese government has attached importance to the green economy, which requires a strong transfer of resources and new technologies to reduce pollution emissions and improve energy efficiency and resource productivity. Therefore, a green development policy could promote corporate ESG performance [37].

Table 6. The Results of Corporate Employment Influence Threshold Regression.

	Single	Double	Triple
EMINF_1	0.340 ***	0.0581	−0.0078
	(7.40)	(0.74)	(−0.11)
EMINF_2	0.432 ***	0.569 ***	0
	(6.05)	(8.51)	(.)
EMINF_3		0.355 ***	0.172 ***
		(7.10)	(3.12)
EMINF_4			0.267 ***
			(5.38)
ANTI	−0.0244 ***	−0.0238 ***	−0.0233 ***
	(−3.32)	(−3.25)	(−3.06)
GOVER	0.0027 ***	0.0028 ***	0.0021 ***
	(9.59)	(9.70)	(7.78)
FCLG	−0.0006	−0.0006	−0.0003
	(−0.16)	(−0.16)	(−0.09)
ROE	0.0026	0.0023	0.0035
	(0.46)	(0.42)	(0.62)
SIZE	0.0312 ***	0.0313 ***	0.0312 ***
	(27.30)	(27.44)	(27.22)
LEV	−0.0512 ***	−0.0499 ***	−0.0551 ***
	(−9.38)	(−9.15)	(−10.55)
AUDITOR	0.0361 ***	0.0355 ***	0.0263 ***
	(6.53)	(6.44)	(4.80)
EPS	0.0015	0.0015	−0.0066 **
	(0.55)	(0.56)	(−2.52)
UCT	0.0270 ***	0.0267 ***	0.0205 ***
	(9.48)	(9.40)	(7.04)
PC	0.0021	0.0023	0.0021
	(0.95)	(1.04)	(0.95)
Industry, year	yes	yes	yes
Constant	−0.417 ***	−0.421 ***	−0.401 ***
	(−17.45)	(−17.62)	(−16.66)
r2_w	0.195	0.198	0.159
r2_b	0.214	0.209	0.230
r2_o	0.193	0.193	0.193
N	9319	9319	9319

Notes: 1. Our model is Equation (2). 2. The t–values are in parentheses. Significance at the 1%, and 5% levels is denoted by ***, and **, respectively.

and Figure 2 show the relationship between corporate employment influence on region and corporate social responsibility behavior through threshold panel regression analysis. When the unemployment rate in the region is lower than the provincial level and is controlled within 0.21, the larger the proportion of employed people in the district units, the better the corporate social responsibility behaviors are performed, which is significant at the 1% level. When the unemployment rate in the area is 0.21 percentage higher than the provincial unemployment rate, $EMINF$ the marginal impact effect coefficient on corporate social responsibility is 0.340 and is significant at the 1% level. This indicates that the greater the influence of the employed population on employment and social stability in the region, and the more the social stability of regional employment depends on the leading enterprises, the greater the demand for CSR that local governments may have because of the political appeal for social stability. The double threshold regression further reveals the relationship between the level of local governments’ control of the unemployment rate and the proportion of the employed population in enterprises on CSR.

4.3. Government Type and Choice of Corporate Social Responsibility

4.3.1. Government Type Classification

Regional economic development and employment government capacity bring about changes in government behavior.

Table 7. Government Type Classification.

Government Type	Standard	Describe	Sample
Predatory Government	PERGDP ≤ 1.359; GOVER > 0.210	Low self-sufficiency rate, low unemployment government capacity and low economic development	223
Collusive Government	1.359 < PERGDP ≤ 10.307; −0.710 < GOVER ≤ 0.210	Middle self-sufficiency rate, middle unemployment government capacity and middle economic development	12,703
Market-Leading Government	GOVER ≤ −0.710; PERGDP > 10.307	Higher self-sufficiency rate, higher unemployment government capacity and higher economic development	711

