To what extent does tax avoidance
interrelate with corporate social
responsibility?
Stapel, F.
University of Groningen
Supervisor: Dr. J.H. von Eije
Article info Abstract
JEL classification:
Recent societal developments have shown that corporate behavior is becoming increasingly important for maintaining a sound business results. Two of the elements under review are a firm’s tax practices and their CSR involvement. Based on existing literature, I develop a model with which their interrelation can be examined. In addition, the interrelation between tax avoidance and three underlying pillars of CSR (social, environmental, and corporate governance) is also examined. I find that firms engaged in tax avoidance are less involved in CSR activities. This finding is consistent with existing literature. In addition, firms with higher social involvement are more likely to engage in tax avoidance. In contrast with the latter, firms with high levels of corporate governance are less likely to engage in tax avoidance practices. These results are, however, not robust among alternative measures of tax avoidance.
H26 K34 M14 M48 Keywords: Tax avoidance Tax aggressiveness Book-tax differences Corporate social responsibility Corporate culture
2 I. Introduction
In a time wherein the focus of MNE’s is shifting from not solely profit-enhancing strategies towards responsible profit generating strategies, firm behavior is
reviewed more comprehensively than before. Two of those reviewed elements which are continuously up for debate are CSR-activities and tax avoidance.
Starting with tax avoidance, in this sense two key defining moments can be pointed out which made the public increasingly attentive on this topic. One of these defining moments was when Margaret Hodge, chair of the public accounts committee in the United Kingdom, called Google and Amazon up to the stand to answer questions regarding their tax practices. Even though the representative of Google justified their actions on the basis of it being in line with the law. Margaret called it “"devious, calculated and, in my view, unethical".1 Another defining moment, or better said moments, was when an international consortium of journalists covered the Luxleaks, Offshoreleaks, Swissleaks, Panama papers and Paradise papers. The data leaks revealed concrete insights on how the rich shelter and shield their
money, not uncommonly in illegal manners. Upon every data leakage, the public was stirred up and called for action. These moments triggered and produced momentum for policy makers to design and enforce broader tax anti-abuse legislation.
Nowadays, at the time of writing this thesis (December 2017), the cooperative initiatives in the international landscape follow each other with rapid pace. The Organisation for Economic Cooperation and Development (“OECD”) and the European Union (“EU”) are mostly leading these initiatives and incentivize coherent action among countries.
When talking about tax avoidance, the extent to which tax planning is perceived as aggressive varies in both literature and media. In this paper, tax avoidance
practices refer to aggressive tax planning, tax minimizing strategies and (legal) tax sheltering. The latter denoted practices are all legal when bearing the law into mind. Tax avoidance therefore does not refer to tax evasion, which concerns illegal practices.
Tax avoidance practices are connected to a wide spectrum of matters in literature. For instance, Desai and Dharmapala (2006) studied the relationship between aggressive tax practices and high-powered incentives. They showed that the extent to which board members are allocated high-powered incentives negatively
correlated with the extent to which a firm is engaged in aggressive tax practices. Corporate social responsibility (“CSR”) has also proved to be a frequently reviewed element and it also became a major research object over recent years with the aim of obtaining insights for accounting and management purposes. For example, Kim
1https://www.theguardian.com/technology/2013/may/16/google-told-by-mp-you-do-do-evil (accessed
3 et al. (2012) study the relationship between a firm’s CSR practices and its earning’s quality. Moreover, Dhaliwal et al. (2011, 2012) examined the (i) relationship
between voluntary nonfinancial disclosure of CSR practices and the cost of equity and (ii) the relationship between voluntary nonfinancial disclosure of CSR practices and analyst forecast accuracy.
Topics related to the topics mentioned above are covered in literature in many occasions but on the relationship between CSR and tax avoidance practices is little known even though the connection is fairly natural. CSR represents the shared belief of a company, a common mentality of how the board and employees of a firm act when facing an issue. These issues can vary from how the firm should deal with increasing public pressure to contribute to a green future but also how to position the firm with regard to their tax practices. In other words, will the firm minimize its taxes paid or will the firm like to pay its ‘fair share’ in each country where it is operating.
Consistent with the above, gaining more insights on the relationship between CSR and tax avoidance practices could offer implicit insights on whether firms only engage in CSR because of window-dressing purposes or because they truly want to make a positive change. Not being engaged in tax avoidance practices as a MNE will increase the taxes a firm has to pay. This a perceived as a cost. If therefore a firm is engaged in positive CSR activities but also avoid taxes on a significant scale, this could imply that firms only engage in positive CSR activities if there is a
financial benefit to be obtained. Of course, this reasoning could also be the other way around.
This study will contribute to the latter debate by examining the relationship
between tax avoidance practices and CSR. The examined sample consists of 444 US-based firms. A model is developed by which this relationship is tested by using three empirical measures for tax avoidance practices, which are explained in Chapter III, and by using the CSR score for the 444 firms in the sample. These alternative tax avoidance measures also allow to test for robustness of the results among different measures of tax avoidance practices.
This investigation also extends to which pillars of CSR would signal tax avoidance practices more accurately. These pillars of CSR are social, environment and
corporate governance. The effect of the individual pillars on tax avoidance practices is covered in existing literature to different extents. For example, Changyuan et al. (2017) studied the effect of social trust on a firm’s tax practices. In addition, the relationship between corporate governance and tax avoidance practices is well-documented. For instance, Desai and Dharmapala (2009) examined the relationship between corporate tax avoidance and firm value. They showed that poorly-governed firms were less able to grasp the benefits of tax avoidance in comparison to well-governed firms than poorly-well-governed firms. This is in line with literature and
4 activities to engage in efficient tax avoidance structuring. In contrast to the pillars social and corporate governance, the relationship between the pillar environmental and tax avoidance practices is barely documented.
This thesis contributes to existing literature by examining whether the tax
avoidance practices of a firm interrelate with the degree of reported CSR of a firm. This relationship has been examined before but only fragmentary. The sample used in this thesis allowed for looking at the whole spectrum of CSR of a firm and not only at, for instance, how negative CSR activities interrelate with tax avoidance practices (i.e. Hoi et al., 2013). In addition, this paper also examined whether multiple parts of CSR (social, environmental and corporate governance) interrelate with tax avoidance practices. This in new to academic literature.
The results of this thesis can be a useful asset for policy makers when designing and enforcing policies to discourage firms to engage in tax avoidance practices. By obtaining more insights on the characteristics of firms that engage in tax avoidance practices, it allows for policy makers to better target the firms that policy makers want to target, i.e. firms that avoid taxes.
This paper proceeds as follows. Section II will present a literature review on the research conducted on both CSR activities and tax avoidance practices and their possible interrelation. Section III describes the research design and, in addition, presents the sample for the regression analysis. In Section IV the empirical results are discussed. Section V concludes and Section VI discuss the limitations of this study.
II. Literature review and hypothesis development
The public debate on tax avoidance practices of MNE’s has intensified over the past several years. This has also been noted in academic literature. Where MNE’s more and more include tax in their published corporate strategy, among other CSR subjects, governments take significant action as well. They target tax sheltering activities and tax avoidance practices at an increasing pace and in a more
cooperative manner, i.e. by cohering national tax systems in the international (i.e. mainly via the OECD or European) context.
As the perception of how a firm should manage its tax affairs changed over time, the relevance of CSR increased as well. Product superiority, innovations and profitability are the main points of interest for investors and consumers when perceiving a company as ‘good’ or ‘bad’. In recent years, corporate social
5 interferences about certain firms if the motives of those firms come across as self-serving.
In the proceedings of this chapter three subjects will be addressed. These serve to derive the hypotheses on how tax avoidance practices interrelate with CSR
activities. The reoccurring topic is which costs and benefits can be assigned to either tax avoidance practices or CSR activities, and after that these costs and benefits will be compared.
