Corporate Charitable Donations, Earnings Performance and Tax Avoidance

Abstract

Under the traditional research framework of corporate social responsibility and tax avoidance, there is no agreement on whether charitable donations constitutes an altruistic behavior or a management tool. Using a sample of Chinese firms, this paper examines the relationship between corporate charitable donations, earnings performance and tax avoidance. The evidence shows that there is a significant negative relationship between corporate charitable donations and tax avoidance. Furthermore, we found that the negative relationship between charitable donations and tax avoidance only exists in enterprises with a good earnings performance, while it is positively correlated with tax avoidance in enterprises with a poor earnings performance. This shows that earnings performance can affect the motivation for corporate charitable donations, as the charitable donations of enterprises with a good performance are mainly an altruistic behavior, while the charitable donations of enterprises with a poor performance are more of a management tool. This conclusion not only enriches and expands the research framework of corporate social responsibility and tax avoidance but also helps to clarify the disputes in the existing literature.

1. Introduction

Corporate social responsibility integrates economic, legal, moral and charitable responsibilities into the business decision-making process of enterprises [1]. It advocates the notion that enterprises should pay attention to the interests of other stakeholders, except for shareholders. As one of the important stakeholders of enterprises, the government participates in the profit distribution of enterprises through taxation. For enterprises, although tax avoidance reduces their tax costs, increases their profits and enhances their value, it hinders the government’s ability to provide public services and has severe negative impacts on society. Thus, it is considered to be a socially irresponsible action [2]. Therefore, enterprises with social responsibility should pay taxes fairly and avoid taxes as little as possible. However, in reality, many enterprises engage in corporate social responsibility activities, such as charitable donations, with great publicity, while they are also quietly carrying out radical tax avoidance. For example, Apple (Cupertino, CA, USA), Google (Mountain View, CA, USA), Amazon (Seattle, WA, USA) and other well-known international enterprises actively carry out offshore tax avoidance activities while fostering a good image of actively fulfilling corporate social responsibility [3]. This contradictory behavior has not only aroused widespread doubts from the media and the public but also attracted strong research interest among scholars. However, the existing research shows mixed results. Based on the traditional shareholder perspective and agency theory, it has been found that there is a substitute relationship between corporate social responsibility and tax payment, that is, corporate social responsibility is significantly positively correlated to tax avoidance [4,5]. Nevertheless, research based on stakeholder perspectives and corporate culture theory has highlighted that corporate social responsibility and tax payment are complementary, and corporate social responsibility and tax avoidance are significantly negatively correlated [6,7].

The conflicting results are explained by three key reasons. First, the research framework of corporate social responsibility and tax avoidance was developed late, and most of the research has focused on samples from a few developed countries, lacking attention to developing countries, especially the relationship between corporate social responsibility and tax avoidance in the context of China, resulting in conclusions that may be affected by sample selection bias [8]. Second, previous research has not reached a general agreement about the most appropriate measure of corporate tax avoidance and has used different proxies for aggressive tax practices [8,9]. Third, most previous studies have used an aggregated measure of corporate social responsibility (CSR). However, an aggregated measure may confound the effects of the individual CSR dimensions, which are not equally important or relevant [10]. Therefore, more research is needed in the future to further investigate the relationship between specific corporate social responsibility activities and tax avoidance [9], and the institutional background of developing countries should be taken into account, especially the relationship between business ethics and corporate social responsibility in the Chinese context [11]. In addition, in view of the differences between different corporate social responsibility activities, it is also necessary to separately examine the relationships between certain dimensions of corporate social responsibility and tax avoidance [12].

In view of this, using the traditional research framework of corporate social responsibility and tax avoidance, this paper uses the data of Chinese listed enterprises to examine the relationship between corporate charitable donations and fair taxation. There are two reasons why we chose China as our research context. First, as the largest emerging economy in the world, China’s business ethics and social responsibility are increasingly a concern of the international community [11]. Second, the evolution of CSR remains in the preliminary stage in China [13], and there are substantial differences in the amount of information on specific CSR activities disclosed by Chinese firms that enable us to determine whether a CSR report reflects a firm’s symbolic exercises or is indicative of substantive activities. Using the Chinese context to conduct research can not only reconcile the controversies of existing research but also has important reference value for the question of how emerging market countries such as China, where corporate tax avoidance is becoming increasingly problematic, supervise enterprises in regard to fair taxation and maintain social equity [11].

As an essential realization of corporate social responsibility, the relationship between charitable donations and tax avoidance is, in theory, unclear. On the one hand, the management tool hypothesis holds that charitable donations are a strategic tool for enterprises to accumulate reputation capital, seek legitimacy and establish a relationship between the government and themselves [5]. Therefore, enterprises may use charitable donations to divert the attention of the government, the public and the media from their tax avoidance activities so as to reduce the risk of exposure to tax avoidance activities and the risk of political, regulatory and social sanctions after exposure. In the emerging and transitional China, because the legal system, market mechanism, corporate governance and other formal systems are not perfect, the weakness of the system may lead to more misconduct or irresponsible behavior, and enterprises’ use of charitable donations for risk (image) management motivation may be more apparent. On the other hand, the altruistic hypothesis holds that charitable donations are an altruistic behavior demonstrated by enterprises to actively act as good corporate citizens and sincerely give back to society [2,14]. Therefore, enterprises will engage in less tax avoidance activities that are inconsistent with the notion of corporate social responsibility. The rich altruism concept in traditional Chinese culture, especially Confucian culture, has become a moral norm and guide of action that is generally respected by entrepreneurs and subtly affects the social responsibility activities of enterprises [15]. Xu et al. [16] also found that the charitable donation behavior of Chinese enterprises under the influence of Confucian culture has a typical altruistic color rather than a tool management purpose. As Lin et al. [11] pointed out, the relationship between charitable donations and tax avoidance is not static, and it needs to be analyzed in combination with specific situations. Therefore, this paper further examines the moderating role of earnings performance.

Based on the data of A-share listed enterprises in China from 2009 to 2019, this paper demonstrates that there is a significant negative relationship between corporate charitable donations and tax avoidance, that is, enterprises with more charitable donations avoid tax less, which supports the view that corporate social responsibility and tax expenditure are complementary. Furthermore, we find that the negative relationship between charitable donations and tax avoidance only exists in enterprises with a good earnings performance, while it is positively correlated with tax avoidance in enterprises with a poor earnings performance. This shows that earnings performance affects the motivation for corporate charitable donations, so that the charitable donations of enterprises with a good performance are mainly an altruistic behavior, while the charitable donations of enterprises with a poor performance are more of a management tool.

Our findings contribute to the CSR literature in several ways. First, there are numerous studies about corporate social responsibility and tax avoidance, but there are relatively few that directly investigate the relationship between them, and most of them are based on empirical evidence on developed capital markets [8]. This paper uses Chinese corporate data to examine the relationship between corporate charitable donations, earnings performance and tax avoidance and, therefore, not only responds to the call of Lin et al. (2017) for more research to pay attention to the relationship between business ethics and corporate social responsibility in emerging capital markets [11] but also enriches and expands the research framework of corporate social responsibility and tax avoidance. Second, previous studies on the relationship between corporate social responsibility and tax avoidance were mostly based on the overall corporate social responsibility rating [5], ignoring the investigation of specific corporate social responsibility activities [12], and did not pay attention to the impact of earnings performance. The findings of this paper compensate for the shortcomings of the existing literature, to a certain extent, and are also helpful in clarifying the controversy in the existing literature. Finally, in the context of the increasingly severe phenomenon of tax avoidance in enterprises, the question of how to supervise enterprises in order to pay taxes fairly and maintain social fairness has important practical significance. This paper finds that enterprises with a poor performance use charitable donations to conceal their tax avoidance activities, which has important policy implications for enhancing tax inspection efficiency and improving social welfare.