illustrates the regression results of the threshold. Economic development is always the top priority for local governments. Therefore, this paper divides the government into 3 types based on the threshold of GDP per capita and the level of governance of the unemployment rate, as shown in Table 7. When $PERGD{P}_{it-1}>10.307$, and $GOVE{R}_{it-1}\le -0.710$, local governments are more manifested in market-leading government. When $1.359<PERGD{P}_{it-1}\le 10.307$, and $-0.710<GOVE{R}_{it-1}\le 0.210$, local governments are more manifested in collusive government. When $PERGD{P}_{it-1}\le 1.359$, and $GOVE{R}_{it-1}>0.210$, the government behavior in areas or periods with low self-sufficiency rate, low unemployment government capacity, and low economic development is defined as predatory government behavior. As can be seen from Table 7, there are 12,703 companies under the regulation of the collusive government. The government reflects a neutral “collusive government” according to economic conditions [42,43,44,45]. Under this type of government behavior, local governments mainly regulate the access and development of the industry through industrial policies and industrial operation licenses and even directly support and control specific enterprises or industries with the “helping hands”; at the same time, it also suppresses the development of certain enterprises or industries through the “grabbing hands”. Therefore, companies will then mainly collude with the government through various political means, such as providing political contributions, political connections, and corporate donations. Collusion with colluding governments results in mutually beneficial industrial regulations that bring more corporate returns and consider the interests of local governments.

Table 8. Descriptive statistical analysis of the different local governments.

Government Type	Statistical	CSR	Government and Public	Employee	Shareholder	Environment	Customer	Partner	Community	GDPINF	EMINF
Predatory Government	Num	223	223	223	223	223	223	223	223	223	223
	Mean	0.16	0.17	0.17	0.54	0.05	0.11	0.36	0.07	0.06	0.02
	Med	0.15	0.13	0.15	0.56	0	0.07	0.42	0.06	0.02	0.01
	Sd	0.07	0.11	0.10	0.22	0.08	0.14	0.22	0.07	0.09	0.02
	Min	0	0	0	0	0	0	0	0	0	0
	Max	0.43	0.61	0.61	0.89	0.45	0.67	0.75	0.33	0.51	0.23
Collusive Government	Num	12,703	12,703	12,703	12,702	12,703	12,703	12,703	12,703	12,677	12,703
	Mean	0.25	0.28	0.29	0.58	0.12	0.21	0.41	0.11	0.04	0.02
	Med	0.22	0.20	0.28	0.56	0.06	0.14	0.44	0.1	0.01	0
	Sd	0.12	0.20	0.15	0.23	0.16	0.21	0.22	0.11	0.08	0.03
	Min	0	0	0	0	0	0	0	0	0	0
	Max	0.8	1	0.81	1	0.94	1	1.89	1.39	0.51	0.23
Market-Leading Government	Num	711	711	711	711	711	711	711	711	711	711
	Mean	0.27	0.31	0.33	0.56	0.11	0.24	0.43	0.13	0.05	0.02
	Med	0.25	0.27	0.33	0.56	0.07	0.2	0.44	0.11	0.02	0
	Sd	0.1	0.19	0.13	0.21	0.13	0.19	0.23	0.11	0.1	0.04
	Min	0.08	0	0.01	0.04	0	0	0	0	0	0
	Max	0.69	0.93	0.81	1	0.71	0.89	1	0.61	0.51	0.23
Total	Num	13,637	13,637	13,637	13,636	13,637	13,637	13,637	13,637	13,611	13,637
	Mean	0.25	0.28	0.29	0.58	0.12	0.21	0.41	0.11	0.04	0.02
	Med	0.22	0.2	0.28	0.56	0.06	0.17	0.44	0.1	0.01	0
	Sd	0.12	0.2	0.15	0.23	0.15	0.2	0.22	0.11	0.09	0.03
	Min	0	0	0	0	0	0	0	0	0	0
	Max	0.8	1	0.81	1	0.94	1	1.89	1.39	0.51	0.23

shows the descriptive statistics of seven dimensionalities of corporate social performance under different local governments. We can see that the mean of the CSR index reaches the maximum in the market-oriented government. The mean of the CSR index in the collusive government is better than that in the Predatory Government, and the mean is 0.27, 0.25, and 0.16, respectively. Moreover, the mean of the employee and environment are 0.29 and 0.12, respectively, in the collusive government, which are comparable to the 0.228 and 0.133 reported by [46]. However, the shareholders’ responsibility index and the environmental responsibility index are best because local governments can cooperate with companies for social responsibility regulation according to their strategic choices; at the same time, companies can maximize their profits and assume strategic CSR [47].