Cost and benefits of tax avoidance practices
When a firm decides whether or not to engage in tax avoidance practices, costs and benefits are weighted. It is obvious that tax avoidance is beneficial because it leads to tax savings. These saving accrue over time and lead to higher net profits,
therefore, directly or indirectly, to higher managerial compensating and higher dividend payouts to shareholders. Another quasi benefit of tax avoidance practices could be the manipulation of cash flows and accounting earnings. This can be done either by postponing costs or temporarily qualifying financing cash inflow as
operating cash inflow and therefore, misclassifying the quality of accounting earnings (Chen et al., 2010).
To get a better grasp of how tax avoidance may arise, some cases are presented. These cases will comprehend the check-the-box regulation for subsidiaries of US firms and IP (intellectual property) structuring. First, the check-the-box regulation. This regulation was designed by the US government to give small firms the option of which organization form they wanted to obtain for tax purposes, as explained by Desai and Dharmapala (2009). The check-the-box regulation, however, also offers US-based firms the opportunity to shelter their non-US profits using hybrid entities in a tax jurisdiction where little or no corporate income tax legislation is present or where profits are taxed at a very favorable tax rate. This offshore money is kept there and as long as this money is not paid out as a dividend, the payment of tax is (infinitively) postponed. Hybrid entities are entities which are perceived as
6 by another tax jurisdiction. This case of tax avoidance costs the US government somewhere around ten billion dollars in lost taxes per year.2
The second case of tax avoidance is the case of IP structuring. Most of the big tech companies nowadays derive a lot of their firm value to their brand name or unique technology. These intangible assets are called IP. By placing IP in low-taxed
jurisdictions (e.g. Bermuda or the Cayman Islands) firms can lower their worldwide taxable amount by licensing the IP to high-taxed jurisdiction. For the licensing a royalty payment is made from the high-tax jurisdiction to the low-taxed jurisdiction. This lowers the profit of a firm in a high-taxed jurisdiction and increases the profit in a low-taxed jurisdiction. In that way high-tax profits are converted to low-taxed profits, leading to a decrease in the worldwide taxable amount of firms. This case of structuring is seen as artificial, because of not reflecting commercial considerations, and denoted as tax avoidance.
There is also a possible cost-side of tax avoidance practices. By pushing the law to its very limits, one may cross the law inevitably at some point in time. The IRS (the US tax authority) will be able to penalize those firms and present them a fine. Graham et al. (2012) show, by providing survey evidence, that tax executives are sensitive to negative sanctions the firm potentially faces when deciding on whether or not to implement a tax avoidance strategy. Also, the transaction costs of
organizing the activities of a firm in the most tax efficient manner, which include time and effort spent by both employees and advisors, can be considered a cost of tax avoidance practices.
Lastly, as seen in recent history with Starbucks, Amazon and other global operating US firms, the reputational costs of tax avoidance practices could be present.
Nevertheless, if present, reputational cost will be higher for business-to-consumer companies than business-to-business companies. The degree of suffered
reputational costs as a result of negative public media coverage on the tax
avoidance practices of a firm are, however, still unknown as these mentioned cases only appeared recently.
2https://www.forbes.com/sites/taxanalysts/2014/02/19/check-the-box-for-tax-avoidance/#6d07d4a2756a
7 Cost and benefits of CSR
CSR as a concept finds its origin in a published work by Bowen (1950) by providing the first comprehensive discussion of social responsibility and business ethics. The definition of this concept evolved over the years. Carroll (1979) defines CSR as the whole set of “economic, legal, ethical and philanthropic expectations that society may have towards a company”. In accordance with Kreps (1990) and Hoi et al. (2013), for the purpose of this paper the corporate culture within a firm is viewed as a consensus on which beliefs will be practiced and about what course of action is acceptable and which course of action is not.
Various authors studied the effects on corporate performance of obtaining an efficient CSR strategy. Turban and Greening (1970) find that increased
sustainability performance could potentially lead to higher-quality employees. In addition, Waddock and Graves (1984) find that a firm can obtain better resources as a result of enhanced sustainability performance. Moreover, the relationship
between CSR and firm performance is studied intensively (e.g. Barnett and Salomon, 2006; Margolis et al., 2009; Waddock and Graves, 1997; Eccles et al., 2015). Meta-analysis shows a small but significant positive relationship between CSR and firm performance.
On the downside, engaging in positive CSR activities costs money. It is argued that sustainability investments for instance could disproportionally increase firm’s cost without actual gain (e.g. McWilliams and Siegel, 2000; Orlitzky et al., 2003). Different reasons are mentioned that may explain this. First of all, managers can derive private benefits from making sustainability investments and therefore making inefficient investment decisions (Cheng, Hong and Shue, 2014). The second reason, as described by Di Giuli and Kostovetsky (2014), could be that political beliefs of managers incentivize them in making sustainability investments. This result was found by comparing red (conservative) and blue (democratic) companies in the US while looking at their degree of sustainability investments.
Tax avoidance as an extension of CSR from a corporate culture perspective.
8 beliefs in a firm which denote the ‘optimal’ course of action or ‘right’ behavior in a firm (Kreps, 1990; Hermalin, 2001). In defining an appropriate corporate culture, the interests of shareholders come first. The culture should facilitate those
shareholders by efficiently organizing coordination in the firm, by naturally negotiating and communicating and therefore saving costs. In addition, corporate culture could reduce transaction cost within a firm (Hermalin, 2001).
As denoted previously in this Section, in existing literature corporate culture is described as the whole set of “economic, legal, ethical and philanthropic
expectations that society may have towards a company” (Carroll, 1979). In that sense, all corporate decisions are systematically affecting by the corporate culture within a firm. CSR influences corporate practices and therefore affects the
stakeholders of a firm. As such, the choice of whether or not to engage in tax avoidance practices is also affected by corporate culture. To elaborate, the
government’s tax revenue and the society’s welfare are affected by the decisions of a firm with regard to their tax practices.
Although this relationship between corporate culture and therefore CSR and tax practices is clearly present, CSR is seldom noted in conjunction with tax avoidance practices in literature. The tax practices of a firm are a fundamental way in which a firm is engaged with society (Hoi et al., 2013). By engaging in tax avoidance
practices as a firm, a firm acts opportunistic and neglects the social contract with society which implicitly states that everybody, as well as every firm, should pay its fair share of tax. For a firm to be activity committed to CSR, it acts contrarily. Namely, it acts as a good servant of society. It naturally follows that tax avoidance practices are negatively related to, and therefore inconsistent, with CSR activities. This implies the following hypothesis:
A. Tax avoidance practices are negatively related to the degree of CSR activities of a firm.
In addition, in this thesis is examined whether particular pillars of CSR are more related to tax avoidance practices than other. These three pillars are social, environmental and corporate governance.
9 more correlated to tax avoidance than other pillars. Nevertheless, several authors managed to capture some of these features in describing corporate behavior.
Xia et al. (2017) examined whether the social trust a society has in firm affects the extent to which a firm is engaged in tax avoidance practices in China. They find that firms with higher levels of social trust are less likely to engage in tax avoidance practices. In addition, several authors document that higher levels of social
awareness enhance the self-perceived trustworthiness of managers and they are, as a result, less likely to engage in tax avoidance practices (Jha & Chen, 2015; Guiso, Sapienza & Zingales, 2004). This implies the following hypothesis:
B. Tax avoidance practices are negatively related to the degree of social awareness (as a pillar of CSR) of a firm.
The second pillar regards to the environmental aspect of CSR. The relationship between environmental practices and the tax avoidance practices of a firm is unnoted in literature. The relationship between environmental practices and tax practices also seems the least natural one of all three pillars. No hypothesis is derived for this relationship.