The remainder of this paper is structured as follows. Section 2 reviews the prior literature and develops our main hypothesis. The detailed research design is reported in Section 3. Section 4 reports and discusses our main empirical results. Section 5 concludes the paper.

2. Literature Review and Hypothesis Development

2.1. Literature Review

According to the traditional shareholder view and agency theory, the only social responsibility of enterprises is to maximize shareholder wealth, and the business objectives of enterprises are only constrained by legal boundaries [17]. Corporate social responsibility activities are a self-interested behavior of managers at the expense of shareholders that are performed in order to enhance their social, political and professional reputations. Such socially responsible but unprofitable investment activities will eventually lead to unsustainable business operations and even bankruptcy [4]. Only when corporate social responsibility activities can serve the goal of profit maximization will enterprises pay attention to the interests of non-shareholder stakeholders [11]. Because of their weaker concern for ethics and social responsibility, these enterprises are also more likely to adopt aggressive tax avoidance activities. However, there are great risks that the discovery of such aggressive tax avoidance activities will damage the enterprise’s reputation and raise public concerns and media pressure, resulting in social and economic sanctions and even consumer boycotts [18,19]. Therefore, corporate social responsibility activities are more of a risk management tool, that is, enterprises participate in social responsibility activities in order to create moral capital, thus reducing the risks of political, regulatory and social sanctions faced by enterprises involved in negative events or with negative public opinions [2]. Based on the management tool hypothesis, relevant empirical research found that corporate social responsibility activities are significantly positively correlated with tax avoidance, that is, enterprises strategically participate in corporate social responsibility activities to build CSR reputation among various stakeholders so as to divert their attention on, and supervision of, corporate tax avoidance and reduce the image and reputation risks posed by tax avoidance activities after exposure [20]. Davis et al. [5] found that companies with more corporate social responsibility activities paid less taxes and lobbied more for tax cuts. Col and Patel [3] found that the CSR rating of large multinational enterprises increased significantly after offshore tax avoidance activities, indicating that enterprises carry out more non-tax corporate social responsibility activities in order to reduce the potential risk of tax avoidance. Hsieh et al. [21] found that enterprises are more likely to engage in tax avoidance activities when they have both overconfident CEOs and overconfident CFOs. Abdelfattah and Aboud [20] found that corporate tax avoidance is positively associated with CSR disclosure, and enterprises with a more sophisticated board of directors provide greater CSR disclosure. Campa et al. [22] found that enterprises with co-opted CFOs exhibit higher levels of tax avoidance, and the board membership of the co-opted CFOs moderates their propensity to engage in tax avoidance. Dang et al. [23] confirmed that the size of the audit committee has a positive correlation with tax avoidance, while the proportion of female members and financial and accounting experts of the audit committee can restrain tax avoidance activities.

In contrast, Freeman’s (1983) stakeholder view recognizes the importance of maintaining a balance between business ethics, economic interests and social responsibility [24]. It posits that enterprises should not only create economic profits within the scope permitted by the law but also pay attention to moral and charitable responsibilities so as to act as good corporate citizens. If enterprises ignore the needs of various stakeholders and cause them to terminate the supply of resources to enterprises, this will inevitably threaten the survival and development of enterprises [24]. Therefore, enterprises must actively practice corporate social responsibility activities, including fair taxation. Corporate culture theory further emphasizes that corporate social responsibility is a common belief among enterprises about the correct behavior of assuming economic, social, legal and environmental responsibilities. As an important aspect of corporate social responsibility activities, fair tax payment is also the most basic way for corporate citizens to participate in social interaction [25]. As aggressive tax avoidance strategies severely damage the government’s ability to provide public services and reduce social welfare levels, they are often seen by the public and the media as irresponsible, immoral and even unpatriotic, opportunistic behavior [2]. Therefore, enterprises with a sense of social responsibility will be less engaged in tax avoidance activities that are inconsistent with the cultural notion of corporate social responsibility. Relevant empirical studies based on stakeholder views and corporate culture theory have confirmed that corporate social responsibility is significantly negatively correlated with tax evasion [7]. For example, Hoi et al. [2] found that irresponsible social activities are significantly positively correlated with book–tax difference, indicating that enterprises without social responsibility engage in more aggressive tax avoidance activities. Lanis and Richardson [6] found that socially responsible companies are more cautious in tax planning and pay more taxes, because they worry that tax avoidance activities will damage their consistent and responsible reputation and good image. López-González et al. [7] documented that social and environmental performance is negatively related to tax avoidance, so that firms with a more socially responsible performance show a lower rate of tax-saving practices.

Notably, a significant amount of research has examined the relationship between corporate social responsibility activities and tax avoidance, and the conclusions of this research are still controversial. Moreover, these papers are mostly based on the empirical evidence of developed countries, but there are relatively few papers on business ethics and corporate social responsibility in developing countries. Whait et al. [8] pointed out that, in the future, more research is needed to examine the relationship between corporate social responsibility and tax avoidance based on the institutional scenarios of developing countries, especially the impacts of specific types of corporate social responsibility activities on tax avoidance strategies. Lin et al. [11] specifically mentioned that the unique scenario of China’s transitional economy could provide direct evidence that can aid in clarifying the nature of corporate social responsibility activities and stressed that the relationship between corporate social responsibility and tax avoidance cannot be generalized but needs to be analyzed in line with specific scenarios. Therefore, this paper examines the relationship between corporate charitable donations and tax avoidance in the context of China and further examines the moderating effect of earnings performance on the relationship between them from the perspective of slack resource theory. Our research contributes to the literature by exploring the relationship between charitable donations and tax avoidance in emerging economies, and it also helps to clarify the differences in the existing literature.

2.2. Hypothesis Development

Based on the traditional shareholder view and agency theory, corporate philanthropy is essentially a risk (image) management tool and political rent seeking tool. On the one hand, charitable donations can establish a good corporate citizen image and enhance the corporate reputation for CSR and moral capital so as to reduce the potential risk of political, regulatory and social sanctions on tax avoidance [5]. Godfrey et al. [26] found that the tax avoidance behavior of enterprises with a high reputation for CSR is usually considered as an unintentional mistake, and the public, the media and the government tend to be more tolerant of it. After the exposure of tax avoidance activities, enterprises face relatively little media pressure, and the government penalty risk, customer resistance risk and damage to corporate value will also be significantly reduced. In addition, charitable donations can divert the attention of the public and the media to unethical acts, such as tax avoidance, and play roles in reputation protection and risk mitigation. Col and Patel [3] pointed out that internationally renowned companies such as Apple, Google and Amazon use corporate social responsibility activities such as charitable donations to divert the attention of the government and consumers from their offshore tax avoidance activities, thus reducing the potential risk of political censorship and negative public opinion pressure regarding tax avoidance and maintaining their corporate image and reputation. In China, which is in the emerging and transitional stage, due to imperfect formal systems such as the legal system, market mechanism and corporate governance, the weakness of the system may lead to more immoral or irresponsible behavior, and the motivation of enterprises to use charitable donations for risk (image) management may be more apparent. As Lin et al. [11] found, the weakness of China’s formal system often leads to more aggressive tax avoidance activities by enterprises claiming social responsibility. On the other hand, local governments in China not only control the allocation of most of the scarce resources needed for the survival and development of enterprises but also have the discretion to intervene in determining the intensity of tax collection and management in their jurisdictions [27]. Therefore, enterprises have an incentive to cooperate with local governments in order to complete certain social and political tasks through charitable donations and, thus, to build the necessary political connections [28] required to obtain a more relaxed tax inspection environment and a stronger political asylum effect, thus facilitating them in taking more radical tax avoidance measures [29]. In short, the charitable donations of Chinese enterprises are only a management tool used by enterprises to disguise tax avoidance activities and reduce the risk and cost of tax avoidance, while enterprises with more charitable donations carry out more aggressive tax avoidance.