4.3.2. Government Type and Choice of Corporate Social Responsibility

Given the types of government, enterprises always make corresponding strategic choices based on their own resources. According to Equations (3) and (4), this paper tests the relationship between the economic influence of enterprises and the dimensions of CSR under the different types of government.

Corporate Social Responsibility Choice in Predatory Government

Table 9. Government Types and Choices of CSR.

Panel A	Predatory Government
	CSR	Government and Public	Employee	Shareholder	Environment	Customer	Partner	Community
GDPINF	−0.632 **	−1.678 ***	−1.810 ***	0.637	−0.871 **	0.952 *	−0.332	−0.352
	(−2.161)	(−3.585)	(−3.388)	(0.367)	(−2.407)	(1.954)	(−0.668)	(−0.706)
EMGDP	1.532	7.018 **	5.469	−6.821	3.299	−2.506	0.297	−0.483
	(0.852)	(2.438)	(1.664)	(−0.640)	(1.481)	(−0.837)	(0.097)	(−0.157)
ANTI	0.050	−0.013	−0.129	0.223	0.121	0.117	0.034	−0.074
	(0.490)	(−0.082)	(−0.695)	(0.370)	(0.963)	(0.689)	(0.196)	(−0.428)
PERGDP	0.197	0.117	0.258	0.435	0.308 *	−0.053	0.238	−0.006
	(1.421)	(0.526)	(1.022)	(0.530)	(1.799)	(−0.231)	(1.014)	(−0.024)
GOVER	−0.000	−0.000	−0.001	−0.005	0.001	0.001	0.002	0.001
	(−0.071)	(−0.153)	(−0.604)	(−1.113)	(0.632)	(0.401)	(1.394)	(0.690)
FCLG	−0.113	−0.434	−0.364	−0.792	0.012	−0.018	−0.378	0.433
	(−0.293)	(−0.702)	(−0.515)	(−0.346)	(0.026)	(−0.028)	(−0.575)	(0.657)
ROE	0.007	0.104	0.079	−0.565	0.071	−0.134	0.294 ***	−0.041
	(0.125)	(1.137)	(0.757)	(−1.661)	(1.005)	(−1.399)	(3.019)	(−0.421)
SIZE	−0.000	−0.033	−0.056	0.062	−0.036	0.158	−0.192	0.112
	(−0.004)	(−0.220)	(−0.328)	(0.112)	(−0.313)	(1.006)	(−1.198)	(0.698)
LEV	0.052	0.500	0.537	−1.049	0.044	−0.501	0.484	−0.316
	(0.255)	(1.516)	(1.424)	(−0.858)	(0.171)	(−1.460)	(1.380)	(−0.899)
AUDITOR	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000
	(.)	(.)	(.)	(.)	(.)	(.)	(.)	(.)
EPS	0.033	0.052	0.117 ***	−0.041	0.034	−0.071 *	0.041	0.051
	(1.556)	(1.565)	(3.058)	(−0.330)	(1.297)	(−2.049)	(1.143)	(1.416)
UCT	0.023	0.014	0.043	0.027	0.016	0.005	−0.005	0.030
	(0.624)	(0.239)	(0.641)	(0.124)	(0.344)	(0.085)	(−0.083)	(0.482)
PC	−0.001	0.015	0.046	−0.043	0.012	−0.014	−0.014	−0.038
	(−0.047)	(0.312)	(0.819)	(−0.236)	(0.314)	(−0.268)	(−0.262)	(−0.737)
Industry, year	yes	yes	yes	yes	yes	yes	yes	yes