The third pillar of CSR under review in this thesis is corporate governance. Several authors reported on the interrelation between tax avoidance practices and a specific feature of corporate governance. For instance, Khurana and Moser (2012) examined the relationship between the degree of institutional ownership and the degree of tax avoidance of a firm. They find that institutional ownership mitigates the extent to which a firm is engaged in tax avoidance practices. In addition, Desai and
Dharmapala (2006) examined if properly incentivized managers will be less engaged in tax avoidance. They find that this is the case as it reduces the agency problem and therefore, these managers are less incentivized to increase their bonuses by engaging in tax avoidance practices. Several other authors studied the relationship between particular features of corporate governance and the degree of tax avoidance practices and their conclusions were consistent with the above (Chen et al., 2010; Desai and Dharmapala, 2008). This above implies the following
10 C. Tax avoidance practices are negatively related to the degree of corporate
governance (as a pillar of CSR) of a firm.
III. Research design and sample III.1 Research design
Measures of aggressive tax practices
To estimate the extent to which companies are engaged in tax avoidance practices, a proxy is needed, since capturing tax avoidance practices is not straight-forward and it can manifest itself in various ways. Following existing literature, three measures of tax avoidance practices are obtained. One measure is the default measure. In this sense, the results derived for the two other, alternative, measures can be used as a robustness test on whether the results are robustness among the alternative
measures.
Two of the measures to capture tax avoidance are adjusted book-tax measures and a cash effective tax rate measure. The default measure to capture tax avoidance will be the permanent book-tax difference for firm i in year t, which is introduced by Frank et al. (2009). The measure by Frank et al. is denoted as 𝐷𝑇𝐴𝑋𝑖𝑡. This measure is therefore described in depth here. For a detailed description of the two other measures, which are briefly introduced below, I refer to Appendix A.
The degree of tax avoidance practices for firm i in year t, 𝐷𝑇𝐴𝑋𝑖𝑡, is 𝜀𝑖𝑡 from the following regression:
𝑃𝐸𝑅𝑀𝐷𝐼𝐹𝐹𝑖𝑡 = 𝛽0+ 𝛽1𝐼𝑁𝑇𝐴𝑁𝐺𝑖𝑡−1 + 𝛽2𝑈𝑁𝐶𝑂𝑁𝑖𝑡−1+ 𝛽3𝑀𝐼 𝑖𝑡−1+ 𝛽4𝐶𝑆𝑇𝐸𝑖𝑡−1 + 𝛽5∆𝑁𝑂𝐿𝑖𝑡−1+ 𝛽6𝐿𝐴𝐺𝑃𝐸𝑅𝑀𝑖𝑡−1+ 𝜀𝑖𝑡
Whereby Hoi et al. (2013) define (the codes between the parentheses represent the codes used in the Standard & Poor Compustat database):
𝑃𝐸𝑅𝑀𝐷𝐼𝐹𝐹𝑖𝑡 = 𝐵𝐼𝑖𝑡− [(𝐶𝐹𝑇𝐸𝑖𝑡+ 𝐶𝐹𝑂𝑅𝑖𝑡)/𝑆𝑇𝑅𝑖𝑡] − (𝐷𝑇𝐸𝑖𝑡/𝑆𝑇𝑅 𝑖𝑡); and • 𝐵𝐼𝑖𝑡 represents the pre-tax book income (PI) for firm i in year t;
• 𝐶𝐹𝑇𝐸𝑖𝑡 represents the current federal tax expense (TXFED) for firm i in year t; • 𝐶𝐹𝑂𝑅𝑖𝑡 represents the current foreign tax expense (TXFO) for firm i in year t; • 𝐷𝑇𝐸𝑖𝑡 represents the deferred tax expense (TXDI) for firm i in year t;
• 𝑆𝑇𝑅𝑖𝑡 represents the statutory tax rate in year t;
• 𝐼𝑁𝑇𝐴𝑁𝐺𝑖𝑡−1 represents the goodwill and other intangibles (INTAN) for firm i in year t;
• 𝑈𝑁𝐶𝑂𝑁𝑖𝑡−1 represents the income (loss) reported under the equity method (ESUB) for firm i in year t;
11 • 𝐶𝑆𝑇𝐸𝑖𝑡−1 represents the current state income tax expense (TXS) for firm i in year
t;
• ∆𝑁𝑂𝐿𝑖𝑡−1 represents the change in net operating loss carryforwards (TLCF) for firm i in year t;
• 𝐿𝐴𝐺𝑃𝐸𝑅𝑀𝑖𝑡−1 represents the one-year lagged PERMDIFF for firm i in year t. For missing values the methodology of Frank et al. (2009) is followed. This
methodology mainly encompasses assigning other proxies for the estimates needed to obtain an estimate on 𝜀𝑖𝑡, which is 𝐷𝑇𝐴𝑋𝑖𝑡, the default measure of tax avoidance practices in this paper. The 𝜀𝑖𝑡 is obtained by the above stated OLS regression. The second measure concerns the discretionary book-tax difference for firm i in year t as described by Desai and Dharmapala (2006) which is denoted as 𝐷𝐷_𝐵𝑇𝑖𝑡. This measure also captures the residual of an OLS regression. For this measure a regression is run in which the book-tax difference is explained by total accruals (total accruals is defined as the sum of discretionary accruals and non-discretionary accruals) and a residual. This residual captures the extent to which a company is engaged in tax avoidance practices. For a detailed description of the 𝐷𝐷_𝐵𝑇𝑖𝑡 measure, I refer to Appendix A.
The third measure to capture tax avoidance practices is more basic and straight-forward measure, namely the cash effective tax rate (𝐶𝐸𝑇𝑅𝑖𝑡). This measure
encompasses the cash effective tax rate for firm i in year t. The measure is defined as the cash tax paid divided by pre-tax book income less special items. In contrast with prior studies, I chose not to truncate the 𝐶𝐸𝑇𝑅𝑖𝑡 to the range of 0 to 1. This gives a better representation of real-life tax affairs because of non-deductible costs, the cash effective tax rate could rise above one or because of preferential regimes (e.g. investment subsidies) the tax rate could be below zero. This, on the other hand, results in a higher standard deviation for 𝐶𝐸𝑇𝑅𝑖𝑡. To arrive at a measure for tax avoidance practices, the cash effective tax rate is subtracted from the statutory tax rate in the US, which is 35%.3 This brings consistently for the interpretation the regression results. Now all three measures with report a higher number if the
degree to which a firm is engaged in tax avoidance practices rises. Without this data alteration, the score on 𝐶𝐸𝑇𝑅𝑖𝑡 would yield a opposite result.
Measuring corporate social responsibility
Corporate social responsibility ratings, i.e. ESG scores, are used to measure corporate activities which are widely recognized as having impact on both
shareholders and non-shareholder activities. These ratings are derived from the ASSET4 ESG database. This database contains firm-year data on 34 binary scores. In this case, firm-year data for US-based firms is collected. The 34 binary scores can be divided into seven categories, which are employee relations, community,
corporate governance, product quality and safety, environment, human rights and diversity. For this paper, the average score of these categories, 𝐸𝑆𝐺_𝑆𝑐𝑜𝑟𝑒𝑖𝑡−1, is
12 used as a variable for the first regression model. In addition, the individual score for the three main pillars of CSR, i.e. social, environmental and corporate governance, are also used to run a second regression to estimate whether a certain pillar of CSR affects the extent to which a firm is engaged in tax avoidance practices differently compared to the other pillars.
These variables are 𝐸𝑁𝐶𝑖𝑡−1, 𝐸𝑁𝑉𝑖𝑡−1 and 𝐺𝑂𝑉𝑖𝑡−1. The variable 𝐸𝑁𝐶𝑖𝑡−1 gives the score on the pillar social, i.e. the total number of these scores for firm i in year t. The variable 𝐸𝑁𝑉𝑖𝑡−1 gives the score on the pillar environment, i.e. the total number of these scores for firm i, in year t. The variable 𝐺𝑂𝑉𝑖𝑡−1 gives the score on the pillar governance, i.e. the total number of these scores for firm i in year t.
