A self-interested gesture? Corporate charitable giving in response to government fiscal pressure

2.1 Corporate charitable giving in China

In this subsection, we briefly introduce the institution background of CCG in China. According to the China Philanthropy Development Report, CCG demonstrates the following features (Chen et al., 2015):

Private donations mainly come from corporations in China, opposed to the case in many developed counties where individuals contribute most to philanthropy [1]. CCG takes different forms, either cash or in-kind benefits.

A large share of donations goes to government fiscal revenue, public charities and government-backed non-governmental organizations (NGOs). Donations are mainly used in four public areas: education, environment protection, poverty alleviation and disaster relief. Corporate donations play a crucial role in assisting the government provide public services.

Corporations can donate through many channels. They can donate directly to the government, public charities or NGOs. In practice, most public charities are supported and regulated by the government. Even for the NGOs, the government has great influence on their operations, especially those large and national organizations. The donors can donate through local charities or through grassroots organizations, the proportion of donations which goes to these two channels is actually very small. In practice, there are three potential channels for firms to relieve local fiscal pressure through charitable donations: First, some of the funds are directly donated to the local governments or local charitable organizations, and then play a role in the fiscal expenditures that fulfill the social responsibility of local governments. Second, although some of the funds are directly donated to national charitable organizations, they are substantially allocated to local social affairs through special purpose forms with certain conditions attached, which are essentially the same as the former. Third, there are also some funds donated directly to national charities, and there is no conditional use specified. In the event of a need for charitable organizations to grant assistance, places with weaker fiscal resources will also receive more allocations. Therefore, from the above three ways, the charitable donations of firms will directly or indirectly affect and transmit to the fiscal pressure of the local government.

The government ultimately determines the usage of most donation money.

2.2 The tax revenue sharing system in China

Government fiscal pressure at city-level has formed since China's 1994 tax revenue-sharing reform. The reform is essentially a process of tax revenue centralization. Value-added tax (VAT), corporate income tax (CIT), individual income tax (IIT) are shared taxes, and tax revenues are shared by the central, provincial and city governments. The business tax (BT) belongs to sub-national tax, and tax revenue is only shared between the provincial and the city governments.

The central government shares constant proportions in revenues of the three major taxes (Wang & Herd, 2013). The revenue sharing rate of VAT for the central government is 75%, while sub-national governments (including both provinces and cities) take the rest 25%. With regards to CIT and IIT, the central government takes 60% of the revenues, while the rest 40% belongs to sub-national governments [2]. BT is a pure local tax; the central government does not share the revenue.

How the provincial government assigns the remaining tax revenues is unclear. In fact, the central government grants the provincial government discretionary powers to determine how to allocate the remaining tax income with cities (Liu, Martinez-Vazquez, & Qiao, 2014). Each province can make its own tax revenue sharing arrangement. All cities within the province should follow the rules with few exceptions. Since tax revenue-sharing rules are determined, the information will be delivered through several channels, such as TV, newspapers, Internet, etc. Under this circumstance, firms are informed that local governments become constrained in terms of fiscal revenue, while local governments would ask firms to apportion the sudden increased fiscal burden.

Provincial governments can change tax revenue sharing arrangement without much difficulty if they consider it a necessity. In a highly centralized political system, cities are not powerful enough to affect upper-level governments' decision-making. As a result, an unexpected change made by the provincial government to increase its tax shares can be identified as an exogenous shock to cities' fiscal revenue pressure.

Table 1 demonstrates sub-national tax revenue sharing rates of the four taxes for sample provinces. We also display tax revenue sharing changes over sample period. We see considerable differences in tax revenue sharing arrangement both across province and over time. It provides strong evidence that provincial governments actually have discretion powers in determining tax sharing rules within its jurisdiction. At the end of our sample period, the average share of provincial government in VAT revenue is 4.44%, 9.31% for CIT revenue, 10.31% for IIT revenue and 15.77% for BT revenue. We also describe the distribution of sample provinces (cities, firms) that experience tax revenue sharing changes and those not. See more details in Appendix Table A1.

In comparison, there is far less centralization in expenditures than in revenues. City governments maintain most responsibilities in a variety of public areas, such as education, public health, urban development expenditure, etc. [3]. The large gap between fiscal revenue and expenditure leads to mounting local fiscal pressure (Martinez-Vazquez, Qiao, Wang, & Zou, 2014).

