Corporate Social Responsibility, Tax Aggressiveness, and Firm Market Value†
Abstract
enThis paper examines the relationship of corporate social responsibility (CSR), tax aggressiveness, and firm market value. An economic model has been developed to show that profit-maximization firms are willing to incur additional costs in CSR, such as paying more taxes, as long as they can differentiate their products from non-CSR firms, and that socially conscious consumers will buy products from CSR firms at prices higher than those of non-CSR firms. The empirical study in this paper indicates that the higher the CSR ranking of a firm, the less likely a firm is to engage in tax aggressiveness. It also indicates that a reputation of higher CSR will enhance firm market value. Using Canadian companies listed in the S&P/TSX 60 index, I find that both firms’ five-year effective tax rates and annual effective tax rates are positively associated with their overall CSR scores as well as with their social scores. Firms’ five-year effective tax rates are also positively associated with their governance index. I also find that firms’ overall CSR ranking and governance scores are positively associated with their market value.
Résumé
frL'auteur examine la relation entre la responsabilité sociale de l'entreprise (RSE), l'audace de ses positions fiscales et sa valeur de marché. Il élabore un modèle économique montrant que les entreprises qui cherchent à maximiser leurs profits sont disposées à engager des coûts additionnels liés à la RSE, notamment à payer davantage d'impôts, dans la mesure où elles pourront ainsi différencier leurs produits de ceux des entreprises qui ne se soucient pas de la RSE, et que les consommateurs qui attachent de l'importance à la RSE accepteront d'acheter les produits des entreprises qui sont socialement responsables à des prix supérieurs à ceux des entreprises qui ne le sont pas. L’étude empirique de l'auteur révèle que plus le classement d'une entreprise est élevé en ce qui a trait à la responsabilité sociale, moins cette dernière est susceptible d'user d'audace dans ses positions fiscales. Elle révèle également qu'une bonne réputation en matière de responsabilité sociale accroît la valeur de marché de l'entreprise. L’étude des sociétés canadiennes de l'indice S&P/TSX 60 permet de constater que les taux d'imposition effectifs des entreprises sur une période cinq ans, de même que leurs taux d'imposition effectifs annuels, sont en relation positive avec leur cote globale en matière de RSE et leur cote au chapitre du comportement social. Leurs taux d'imposition effectifs quinquennaux sont aussi en relation positive avec leur indice de gouvernance. L'auteur observe également que le classement global des entreprises en matière de RSE et leur cote au chapitre de la gouvernance sont en relation positive avec leur valeur de marché.
Introduction
The question of why there is a wide variety in tax payments across firms has always been an agenda item for both policy makers and academics. Corporate tax aggressiveness has been a popular issue in recent years, and more tax studies are investigating what is associated with tax-aggressive activities. To date, the literature has shown that corporate tax aggressiveness is associated with a number of firm characteristics (size, prior losses, leverage, intensity of fixed assets, etc.), ownership structure (family-owned firms, institutional shareholding, inside shareholding, ownership concentration, etc.), the weakness of internal control, the tax expertise of external audit firms, and the CEO's educational background, to name a few (Stickney and McGee, 1982; Zimmerman, 1983; Porcano, 1986; Omer et al., 1993; Gupta and Newberry, 1997; Klassen, 1997; Adhikari et al., 2006; Dyreng et al., 2008; Badertscher et al., 2009; Robinson et al., 2010; Zeng, 2010, 2011; Bauer, 2011; Armstrong et al., 2012, among others).
Corporate social responsibility (CSR) has also received increased attention from both businesses and academics. There are a number of definitions applied to CSR. According to McWilliams and Siegel (2001), CSR means actions that enhance social goods, beyond any legal requirements and even beyond a firm's interests. Carroll (1979) stated that CSR means engaging in activities that respond to economic, legal, ethical, and discretionary expectations as well as to expectations from broader society and other stakeholders (including consumers, suppliers, communities, governments, employees, and shareholders).
Numerous studies have examined the relationship between CSR and firm performance, stock returns, and the cost of capital (Waddock and Graves, 1997; Stanwick and Stanwick, 1998; Forster and Meinhard, 2002; Bassen et al., 2006; Brammer et al., 2006; Paul and Siegel, 2006; Piercy and Lane, 2009; Ghoul et al., 2011; Dhaliwal et al., 2011, to name a few). However, only a few studies have explored the relationship between CSR and tax reporting practices. These studies focus on U.S firms and the results are mixed (Davis et al., 2013; Watson, 2011a,b). As argued by Christensen and Murphy (2004), current CSR studies have touched on almost every other area of corporate engagement, except the one where a company's corporate citizenship is most tangible and important: corporate tax avoidance.