Regression models
Following Hoi et al. (2013), Manzon and Plesko (2002), and Chen et al. (2010), the following two regression model are used to test the hypotheses:
A. Tax avoidance practices are negatively related to the degree of CSR activities of a firm. 𝐴𝐺𝐺𝑅𝐸𝑆𝑆𝐼𝑉𝐸𝑖𝑡 = 𝛽0+ 𝛽1𝐸𝑆𝐺_𝑆𝑐𝑜𝑟𝑒𝑖𝑡−1 + 𝛽2𝐶𝐴𝑆𝐻𝑖𝑡−1+ 𝛽3𝑅𝑂𝐴 𝑖𝑡−1+ 𝛽4𝐿𝐸𝑉𝑖𝑡−1 + 𝛽5𝑁𝑂𝐿𝑖𝑡−1+ 𝛽6∆𝑁𝑂𝐿𝑖𝑡−1+ 𝛽7𝐹𝐼𝑖𝑡−1+ 𝛽8𝑃𝑃𝐸𝑖𝑡−1+ 𝛽9𝐼𝑁𝑇𝐴𝑁𝐺𝑖𝑡−1 + 𝛽10𝐸𝑄𝐼𝑁𝐶𝑖𝑡−1+ 𝛽11𝑅&𝐷𝑖𝑡−1+ 𝛽12𝐸𝑀𝑃𝑖𝑡−1+ 𝛽13∆𝑆𝑎𝑙𝑒𝑖𝑡−1 + 𝛽14𝑆𝑖𝑧𝑒𝑖𝑡−1+ 𝛽15𝑀𝐵𝑖𝑡−1+ 𝛽16𝐿𝑎𝑔𝑖𝑡−1 (𝑑𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑡 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒) + 𝑌𝑒𝑎𝑟 𝑑𝑢𝑚𝑚𝑖𝑒𝑠 + 𝜀𝑖𝑡
B. The degree of social awareness (as a pillar of CSR) of a firm is negatively related to the tax avoidance practices of a company.
C. The degree of corporate governance (as a pillar of CSR) of a firm is negatively related to the tax avoidance practices of a company.
𝐴𝐺𝐺𝑅𝐸𝑆𝑆𝐼𝑉𝐸𝑖𝑡 = 𝛽0+ 𝛽1𝐸𝑁𝐶𝑖𝑡−1+ 𝛽2𝐸𝑁𝑉𝑖𝑡−1+ 𝛽3𝐺𝑂𝑉𝑖𝑡−1 + 𝛽4𝐶𝐴𝑆𝐻𝑖𝑡−1+ 𝛽5𝑅𝑂𝐴 𝑖𝑡−1 + 𝛽6𝐿𝐸𝑉𝑖𝑡−1+ 𝛽7𝑁𝑂𝐿𝑖𝑡−1+ 𝛽8∆𝑁𝑂𝐿𝑖𝑡−1+ 𝛽9𝐹𝐼𝑖𝑡−1+ 𝛽10𝑃𝑃𝐸𝑖𝑡−1 + 𝛽11𝐼𝑁𝑇𝐴𝑁𝐺𝑖𝑡−1+ 𝛽12𝐸𝑄𝐼𝑁𝐶𝑖𝑡−1+ 𝛽13𝑅&𝐷𝑖𝑡−1+ 𝛽14𝐸𝑀𝑃𝑖𝑡−1 + 𝛽15∆𝑆𝑎𝑙𝑒𝑖𝑡−1+ 𝛽16𝑆𝑖𝑧𝑒𝑖𝑡−1+ 𝛽17𝑀𝐵𝑖𝑡−1 + 𝛽18𝐿𝑎𝑔𝑖𝑡−1 (𝑑𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑡 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒) + 𝑌𝑒𝑎𝑟 𝑑𝑢𝑚𝑚𝑖𝑒𝑠 + 𝜀𝑖𝑡
where for both regression models 𝐴𝐺𝐺𝑅𝐸𝑆𝑆𝐼𝑉𝐸𝑖𝑡 represent the three tax avoidance measures as described above. The control variables 𝐶𝐴𝑆𝐻𝑖𝑡−1, 𝑅𝑂𝐴 𝑖𝑡−1, 𝐿𝐸𝑉𝑖𝑡−1, 𝑁𝑂𝐿𝑖𝑡−1, ∆𝑁𝑂𝐿𝑖𝑡−1, 𝐹𝐼𝑖𝑡−1, 𝑃𝑃𝐸𝑖𝑡−1, 𝐼𝑁𝑇𝐴𝑁𝐺𝑖𝑡−1, 𝐸𝑄𝐼𝑁𝐶𝑖𝑡−1, 𝑅&𝐷𝑖𝑡−1, 𝐸𝑀𝑃𝑖𝑡−1, ∆𝑆𝑎𝑙𝑒𝑖𝑡−1, 𝑆𝑖𝑧𝑒𝑖𝑡−1 and 𝑀𝐵𝑖𝑡−1 will control for the effects of earnings quality, firm performance, and other firm characteristics. A description of the control variables can be found in Appendix A.
13 for the first regression model. I expect opposite signs if firms with a high score on their CSR activities are more likely to engage in tax avoidance practices.
With regard to the pillars, social, environmental and corporate governance, of CSR (𝐸𝑁𝐶𝑖𝑡−1, 𝐸𝑁𝑉𝑖𝑡−1, 𝐺𝑂𝑉𝑖𝑡−1), if one of those three pillars reports a high score and as a result a firm is less likely to engage in tax avoidance practices, I expect a negative coefficient on that particular variable, 𝛽1, 𝛽2 or 𝛽3, for the second regression model. To control for various other elements effecting the dependent variable, control variables are included. The control variables are derived from the previous literature on this topic. In the two regression models I control for firm
characteristics that based on existing literature correlate with tax avoidance
measures to ensure that the results are not primarily driven by differences between the characteristics of the firms in the sample.
The first set of control variables (𝐶𝐴𝑆𝐻𝑖𝑡−1, 𝑅𝑂𝐴 𝑖𝑡−1, 𝐿𝐸𝑉𝑖𝑡−1, 𝑁𝑂𝐿𝑖𝑡−1, ∆𝑁𝑂𝐿𝑖𝑡−1) captures the firm’s profitability, the leverage of a firm and the amount of foreign income. Firms with more profits and subsequently, less tax loss carry-forward have, in general, higher effective tax rates. Not controlling for those firm characteristics would affect the coefficients for the CSR estimate.
For example, Graham and Tucker (2006) constructed a sample of 44 corporate tax shelter cases to search for common firm characteristics of firms engaged in these corporate tax shelter cases. They find that, amongst others, size and profitability are common firm characteristics to those firms. Being a firm of significant size allows a firm to allocate activities to a greater number of jurisdictions which makes it easier to cherry-pick favorable tax legislation. Subsequently, profitability enables firms to profit from tax avoidance practices, which is less applicable when having zero or negative profits. The same holds for firms with growth (proxied by 𝑀𝐵𝑖𝑡−1) as these firms are able to make more investments in tax-favored assets that could generate time differences in recognition of expenses (i.e. by, for instance, the possibility of accelerated depreciation and pushing costs forward in time). Following existing literature, the lagged dependent variable is added to the regression model. In this way tax avoidance that persistence of time is captured (e.g. Manzon and Plesko, 2002).
Also for the remainder of the control variables, existing literature shows that those variables could influence the degree to which a firm is engaged in tax avoidance practices. Not controlling for these variables could affect the CSR coefficients.