A caveat should be mentioned. Once provincial government makes an unexpected increase in its tax shares, sudden increased fiscal pressure would urge local governments to raise non-tax resources to finance public welfare issues. City governments have several non-tax financial resources, the sale of land-use rights comes on top of the non-tax revenue sources. However, governments have to take the cost of installing the road, water, electricity, etc., which means they do not always generate fiscal surplus from land sales revenue (Wang & Herd, 2013). Moreover, governments are constrained from making the use of land sales revenue freely. The majority of the revenue is used for intended purposes, such as the construction of physical assets (Lu & Sun, 2013). However, we include land sales revenue into regressions where we can isolate the potential impact of land sales revenues on CCG.

Overall, corporate donations play an irreplaceable role in relieving government fiscal pressure. On the one hand, governments pay little compensation to use corporate donations. On the other hand, corporate donation finances major public welfare issues.

3.1 The relationship between government fiscal pressure and corporate charitable giving

In the context of China's tax revenue sharing system, local governments obtain limited financial resources from tax revenue. Because local officials are promoted on local economic performance, they finance capital construction projects in priority but allocate fiscal revenue to public welfare areas with a relatively small weight (Li & Zhou, 2005; Jia, Guo, & Zhang, 2014).

On the other side, local officials are held accountable for maintaining social stability as they seek for career promotion (Chan and Gao, 2009). To finance public welfare business, local governments expect to collect additional funding from other channels. CCG involves considerable contributions to social welfare services. Governments reach out to firms for charitable donations, and firms can help reduce governmental burdens in social areas through charitable giving (Dickson, 2003; Wang & Qian, 2011). To summarize, local governments demand more private donations to meet the ongoing commitments in financing welfare services, especially in fiscal hardship.

Several studies provide evidence that private donations serve the role as a good substitute for public expenditure. Melkote (2015) empirically examined the impact of the Great Recession on charitable giving. He found that donors contributed more during economic hardship with more concerns on the poverty and unemployment reduction. Garrett and Rhine (2009) found a negative relationship between charitable contributions and government spending. Roberts (1984), Bergstrom et al. (1986), Payne (1998), Payne (2001), Andreoni and Payne (2011) found the crowd-out effect where increasing (decreasing) government-provided grants reduced (increased) private donations.

Our next concern is why for-profit firms engage in charity business. Several studies highlight the importance of corporate charity to firms' future financial performance. Another strand of literature suggests governments offer substantial incentives to encourage private donations. For instance, governments grant firms the access to tax preferential treatment where firms can deduct donations from tax liability (Carroll & Joulfaian, 2005; Webb, 1996; Gautier & Pache, 2015; Duquette, 2016). Collectively, these studies suggest that CCG acts as a favored strategy choice.

Hence, we give our first hypothesis.

3.2 Underlying mechanisms

In the following subsections, we discuss two motivations that drive firms to participate in philanthropic activities, i.e. tax evasion and building political connection.

Tougher tax enforcement largely offsets tax revenue loss of the government. As a result, corporate ETR is enhanced to raise enough fiscal revenue during fiscal hardship. According to China Corporate Income Tax Law, charitable donations not exceeding 12% of pre-tax accounting income can be fully deducted from the tax liability. Transferring charity giving to fiscal hardship brings firms more tax savings when the ETR climbs up.

Furthermore, local government has a great influence on the tax enforcement that the firm may face. Usually, local governments are more likely to provide tax incentives and shelters to firms that implement social responsibilities. Firms signal the government that they have good financial performance and a strong sense of social responsibilities by donating in fiscal pressure periods (Chen et al., 2015). In return, local governments would reduce the frequencies of tax inspection while providing tax credits to these firms. In a word, firms enjoy lower effective tax rate than those which do not engage in philanthropic activities, resulted from a better mutual relationship with local governments through charitable contributions.

As discussed above, firms have a lower effective tax rate once they form a political connectedness through charitable giving. This implies that political connection plays an important role in the firm's development, especially in transitional economies that are featured with extensive government interference. Governments control key economic resource allocations in transitional countries like China, such as preferential tax treatments (Wu, Wu, Zhou, & Wu, 2012), bank loans (Cull et al., 2015), land usage (Liu & Alm, 2016), trade protection, etc. Firms are willing to establish a closer relationship with the government by engaging in charitable activities, which facilitates them to gain critical economic resources (Leuz & Oberholzer-Gee, 2006; Bai, Lu, & Tao, 2006a, b; Faccio, 2006; Li, Meng, & Zhang, 2006).