This paper examines the relationship of CSR, tax aggressiveness, and firm market value. An economic model has been developed to show that profit-maximization firms are willing to incur additional costs in CSR, such as paying more taxes, as long as they can differentiate their products from non-CSR firms, and that socially conscious consumers will buy products from CSR firms at prices higher than those of non-CSR firms. This paper also links CSR to firm market value and argues that paying more taxes can enhance firm market value by contributing to higher CSR rankings.
The empirical study in this paper indicates that socially responsible firms are less likely to undertake aggressive tax activities, while those firms that are less interested in being socially responsible are more likely to undertake aggressive tax activities. It also indicates that a reputation of higher CSR will enhance firm market value. Using Canadian companies listed in the S&P/TSX 60 index, I find that both firms’ five-year effective tax rates and annual effective tax rates are positively associated with their overall CSR scores as well as with their social scores. Firms’ five-year effective tax rates are also positively associated with their governance index. I also find that firms’ overall CSR ranking and governance scores are positively associated with their market value. Overall, the empirical results show that paying more taxes leads to higher CSR rankings and thus enhances firm market value.
The remainder of this paper is organized as follows. Section 2 reviews the current literature on tax aggressiveness and CSR. Section 3 develops a theoretical model showing that paying more taxes and maximizing profits are compatible. It also develops two hypotheses for empirical tests. Section 4 designs the empirical models and describes sample selection and variables. Section 5 presents the main testing results as well as the supplementary testing results. Finally, section 6 concludes the study.
Literature Review
There are several accounting studies examining CSR and tax aggressiveness as well as other firm characteristics in recent years. Dhaliwal et al. (2011) find that initiating the voluntary disclosure of CSR will attract dedicated institutional investors and analysis coverage, and reduce a firm's cost of capital. Richardson and Lanis (2012) argue that socially responsible firms are likely to deter tax-aggressive behavior. Using a sample of Australian public companies, they find a negative relationship between CSR disclosure and tax aggressiveness. Watson (2011a) finds that in the United States, firms with low CSR scores undertake more aggressive tax activities such as more book-tax differences, high effective tax rates, and more tax-sheltering activities. In addition, Watson (2011b) finds that socially irresponsible firms generally have larger unrecognized tax benefits than socially conscious firms. He argues that socially responsible firms attract consumers and investors with similar norm and value and therefore deter these firms’ tax aggressiveness activities. Hoi et al. (2013) also find that firms with excessive socially irresponsible activities are more likely to engage in tax aggressiveness. However, Davis et al. (2013) analyze whether paying corporate tax is viewed as socially responsible. Using U.S. public firms, they find that CSR is negatively associated with effective tax rates, and is positively associated with tax-lobbying expenditures. Their results show that socially responsible firms pay less tax and engage in more tax lobbying.
Given the mixed results on the relationship between CSR and tax aggressiveness, this paper examines the Canadian scenario. As Foster and Meinhard (2002) state, the Canadian government, unlike the U.S. government, plays a significant role in supporting universal social programs. As a result, it is expected that Canadian firms will act differently from their U.S. counterparts when considering CSR activities and that the relationship between CSR activities and tax aggressiveness is different in the two countries. Another factor to consider is that, while Canada's cultural environment is similar to that of the United States, its institutional environment is not. For example, the corporate tax rate in the United States is 40 percent, while in Canada it is around 30 percent. Prior studies suggest that a firm's national institutional context, including legal, regulatory, and professional structures, affect its sensitivities toward its social performance (Mahoney and Thorn, 2006)
In addition, Canada has a very different industry mix than that of the United States. For example, Canada has the most number of publicly listed oil and gas companies, and Canadian stock exchanges raise the most capital for the mining sector in the world (Li et al., 2014). Given the fact that companies in high-profile, environmentally sensitive sectors such as oil, gas, and mining would have more incentives to build up a positive image and give prominence to CSR, Canada provides an interesting scenario in which to examine the effect of CSR on firms' activities, such as tax reporting.
To make a comparison between the U.S. firms and Canadian firms, in a supplementary test, a Canadian indicator variable is included to test whether or not there is a difference in the Canadian context. I find that in Canada, paying more taxes contributes to a higher overall CSR ranking.
In the next section, I develop an economic model showing that a profit-maximization firm is willing to incur additional costs in CSR, such as paying more taxes, as long as it can differentiate its products from non-CSR firms, and that socially conscious consumers will buy products from CSR firms at prices higher than those of non-CSR firms.