_cons	0.020	0.633	0.993	−0.161	0.487	−2.952	4.070	−2.197
	(0.010)	(0.209)	(0.288)	(−0.014)	(0.208)	(−0.938)	(1.268)	(−0.681)
N	56	56	56	56	56	56	56	56
R-sq	0.440	0.612	0.555	0.422	0.452	0.617	0.710	0.342
Panel B	Collusive Government
	CSR	Government and public	Employee	Shareholder	Environment	Customer	Partner	Community
GDPINF	−0.084 ***	−0.152 ***	−0.069 *	−0.325 ***	−0.100 **	−0.062	0.010	0.023
	(−2.764)	(−2.940)	(−1.648)	(−3.294)	(−2.374)	(−1.199)	(0.147)	(0.631)
EMGDP	−0.088	−0.194 *	−0.014	−0.083	−0.035	−0.033	−0.137	−0.009
	(−1.418)	(−1.835)	(−0.168)	(−0.412)	(−0.402)	(−0.312)	(−0.977)	(−0.121)
ANTI	0.007	0.021	0.007	−0.054 **	0.010	0.004	0.018	0.025 ***
	(0.832)	(1.546)	(0.635)	(−2.092)	(0.871)	(0.290)	(0.973)	(2.676)
PERGDP	0.005 ***	0.011 ***	0.007 ***	−0.005 ***	0.004 ***	0.011 ***	0.000	0.002 ***
	(14.914)	(17.298)	(14.306)	(−3.844)	(8.346)	(17.486)	(0.498)	(5.071)
GOVER	0.000	0.000	0.000	0.000	0.000	0.000	0.001	0.000
	(1.144)	(0.787)	(0.605)	(0.148)	(0.697)	(0.659)	(0.939)	(0.342)
FCLG	−0.005	−0.007	−0.003	−0.032 *	−0.011	−0.003	0.018	−0.014 **
	(−0.899)	(−0.727)	(−0.420)	(−1.797)	(−1.414)	(−0.293)	(1.405)	(−2.192)
ROE	0.001	−0.001	0.009	−0.016	−0.004	−0.002	0.013	0.006
	(0.168)	(−0.093)	(1.395)	(−1.091)	(−0.629)	(−0.241)	(1.327)	(1.128)
SIZE	0.025 ***	0.048 ***	0.028 ***	−0.013 ***	0.023 ***	0.037 ***	0.012 ***	0.018 ***
	(16.726)	(19.094)	(13.839)	(−2.776)	(11.338)	(14.680)	(3.717)	(10.174)
LEV	−0.041 ***	−0.064 ***	−0.037 ***	0.003	−0.050 ***	−0.053 ***	−0.055 ***	−0.017 **
	(−7.212)	(−6.715)	(−4.720)	(0.139)	(−6.379)	(−5.553)	(−4.315)	(−2.537)
AUDITOR	−0.015 **	−0.007	0.008	−0.014	−0.017 *	−0.014	−0.070 ***	−0.003
	(−2.098)	(−0.564)	(0.818)	(−0.596)	(−1.726)	(−1.103)	(−4.235)	(−0.382)
EPS	−0.004 *	−0.004	−0.001	−0.019 **	0.001	−0.020 ***	−0.009 *	−0.001
	(−1.799)	(−1.046)	(−0.180)	(−2.491)	(0.229)	(−5.110)	(−1.693)	(−0.262)
UCT	−0.015 ***	−0.020 **	−0.020 ***	−0.018	−0.011	−0.024 ***	0.001	−0.007
	(−2.999)	(−2.392)	(−3.024)	(−1.143)	(−1.615)	(−2.922)	(0.072)	(−1.218)
PC	0.006 ***	0.006 *	0.003	−0.000	0.005	0.011 ***	0.003	0.009 ***
	(2.976)	(1.735)	(1.106)	(−0.013)	(1.621)	(3.170)	(0.602)	(3.818)
Industry, year	yes	yes	yes	yes	yes	yes	yes	yes
_cons	−0.265 ***	−0.739 ***	−0.297 ***	0.982 ***	−0.342 ***	−0.595 ***	0.168 **	−0.265 ***