Increasing donations in fiscal pressure periods makes it easier for firms to cultivate political connections with local governments (Goldman, Rocholl, & So, 2013). Firms are easier to please local officials when the government needs external financial resources. In other words, firms build stronger political connectedness during fiscal hardship because the government attaches greater importance to outside financial sources when they need assistance. Therefore, if the relation between CCG and government fiscal pressure is due to building political networks, we suppose firms would build stronger political connectedness since they donate in fiscal pressure periods.

We then develop our third hypothesis as follows:

3.3 Corporate charitable giving and firm performance

Prior studies suggest that political connections affect firms' performance (Li & Zhou, 2005), we then investigate the following argument: do firms that increase donation in government fiscal pressure periods outperform those that do not. Therefore, we develop the fourth hypothesis:

3.4 Heterogeneous effect

The extent to which firms benefit from engaging in philanthropic activities varies. If H3 holds, firms that desire more political connections will spend more on charity business. Prior studies pointed out that three kinds of firms are more eager to build political connections, i.e. non-SOEs, small firms and firms in regions where institutions are weak. SOEs are controlled by governments, and they are naturally closer to governments, while non-SOEs are privately owned and they seek for more political protection. Similarly, small firms are prone to build bond with governments as they are in greater need for political protection. Finally, institutional quality represents the extent of government interference (e.g. corruption) as well as property rights protection. Firms in weak institution regions have to invest more relationship-based transactions, therefore, are more likely to engage in charity programs to bond with governments.

Finally, we propose our fifth hypothesis as follows:

4.1 Data and sample

Our sample consists of publicly A-shared corporations listed on the Shanghai and Shenzhen Stock Exchanges. We obtain the corporate financial data, corporate governance data as well as its charitable giving information from the CSMAR database.

The database contains more than 2,000 listed firms and spans 31 provinces or province-equivalent municipalities. It covers all kinds of industries, ensuring its national representativeness. We start data collection from year 2003 when all the firms are required by the Ministry of Finance to disclose charitable donation information. The data on charitable giving, displayed as one item of “non-operating expenses”, are hand collected from the footnotes of firms' financial statements. Each firm is matched to its host city by corporate headquarter location.

To make firms more comparable, we implement several data filters. First, we exclude firms that are specially treated, namely, the firms with “ST” (specially treated) hats or “ST*” hats. Second, we exclude firms if headquarter locations are not in accordance with corporate registration locations. Third, we exclude firms if headquarter locales have been changed during the sample period. Fourth, we exclude firms if they issue B shares or H shares. Fifth, we exclude firms located in Sichuan in 2008 and in Qinghai in 2011. These two provinces suffered from catastrophic earthquakes during sample period. Firms contributing donations in these two provinces are mostly driven by altruistic motivations or public media pressures. Seventh, we exclude firms from Xizang autonomous region due to poor data quality. We finally have 14,168 firm-year observations.

In addition, city-level economic and financial data are provided by China City Statistical Yearbook (2004–2013).

4.2 Empirical measures – corporate charitable giving

We use two measures of CCG. One is the total amount of CCG, and we use this measure in our baseline regressions. The second measure is the amount of corporate public welfare donations (CPWD), and we use this measure in the robustness check. According to the tax credits for giving, we define CPWD as charitable donations that firms donate through authorized charitable social organizations or directly to county-level (or above) government. Corporate public welfare donations can be fully deductible from the tax liability since the amount of CPWD does not exceed 12% of pre-tax accounting income. We find that the amount of corporate public welfare donations is equal to that of CCG for most sample firms, implying that firms may donate to achieve intended tax obligations.

4.3 Firm characteristics, city-level variables and other relevant measurements

We include a set of firm-level controlling variables that previous studies use as factors that affect firms' donating decisions. These variables are SIZE (measured by total assets, in logs), EMPLOYMENT (measured by the number of employees, in logs), ROA (measured by the ratio of net profits to assets), LEVERAGE (measured as firm's debt over shareholders' equity), COMPETITIVENESS (measured by the squared term of the ratio between firm output and the total industry output) and SSM (the percentage of stock shares held by the top ten executives).

We also include a set of city-level controlling variables which are supposed to affect province's decision-making process as well as firms' CCG decisions. They are GROWTH (gross domestic product [GDP] growth measured over the previous year), AGDP (GDP divided by the population), DENSITY (population divided by the jurisdictional land area), INDUSTRY (measured by the ratio of added value of the manufacturing industry to GDP), FE (measured by fiscal expenditures per capita, scaled by fiscal revenues per capita) and LSR (measured by the amount of land sales revenue).

To test potential mechanisms, we construct another two explanatory variables. We describe them as follows.