Models and Hypotheses
A number of studies on CSR argue that firms’ involvement in CSR activities may contribute to enhancing profitability. In these studies, the firms’ investment in CSR is a strategy through which firms can build up their reputation and differentiate their products from those of their competitors. Keith and Henderson (2010) argue that the adoption of a CSR program may improve a firm's image, with potential attendant business upsides. According to a survey conducted by the Canadian Democracy and Corporate Accountability Commission (CDCAC), 75 percent of Canadians agree that the government should not purchase from companies with poor CSR records; 72 percent of Canadians agree that firms should have accountability beyond their profit margins; and 74 percent of shareholders share the same opinions (Keith and Henderson, 2010). Also, a Canadian national poll (Abacus Data, 2014) survey found that over 60 percent of Canadians are willing to pay more for products and services from a socially responsible firm. A 2001 Hill & Knowlton/Harris Interactive poll shows that 79 percent of Americans take CSR into account when they decide whether or not to purchase a particular firm's goods (Redman 2005). Marketing studies show that firms’ social performance shapes consumers’ purchasing intentions (e.g., Brown and Dacin, 1997; Creyer and Ross, 1997; Mohr and Webb 2005). Furthermore, Boehe and Cruz (2010) find that CSR contributes to product differentiation in the Brazilian export market and thus improves export performance. Flammer (2014) examines the causal mechanisms through which CSR can improve competitiveness in the U.S. manufacturing industry. Her findings suggest that firms engage in CSR in order to signal product quality and differentiate themselves from others. Lev et al. (2010) find that firms’ charitable donations are significantly associated with their revenue growth and customer satisfaction.
To incorporate the above arguments, I develop an economic model under a few assumptions; that is, socially conscious consumers exist, and will incur a disutility if they purchase goods produced by non-CSR firms, based on Barboza and Trejos (2011); socially conscious consumers are rational and maximize utility subject to budget constraints; and both tax aggressive firms and tax compliant firms are profit maximizers.
Socially Conscious Consumers
There are two sets of perfect substitute products, those produced by CSR firms XCSR and those produced by non-CSR firms XnCSR. The products are perfect substitutes in the physical sense, but are differentiated by their producers’ reputation as CSR or non-CSR firms.
Equation 6 shows that the proportion of CSR goods to non-CSR goods purchased by the socially conscious consumer () is an inverse function of the price ratio , and is positively related to disutility parameter η. If the price ratio between CSR goods and non-CSR goods () is equal to , a consumer would purchase a combination of the two goods. Since for η ∈ (0, 1), a CSR firm's goods can be priced higher than a non-CSR firm's goods. This is consistent with the arguments that one benefit for a firm that pursues CSR is gaining a good reputation and therefore having the potential to charge a premium price for its products.1
However, if the price ratio between CSR goods and non-CSR goods () is much too high, i.e., the price for CSR goods is much too high relative to non-CSR goods, the consumer will purchase more non-CSR goods; otherwise, if the price ratio is too low, the consumer will purchase more CSR goods.
Parameter η actually measures the degree of disutility incurred for the socially conscious consumer to purchase non-CSR goods. As η increases, the socially conscious consumer is willing to pay more for CSR goods, thereby creating opportunities for firms to behave in a socially responsible manner.
If the extreme case of η = 0 exists, that is, the consumer does not incur disutility by purchasing non-CSR goods, or he/she cannot differentiate between CSR goods and non-CSR goods, CSR goods cannot be sold at a price higher than non-CSR goods. In the other extreme case, where η = 1, that is, purchasing non-CSR goods will not contribute to the socially responsible consumer's utility at all, there is no market for non-CSR goods.
Firms
Equation 7 shows that, even in the presence of extra tax cost ρ, a firm will produce positive output () as long as the marginal cost is less than or equal to marginal revenue.
Equation (9) shows that tax-compliant firms will charge a higher price for their products by ρ, being the cost of paying more tax per unit.
Notwithstanding the fact that tax-compliant firms incur an additional cost of paying more taxes, they can still maximize profits if they are seen as CSR firms, since goods from CSR firms can be priced higher than non-CSR goods, as depicted in equation (6). In other words, the additional cost incurred by tax-compliant firms is in fact the cost of differentiating their goods from other firms.
In fact, whether or not a firm decides to behave in a tax-compliant manner depends on the prices it can charge socially conscious customers. If the price difference between tax-compliant products and tax-aggressive products is equal to the additional cost of paying more taxes by a tax-compliant firm, there will be a combination of tax-compliant and tax-aggressive firms in the market. Alternatively, if the price difference is higher than the additional cost of paying more taxes, all firms will behave with tax compliance, whereas if the price difference is lower than the additional cost of paying more taxes, all firms will behave with tax aggressiveness.