However, the stakeholder view and corporate culture theory emphasize that enterprises need to maintain a balance between business ethics, economic interests and social responsibility, and corporate social responsibility culture advocates the notion that enterprises need not only to achieve the goal of maximizing the interests of shareholders but also to pay attention to the interests of non-shareholder stakeholders. As an important aspect of corporate social responsibility, charity and fair tax payment are the shared beliefs of enterprises about the correct behavior of taking up economic, social, legal and charitable social responsibilities [3,24]. More charitable donations mean that the cultural atmosphere of corporate social responsibility is stronger. Thus, enterprises will firmly adhere to the responsibility and mission of benefiting the world, pay more attention to the maximization of social welfare, and be unlikely to use charitable donations for risk (image) management and political rent seeking [11]. Although tax avoidance can reduce the cost expenditure of enterprises, increase the profits of enterprises and enhance the value of enterprises, tax avoidance will also seriously damage the ability of the government to provide public services, reduce the level of social welfare, and be regarded by the public and the media as an irresponsible, immoral and even unpatriotic, opportunistic behavior [2]. This is not only contrary to the cultural notion of the correct behavior recognized within the enterprise but will also damage the reputation capital and good image established by corporate charitable donations and even, in serious cases, lead to customer resistance and the withdrawal of resources from other stakeholders, ultimately threatening the survival and development of the enterprise [11]. Therefore, enterprises with more charitable donations will be more cautious in tax planning, relatively less engaged in tax avoidance activities that are inconsistent with the cultural notion of corporate social responsibility, and more likely to exhibit a good corporate citizen image of abiding by laws and regulations and morality in tax payment. Therefore, we establish our first hypothesis in a non-directional form as follows:

Hypothesis 1 (H1).

Ceteris paribus, there is an association between charitable donations and tax avoidance.

Charitable donations cause competition in resource allocation between shareholders and other stakeholders, so that the availability of resources has an important impact on the amount and motivation for corporate charitable donations. According to slack resource theory, enterprises with a good earnings performance will pay more attention to the needs of non-shareholder stakeholders, and the frequency and amount of charitable donations will be higher. On the contrary, enterprises with a poor earnings performance tend to keep scarce resources within the enterprise and make less charitable donations [25]. However, there are many examples showing that, in reality, many poorly performing enterprises actively donate to charity in the same way as high-performing enterprises. We believe that the quality of the earnings performance affects motivation for corporate charitable donations. For enterprises with a good earnings performance, the financial pressure is relatively weak, and enterprises do not need to take higher reputation risks and inspection risks to carry out radical tax avoidance activities so as to save cash flow. Therefore, charitable donations made by enterprises with a good earnings performance may be motivated more by altruism and relatively less by engagement in tax avoidance activities [11]. On the contrary, charitable donations made by enterprises with a poor earnings performance may be more of a management tool, which may hide aggressive tax avoidance. This is because tax expenditure will bring greater cost pressures and cash flow pressures to enterprises with a poor performance, as being a good corporate citizen who pays taxes fairly is expensive [25]. Thus, enterprises with a poor performance have stronger tax avoidance incentives. However, considering the potential risk and cost of tax avoidance, it is undoubtedly the best choice to use low-cost charitable donations for risk management and political rent seeking. Davis et al. [5] pointed out that using lower-cost corporate social responsibility activities to save higher-cost tax expenditure has become a common practice of responsible enterprises in the United States. Watson [25] directly showed that responsible enterprises with a good earnings performance do not avoid tax, while responsible enterprises with a poor earnings performance avoid tax to a certain extent. Therefore, we establish our second hypothesis below:

Hypothesis 2 (H2).

Earnings performance has a moderating effect on the relationship between charitable donations and tax avoidance.

3. Data and Methodology

3.1. Data and Sample

All listed enterprises in China from 2009 to 2019 were included in our sample. We chose to start with 2009 to avoid the effects of the 2008 financial crisis. The China Stock Market and Accounting Research (CSMAR) database provides information on charitable donations, corporate governance and firm fundamental analysis. CSMAR is the primary source of information on Chinese stock markets and the financial statements of Chinese listed firms, and this dataset is universally used in research concerning China’s corporate issues. In accordance with previous studies [30,31,32], we performed the following processes to arrange the data: winsorizing all the continuous variables; deleting all outliers such that the total number of assets is negative and income tax expense is negative; eliminating Special Treatment enterprises; and eliminating enterprises with missing data. Finally, we obtained 20,225 firm–year observations to estimate the effect of charitable donations on tax avoidance during the period of 2009–2019.

3.2. Research Design

On the basis of Hypothesis 1, we initially used the following regression to examine the link between charitable donations and tax avoidance:

$$TRD={\alpha }_0+{\alpha }_{1}Charity+{\alpha }_{2}Controls+Year+Industry+Province+\epsilon$$

(1)

where the dependent variable TRD denotes the enterprise’s tax avoidance. Here, we use the book–tax difference to measure the tax avoidance degree of enterprises [2]. Specifically, we define the total profit of an enterprise minus the taxable income divided by the total assets at the end of the period as the proxy of book–tax difference. The greater the value of TRD is, the more serious the degree of tax avoidance of the enterprise is. The independent variable Charity denotes the enterprise’s charitable donations. Referring to Wang et al. [32], we use the donation tendency (Charity1) and donation amount (Charity2), respectively, to measure charitable donations. Charity1 is a dummy variable. If the enterprise makes a donation in the current period, it equals 1; otherwise, it equals 0. Charity2 is the natural logarithm of the donation amount of the enterprise in the current year. The controls denote the set of all the covariates. Referring to Hoi et al. [2] and Li et al. [29], we include the Size, Lev, Roa, Growth, Cash, PPE, INTAN, MB, Soe, Age, Top10, INS and AbsDA. All the variable definitions are presented in

Table 1. Variable definition.

Variable Type	Variable Name	Variable Definition
Dependent variable	TRD	Pretax profit minus the taxable income, divided by total assets
Independent variable	Charity1	Dummy variable. If the enterprise makes a donation in the current period, it equals 1; otherwise, it equals 0
	Charity2	The natural logarithm of the donation amount of the enterprise in the current year
Moderating variable	HP	Good earnings performance
	LP	Poor earnings performance
Control variable	Size	The natural logarithm of total assets
	Lev	Total liabilities divided by total assets
	Roa	Net profit divided by total assets
	Growth	The growth rate of net profit.
	Cash	Cash equivalent divided by total assets
	PPE	Fixed assets divided by total assets
	INTAN	Intangible assets divided by total assets
	MB	Total assets divided by market value
	Soe	A variable that equals 1 if the firm is a state-owned enterprise or 0 otherwise
	Age	The natural logarithm of the year of the firm’s establishment
	Top10	The top ten majority shareholding ratio
	INS	Total shareholding ratio of institutional investors
	AbsDA	Absolute value of discretionary accruals for firm i, year t, where discretionary accruals are computed using the modified Jones model

. Moreover, the year, industry and province-fixed effects are also included in the regression models.