4.4 Statistical summary

Table 2 reports summary statistics of CCG from various perspectives. As shown, the median value of CCG of sample firms is 0, while the mean value is 1,292,155 yuan. The result implies that at least half of sample firms do not donate; however, firms involved in charitable activities donate a considerable amount.

CCG varies considerably across regions. We find that, on average, firms in Eastern China have the highest level of donating (1,601,726 yuan), whereas firms in Middle China have the lowest level (680,653 yuan). The distribution of CCG also demonstrates variation across industries. The wholesale and retail industry reports the highest level of corporate donating (1,767,151 yuan), whereas the transport and storage industry report the lowest level (633,556 yuan). These results suggest that there are both industry-level and regional-level effects to control in estimating the effect of government fiscal pressure on CCG.

Table 3 reports the descriptive statistics of our controlling variables. The first panel presents the descriptive statistics of firm-level controlling variables. The sample corporations are very large, with average assets of 922.8 million yuan. Firms have approximately 3,723 employers on average, and most of them adopt modest financing strategies (the mean value of the leverage ratio is 0.462). The firm profitability is reasonable (3.85%), and the industry is low-concentrated (11.7%). The top ten executives hold 4.86% of corporations' stocks, suggesting the ownership of sample firms is dispersed. The effective corporate tax rate (ETR) is 2.76%, while the average political connection score is 5.02.

The second panel presents the descriptive statistics of city-level controlling variables. The annual GDP growth rate is 12.28%, and the manufacturing industry plays a key role in prompting local economic growth. The mean value of the fiscal gap indicator is 1.8, which implies that fiscal expenditure exceeding revenues by roughly 80%. Land sales revenue contributes a considerable amount to local government revenue with about five million yuan on average.

Subsequently, our sample firms are split into two groups. The (4)–(3) difference column indicates that significant differences exist between donating firms and non-donating firms. Firms involved in charitable contributions tend to be larger with respect to the employment scale and more profitable. Donating firms are more aggressive in financing policies, more diversified in the ownership and closer to governments. We also find that firms located in more developed and industrialized areas donate more than their counterparts.

4.5 Empirical framework

We begin our empirical analysis by estimating the following model with ordinary least square (OLS) regressions:

To test the robustness of our results, we also focus on whether firms are responsive to a marginal change in local government fiscal pressure. We thus exclude firms locating in provinces that shared cities' tax revenue but never changed the tax revenue sharing arrangement. This modification makes our approach more akin to difference-in-differences (DID) design. Firms that experience tax revenue sharing arrangement changes serves as the “treatment” group, while firms with the provincial government never sharing tax revenue serves as the “control” group. In the baseline result, we took the regions where the provincial government's participation in the city's tax share changed during the sample period as the treated group, and the other regions with no change in the share status were used as the control group. At the meanwhile, the control group contains two parts: one is the sample that the provincial government never shares from the tax revenue of the city government, and the other is the sample that the provincial government shares from the tax revenue of the city government, but never changed the tax revenue sharing arrangement. In fact, the former one is the perfect control group; while the latter one also means that the city has “wasted” a part of the tax revenue and lead to fiscal pressure to some extent. Based on this point, this paper excluded firms locating in provinces that shared cities' tax revenue but never changed the tax revenue sharing arrangement from control group as a robustness check.

We assume CCG responsiveness is a short-term effect, because city governments are very likely to exploit other channels to relieve fiscal pressure in the long term. Therefore, the analysis window is restricted to only contain years before the tax revenue sharing change event to three years after. However, we also run the regressions with modifications to the analysis event window, extending it to 4 years, 5 years, 6 years, 7 years and 8 years, respectively. We will discuss these in robustness checks.

5.1 Results for H1

We first examine whether the government fiscal pressure imposed an impact on CCG. The results are reported in Table 4.

In columns 1 and 6, we estimate the baseline regression considering the overall tax revenue sharing status. From columns 2 to 5, we examine the impact of the government fiscal pressure on CCG as the province suddenly shares the revenue of a specific tax. From columns 7 to 10, we estimate the marginal responsiveness effect of the revenue sharing change of a specific tax.

We find the key variable of our interest has a positive and statistically significant effect on CCG in most regressions. Our results also have economic significance. On average, firms increase charitable donations by 136,773 yuan if the provincial government shares the revenue of an additional tax.