Firms are subject to the expectations of all their stakeholders including employees, shareholders, consumers, communities, as well as governments; that is, firms are expected to consider the concerns of the broader society. Complying with tax obligations, refraining from using aggressive or abusive avoidance techniques, and disclosing detailed tax-planning transaction are all important tools used by firms to engage with the broader society. Tax aggressiveness, on the other hand, is perceived by the broader society as a firm not paying its fair share of taxes thereby transferring a larger share of the overall tax burden onto others. Tax aggressiveness enables a firm to become a free-rider, i.e., it enjoys benefits and resources without paying (Christensen, 2004). Tax aggressiveness can also harm firms’ reputations and branding images (Fisher, 2014). Tax aggressiveness is therefore not compatible with good corporate citizenship and can be viewed as being socially irresponsible.
Therefore, I expect that paying more taxes (ρ > 0 in the above models) would increase the reputation of firms for being socially responsible and hence differentiate goods produced by these firms from others to maximize profit and market value. I thus specify the following two hypotheses.
Hypothesis 1. There is a positive/negative relationship between CSR and tax compliance/aggressiveness, ceteris paribus
Hypothesis 2.There is a positive relationship between CSR and firm market value, ceteris paribus.
Empirical Models
Sample and Data Collection
Financial data have been collected from the System for Electronic Document Analysis and Retrieval (SEDAR) (CSA, 2014) for five years from 2005 to 2009. SEDAR reports the financial statements of all Canadian public companies and venture capitals.
CSR index data have been collected from the 2010 CSR Ranking from Corporate Knights Research Group. Corporate Knights Research Group evaluates the environmental, social, and governance activities for TSX 60 firms in Canada based on internationally recognized standards. The index is measured between 0 percent and 100 percent. The data include not only overall CSR scores but also the scores of the individual components of CSR, such as social, governance, and environmental scores.
The environmental score measures a firm's energy, carbon, water, and waste productivities.2 The governance score measures a firm's sustainability leadership, leadership diversity, and sustainability remuneration. The social score measures a firm's employee safety, compensations between CEO and workers, percentage of tax paid, pension plan, and pension funding.
Since the CSR Ranking from Corporate Knights Research Group is published in the Global and Mail every year and is also available on-line, a large percentage of the population has access to it.
Regression Models
To determine the relationship between a firm's CSR and tax aggressiveness and to test Hypothesis 1, I design the following regression models:
where
ETRi = effective tax rate for firm i
CSRi = overall corporate social responsibility index for firm i
GOVERNi = corporate governance index for firm i
SOCIALi = corporate social index for firm i
CONTROLit = a set of control variables
The numerator is cash tax paid, shown on the cash flow statement.
I control for other firm characteristics that could potentially affect a firm's CSR activities, as documented in previous studies (Waddock and Graves, 1997; Stanwick and Stanwick, 1998; Foster and Meinhard, 2002; Brammer et al., 2006; Paul and Siegel, 2006; Piercy and Lane, 2009, to name a few). They are, size (SIZE), measured as the log of total assets; return on assets (ROA), measured as net earnings over total assets; losses (LOSS), an indicator variable, equal to 1 for firms with loss carryovers at the beginning of 2005; leverage (LEV), measured as the sum of short-term and long-term debts over total assets; capital intensity (FIX), the ratio of fixed assets to total assets; and inventory intensity (INV), the ratio of inventory to total assets. Hypothesis 1 predicts α1 > 0.
The second hypothesis is on the association between CSR and firm market value. Based on Ohlson's (1995) residual income model, firm market value can be expressed as firm book value and earnings. Therefore, I design the following regression models:
where
VALi = averaged fiscal-year-end share price from 2005 to 2009 for firm i
CSRi = overall corporate social responsibility index for firm i
GOVERNi = corporate governance index for firm i
SOCIALi = corporate social index for firm i
BOOKi = shareholders’ equity, deflated by total assets for firm i
ROAi = net earnings over total assets for firm i
Hypothesis 2 predicts β1 > 0.
The original sample includes TSX 60 firms. I delete five companies which are trusts and are taxed differently than the other firms.
When using the annual effective tax rate, I further delete 38 observations with either a negative pre-tax income or with the annual effective tax rate falling outside [0, 1]. As a result, there are 237 observations left.
When using the five-year effective tax rate, I delete two companies with the five-year effective tax rate falling outside [0, 1], following other studies of effective tax rates (Dyreng et al., 2008). As a result, the final sample includes 53 firms.