	(−8.429)	(−13.868)	(−6.874)	(9.631)	(−7.882)	(−11.124)	(2.374)	(−7.114)
N	8623	8623	8623	8622	8623	8623	8623	8623
R-sq	0.131	0.162	0.105	0.013	0.058	0.133	0.009	0.039
Panel C	Market-Leading Government
	CSR	Government and public	Employee	Shareholder	Environment	Customer	Partner	Community
GDPINF	−0.253	−0.739 *	−0.153	−1.580 **	−0.479	0.181	−0.069	0.290
	(−1.065)	(−1.786)	(−0.450)	(−2.193)	(−1.457)	(0.475)	(−0.115)	(0.981)
EMGDP	−0.338	−0.520	−0.318	−0.226	−0.067	−0.702 *	−0.320	−0.004
	(−1.491)	(−1.321)	(−0.980)	(−0.330)	(−0.214)	(−1.936)	(−0.560)	(−0.013)
ANTI	0.096	0.096	0.098	0.392 **	0.123	−0.020	0.301 *	0.023
	(1.553)	(0.895)	(1.111)	(2.105)	(1.454)	(−0.198)	(1.937)	(0.297)
PERGDP	0.001	0.004 **	0.002	−0.007 *	0.002	0.001	−0.002	0.002
	(0.620)	(2.120)	(1.364)	(−1.934)	(1.005)	(0.591)	(−0.783)	(1.131)
GOVER	−0.016	−0.008	0.020	−0.002	−0.000	−0.049 ***	−0.025	−0.014
	(−1.597)	(−0.481)	(1.380)	(−0.077)	(−0.029)	(−3.046)	(−0.998)	(−1.120)
FCLG	0.011	−0.006	−0.035	0.045	−0.005	0.023	−0.029	0.037
	(0.626)	(−0.202)	(−1.349)	(0.822)	(−0.198)	(0.798)	(−0.626)	(1.633)
ROE	−0.043	−0.112	−0.053	0.071	−0.025	−0.107	0.394 ***	−0.126 **
	(−0.997)	(−1.505)	(−0.864)	(0.544)	(−0.424)	(−1.556)	(3.638)	(−2.359)
SIZE	0.007	0.069 ***	0.012	0.018	0.010	0.024	−0.075 ***	−0.007
	(0.716)	(3.828)	(0.832)	(0.559)	(0.691)	(1.423)	(−2.859)	(−0.504)
LEV	0.093 *	−0.042	0.095	0.075	0.037	0.020	0.316 **	0.050
	(1.892)	(−0.488)	(1.345)	(0.501)	(0.547)	(0.251)	(2.541)	(0.809)
AUDITOR	−0.031	−0.060	−0.003	0.028	−0.056	−0.006	−0.064	0.021
	(−0.937)	(−1.036)	(−0.064)	(0.283)	(−1.230)	(−0.111)	(−0.765)	(0.503)
EPS	−0.003	0.004	0.007	−0.062	0.008	−0.024	−0.017	0.021
	(−0.212)	(0.148)	(0.312)	(−1.405)	(0.399)	(−1.024)	(−0.474)	(1.147)
UCT	−0.018	0.135 **	−0.051	0.004	0.004	−0.124 **	−0.122	0.029
	(−0.558)	(2.385)	(−1.098)	(0.040)	(0.081)	(−2.378)	(−1.481)	(0.706)
PC	0.008	0.035 **	0.009	−0.026	0.015	−0.015	0.028	0.006
	(0.831)	(2.154)	(0.705)	(−0.923)	(1.197)	(−1.019)	(1.184)	(0.534)
Industry, year	yes	yes	yes	yes	yes	yes	yes	yes
_cons	0.025	−1.319 ***	0.052	0.155	−0.153	−0.287	1.938 ***	0.136
	(0.117)	(−3.560)	(0.172)	(0.241)	(−0.520)	(−0.841)	(3.605)	(0.513)
N	620	620	620	620	620	620	620	620
R-sq	0.066	0.156	0.040	0.048	0.034	0.088	0.109	0.049