To examine whether the association between charitable donations and tax avoidance varies according to earnings performance (Hypothesis 2), the following model was estimated:

$$\begin{gathered} TRD={\alpha }_0+{\alpha }_{1}Charity\times HP\left(LP\right)+{\alpha }_{2}Charity+{\alpha }_{3}HP\left(LP\right)+{\alpha }_{n}Controls \\ +Year+Industry+Province+\epsilon \end{gathered}$$

(2)

where the moderate variable HP(LP) denotes the earnings performance. Referring to Watson [25], we rank Roa from small to large. If Roa is within the first third of the range, LP takes 1. Otherwise, it takes 0, which is defined as a poor earnings performance. If Roa is within the last third of the range, HP takes 1. Otherwise, it takes 0, which is defined as a good earnings performance. The other variables are defined in the same way as in model (1).

Table 1 shows the definitions of the variables in this paper.

4. Empirical Results

4.1. Descriptive Statistics and Correlations

Table 2. Descriptive statistics.

Variable	Observations	Mean	S.D.	Minimum	Median	Maximum
TRD	20,225	0.070	0.241	−0.590	0.035	0.947
Charity1	20,225	0.733	0.442	0.000	1.000	1.000
Charity2	20,225	9.296	5.873	0.000	11.775	17.228
HP	20,225	0.333	0.471	0.000	0.000	1.000
LP	20,225	0.333	0.471	0.000	0.000	1.000
Size	20,225	22.212	1.291	19.960	22.029	26.217
Lev	20,225	0.425	0.202	0.052	0.420	0.859
Roa	20,225	0.051	0.040	0.001	0.042	0.196
Growth	20,225	0.208	0.424	0.433	0.126	2.870
Cash	20,225	0.052	0.069	0.152	0.050	0.246
PPE	20,225	0.220	0.165	0.002	0.186	0.711
INTAN	20,225	0.047	0.051	0.000	0.034	0.323
MB	20,225	0.616	0.242	0.130	0.614	1.141
Soe	20,225	0.411	0.492	0.000	0.000	1.000
Age	20,225	1.949	0.942	0.000	2.197	3.219
Top10	20,225	0.589	0.150	0.239	0.600	0.903
INS	20,225	0.460	0.245	0.004	0.487	0.908
AbsDA	20,225	0.049	0.051	0.033	0.001	0.280
AGE	20,225	3.873	0.421	0.000	3.932	4.190
MALE	20,225	0.927	0.260	0.000	1.000	1.000

presents the descriptive statistics for the variables used in our main regression model. As shown, the mean value of TRD is 0.070. In addition, the mean value of Charity1 is 0.77, indicating that most enterprises participated in charitable donations during the sample period. The mean value of Charity2 is 9.296, and the minimum (maximum) value of Charity2 is 0.000(17.228), which means that there are huge differences in the amount of donations between different enterprises. Moreover, all the variables in Table 2 are comparable with the data obtained in previous studies.

Appendix A reports the correlation coefficients of the main variables. As revealed in the table, except for Charity1 and Charity2, Roa, HP and LP, the absolute values of the correlation coefficients between the other variables are less than 0.55. This indicates that there is no serious multicollinearity between the variables.

4.2. Regression Analysis

The regression results of model (1) are reported in columns (1) and (2) of

Table 3. Regression analysis.

Variable	(1)	(2)	(3)	(4)	(5)	(6)
	TRD	TRD	TRD	TRD	TRD	TRD
Charity1	−0.022 ***		−0.031 ***	−0.011 ***
	(−6.648)		(−6.876)	(−3.500)
Charity2		−0.002 ***			−0.003 ***	−0.001 ***
		(−7.149)			(−7.169)	(−4.447)
Charity1 × LP			0.023 ***
			(3.961)
Charity1 × HP				−0.037 ***
				(−4.337)
Charity2 × LP					0.002 ***
					(3.487)
Charity2 × HP						−0.002 ***
						(−3.839)
LP			−0.053 ***		−0.051 ***
			(−9.099)		(−8.884)
HP				0.035 ***		0.030 ***
				(4.040)		(3.620)
Size	0.005 **	0.006 ***	0.005 **	0.005 **	0.007 ***	0.007 ***
	(2.326)	(2.998)	(2.346)	(2.372)	(3.073)	(3.084)
Lev	−0.024 **	−0.024 **	−0.017	−0.024 **	−0.017	−0.024 **
	(−2.131)	(−2.149)	(−1.501)	(−2.182)	(−1.508)	(−2.215)
Roa	2.924 ***	2.933 ***	2.656 ***	2.851 ***	2.664 ***	2.862 ***
	(39.801)	(39.839)	(29.605)	(27.367)	(29.662)	(27.421)
Growth	−0.008 *	−0.008 *	−0.009 **	−0.008 *	−0.009 **	−0.008 *
	(−1.869)	(−1.887)	(−2.173)	(−1.879)	(−2.207)	(−1.903)
Cash	−0.313 ***	−0.313 ***	−0.303 ***	−0.310 ***	−0.303 ***	−0.310 ***
	(−11.024)	(−11.034)	(−10.656)	(−10.944)	(−10.657)	(−10.928)
PPE	0.034 ***	0.034 ***	0.035 ***	0.035 ***	0.035***	0.034 ***
	(2.738)	(2.761)	(2.817)	(2.781)	(2.823)	(2.771)
INTAN	−0.068 **	−0.066 *	−0.071 **	−0.070 **	−0.068 **	−0.068 **
	(−2.010)	(−1.939)	(−2.080)	(−2.059)	(−2.013)	(−2.011)
MB	0.013	0.012	0.009	0.011	0.008	0.010
	(1.260)	(1.152)	(0.898)	(1.102)	(0.757)	(0.937)
Soe	0.005	0.004	0.005	0.005	0.004	0.004
	(1.208)	(1.062)	(1.255)	(1.159)	(1.090)	(1.005)
Age	0.009 ***	0.009 ***	0.010 ***	0.009 ***	0.010 ***	0.009 ***
	(3.675)	(3.581)	(4.200)	(3.733)	(4.111)	(3.678)
Top10	−0.013	−0.013	−0.015	−0.014	−0.015	−0.014
	(−0.890)	(−0.910)	(−1.006)	(−0.938)	(−1.041)	(−0.942)
INS	0.002	0.001	0.000	0.002	−0.000	0.001
	(0.176)	(0.111)	(0.024)	(0.216)	(−0.041)	(0.141)
AbsDA	−0.184 ***	−0.185 ***	−0.175 ***	−0.181 ***	−0.175 ***	−0.181 ***
	(−5.437)	(−5.455)	(−5.199)	(−5.369)	(−5.209)	(−5.368)
Constant	0.217 ***	0.187 ***	0.246 ***	0.209 ***	0.214 ***	0.179 ***
	(4.343)	(3.720)	(4.912)	(4.179)	(4.240)	(3.555)
Year	Yes	Yes	Yes	Yes	Yes	Yes
Ind	Yes	Yes	Yes	Yes	Yes	Yes
Province	Yes	Yes	Yes	Yes	Yes	Yes
Adj-R2	0.304	0.304	0.307	0.305	0.308	0.305
N	20,225	20,225	20,225	20,225	20,225	20,225

Note: t statistics based on robust standard errors clustered by firm are reported in brackets. ***, ** and * represent statistical significance at the 1%, 5% and 10% levels, respectively.

. The independent variables are Charity1 and Charity2, respectively, and the dependent variable is TRD. In column (1), the estimated coefficient is −0.022, and it is statistically significant. In column (2), the estimated coefficient is −0.002, and it is also statistically significant. These results are consistent with the stakeholder view and corporate culture theory, indicating that corporate charitable donations generally constitute an altruistic behavior rather than a management tool. In other words, enterprises with more charitable donations will be less engaged in tax avoidance.