Table 1 shows the distribution of the sample firms by industry according to the Global Industry Classification (GIC), which is collected from the Corporate Knights Research Group database. The major industries are energy (11 firms), financials (11 firms), materials (10 firms) and consumer staples and discretionary (10 firms).
| Industry (GICS sector) | Numbers of firms |
|---|---|
| Energy | 11 |
| Consumer staple and discretionary | 11 |
| Materials | 10 |
| Financials | 10 |
| Industrials | 4 |
| Telecommunications | 3 |
| Utilities | 2 |
| Information Technology | 1 |
| Healthcare | 1 |
| Total | 53 |
The four major industries are included to control for potential industry fixed effects in all the above regression models.
Empirical Results
Panel A from Table 2 presents the mean and median values of CSR ratings and ETR for each of the four major industries. It shows that energy has the highest mean value of the overall CSR scores; financials has the highest mean and median values of social scores; and materials has the highest mean and median values of governance scores. Financials has the lowest mean and median values of five-year ETR as well as annual ETR.
| Panel A: Descriptive statistics by industries | ||||||||
|---|---|---|---|---|---|---|---|---|
| Variable | Energy | Consumer | Material | Financials | ||||
| Mean | Median | Mean | Median | Mean | Median | Mean | Median | |
| CSR | 0.556 | 0.555 | 0.427 | 0.442 | 0.498 | 0.51 | 0.536 | 0.563 |
| SOCIAL | 0.656 | 0.687 | 0.624 | 0.61 | 0.453 | 0.434 | 0.733 | 0.72 |
| GOVERN | 0.549 | 0.681 | 0.186 | 0.042 | 0.559 | 0.683 | 0.341 | 0.37 |
| ETR1 | 0.242 | 0.246 | 0.275 | 0.282 | 0.226 | 0.217 | 0.18 | 0.183 |
| ETR2 | 0.292 | 0.239 | 0.431 | 0.296 | 0.496 | 0.191 | 0.231 | 0.207 |
| Panel B: Summary statistics for all firms | |||||||
|---|---|---|---|---|---|---|---|
| Variable | Mean | Median | 1st quartile | 3rd quartile | SD | Minimum | Maximum |
| ETR1 | 0.215 | 0.215 | 0.144 | 0.282 | 0.114 | 0.012 | 0.556 |
| ETR2 | 0.2434 | 0.2338 | 0.129 | 0.3165 | 0.1633 | 0.0021 | 0.968 |
| CSR | 0.493 | 0.506 | 0.396 | 0.617 | 0.157 | 0.026 | 0.808 |
| SOCIAL | 0.612 | 0.644 | 0.506 | 0.739 | 0.191 | 0.03 | 0.948 |
| GOVERN | 0.379 | 0.365 | 0.042 | 0.685 | 0.28 | 0 | 0.738 |
| ROA | 0.084 | 0.064 | 0.031 | 0.126 | 0.066 | 0.003 | 0.278 |
| LEV | 0.174 | 0.172 | 0.039 | 0.274 | 0.145 | 0 | 0.53 |
| INV | 0.052 | 0.016 | 0 | 0.063 | 0.073 | 0 | 0.276 |
| FIX | 0.44 | 0.469 | 0.141 | 0.677 | 0.302 | 0 | 0.929 |
| LOSS | 0.736 | 1 | 0 | 1 | 0.445 | 0 | 1 |
| SIZE | 10.919 | 10.917 | 10.488 | 11.214 | 0.685 | 9.641 | 12.475 |
| BOOK | 0.081 | 0.084 | 0.052 | 0.109 | 0.043 | 0.008 | 0.159 |
| VAL | 36.97 | 34.81 | 26.64 | 48.16 | 16.5 | 4.384 | 82.05 |
Notes
- Table reports the summary statistics for the sample firms during the fiscal years 2005–2009.
- The variables are defined as follows: ETR1 is five-year average effective tax rate, ETR2 is annual effective tax rate, CSR is overall 2010 corporate social responsible index, SOCIAL is 2010 corporate social index, GOVERN is 2010 corporate governance index, BOOK is measured as the sum of shareholders’ equity over the sum of total assets, and VAL is averaged fiscal-year-end share price from 2005 to 2009.
- Firm-specific control variables are defined as follows. ROA: measured as the sum of earnings over the sum of total assets for five years from 2005 to 2009; LEV: measured as the sum of short- and long-term debts over the sum of total assets for five years from 2005 to 2009; FIX: capital intensity, the ratio of the sum of fixed assets to the sum of total assets for five years from 2005 to 2009; INV: inventory intensity, the ratio of the sum of inventory to the sum of total assets for five years from 2005 to 2009; LOSS: indicator variable, equal to 1 with positive loss carryovers at the beginning of 2005; SIZE: measured as log of the average total assets for five years from 2005 to 2009.