Notes: Significance at the 1%, 5%, and 10% levels is denoted by ***, **, and *, respectively.

Panel A presents how the GDPINF and the EMINF affect the choice of CSR in predatory governments. For the predatory government, the coefficient of impact of $GDPINF$ on corporate social responsibility is −0.632 with a significance of 5%. In other words, the greater the influence of the main business of a company on the local economy, the collusive between the local government and the leading enterprises, the reduction of the requirements of corporate social responsibility regulation, or the decrease in the frequency of supervision and the increase of the possibility of reducing the penalties. This is consistent with [8]. Therefore, the worse the corporate performance of social responsibility is. This result verifies hypothesis H4A. $GDPINF$ only has a non-significant positive correlation with shareholder responsibility and consumer responsibility, but a negative correlation with the other dimensions of social responsibility, especially a significant negative correlation between government responsibility, employee responsibility, and environmental responsibility. However, the employment influence $EMINF$ of the enterprises on the working population in the region has a positive correlation with the government and the public, as well as employee responsibility and environmental responsibility dimensions, while there is an insignificantly negative correlation with the responsibility of the shareholders, the customer, partners, and the community responsibility. One possible explanation is the influence of the political red line that “social stability is overwhelming”. Local government officials who transform into predatory governments with poor social governance (such as poor economic development and employment) may abandon economic competition and join the “convergence club”. The goal pursued by these officials in their political pursuit is to maintain social stability in the region and, therefore, to emphasize the responsibility of employees, consumers, or others who have a significant impact on employment in the region. In the predatory government, the government may exhibit more predation of the enterprise value. The greater influence the enterprise exerts on employment and stability in the area, the more employees or governance is needed to gain public responsibility.

Choices of Corporate Social Responsibility in Collusive Government

Panel B of Table 9 presents the results of the relationship between corporate economic influence and the CSR choices of the collusive government sub-sample. In the collusive government, officials not only pursue economic goals but also maintain social stability and the public interest, so officials need to frequently seek a balance between economic interests and public interests and between long-term interests and short-term interests. The pursuit of economic development would often outpace the pursuit of public interest and CSR requirements. Therefore, the results show that the economic influence of enterprises $EMINF$ has not had a significantly negative correlation with CSR; meanwhile, $GDPINF$ shows a significantly negative correlation with corporate social responsibility. In the collusive government, the marginal impact of $GDPINF$ on corporate social responsibility is −0.084, and is negatively correlated at 1%; and $GDPINF$ is in a significant negative correlation with government responsibility, employee responsibility, environmental responsibility, and shareholder responsibility and is not significantly positively correlated with partner responsibility and community responsibility. However, the relationship between the $EMINF$ and CSR, the government, shareholders, consumers, and so on are not in a significantly negative relationship, showing a completely opposite relationship with the predatory government. This indicates in another way that the concessions of various rules of social responsibility by the government are based on rational games of local government interests.

Choice of Corporate Social Responsibility in Market-Leading Government

Panel C of Table 9 shows the relationship between corporate economic influence and the CSR choice of market-oriented governments. In market-oriented governments, “economic growth—political promotion” reduces the marginal benefit of the political promotion of local officials. The development of a regional economy is no longer the only goal pursued by officials. Instead, officials began to pursue various goals of environmental and social development in order to have a highlight in political promotion. As a result, the marginal impact of $GDPINF$ on CSR is decreasing. In the predatory government, the coefficient was −0.632 at the 5% level. In the collusive government, the coefficient was −0.084 at the 1% level and the coefficient was −0.253 in the market-oriented government. On the contrary, the government may require enterprises to shoulder more social responsibility because the enterprises have a larger scale and greater social influence. For the market-oriented government, corporate profit depends on product quality and a good reputation, so the companies must meet the demands of the market and pay more attention to customer responsibility and community responsibility strategically.

The $EMINF$ is always negatively correlated with corporate social responsibility (CSR) but not significantly in market-oriented government. The economic influence of enterprises on regional employment $EMINF$ is in a positive correlation with government responsibility, and the coefficient was 7.018 in predatory government. The coefficient was −0.194 at the 10% level in collusive government; moreover, the coefficient was −0.739 at the 10% level in market-oriented government [48]. A larger gap between meeting moral needs and ignoring direct and economic needs may cause negative effects. Different from the motives of the predatory government, the motives of the market-oriented government are demonstrated in the improvement of market-oriented government services and the pursuit of public welfare.