Columns (3) to (6) further report the regression results of model (2). The results show that the coefficients of Charity1 × LP and Charity2 × LP are 0.023 and 0.002, respectively, which are statistically significant. The coefficients of Charity1 × HP and Charity2 × HP are −0.037 and −0.002, respectively, which are also statistically significant. These results show that the earnings performance can affect the motivation for corporate charitable donations. The charitable donations of enterprises with a good performance are mainly an altruistic behavior, while the charitable donations of enterprises with a poor performance are more of a management tool. That is to say, the earnings performance has a moderating effect on the relationship between charitable donations and tax avoidance, thus supporting the second hypothesis, H2.

4.3. Robustness Analyses

We conducted a few robustness tests to verify the sensitivity of our main results.

First, following prior studies [11], we used the propensity score matching (PSM) procedure to eliminate sample selection issues. Specifically, in this process, according to whether or not the enterprises make charitable donations, the enterprises that make charitable donations are defined as the experimental group, and the enterprises that do not make charitable donations are defined as the control group. Then, the propensity score is calculated by logit regression, and all the control variables in the main regression are used as the matching criteria. Based on the scores of each observation value, the samples of the experimental group and the control group are matched by radius.

Table 4. Equilibrium hypothesis testing: propensity score matching.

Covariate	Sample	Mean Difference		t Value (p Value)	Standard Deviation (%)
		Treatment Group	Control Group		Standard Deviation	Decrease (%)
Size	Before matching	22.328	21.892	21.46 (0.000)	33.9
	After matching	22.143	22.136	0.21 (0.837)	0.5	98.6
Lev	Before matching	0.433	0.404	9.02 (0.000)	14.2
	After matching	0.421	0.420	0.25 (0.805)	0.6	95.6
Roa	Before matching	0.053	0.046	10.79 (0.000)	17.5
	After matching	0.051	0.050	1.01 (0.312)	2.5	85.8
Growth	Before matching	0.209	0.203	0.86 (0.388)	1.3
	After matching	0.201	0.202	−0.10 (0.904)	−0.2	83.2
Cash	Before matching	0.053	0.047	6.12 (0.000)	9.7
	After matching	0.050	0.053	−1.35 (0.179)	−3.3	65.9
PPE	Before matching	0.218	0.226	−2.95 (0.003)	−4.6
	After matching	0.221	0.223	−0.38 (0.704)	−0.9	79.3
INTAN	Before matching	0.048	0.043	5.66 (0.000)	9.0
	After matching	0.046	0.045	0.51 (0.609)	1.3	86.0
MB	Before matching	0.631	0.575	14.81 (0.000)	23.4
	After matching	0.621	0.627	−1.00 (0.319)	−2.4	89.8
Soe	Before matching	0.387	0.477	−11.63 (0.000)	−18.4
	After matching	0.409	0.393	1.29 (0.196)	3.2	82.4
Age	Before matching	1.941	1.970	−1.92 (0.055)	−3.0
	After matching	1.894	1.863	1.29 (0.197)	3.3	−10.1
Top10	Before matching	0.591	0.587	1.91 (0.056)	3.1
	After matching	0.589	0.593	−1.17 (0.243)	−2.9	4.2
INS	Before matching	0.459	0.462	−0.78 (0.437)	−1.3
	After matching	0.454	0.449	0.76 (0.448)	2.0	56.2
AbsDA	Before matching	0.048	0.052	−4.85 (0.000)	−7.6
	After matching	0.049	0.048	1.07 (0.286)	2.5	66.3

Note: t statistics based on robust standard errors clustered by firm are reported in brackets.

reports the equilibrium hypothesis testing of the PSM results. It shows that there are no significant differences between the treatment and control groups after matching.

Table 5. Propensity score matching (PSM) test.

Variable	(1)	(2)	(3)	(4)	(5)	(6)
	TRD	TRD	TRD	TRD	TRD	TRD
Charity1	−0.032 ***		−0.041 ***	−0.017 ***
	(−5.844)		(−5.645)	(−3.108)
Charity2		−0.003 ***			−0.003 ***	−0.002 ***
		(−5.779)			(−5.398)	(−3.534)
Charity1 × LP			0.026 ***
			(2.644)
Charity1 × HP				−0.046 ***
				(−3.483)
Charity2 × LP					0.002 **
					(2.102)
Charity2 × HP						−0.003 ***
						(−2.862)
LP			−0.040 ***		−0.037 ***
			(−4.097)		(−3.815)
HP				0.025 *		0.020
				(1.919)		(1.552)
Size	−0.002	−0.001	−0.002	−0.002	−0.000	−0.001
	(−0.475)	(−0.208)	(−0.377)	(−0.437)	(−0.105)	(−0.157)
Lev	−0.016	−0.016	−0.011	−0.016	−0.011	−0.016
	(−0.772)	(−0.771)	(−0.539)	(−0.755)	(−0.541)	(−0.778)
Roa	3.260 ***	3.270 ***	3.062 ***	3.262 ***	3.071 ***	3.268 ***
	(23.233)	(23.277)	(17.495)	(16.075)	(17.541)	(16.094)
Growth	0.003	0.003	0.002	0.003	0.002	0.002
	(0.334)	(0.318)	(0.242)	(0.311)	(0.222)	(0.298)
Cash	−0.270 ***	−0.270 ***	−0.265 ***	−0.266 ***	−0.265 ***	−0.266 ***
	(−4.979)	(−4.970)	(−4.881)	(−4.923)	(−4.867)	(−4.909)
PPE	0.001	0.001	0.002	0.002	0.001	0.002
	(0.042)	(0.036)	(0.074)	(0.099)	(0.058)	(0.069)
INTAN	−0.027	−0.025	−0.029	−0.024	−0.028	−0.025
	(−0.407)	(−0.386)	(−0.438)	(−0.371)	(−0.421)	(−0.375)
MB	0.028	0.028	0.026	0.027	0.025	0.026
	(1.476)	(1.458)	(1.342)	(1.424)	(1.317)	(1.360)
Soe	0.020 **	0.019 **	0.020 **	0.019 **	0.019 **	0.019 **
	(2.491)	(2.420)	(2.460)	(2.403)	(2.396)	(2.346)
Age	0.014 ***	0.014 ***	0.014 ***	0.013 ***	0.014 ***	0.013 ***
	(2.908)	(2.884)	(3.016)	(2.880)	(3.002)	(2.880)
Top10	−0.021	−0.021	−0.024	−0.024	−0.024	−0.023
	(−0.748)	(−0.765)	(−0.848)	(−0.844)	(−0.858)	(−0.828)
INS	−0.018	−0.018	−0.018	−0.016	−0.018	−0.017
	(−1.045)	(−1.063)	(−1.040)	(−0.941)	(−1.062)	(−0.977)
AbsDA	−0.046	−0.046	−0.035	−0.043	−0.035	−0.042
	(−0.695)	(−0.693)	(−0.538)	(−0.649)	(−0.533)	(−0.645)
Constant	0.433 ***	0.410 ***	0.449 ***	0.425 ***	0.425 ***	0.403 ***
	(4.459)	(4.211)	(4.609)	(4.349)	(4.352)	(4.118)
Year	Yes	Yes	Yes	Yes	Yes	Yes
Ind	Yes	Yes	Yes	Yes	Yes	Yes
Province	Yes	Yes	Yes	Yes	Yes	Yes
Adj-R2	0.325	0.325	0.327	0.327	0.327	0.326
N	5630	5630	5630	5630	5630	5630

Note: t statistics based on robust standard errors clustered by firm are reported in brackets. ***, ** and * represent statistical significance at the 1%, 5% and 10% levels, respectively.

reports the matched regression results. Consistent with our main findings, the coefficients of Charity1 and Charity2 are significantly negative, the coefficients of Charity1 × LP and Charity1 × LP are significantly positive, and the coefficients of Charity1 × HP and Charity1 × HP are significantly negative.