Panel B from Table 2 presents the descriptive statistics of all major variables. It shows that on average, the five-year ETR is 21.5 percent, which is less than the statutory rate (around 30 percent from 2005 to 2009). The annual ETR, on average, is 24.34 percent, which is also less than the statutory rate.
Panel B from Table 2 also reveals that, on average, TXS 60 firms have an overall CSR index of 49.3 percent, a governance index of 37.9 percent and a social index of 61.2 percent. The maximum overall CSR index is 80.8 percent, while the minimum overall CSR index is 2.6 percent.
In addition, Panel B from Table 2 documents that, on average, profitability (ROA) is 0.084, leverage (LEV) is 0.174, average share price is $36.97, and more than 75 percent of firms have a loss carry-over from prior years.
Table 3 reports the Pearson correlation matrix. It shows the correlations between major variables. The maximum absolute value of the correlation is 0.808, between BOOK and SIZE. The minimum absolute value of the correlation is 0.0001, between ROA and LEV.
| Variable | CSR | ETR | ROA | LEV | INV | FIX | LOSS | SIZE | BOOK | VAL |
|---|---|---|---|---|---|---|---|---|---|---|
| CSR | 1 | |||||||||
| ETR | 0.32 | 1 | ||||||||
| ROA | −0.155 | 0.116 | 1 | |||||||
| LEV | −0.016 | −0.179 | −0.000 | 1 | ||||||
| INV | −0.145 | 0.266 | 0.244 | 0.012 | 1 | |||||
| FIX | 0.051 | −0.05 | 0.336 | 0.414 | −0.099 | 1 | ||||
| LOSS | −0.012 | −0.137 | 0.083 | 0.313 | 0.004 | 0.115 | 1 | |||
| SIZE | 0.372 | −0.096 | −0.516 | −0.192 | −0.467 | −0.385 | −0.242 | 1 | ||
| BOOK | −0.236 | 0.142 | 0.561 | −0.092 | 0.328 | 0.378 | 0.236 | −0.808 | 1 | |
| VAL | 0.293 | 0.041 | −0.031 | −0.168 | −0.059 | −0.232 | −0.294 | 0.375 | −0.258 | 1 |
Notes
- Table reports the summary statistics for the sample firms during the fiscal years 2005–2009.
- The variables are defined as follows: ETR is five-year average effective tax rate, CSR is overall 2010 corporate social responsible index, BOOK is measured as the sum of shareholders’ equity over the sum of total assets, and VAL is averaged fiscal-year-end share price from 2005 to 2009.
- Firm-specific control variables are defined as follows. ROA: measured as the sum of earnings over the sum of total assets for five years from 2005 to 2009; LEV: measured as the sum of short- and long-term debts over the sum of total assets for five years from 2005 to 2009; FIX: capital intensity, the ratio of the sum of fixed assets to the sum of total assets for five years from 2005 to 2009; INV: inventory intensity, the ratio of the sum of inventory to the sum of total assets for five years from 2005 to 2009; LOSS: indicator variable, equal to 1 with positive loss carryovers at the beginning of 2005; SIZE: measured as log of the average total assets for five years from 2005 to 2009.
Consistent with the hypotheses, ETR is positively associated with CSR; and CSR is positively associated with VAL.
Table 4 presents the results from models 10 to 12, where ETR is measured as the five-year cash ETR. Column 3 shows the results from model 10, where the overall CSR is regressed on ETR and other control variables. Column 4 shows the results from model 11, where a component of the CSR index, the social index, is regressed on ETR and other control variables. Column 5 shows the results from model 12, where a component of the CSR index, the governance index, is regressed on ETR and other control variables.