14
TABLE 1 Descriptive Statistics
Panel A: Measures of tax avoidance practices
Variable n Mean Std. Dev. P25 Median P75
𝐷𝑇𝐴𝑋𝑖𝑡 6,093 0.000 0.169 -0.015 0.001 0.018
𝐷𝐷_𝐵𝑇𝑖𝑡 2,855 0.019 0.080 -0.004 0.011 0.031
𝐶𝐸𝑇𝑅𝑖𝑡 6,234 0.263 2.945 0.075 0.206 0.310
Panel B: CSR Measures and Control Variables
Variable n Mean Std. Dev. P25 Median P75
𝐸𝑆𝐺_𝑆𝑐𝑜𝑟𝑒𝑖𝑡−1 5,947 181.753 62.074 132.576 178.880 240.193 𝐸𝑁𝐶𝑖𝑡−1 5,943 57.373 27.750 35.290 60.730 83.210 𝐸𝑁𝑉𝑖𝑡−1 5,943 48.844 32.278 17.690 42.090 84.250 𝐺𝑂𝑉𝑖𝑡−1 5,943 75.659 16.100 68.350 79.270 87.543 𝐶𝐴𝑆𝐻𝑖𝑡−1 5,737 0.142 0.171 0.033 0.085 0.184 𝑅𝑂𝐴 𝑖𝑡−1 5,737 0.085 0.125 0.024 0.071 0.134 𝐿𝐸𝑉𝑖𝑡−1 5,737 0.239 0.202 0.102 0.202 0.334 𝑁𝑂𝐿𝑖𝑡−1 6,203 0.451 0.498 0.000 0.000 1.000 ∆𝑁𝑂𝐿𝑖𝑡−1 5,737 0.01 0.353 0.000 0.000 0.001 𝐹𝐼𝑖𝑡−1 5,737 0.0271 0.049 0.000 0.004 0.042 𝑃𝑃𝐸𝑖𝑡−1 5,737 0.26 0.275 0.054 0.166 0.385 𝐼𝑁𝑇𝐴𝑁𝐺𝑖𝑡−1 5,737 0.213 0.282 0.018 0.122 0.343 𝐸𝑄𝐼𝑁𝐶𝑖𝑡−1 5,737 0.002 0.007 0.000 0.000 0.001 𝑅&𝐷𝑖𝑡−1 5,737 0.022 0.054 0.000 0.000 0.020 𝐸𝑀𝑃𝑖𝑡−1 6,117 2.797 1.590 1.826 2.833 3.896 ∆𝑆𝑎𝑙𝑒𝑖𝑡−1 5,662 0.0587 0.302 -0.006 0.026 0.095 𝑆𝑖𝑧𝑒𝑖𝑡−1 6,140 0.181 1.306 8.328 9.080 9.991 𝑀𝐵𝑖𝑡−1 6,171 3.177 26.145 1.470 2.358 3.906 n = number of observations.
Panel A presents descriptive statistics on the aggressive tax avoidance measures and cash effective tax rate for the observations used in the baseline regression analysis. The initial sample consists of 6,659 firm-year observations for which data for CSR variables, control variables of the baseline regressions, and at least one tax avoidance variable are available. Appendix A provides detailed
information on all the variables.
Robustness test
15 checking whether the regression outcomes are robust for different measures of aggressive tax planning.
III. 2 Sample and descriptive statistics Sample Selection
The sample is compiled using two different databases and matching the pairs out of the available firms in those databases. Financial data of US resident companies is obtained from the Standard & Poor Compustat database. The ESG data, i.e. CSR score, is obtained from Datastream which retrieves the ESG data from the ASSET4 ESG database. The regression model is estimated using these two data sources. The data sample is restricted by the environmental, social and corporate governance data on approximately 3400 companies worldwide covering FTSE 250, S&P 500, NASDAQ 100, DJ STOXX, Russell 1000, S&P ASX 200 and MSCI World indices of the ASSET4 ESG database.
Consistent with prior research (e.g. Desai and Dharmapala, 2006; Hoi et al., 2013), in both data sets is searched for matching pairs, i.e. US firms which appear in both data sets. By linking the companies in the ASSET4 ESG database with the financial information provided by the Standard & Poor Compustat database, I obtain a
sample of 444 unique firms with 6,659 firm-observations covering the period 2002-2016.
Descriptive Statistics
In Table 1, Panel A the descriptive statistics of the measures that capture tax avoidance practices is presented. These statistics include the number of
observations, the mean, the standard deviation, the 25th percentile, the median and the 75th percentile. Due to the different methodologies of estimating 𝐷𝑇𝐴𝑋
𝑖𝑡 𝐷𝐷_𝐵𝑇𝑖𝑡 and 𝐶𝐸𝑇𝑅𝑖𝑡, the number of observations vary significantly. The firm-year
observations for 𝐷𝑇𝐴𝑋𝑖𝑡 is 6,093. The number firm-year observations for 𝐷𝐷_𝐵𝑇𝑖𝑡 and 𝐶𝐸𝑇𝑅𝑖𝑡 are respectively 2,855 and 6,234.
16 from a sample of 6,234 firm-year observations. To put this in perspective, the US statutory tax rate is 35%.
In Table 1, Panel B the descriptive statistics of the CSR measures and the control variables are presented. These CSR measures consist of 𝐸𝑆𝐺_𝑆𝑐𝑜𝑟𝑒𝑖𝑡−1,
𝐸𝑁𝐶𝑖𝑡−1, 𝐸𝑁𝑉𝑖𝑡−1 and 𝐺𝑂𝑉𝑖𝑡−1. The latter variables, 𝐸𝑁𝐶𝑖𝑡−1, 𝐸𝑁𝑉𝑖𝑡−1 and 𝐺𝑂𝑉𝑖𝑡−1, present resp. the social score, environmental score and the corporate governance score. These scores are scaled from 1 (low score) till 100 (high score). The
𝐸𝑆𝐺_𝑆𝑐𝑜𝑟𝑒𝑖𝑡−1 presents the CSR score for all three pillars, social, environmental and corporate governance. The score for corporate governance, with a mean of 75.659, stands out compared to the social score and environmental score, suggesting that the firms in the sample are well-governed but allocate less attention to social and environmental issues.
IV. Empirical results
In this chapter the empirical results of the regression are discussed. While the central hypothesis concerns the relationship between tax avoidance practices and the CSR score for the firms in the sample, the question whether the individual pillars of CSR, social, environmental and corporate governance, affect the degree of tax avoidance of a firm is also of significant interest.
Regression results
Table 2 presents the results for the estimated regression models to test the hypotheses. The tax avoidance measures, 𝐷𝑇𝐴𝑋𝑖𝑡, 𝐷𝐷_𝐵𝑇𝑖𝑡 and 𝐶𝐸𝑇𝑅𝑖𝑡, are the dependent variables for the OLS regressions. I use four proxies to capture different aspects of the degree of which a firm is involved in CSR activities. These are
17 Tax avoidance and the CSR-score (𝐸𝑆𝐺_𝑆𝑐𝑜𝑟𝑒𝑖𝑡−1)
The coefficient on 𝐸𝑆𝐺_𝑆𝑐𝑜𝑟𝑒𝑖𝑡−1 is significant and negative for the default measure on tax avoidance (𝐷𝑇𝐴𝑋𝑖𝑡). This shows that firms with a high degree of involvement in CSR in a given year avoid tax more than in other firm-years with less
involvement in CSR in that they have a higher degree of tax avoidance. Given the significance of the lagged dependent variable, this result is persistent over time. The results are marginal economically significant. Based on the parameter estimate in Model 1, for firms with a high degree of CSR involvement the likelihood
decreases that they engage in tax avoidance practices when the CSR score increases. However, this likelihood decreased with a small magnitude.
Tax avoidance and the stand-alone CSR pillars (𝐸𝑁𝐶𝑖𝑡−1, 𝐸𝑁𝑉𝑖𝑡−1 and 𝐺𝑂𝑉𝑖𝑡−1) The coefficients on 𝐸𝑁𝐶𝑖𝑡−1, 𝐸𝑁𝑉𝑖𝑡−1 and 𝐺𝑂𝑉𝑖𝑡−1 (Model 2) yield different results. The coefficient on 𝐸𝑁𝐶𝑖𝑡−1 (social pillar of CSR) and 𝐸𝑁𝑉𝑖𝑡−1 (environmental pillar of CSR) are significant for the default measure for tax avoidance (𝐷𝑇𝐴𝑋𝑖𝑡). The coefficient on 𝐺𝑂𝑉𝑖𝑡−1 (corporate governance pillar of CSR) is not significant for the default measure for tax avoidance.