Second, we controlled the executive characteristics. Prior research found that executives play an important role in corporate tax avoidance [33]. Referring to Davis et al. [5], we further controlled for CEO age (AGE) and gender (MALE) in models (1) and (2).

Table 6. Controlling the executive characteristics.

Variable	(1)	(2)	(3)	(4)	(5)	(6)
	TRD	TRD	TRD	TRD	TRD	TRD
Charity1	−0.022 ***		−0.031 ***	−0.011 ***
	(−6.552)		(−6.823)	(−3.435)
Charity2		−0.002 ***			−0.003 ***	−0.001 ***
		(−7.075)			(−7.128)	(−4.415)
Charity1 × LP			0.024 ***
			(4.000)
Charity1 × HP				−0.036 ***
				(−4.286)
Charity2 × LP					0.002 ***
					(3.513)
Charity2 × HP						−0.002 ***
						(−3.779)
LP			−0.053 ***		−0.051 ***
			(−9.119)		(−8.895)
HP				0.035 ***		0.030 ***
				(4.068)		(3.644)
Size	0.005 **	0.006 ***	0.005 **	0.005 **	0.006 ***	0.006 ***
	(2.263)	(2.933)	(2.287)	(2.310)	(3.012)	(3.020)
Lev	−0.026 **	−0.026 **	−0.019 *	−0.026 **	−0.019 *	−0.027 **
	(−2.306)	(−2.325)	(−1.691)	(−2.358)	(−1.698)	(−2.390)
Roa	2.909 ***	2.918 ***	2.642 ***	2.831 ***	2.649 ***	2.842 ***
	(39.395)	(39.432)	(29.334)	(27.059)	(29.390)	(27.112)
Growth	−0.008 *	−0.008 *	−0.009 **	−0.008 *	−0.009 **	−0.008 *
	(−1.839)	(−1.859)	(−2.144)	(−1.845)	(−2.180)	(−1.871)
Cash	−0.311 ***	−0.312 ***	−0.301 ***	−0.308 ***	−0.301 ***	−0.308 ***
	(−10.823)	(−10.834)	(−10.455)	(−10.733)	(−10.457)	(−10.721)
PPE	0.034 ***	0.034 ***	0.035 ***	0.034 ***	0.035 ***	0.034 ***
	(2.702)	(2.722)	(2.785)	(2.740)	(2.788)	(2.729)
INTAN	−0.076 **	−0.073 **	−0.078 **	−0.077 **	−0.075 **	−0.076 **
	(−2.195)	(−2.123)	(−2.257)	(−2.247)	(−2.187)	(−2.197)
MB	0.014	0.013	0.011	0.013	0.009	0.011
	(1.381)	(1.277)	(1.031)	(1.228)	(0.893)	(1.069)
Soe	0.005	0.004	0.005	0.005	0.005	0.004
	(1.254)	(1.110)	(1.303)	(1.214)	(1.141)	(1.060)
Age	0.009 ***	0.009 ***	0.010 ***	0.009 ***	0.010 ***	0.009 ***
	(3.770)	(3.678)	(4.299)	(3.825)	(4.209)	(3.771)
Top10	−0.008	−0.009	−0.010	−0.009	−0.011	−0.009
	(−0.574)	(−0.598)	(−0.693)	(−0.631)	(−0.732)	(−0.638)
INS	0.003	0.002	0.001	0.003	0.001	0.002
	(0.276)	(0.214)	(0.134)	(0.323)	(0.073)	(0.251)
AbsDA	−0.184 ***	−0.184 ***	−0.174 ***	−0.181 ***	−0.175 ***	−0.181 ***
	(−5.331)	(−5.351)	(−5.086)	(−5.261)	(−5.099)	(−5.263)
AGE	−0.000	−0.000	−0.000	−0.000	−0.000	−0.000
	(−0.935)	(−0.942)	(−1.023)	(−1.007)	(−1.034)	(−1.017)
MALE	−0.000	−0.001	−0.001	−0.001	−0.001	−0.001
	(−0.039)	(−0.082)	(−0.089)	(−0.084)	(−0.141)	(−0.143)
Constant	0.230 ***	0.200 ***	0.259 ***	0.222 ***	0.227 ***	0.193 ***
	(4.467)	(3.866)	(5.039)	(4.328)	(4.392)	(3.729)
Year	Yes	Yes	Yes	Yes	Yes	Yes
Ind	Yes	Yes	Yes	Yes	Yes	Yes
Province	Yes	Yes	Yes	Yes	Yes	Yes
Adj-R2	0.303	0.304	0.306	0.304	0.307	0.304
N	20,011	20,011	20,011	20,011	20,011	20,011

Note: t statistics based on robust standard errors clustered by firm are reported in brackets. ***, ** and * represent statistical significance at the 1%, 5% and 10% levels, respectively.

reports the regression results after adding the control variables. Clearly, the results are consistent with our main regression results reported in Table 3.

Third, although propensity score matching (PSM) and the addition of control variables can alleviate the problem of missing variables to a certain extent, there may still be some factors that are difficult to directly observe or measure, which will cause endogenous problems. Therefore, this paper uses the instrumental variable method of two-stage least squares regression. Specifically, following Zhang et al. [27] and Li et al. [29], we use R&D investment (R&D expenditure/operating income) and advertising investment (sales expenses/operating income) as instrumental variables. Meanwhile, by running the following model (3), we obtain the fitting values Donate1 and Donate2 for Charity1 and Charity2. Finally, the fitted values Donate1 and Donate2 were substituted into model (1) and model (2) for regression.

Table 7. Instrumental variable (IV) test.