| Variable | Exp. sign | Model 10 | Model 11 | Model 12 |
|---|---|---|---|---|
| Intercept | ? | −1.4676** | −1.091* | −2.806*** |
| (−2.24) | (−1.545) | (−2.58) | ||
| ETR | + | 0.570*** | 0.648*** | 0.480* |
| (−2.841) | (−2.998) | (−1.441) | ||
| ROA | + | −0.06 | 0.373 | 0.398 |
| (−0.153) | (−0.884) | (−0.613) | ||
| LEV | ? | 0.055 | 0.304* | 0.135 |
| (−0.26) | (−1.321) | (−0.381) | ||
| INV | ? | 0.233 | 0.273 | 0.181 |
| (−0.655) | (−0.71) | (−0.306) | ||
| FIX | ? | 0.127 | 0.304*** | 0.07 |
| (−1.216) | (−2.695) | (−0.405) | ||
| LOSS | ? | 0.028 | 0.155*** | −0.111* |
| (−0.547) | (−2.823) | (−1.311) | ||
| SIZE | + | 0.161*** | 0.108* | 0.280*** |
| (−2.658) | (−1.66) | (−2.786) | ||
| Energy | + | −0.024 | −0.071 | 0.085 |
| (−0.31) | (−0.854) | (−0.669) | ||
| Materials | + | 0.062 | −0.117* | 0.348*** |
| (−0.876) | (−1.539) | (−2.956) | ||
| Financial | + | −0.053 | 0.314** | −0.272 |
| (−0.396) | (−2.185) | (−1.23) | ||
| Consumer | + | −0.051 | 0.057 | −0.125 |
| (−0.723) | (−0.741) | (−1.057) | ||
| R2 (adj) | 0.23 | 0.39 | 0.33 | |
| Observations | 53 | 53 | 53 |
Notes
-
The table reports the testing results from the following regression models:
where ETR is five-year average effective tax rate, CSR is 2010 overall corporate social responsible index, SOCIAL is 2010 corporate social index, GOVERN is 2010 corporate governance index.
- Firm-specific control variables are defined as follows. ROA: measured as the sum of earnings over the sum of total assets for five years from 2005 to 2009; LEV: measured as the sum of short- and long-term debts over the sum of total assets for five years from 2005 to 2009; FIX: capital intensity, the ratio of the sum of fixed assets to the sum of total assets for five years from 2005 to 2009; INV: inventory intensity, the ratio of the sum of inventory to the sum of total assets for five years from 2005 to 2009; LOSS: indicator variable, equal to 1 with positive loss carryovers at the beginning of 2005; SIZE: measured as log of the average total assets for five years from 2005 to 2009.
- ***, **, and * denote significance at the 0.01, 0.05, and 0.10 levels, respectively, based on one-tailed t-test.
Column 3 shows that, consistent with Hypothesis 1, the coefficient on ETR is positive and statistically significant at the 0.01 level, which implies that paying more taxes contributes to a higher overall CSR ranking. Similarly, column 4 shows that the coefficient on ETR is positive and statistically significant at the 0.01 level, which implies that firms that pay more taxes enhance their social scores. Finally, column 5 shows that the coefficient on ETR is positive and significant at the 0.1 level, which implies that paying more taxes also helps enhance governance scores. In summary, Table 4 indicates that a high effective tax rate is positively associated with the overall CSR ranking as well as all the three CSR components.
A few control variables are also significant. For example, the coefficient on SIZE is positive and significant for all the three models, which implies that large firms have high CSR scores as well as high social and governance scores. This is consistent with the argument that large firms are most likely to face public scrutiny and be forced to invest heavily in CSR. In addition, large firms generally have more resources and are therefore better able to make CSR a priority.
LEV and FIX are positively associated with the social index. LOSS is positively associated with social scores but negatively associated with governance scores.
Finally, Table 4 shows that industry effect exists. Financial institutions are positively associated with social scores. Materials sector is positively associated with governance scores but negatively associated with social scores.
Table 5 presents the results from models 10 to 12, where ETR is measured as annual cash ETR. Column 3 shows the results from model 10, where the overall CSR is regressed on annual cash ETR and other control variables. Column 4 shows the results from model 11, where a component of the CSR index, the social index, is regressed on annual cash ETR and other control variables. Column 5 shows the results from model 12, where a component of the CSR index, the governance index, is regressed on annual cash ETR and other control variables.
| Variable | Exp. sign | Model 10 | Model 11 | Model 12 |
|---|---|---|---|---|
| Intercept | ? | −1.1*** | −0.788*** | −1.986*** |
| (−4.327) | (−2.827) | (−5.076) | ||
| ETR | + | 0.087* | 0.151** | 0.082 |
| (−1.463) | (−2.317) | (−0.894) | ||
| ROA | + | 0.02 | 0.281** | 0.275* |
| (−0.15) | (−1.874) | (−1.31) | ||
| LEV | ? | 0.029 | 0.213** | 0.05 |
| (−0.329) | (−2.24) | (−0.376) | ||
| INV | ? | 0.149 | 0.145 | 0.089 |
| (−0.919) | (−0.817) | (−0.356) | ||
| FIX | ? | 0.062* | 0.208*** | 0.028 |
| (−1.306) | (−4.013) | (−0.386) | ||
| LOSS | ? | 0.008 | 0.12*** | −0.119*** |
| (−0.364) | (−4.99) | (−3.52) | ||
| SIZE | + | 0.149*** | 0.105*** | 0.232*** |
| (−5.899) | (−3.799) | (−5.97) | ||
| Energy | + | 0.03 | −0.013 | 0.14*** |
| (−0.034) | (−0.342) | (−2.653) | ||
| Materials | + | 0.092*** | −0.072** | 0.349*** |
| (−2.895) | (−2.06) | (−7.145) | ||
| Financials | + | −0.067 | 0.233*** | −0.261*** |
| (−1.215) | (−3.843) | (−3.073) | ||
| Consumer | + | −0.009 | 0.102*** | −0.111** |
| (−0.282) | (−2.865) | (−2.212) | ||
| R2 (adj) | 0.24 | 0.38 | 0.45 | |
| Observations | 237 | 237 | 237 |
Notes
-
The table reports the testing results from the following regression models:
where ETR is annual effective tax rate, CSR is 2010 overall corporate social responsible index, SOCIAL is 2010 corporate social index, and GOVERN is 2010 corporate governance index.