5. Discussion

This paper addresses the limited research on the influence of governments on the regulation levels of leading enterprises regarding CSR. We explore the economic factors that shape CSR behavior across different types of governments. In contrast to previous studies of references [1,2,4], we make two key assumptions: (1) the interests of the government and leading enterprises do not always align, and (2) the government must respond to the influence exerted by enterprises. Leading enterprises can exert influence over local governments to reduce or relax CSR regulations. However, the marginal utility of regional governments’ dependency on economic influence diminishes over time. In fact, the interests of local governments evolve in response to changes in government capacity. When local economic development reaches a certain threshold, the total regional output generated by enterprises no longer holds a significant sway in discussions with governments. On the contrary, local governments may demand that leading enterprises take on greater CSR responsibilities. This study contributes to the existing literature by proposing the existence of “thresholds” in the relationship between regional economic growth and CSR regulations, as indicated by previous studies [21,22,24,25].

Furthermore, we demonstrate that the link between corporate influence through GDP and CSR varies across different government types, as evidenced by the regression coefficients in subsample analyses. We find a significant negative correlation between predatory and collusive governments, while no significant negative correlation is observed in intensified market-oriented governments. This suggests that leading enterprises should prioritize government and public responsibilities in predatory and collusive governments, while the opposite holds true in market-oriented governments. These results align with previous research [8] that highlights a negative relationship between local state capture (corruption) and CSR.

Lastly, by utilizing indicators such as GDPINF and EMINF, this paper dynamically illustrates the impact of economic influence on CSR behavior under different government actions. We showcase how changes in local government types reverberate in local CSR decision-making.

6. Conclusions

This paper utilizes data from Chinese firms spanning the period of 2004 to 2016 to examine the relationship between CSR and government types using the panel threshold model. The study begins by considering the autonomy of the government and regional economic development, aiming to identify the critical point at which government behavior undergoes a qualitative change. Furthermore, this paper explores the influence of the economic power of enterprises on CSR behavior. Based on the qualitative change point obtained through threshold regression, the governments in the study are classified into three types: market-oriented government, collusive government, and predatory government. The research findings reveal that different government types exhibit distinct strategies concerning CSR behavior in companies.

The paper emphasizes that regardless of whether it is a predatory government, collusive government, or market-oriented government, local economic development plays a significant role in government regulation and decision-making. As leading enterprises have a greater impact on the GDP, it becomes more likely for local governments to relax social responsibility regulations, resulting in a decline in CSR. From the perspective of the influence of enterprises on local economic growth, the study demonstrates that as the local economy develops, the relationship between “economic growth-political promotion” becomes less pronounced. Consequently, the marginal utility of enterprises’ economic influence in reducing CSR diminishes. In the case of a predatory government, there is a positive correlation between employment, regional employment, and social stability due to the government’s adherence to the ideology of maintaining stability. In a collusive government, stricter CSR regulations are necessary for political advancement. Conversely, in a market-oriented government, a similar phenomenon arises as the government pursues public welfare. Overall, the paper highlights the diverse approaches adopted by governments to influence CSR behavior in companies, with local economic development being a crucial consideration in government decision-making.

This study has some managerial and theoretical implications. First, different types of government can influence CSP as government goals change. Second, the communicative relationship between the government and leading companies is a major external motivator for CSR performance in China. When investors, creditors, and regulatory makers understand the motivations behind CSR performance, this can enable them to create better corporate valuations and effective CSR strategies in transitional countries. However, this study is subject to several limitations. First, due to data limitations, this paper only analyzes and explores the status quo of Chinese firms. This study fails to examine the impact of different corporate influences in the same country, nor does it examine the impact of different regimes and types of power in different countries. Future research should explore the relationship between the other corporate influences (such as political influence, industrial association influence, etc.) and CSR performance in other countries, as well as at a global level. Second, a renewed search should use dynamic panel threshold models to examine the ongoing impact of government CSR policies.