Variable	(1)	(2)	(3)	(4)	(5)	(6)
	TRD	TRD	TRD	TRD	TRD	TRD
Donate1	−1.253 ***		−1.312 ***	−1.196 ***
	(−3.100)		(−3.372)	(−3.053)
Donate2		−0.035 **			−0.036 **	−0.032 **
		(−2.219)			(−2.349)	(−2.095)
Donate1 × LP			0.085 ***
			(3.538)
Donate1 × HP				−0.131 ***
				(−4.040)
Donate2 × LP					0.004 ***
					(2.643)
Donate2 × HP						−0.006 ***
						(−3.220)
LP			−0.101 ***		−0.072 ***
			(−5.633)		(−5.476)
HP				0.092 ***		0.051 ***
				(3.691)		(2.736)
Size	0.100 ***	0.058 **	0.103 ***	0.099 ***	0.058 **	0.057 **
	(3.067)	(2.179)	(3.256)	(3.148)	(2.230)	(2.198)
Lev	0.058 **	−0.003	0.066 **	0.055 **	0.003	−0.005
	(2.009)	(−0.190)	(2.336)	(1.960)	(0.213)	(−0.330)
Roa	4.189 ***	3.502 ***	3.943 ***	4.256 ***	3.223 ***	3.566 ***
	(10.077)	(12.782)	(9.729)	(10.468)	(11.790)	(12.943)
Growth	−0.017 **	−0.011 *	−0.020 ***	−0.018 ***	−0.013 **	−0.012 *
	(−2.545)	(−1.817)	(−2.949)	(−2.652)	(−2.163)	(−1.885)
Cash	−0.239 ***	−0.252 ***	−0.223 ***	−0.236 ***	−0.237 ***	−0.249 ***
	(−6.423)	(−6.829)	(−6.029)	(−6.366)	(−6.461)	(−6.785)
PPE	0.058 ***	0.048 ***	0.060 ***	0.057 ***	0.050 ***	0.048 ***
	(3.534)	(3.074)	(3.712)	(3.523)	(3.210)	(3.043)
INTAN	0.438 **	0.116	0.447 **	0.425 **	0.112	0.105
	(2.414)	(1.046)	(2.543)	(2.417)	(1.031)	(0.967)
MB	0.127 ***	0.046 ***	0.124 ***	0.119 ***	0.041 ***	0.039 **
	(3.596)	(2.960)	(3.609)	(3.461)	(2.649)	(2.549)
Soe	−0.145 ***	−0.056 *	−0.148 ***	−0.143 ***	−0.055 *	−0.054 *
	(−2.947)	(−1.942)	(−3.105)	(−2.992)	(−1.953)	(−1.923)
Age	−0.019 **	−0.004	−0.018 **	−0.018 **	−0.002	−0.003
	(−2.053)	(−0.604)	(−2.010)	(−2.017)	(−0.363)	(−0.474)
Top10	−0.187 ***	−0.075 **	−0.191 ***	−0.184 ***	−0.076 **	−0.073 **
	(−3.291)	(−2.450)	(−3.461)	(−3.333)	(−2.501)	(−2.429)
INS	−0.089 ***	−0.034	−0.093 ***	−0.090 ***	−0.035 *	−0.034
	(−2.698)	(−1.567)	(−2.917)	(−2.802)	(−1.664)	(−1.606)
AbsDA	−0.477 ***	−0.258 ***	−0.475 ***	−0.471 ***	−0.248 ***	−0.253 ***
	(−4.201)	(−3.959)	(−4.314)	(−4.269)	(−3.871)	(−3.938)
Constant	−0.598 *	−0.455	−0.580 *	−0.625 **	−0.419	−0.457
	(−1.814)	(−1.156)	(−1.812)	(−1.965)	(−1.091)	(−1.198)
Year	Yes	Yes	Yes	Yes	Yes	Yes
Ind	Yes	Yes	Yes	Yes	Yes	Yes
Province	Yes	Yes	Yes	Yes	Yes	Yes
Adj-R2	0.318	0.317	0.322	0.319	0.320	0.317
N	15,294	15,294	15,294	15,294	15,294	15,294

Note: t statistics based on robust standard errors clustered by firm are reported in brackets. ***, ** and * represent statistical significance at the 1%, 5% and 10% levels, respectively. Due to the lack of R&D expenditure data, there is a certain lack of samples.

reports the regression results. Columns (1) and (2) show that the coefficients of Donate1 and Donate2 are −1.253 and −0.035, respectively, and are significantly negative, at least at the 5 % level. Columns (3)–(6) show that the coefficients of Donate1 × LP and Donate2 × LP are 0.085 and 0.004, respectively, which are significantly positive at the 1 % level. The coefficients of Donate1 × HP and Donate2 × HP are −0.131 and −0.006, respectively, which are significantly negative at the 1 % level, and the conclusion remains robust.

$$Charity={\alpha }_0+{\alpha }_{1}RD+{\alpha }_{2}AD+{\alpha }_{n}Controls+\sum Year+\sum Ind+\epsilon$$

(3)

Finally, we consider two alternative variables. First, we use ETR instead of TRD. Following prior studies [11,34], we define ETR as the income tax expense divided by the total assets. Second, we use Roe instead of Roa. Roe was computed as a ratio of the net income after tax against the average equity value.

Table 8. Using ETR to measure corporate tax avoidance.

Variable	(1)	(2)	(3)	(4)	(5)	(6)
	ETR	ETR	ETR	ETR	ETR	ETR
Charity1	0.045 ***		0.071 ***	0.024
	(3.059)		(3.844)	(1.452)
Charity2		0.005 ***			0.007 ***	0.003 ***
		(4.516)			(4.905)	(2.665)
Charity1 × LP			−0.071 ***
			(−2.596)
Charity1 × HP				0.070 **
				(2.195)
Charity2 × LP					−0.005 **
					(−2.359)
Charity2 × HP						0.006 **
						(2.333)
LP			0.077 ***		0.072 ***
			(2.973)		(2.855)
HP				−0.032		−0.032
				(−0.976)		(−1.013)
Size	−0.068 ***	−0.074 ***	−0.068 ***	−0.069 ***	−0.074 ***	−0.074 ***
	(−7.382)	(−7.877)	(−7.390)	(−7.404)	(−7.903)	(−7.926)
Lev	0.298 ***	0.297 ***	0.295 ***	0.300 ***	0.294 ***	0.300 ***
	(6.219)	(6.211)	(6.162)	(6.281)	(6.146)	(6.288)
Roa	16.046 ***	16.006 ***	16.245 ***	15.867 ***	16.212 ***	15.820 ***
	(56.788)	(56.591)	(48.797)	(41.447)	(48.692)	(41.298)
Growth	−0.026	−0.026	−0.026	−0.026	−0.025	−0.026
	(−1.326)	(−1.303)	(−1.298)	(−1.322)	(−1.265)	(−1.296)
Cash	1.251 ***	1.251 ***	1.239 ***	1.246 ***	1.239 ***	1.244 ***
	(10.809)	(10.817)	(10.675)	(10.781)	(10.675)	(10.770)
PPE	0.402 ***	0.402 ***	0.401 ***	0.401 ***	0.401 ***	0.401 ***
	(7.338)	(7.328)	(7.305)	(7.314)	(7.307)	(7.318)
INTAN	−1.448 ***	−1.462 ***	−1.447 ***	−1.448 ***	−1.460 ***	−1.459 ***
	(−7.300)	(−7.370)	(−7.299)	(−7.297)	(−7.364)	(−7.355)
MB	−0.141 ***	−0.140 ***	−0.136 ***	−0.137 ***	−0.134 ***	−0.134 ***
	(−3.092)	(−3.060)	(−2.982)	(−3.009)	(−2.926)	(−2.931)
Soe	0.053 ***	0.056 ***	0.053 ***	0.053 ***	0.056 ***	0.056 ***
	(2.843)	(2.992)	(2.838)	(2.856)	(2.996)	(3.014)
Age	−0.040 ***	−0.039 ***	−0.041 ***	−0.040 ***	−0.040 ***	−0.039 ***
	(−4.116)	(−4.005)	(−4.229)	(−4.087)	(−4.126)	(−3.997)
Top10	0.301 ***	0.304 ***	0.303 ***	0.303 ***	0.307 ***	0.305 ***
	(4.881)	(4.933)	(4.913)	(4.910)	(4.974)	(4.956)
INS	0.019	0.023	0.020	0.018	0.023	0.021
	(0.567)	(0.659)	(0.591)	(0.529)	(0.680)	(0.621)
AbsDA	1.094 ***	1.101 ***	1.086 ***	1.093 ***	1.091 ***	1.097 ***
	(7.684)	(7.734)	(7.632)	(7.678)	(7.672)	(7.711)
Constant	0.199	0.295	0.163	0.217	0.263	0.318
	(0.990)	(1.461)	(0.813)	(1.080)	(1.301)	(1.568)
Year	Yes	Yes	Yes	Yes	Yes	Yes
Ind	Yes	Yes	Yes	Yes	Yes	Yes
Province	Yes	Yes	Yes	Yes	Yes	Yes
Adj-R2	0.437	0.437	0.437	0.437	0.437	0.437
N	20,225	20,225	20,225	20,225	20,225	20,225

Note: t statistics based on robust standard errors clustered by firm are reported in brackets. ***, ** represent statistical significance at the 1% and 5% levels, respectively.

and

Table 9. Using Roe to measure earnings performance.