The coefficient on 𝐸𝑁𝐶𝑖𝑡−1 shows that firms with a high degree of social involvement in a given year are more likely to engage in tax avoidance practices than in other firm-years with low social involvement. The coefficient on 𝐸𝑁𝑉𝑖𝑡−1 offers an opposite view. The coefficient on 𝐸𝑁𝑉𝑖𝑡−1 shows that firms with a high degree of
environmental involvement in a given year are less likely to engage in tax
avoidance practices than in other firm-years with low environmental involvement. Despite the coefficient on 𝐺𝑂𝑉𝑖𝑡−1, the coefficient shows that firms with higher levels of corporate governance in given year are less likely to engage in tax avoidance practices than other firm-year with a low level of corporate governance. Given the significance of the above lagged dependent variables, these results are persistent of time.
18
TABLE 2
CSR and Tax avoidance
Evidence from Firm-Year Level Regressions
Regression Results
𝐷𝑇𝐴𝑋𝑖𝑡 𝐷𝐷_𝐵𝑇𝑖𝑡 𝐶𝐸𝑇𝑅𝑖𝑡
OLS OLS OLS
Dependent Variable Model 1 Model 2 Model 3 Model 4 Model 5 Model 6
19
0.0142** 0.0112* 0.0060 0.0061 -0.2338 -0.2360* (2.4382) (1.8657) (1.5451) (1.5823) (-1.9087) (-1.9253)
-0.0001 -0.0000 -0.0000 -0.0000 0.0002* 0.0002
(-1.0559) (-0.3676) (-0.4774) (-0.4929) (0.1233) (0.1363) Lag (Dependent Variable) 0.1165*** 0.0990*** 0.2156*** 0.2166*** 0.1217*** 0.1218*** (-8.7310) (-7.0920) (8.0595) (8.0964) (-8.3532) (-8.3543) Intercept -0.0602 -0.0682 0.0054 -0.0013 2.9345*** 2.9532*** (-1.2272) (-1.3170) (0.1663) (-0.0395) (2.8613) (2.7993) F-statistic 3.9035 3.1323 14.1549 14.0774 1.3930 1.3870 Prob(F-statistic) 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 Adjusted R2 0.1976 0.1546 0.6245 0.6245 0.0342 0.0338 Number of Observations 5,437 5,434 2,366 2,366 5,112 5,111 *, **, *** Indicate significate at the 10 percent, 5 percent and 1 percent levels,
respectively.
t-statistics and z-statistics, reported in parentheses, are based on standard errors clustered at the firm level. The initial sample consists of 6,659 firm-year observations covering the period 2002-2016 for which data for CSR variables, control variables of the regression, and at least one tax avoidance variable are available. The number of observations used in each regression varies due to additional data requirements for estimating the three measures of tax avoidance practices. The regression models are: 𝐴𝐺𝐺𝑅𝐸𝑆𝑆𝐼𝑉𝐸𝑖𝑡 = 𝛽0+ 𝛽1𝐸𝑆𝐺_𝑆𝑐𝑜𝑟𝑒𝑖𝑡−1 + 𝛽2𝐶𝐴𝑆𝐻𝑖𝑡−1+ 𝛽3𝑅𝑂𝐴 𝑖𝑡−1+ 𝛽4𝐿𝐸𝑉𝑖𝑡−1+ 𝛽5𝑁𝑂𝐿𝑖𝑡−1 + 𝛽6∆𝑁𝑂𝐿𝑖𝑡−1+ 𝛽7𝐹𝐼𝑖𝑡−1+ 𝛽8𝑃𝑃𝐸𝑖𝑡−1+ 𝛽9𝐼𝑁𝑇𝐴𝑁𝐺𝑖𝑡−1+ 𝛽10𝐸𝑄𝐼𝑁𝐶𝑖𝑡−1+ 𝛽11𝑅&𝐷𝑖𝑡−1 + 𝛽12𝐸𝑀𝑃𝑖𝑡−1+ 𝛽13∆𝑆𝑎𝑙𝑒𝑖𝑡−1+ 𝛽14𝑆𝑖𝑧𝑒𝑖𝑡−1+ 𝛽15𝑀𝐵𝑖𝑡−1 + 𝛽16𝐿𝑎𝑔𝑖𝑡−1 (𝑑𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑡 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒) + 𝑌𝑒𝑎𝑟 𝑑𝑢𝑚𝑚𝑖𝑒𝑠 + 𝜀𝑖𝑡 𝐴𝐺𝐺𝑅𝐸𝑆𝑆𝐼𝑉𝐸𝑖𝑡 = 𝛽0+ 𝛽1𝐸𝑁𝐶𝑖𝑡−1+ 𝛽2𝐸𝑁𝑉𝑖𝑡−1+ 𝛽3𝐺𝑂𝑉𝑖𝑡−1 + 𝛽4𝐶𝐴𝑆𝐻𝑖𝑡−1+ 𝛽5𝑅𝑂𝐴 𝑖𝑡−1+ 𝛽6𝐿𝐸𝑉𝑖𝑡−1 + 𝛽7𝑁𝑂𝐿𝑖𝑡−1+ 𝛽8∆𝑁𝑂𝐿𝑖𝑡−1+ 𝛽9𝐹𝐼𝑖𝑡−1+ 𝛽10𝑃𝑃𝐸𝑖𝑡−1+ 𝛽11𝐼𝑁𝑇𝐴𝑁𝐺𝑖𝑡−1 + 𝛽12𝐸𝑄𝐼𝑁𝐶𝑖𝑡−1+ 𝛽13𝑅&𝐷𝑖𝑡−1+ 𝛽14𝐸𝑀𝑃𝑖𝑡−1+ 𝛽15∆𝑆𝑎𝑙𝑒𝑖𝑡−1+ 𝛽16𝑆𝑖𝑧𝑒𝑖𝑡−1 + 𝛽17𝑀𝐵𝑖𝑡−1+ 𝛽18𝐿𝑎𝑔𝑖𝑡−1 (𝑑𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑡 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒) + 𝑌𝑒𝑎𝑟 𝑑𝑢𝑚𝑚𝑖𝑒𝑠 + 𝜀𝑖𝑡
where 𝐴𝐺𝐺𝑅𝐸𝑆𝑆𝐼𝑉𝐸𝑖𝑡 represents the three measures of tax avoidance practices and 𝐸𝑆𝐺_𝑆𝑐𝑜𝑟𝑒𝑖𝑡−1, 𝐸𝑁𝐶𝑖𝑡−1, 𝐸𝑁𝑉𝑖𝑡−1 and 𝐺𝑂𝑉𝑖𝑡−1 are the measures for CSR activities. Year dummies are included in each specification. This table presents the OLS regression results.
Appendix A provides detailed information on all the variables.
In addition, based on the parameter estimate in Model 1, firms with a higher degree of environmental involvement and higher degree of corporate governance are both less likely to engage in tax avoidance practices when the score on the environmental involvement or the degree of corporate governance of a firm increases.
20 Robustness of the regression results
In this part the regression results on Model 3, 4, 5 and 6 will be discussed. They function as a robustness check on whether the results on the default measure (𝐷𝑇𝐴𝑋𝑖𝑡) of tax avoidance practices are robust along the two other alternative
measures for tax avoidance measures. The two measures to check for robustness are 𝐷𝐷_𝐵𝑇𝑖𝑡 and 𝐶𝐸𝑇𝑅𝑖𝑡. The results of the 𝐷𝐷_𝐵𝑇𝑖𝑡 measure are estimated using Model 3 and Model 4. The results of the 𝐶𝐸𝑇𝑅𝑖𝑡 measure are estimated using Model 5 and Model 6.