Furthermore, we find the magnitude of the coefficient of our key variable is greatest for VAT tax revenue sharing change. The result is plausible as VAT is the largest tax in China. Firms increase donations by 461,296 yuan if the provincial government shares the VAT tax revenue of an additional tax. With regards to BT and IIT, the government fiscal pressure effect holds but the magnitude of the effect is relatively smaller. With respect to CIT, the effect is not statistically significant, while the sign of the coefficient of our key variable is positive.

Finally, we find the impact is more pronounced in terms of both magnitude and statistical significance where we estimate CCG response to a marginal increase in government fiscal pressure. Our result is consistent with H1.

5.2 Results for H2 and H3

In this subsection, we investigate motivations that drive firms to increase charitable giving in government fiscal pressure periods. According to H2, we propose that corporate effective tax rate increases when government fiscal pressure rises, and then firms are more likely to reduce effective tax rates if they donate. We construct Eq. (2) to test whether firms reduce tax burden if they donate in government fiscal pressure periods.

Similarly, as we discuss in H3, firms are more likely to build political connectedness once they donate to relieve government fiscal pressure responsively. Therefore, we construct Eq. (3) to test whether firms are rewarded in terms of political connection.

Here we also consider both the overall tax revenue sharing status and the tax revenue sharing change of a single tax. If H3 holds, we suppose c₂ to be positive. Results are reported in Table 5.

Columns (1)–(5) report estimations of Eq. (2). In all regressions, coefficients of the binary variable to measure tax revenue sharing change are statistically positive. It suggests that corporate effective tax rate increases due to tougher government fiscal pressure. On the other hand, coefficients of the interaction term (∑Dpt−1,n(n=1,2,3,4)×Donicpt) are statistically negative. Donating firms are more capable to reduce corporate tax burden. Results confirm tax evasion as an important motivation to explain CCG behavior.

Columns (6)–(10) report estimations of Eq. (3). In all specifications, the coefficients of the interaction term (∑Dpt−1,n(n=1,2,3,4)×Donicpt) are statistically positive. The results confirm H3, suggesting that firms build stronger political connection once they donate in response to increased government fiscal pressure.

5.3 Results for H4

Our next concern is do firms that donate subsequent to tax revenue sharing changes tend to outperform firms that do not. We estimate Eq. (4) to test H4:

We use several measures to estimate the firm performance, including return on assets (ROA), return on sales (ROS), sale fees (SR), short-term loans (SL), debt-to-equity ratio (LEVERAGE). ROA and ROS are well-known accounting performance measures. The reduction of sale fees is treated as market reputation rewards for the firm's charitable giving. While several studies find that donating firms are more accessible to bank loans that are strictly regulated by the government in China. Estimation results of Eq.(4) are reported in Table 6.

Table 6 represents the regression results of Eq. (4). Results suggest that firms that donate subsequent to tax revenue sharing changes outperform firms that do not donate in terms of market reputation and bank loan access. However, we do not find donating firms outperform their counterparts with respect to the accounting performance. We conjecture that firms may need a longer period to translate the political connection and the tax shelter advantages into performance achievement.

5.4 Results for H5

According to H5, non-SOEs, small firms and firms in areas with heavy government interference are supposed to donate more in response to government fiscal pressure. We define SOEs as firms who are ultimately controlled by the government. Firms with assets below the mean value of assets of sample firms are categorized as small firms. And we classify firms into intensive government interference group on the basis of the marketization index value. The index is reported by China's National Economic Research Institution. Cities with the marketization score of the initial sample year lower than the mean value of sample cities of that year are categorized into intensive government interference group. As we divide the total sample into these sub-samples, we re-run Eq.(1) and results are reported in Table 7.

As shown, the positive impact of government fiscal pressure on CCG is more pronounced for small firms and firms facing heavy government interference, in terms of both statistical and economic significance.

Strikingly, we do not find any statistically significant impact of government fiscal pressure in SOEs and non-SOEs. We are concerned whether the results are caused by insufficient samples. Therefore, we include the interaction term (∑Dpt−1,n(n=1,2,3,4)×D_soe) into baseline regression and re-estimate. The D_soe variable equals 1 for SOEs and 0 for non-SOEs. We find that SOEs donate more than their counterparts in fiscal pressure periods. SOEs have to take on more social responsibilities when the government is unable to provide sufficient public services.

5.5 Robustness checks

In the final subsection, we conduct several robustness checks. We first change measures of CCG and offer an alternative method to gauge tax revenue sharing change. We then check whether sample restrictions would bias our main findings. Subsequently, we extend the analysis window to investigate both the short-term and the long-term effects of government fiscal pressure on charitable giving. Finally, we rule out the possibility of potential bias caused by reverse causality.