Variable	(1)	(2)	(3)	(4)
	TRD	TRD	TRD	TRD
Charity1	−0.030 ***	−0.013 ***
	(−6.478)	(−4.100)
Charity2			−0.002 ***	−0.001 ***
			(−6.647)	(−5.049)
Charity1 × LP	0.019 ***
	(3.240)
Charity1 × HP		−0.032 ***
		(−3.656)
Charity2 × LP			0.001 **
			(2.422)
Charity2 × HP				−0.002 ***
				(−3.006)
LP	−0.049 ***		−0.045 ***
	(−8.181)		(−7.737)
HP		0.028 ***		0.022 ***
		(3.205)		(2.644)
Size	0.005 **	0.005 **	0.006 ***	0.007 ***
	(2.191)	(2.459)	(2.931)	(3.203)
Lev	−0.046 ***	−0.029 **	−0.046 ***	−0.029 **
	(−3.918)	(−2.373)	(−3.939)	(−2.401)
Roa	2.627 ***	2.883 ***	2.634 ***	2.889 ***
	(27.873)	(29.437)	(27.927)	(29.448)
Growth	−0.009 **	−0.008 *	−0.009 **	−0.008 *
	(−2.078)	(−1.869)	(−2.116)	(−1.897)
Cash	−0.304 ***	−0.312 ***	−0.304 ***	−0.312 ***
	(−10.725)	(−11.017)	(−10.736)	(−11.020)
PPE	0.035 ***	0.035 ***	0.035 ***	0.035 ***
	(2.828)	(2.807)	(2.828)	(2.787)
INTAN	−0.072 **	−0.070 **	−0.069 **	−0.068 **
	(−2.112)	(−2.045)	(−2.038)	(−1.990)
MB	0.010	0.011	0.009	0.010
	(1.009)	(1.101)	(0.871)	(0.948)
Soe	0.005	0.005	0.004	0.004
	(1.184)	(1.174)	(1.020)	(1.018)
Age	0.010 ***	0.009 ***	0.010 ***	0.008 ***
	(4.201)	(3.593)	(4.100)	(3.505)
Top10	−0.017	−0.014	−0.018	−0.014
	(−1.185)	(−0.945)	(−1.200)	(−0.947)
INS	0.002	0.002	0.001	0.001
	(0.230)	(0.244)	(0.153)	(0.151)
AbsDA	−0.174 ***	−0.183 ***	−0.175 ***	−0.184 ***
	(−5.181)	(−5.425)	(−5.194)	(−5.446)
Constant	0.265 ***	0.209 ***	0.231 ***	0.177 ***
	(5.258)	(4.156)	(4.564)	(3.503)
Year	Yes	Yes	Yes	Yes
Ind	Yes	Yes	Yes	Yes
Province	Yes	Yes	Yes	Yes
Adj-R2	0.307	0.305	0.307	0.305
N	20,225	20,225	20,225	20,225

Note: t statistics based on robust standard errors clustered by firm are reported in brackets. ***, ** and * represent statistical significance at the 1%, 5% and 10% levels, respectively.

, respectively, report the regression results. Clearly, these results are consistent with our main findings.

5. Discussion

In emerging market countries, because the legal system, market mechanism, corporate governance and other formal systems are not perfect, the system’s weakness may lead to more misconduct or irresponsible behavior. With the rapid development of China’s economy, China’s business ethics and social responsibility are increasingly a concern of the international community. In particular, in the case of serious tax avoidance by Chinese enterprises, it is not clear whether the charitable donations of Chinese enterprises are an altruistic behavior or a management tool. Since previous studies mainly focused on the relationship between corporate social responsibility and tax avoidance in developed countries and on the overall corporate social responsibility rating, this has led to inconsistent conclusions, and it is difficult to accurately answer questions regarding the business ethics and social responsibility of Chinese enterprises. Based on this observation, this paper uses Chinese enterprise data to examine the relationship between charitable donations and tax avoidance and further examines the moderating effect of earnings performance. This paper shows that, on the whole, the corporate social responsibility behavior of emerging market enterprises is an altruistic behavior rather than a management tool. However, earnings performance affects enterprises’ attention to the needs of non-shareholder stakeholders, thus affecting the motivation for corporate social responsibility activities. Therefore, the controversy regarding previous studies may lie in the lack of consideration of the availability of enterprise resources, which also verifies Lin et al.’s (2017) view that the relationship between corporate social responsibility and tax avoidance is not static and needs to be analyzed in line with specific situations. This paper contributes to the CSR literature in several ways. First, this paper uses Chinese corporate data to examine the relationship between corporate charitable donations, earnings performance and tax avoidance, which enriches and expands the research framework of corporate social responsibility and tax avoidance. Second, the findings of this paper compensate for the shortcomings of the existing literature, to a certain extent, and are also helpful in clarifying the controversy in the existing literature. Finally, this paper demonstrates that enterprises with a poor performance use charitable donations to hide their tax avoidance activities, which has important policy implications for enhancing the efficiency of tax inspection and improving the level of social welfare.

6. Conclusions, Implications and Limitations

6.1. Conclusions

Corporate social responsibility and tax avoidance have always been two important areas of academic research, but there are relatively few studies directly addressing the relationship between them, and the research conclusions are inconsistent. Based on the data of Chinese A-share listed enterprises, this paper studied the relationship between corporate charitable donations, earnings performance and tax avoidance. This paper found that there is a significant negative relationship between corporate charitable donations and tax avoidance. Furthermore, we found that the negative relationship between charitable donations and tax avoidance only exists in enterprises with a good earnings performance, while it is positively correlated with tax avoidance in enterprises with a poor earnings performance. This shows that the earnings performance affects the motivation for corporate charitable donations, and the charitable donations of enterprises with a good performance are mainly an altruistic behavior, while the charitable donations of enterprises with a poor performance are more of a management tool.

6.2. Implications

This study has the following implications. First, in theory, the research reported in this paper helps to reconcile the controversy regarding the existing research, that is, the relationship between corporate social responsibility and tax avoidance is not static. It is affected by the availability of corporate resources, a finding which also echoes the research of Lin et al. (2017) to a certain extent. Second, in the context of increasingly severe tax avoidance in emerging markets such as China, this paper found that enterprises with a poor performance are motivated to use charitable donations for tax avoidance, while enterprises with a good performance avoid tax less. This shows that it is necessary for the tax department to carry out additional tax inspections of enterprises with a poor performance who claim to be engaged in substantive corporate social responsibility activities, which could reduce the input of inspection resources for enterprises with a good performance and social responsibility so as to improve the efficiency of tax inspection and promote fair tax payment.

6.3. Limitations

The limitations of this study include the following. Firstly, previous research has not reached a general agreement about the most appropriate measure of corporate tax avoidance, and this paper only uses two indicators to measure corporate tax avoidance, which may lead to some differences between the conclusions and the real situation. In our future research, we will attempt to use an increased number of reasonable indicators to measure corporate tax avoidance, thereby increasing the reliability of the conclusions. Secondly, although this study included the relevant endogenous test, we cannot completely rule out the impact of endogenous problems. In future research, if we can find a perfect exogenous shock to construct a difference-in-differences model, this will undoubtedly solve the endogenous problem perfectly.