Tax avoidance and the CSR-score (𝐸𝑆𝐺_𝑆𝑐𝑜𝑟𝑒𝑖𝑡−1)
In first regression model for 𝐷𝐷_𝐵𝑇𝑖𝑡 (Model 3), the coefficient on 𝐸𝑆𝐺_𝑆𝑐𝑜𝑟𝑒𝑖𝑡−1 is not significant for the measure on tax avoidance (𝐷𝐷_𝐵𝑇𝑖𝑡). However, the coefficient is negative, which is consistent with the default measure (𝐷𝑇𝐴𝑋𝑖𝑡). Given the
significance of the lagged dependent variable, this result is persistent over time. The results on the second alternative measure of tax avoidance (𝐶𝐸𝑇𝑅𝑖𝑡), Model 5, shows a divergent view. Consistent with 𝐷𝐷_𝐵𝑇𝑖𝑡, the coefficient on 𝐸𝑆𝐺_𝑆𝑐𝑜𝑟𝑒𝑖𝑡−1 is not significant for the second alternative measure on tax avoidance (𝐶𝐸𝑇𝑅𝑖𝑡). In contrast with the results on the previous measures, the coefficient on 𝐸𝑆𝐺_𝑆𝑐𝑜𝑟𝑒𝑖𝑡−1 is positive for 𝐶𝐸𝑇𝑅𝑖𝑡.
Consistent with the economic significance of the default measure for tax avoidance practices (𝐷𝑇𝐴𝑋𝑖𝑡), the measures that serve as a robustness check for the default measure are only marginal economically significant. Based on the parameter estimate in Model 3 (𝐷𝐷_𝐵𝑇𝑖𝑡), firms with a high degree of CSR involvement are equally likely (coefficient is 0.0000) to engage in tax avoidance practices when the CSR score of a firm increases compared to other firm-years in which the CSR score decreases. Based on the parameter estimate in Model 5 (𝐶𝐸𝑇𝑅𝑖𝑡), firms with a higher degree of CSR involvement are more likely to engage in tax avoidance practices (coefficient is 0.0060) when the CSR score increases compare to firm-year in which the CSR score decreases.
21 Tax avoidance and the stand-alone CSR pillars (𝐸𝑁𝐶𝑖𝑡−1, 𝐸𝑁𝑉𝑖𝑡−1 and 𝐺𝑂𝑉𝑖𝑡−1) The coefficients for 𝐷𝐷_𝐵𝑇𝑖𝑡 and 𝐶𝐸𝑇𝑅𝑖𝑡 on 𝐸𝑁𝐶𝑖𝑡−1, 𝐸𝑁𝑉𝑖𝑡−1 and 𝐺𝑂𝑉𝑖𝑡−1 (Model 4 and Model 6) yield diverging results. The coefficients on 𝐸𝑁𝐶𝑖𝑡−1 (social pillar of
CSR), 𝐸𝑁𝑉𝑖𝑡−1 (environmental pillar of CSR) and 𝐺𝑂𝑉𝑖𝑡−1 (corporate governance pillar of CSR) are all not significant for both alternative measures for tax avoidance practices (𝐷𝐷_𝐵𝑇𝑖𝑡 and 𝐶𝐸𝑇𝑅𝑖𝑡).
The coefficient on 𝐸𝑁𝐶𝑖𝑡−1 shows that firms with a high degree of social involvement in a given year are less likely to engage in tax avoidance practices than other firm-years with low social involvement for the measure 𝐷𝐷_𝐵𝑇𝑖𝑡 (Model 4). The coefficient on 𝐸𝑁𝐶𝑖𝑡−1 for 𝐶𝐸𝑇𝑅𝑖𝑡 offers an opposite view. The coefficient on 𝐸𝑁𝑉𝑖𝑡−1 for 𝐶𝐸𝑇𝑅𝑖𝑡 shows that firms with a high degree of social involvement in a given year are more likely to engage in tax avoidance practices than other firm-years with low social involvement (Model 6).
The coefficient on 𝐸𝑁𝑉𝑖𝑡−1 shows that firms with a high degree of social involvement in a given year are more likely to engage in tax avoidance practices than other firm-years with low social involvement for both alternative measure of tax avoidance (𝐷𝐷_𝐵𝑇𝑖𝑡 and 𝐶𝐸𝑇𝑅𝑖𝑡) but the likelihood increases only with a small magnitude (Model 4 and Model 6).
The coefficient on GOV𝑖𝑡−1 shows that firms with a high degree of corporate
governance in a given year are more likely to engage in tax avoidance practices than other firm-years with low social involvement for both alternative measure of tax avoidance (𝐷𝐷_𝐵𝑇𝑖𝑡 and 𝐶𝐸𝑇𝑅𝑖𝑡) but the likelihood increases only slightly (Model 4 and Model 6).
Again, the latter mentioned coefficients are not significant. The robustness check for the default measure with regard to relationship between tax avoidance and social involvement, which showed a significant, positive relationship, fails because of the insignificant coefficients of the test variables for the alternative measures. In addition, the robustness check for the default measure with regard to relationship between tax avoidance and the degree of corporate governance within a firm (which shows a significant, negative relationship) fails because of the insignificant
22
V. Conclusion
The literature on tax avoidance practices of firms has focused on a variety of matters concerning firm behavior and firm valuation. However, it has
underemphasized whether the choice to engage in tax avoidance practices is part of the general decision-making mechanism of a firm. That is, whether the framework wherein corporate decisions are made and how the shared beliefs within a firm can be an explaining common denominator in making tax strategy decisions. Those factors that may drive the general decision-making mechanism within a firm include the CSR guidelines in place. In that sense, it would seem odd for a firm to make contrasting decisions regarding tax avoidance and CSR activities.
This thesis contributes to the above debate, and therefore the existing literature, by extensively looking at the relationship between tax avoidance and CSR. In addition, the relationship between tax avoidance and the underlying properties of CSR is also examined. The simple presumption that the degree to which a firm is engaged in tax avoidance can be explained by the extent to which a firm is involved in CSR seems not to be completely validated in the data. Nevertheless, the patterns in the data imply a negative relationship between tax avoidance and CSR but this
relationship is of a small magnitude. This result is reinforced by the default
measure for tax avoidance (the book-tax difference measure as described by Frank et al. (2009)) but is not robust to the alternative measures for tax avoidance (the book-tax difference measure as described by Desai and Dharmapala (2006) and the cash effective tax rate measure). The negative relationship of small magnitude is supported by existing literature.
With regard to the pillars of CSR (social, environmental and corporate governance) it was expected that both a high degree of social involvement and a high degree of corporate governance would be negatively correlated with the extent to which a firm engages in tax avoidance, based on existing literature. A high degree of
environmental involvement in relation to tax avoidance is not emphasized in literature.
In contrast with literature, the findings show that a higher level of social
23 topic is preferred. With regard to the corporate governance pillar, the findings show that a high degree of corporate governance leads to a decreased probability that a firm is engaged in tax avoidance practices. This is consistent with literature. However, these results were not supported, due to insignificant results, by the alternative measures for tax avoidance.
In other words, the hypothesis that tax avoidance is a function of the degree of CSR involvement of a firm is mildly confirmed. This paper thus shows that incorporating CSR standards in policies that target tax avoidance could give divergent results. However, this thesis opens up the debate on whether the implications of (certain pillars of) CSR involvement should be considered when evaluating tax practices, which I leave for further debate.
VI. Limitations
In this section, some caveats with respect to the analysis are discussed.
First of all, due to the nature of the databases that were used to compile the data set (Standard & Poor Compustat database and ASSET4 ESG database), the firms in the sample are relatively big in terms of size. Literature shows that larger firms have more opportunities to avoid taxes. When a sample with a distributed set of firms, in terms of size, was used, the regression outcomes could be affected. This effect is not controlled for in the regression model.
Another bias could be present because certain companies report on their CSR activities and others not. It could be the case that companies who report on their CSR activities would be more engaged in such practices than those who are not reporting.
25 VII. Literature list
Bowen, H.R. 1953. Social responsibilities of the businessman. NewYork, Harper& Row. Barnett, M. L., and R. M. Salomon. 2006. Beyond dichotomy: The curvilinear relationship between social responsibility and financial performance. Strategic Management Journal 27 (11): 1101–1122.
Carroll, A.B. (1979). A Three-Dimensional Conceptual Model of Corporate Performance, The Academy of Management Review, 4(4), 497-505
Changyuan, X., Cao, C., Kam, C. C. 2017. Social trust environment and firm tax avoidance: Evidence from China. North American Journal of Economics and Finance 42: 374–392.
Cheng, I. H., H. Hong, and K. Shue. 2014. Do Managers Do Good with Other People’s Money? Working paper no. w19432, National Bureau of Economic Research.
Dechow, P. M., and R. G. Sloan. 1995. Detecting earnings management. The Accounting Review 70 (2): 193–225.
Desai, M. A., and D. Dharmapala. 2006. Corporate tax avoidance and high-powered incentives. Journal of Financial Economics 79 (1): 145–179.
Desai, M. A., & Dharmapala, D. (2008). Tax and corporate governance. An economic approach. In Tax and corporate governance. Berlin Heidelberg: Springer.
Desai, M. A., and D. Dharmapala. 2009. Corporate tax avoidance and firm value. Review of Economics and Statistics 91 (3): 537–546.
Dhaliwal, D., S. Radhakrishnan, A. Tsang, and Y. Yang. 2012. Nonfinancial disclosure and analyst forecast accuracy: International evidence on corporate social responsibility disclosure. The Accounting Review 87: 723–759.
Dhaliwal, D., O. Li, A. Tsang, and Y. Yang. 2011. Voluntary nonfinancial disclosure and the cost of equity capital: The initiation of corporate social responsibility reporting. The Accounting Review 86: 59–100.
26 Eccles, R., and G., Serafeim. 2013. The performance frontier: Innovating for a sustainable strategy. Harvard Business Review 91 (5): 50 – 60.
Frank, M. M., L. J. Lynch, and S. O. Rego. 2009. Tax reporting aggressiveness and its relation to aggressive financial reporting. The Accounting Review 84 (2): 467–496.
Graham, J. R., M. Hanlon, T. Shevlin, and N. Shroff. 2012. Incentives for Tax Planning and Avoidance: Evidence from the Field. Working paper, Duke University, Massachusetts Institute of Technology, and University of California, Irvine.
Graves, S. B., and S. A. Waddock. 1994. Institutional owners and corporate social performance. Academy of Management Journal 37 (4): 1034–1046.
Guiso, L., Sapienza, P., & Zingales, L. (2004). The role of social capital in financial development. American Economic Review, 94, 526–556.
Hermalin, B. E. 2001. Economics and corporate culture. In The International Handbook of Organizational Culture and Climate, edited by C. L. Cooper, S. Cartwright, and P. C. Earley. Chichester, U.K.: John Wiley & Sons.
Hoi, C.K., Qiang, W., Hao, Z. 2013. Is corporate social responsibility (CSR) aaociated with tax avoidance? Evidence from irresponsible CSR activities. The Accounting Review 88: 2025–2059.
Jha, A., & Chen, Y. (2015). Audit fees and social capital. Accounting Review, 90, 611–639. Kim, Y., M. S. Park, and B. Wier. 2012. Is earnings quality associated with corporate social responsibility? The Accounting Review 87: 761–796.
Khurana, I. K., & Moser, W. J. (2012). Institutional shareholders’ investment horizons and tax avoidance. Journal of the American Taxation Association, 35, 111–134.
Kreps, D. M. 1990. Corporate culture and economic theory. In Perspectives on Positive Political Economy, edited by J. E. Alt, and K. A. Shepsle. Cambridge, U.K.: Cambridge University Press.
Manzon, G. B., and G. A. Plesko. 2002. The relation between financial and tax reporting measures of income. Tax Law Review 55 (2): 175–214.
27 Performance. Working paper, Harvard University, Washington University in St. Louis, and
University of Michigan.
McWilliams, A., and D. Siegel. 2000. Corporate social responsibility and financial
performance: Correlation or misspecification? Strategic Management Journal 21 (5): 603–609. Orlitzky, M., F. L. Schmidt, and S. L. Rynes. 2003. Corporate social and financial
performance: A meta-analysis. Organization Studies 24 (3): 403–441.
Shuping, C., Xia, C., Qiang, C., Shevlin, T. 2010. Are family firms more tax aggressive than non-family firms? Journal of Financial Economics 95 (2010) 41–61.
Turban, D. B., and D. W. Greening. 1997. Corporate social performance and organizational attractiveness to prospective employees. Academy of Management Journal 40 (3): 658–672.
Waddock, S.A., and S. B. Graves. 1997. The corporate social performance-financial performance link. Strategic Management Journal 18 (4): 303–319.
28
APPENDIX A Variable definitions
Variable Definition
Measures of tax avoidance
The discretionary book-tax difference (𝐷𝐷_𝐵𝑇𝑖𝑡) for firm i, in year t is described by Desai and Dharmapala (2006).
f𝐷𝐷_𝐵𝑇𝑖𝑡 is estimated as 𝜇𝑖 + 𝜀𝑖𝑡 from the following firm fixed-effect regression:
𝐵𝑇𝑖𝑡 = 𝛽1𝑇𝐴𝑖𝑡 + 𝜇𝑖 + 𝜀𝑖𝑡,
where 𝐵𝑇𝑖𝑡 is the book-tax difference presented by Manzon and Plesko (2002), which is described below. 𝑇𝐴𝑖𝑡 represents a total accruals measure for firm i, year t, scaled by lagged value of assets (Dechow et al. 1995); 𝜇𝑖 represents the average value of the residual for firm i, year t over the sample period; and 𝜀𝑖𝑡 is the deviation of the residual in year t from firm i's average residual.
BT is defined as follows: (U.S. domestic financial income – U.S. domestic taxable income – Income taxes (State) – Income taxes (Other) – Equity in Earnings) / lagged assets = (PIDOM – TXFED / Statutory tax rate – TXS – TXO -ESUB) / 𝑇𝐴𝑖𝑡−1. When a firm has zero or negative taxable income the firm is assumed to have attenuated incentives, at the margin, to engage in tax avoidance practices. Following existing literature of Desai and Dharmapala (2006), I include only firm-years with positive TXFED.
Control variables
Amount of cash or cash equivalents hold by firm i, year t, defined as cash and marketable securities (CHE) divided by lagged assets (AT);
Return on assets for firm i, year t, measured as operating income (PI - XI) scaled by lagged assets (AT);
Leverage for firm i, year t, measured as long-term debt (DLTT) scaled by lagged assets (AT);
A dummy variable which takes the value of 1 if loss carry forward (TLCF) for firm I is positive at the beginning of the year t;
Change in loss carry forward (TLCF) for firm i, year t, scaled by lagged assets (AT); Foreign income (PIFO) for firm I, year t, scaled by lagged assets (AT). Missing
value for PIFO are set to 0;
Property, plant and equipment (PPENT) for firm I, year t, scaled by lagged assets (AT);
Intangible assets (INTAN) for firm I, year t, scaled by lagged assets (AT);
Equity income in earnings (ESUB) for firm I, year t, scaled by lagged assets (AT);
Research and development expense ratio for firm I, year t, measured as research and development expense (XRD) scaled by lagged assets (AT). Missing values in XRD are set to 0;
29
Natural logarithm of the market value of equity (PRCC_F x CSHO) for firm I at the beginning of the year t;
Market-to-book ratio for firm I, at the beginning of year t, measured as market value of equity (PRCC_F x CSHO), scaled by book value of equity (CEQ).