- Firm-specific control variables are defined as follows. SIZE: measured as log total assets; ROA: measured as net earnings over total assets; LOSS: indicator variable, equal to 1 with positive loss carryovers at the beginning of 2005; LEV: measured as the sum of short- and long-term debts over total assets; FIX: capital intensity, the ratio of fixed assets to total assets; INV: inventory intensity, the ratio of inventory to total assets.
- ***, **, and * denote significance at the 0.01, 0.05, and 0.10 levels, respectively, based on one-tailed t-test.
Column 3 shows that, consistent with Hypothesis 1, the coefficient on annual ETR is positive and statistically significant at the 0.1 level, which implies that paying more taxes contributes to a higher overall CSR ranking. Similarly, column 4 shows that the coefficient on annual ETR is positive and statistically significant at the 0.01 level, which implies that firms that pay more taxes enhance their social scores. However, column 5 shows that the coefficient on annual ETR is not significant, which implies that paying more taxes annually does not enhance a firm's governance scores. In summary, Table 5 indicates that a high effective tax rate is positively associated with the overall CSR ranking and the social scores.
A few control variables are also significant. Some of the results are similar to those in Table 4. For example, the coefficient on SIZE is positive and significant for all three models; LEV and FIX are positively associated with the social index; and LOSS is positively associated with social scores but negatively associated with governance scores.
In addition, Table 5 shows that ROA is positively associated with both the social and governance scores.
Finally, Table 5 shows that industry effects exist. The four industries are all significant for some or all CSR scores.
Table 6 presents the results from models 13 to 15, where I link CSR ranking and its components to firm market value. Column 3 shows the results from model 13, where firm market value is regressed on the overall CSR, book value, and earnings. Column 4 shows the results from model 14, where firm market value is regressed on the social index, book value, and earnings. Column 5 shows the results from model 15, where firm market value is regressed on the governance index, book value, and earnings. It shows that, consistent with Hypothesis 2, the coefficient on CSR is positive and statistically significant at the 0.01 level, which implies that a high overall CSR ranking enhances firm market value. Similarly, column 5 shows that the coefficient on the governance index is positive and statistically significant at the 0.01 level, which implies that a high governance score also enhances firm market value. Column 4, however, shows that the coefficient on social index is not significant. In addition, energy and materials are negatively associated with firm market value, but financials and consumer staple and discretionary are positively associated with firm market value.
| Variable | Exp. sign | Model 13 | Model 14 | Model 15 |
|---|---|---|---|---|
| Intercept | ? | 10.327 | 25.995** | 18.441** |
| (−0.909) | (−2.063) | (−2.217) | ||
| BOOK | + | 11.242 | −27.599 | 35.947 |
| (−0.127) | (−0.294) | (−0.426) | ||
| ROA | + | 65.4* | 56.559* | 49.690* |
| (−1.625) | (−1.318) | (−1.303) | ||
| CSR | + | 37.84*** | ||
| (−2.628) | ||||
| SOCIAL | + | 8.665 | ||
| (−0.641) | ||||
| GOVERN | + | 30.58*** | ||
| (−3.555) | ||||
| Energy | ? | −5.407 | −1.409 | −9.537* |
| (−0.801) | (−0.2) | (−1.433) | ||
| Materials | ? | −9.25 | −4.141 | −17.05** |
| (−1.177) | (−0.502) | (−2.116) | ||
| Financials | ? | 15.52** | 14.58* | 17.04** |
| (−1.896) | (−1.661) | (−2.19) | ||
| Consumer | ? | 7.326 | 7.059 | 8.811* |
| (−1.115) | (−0.998) | (−1.41) | ||
| R2 (adj) | 0.18 | 0.07 | 0.26 | |
| Observations | 53 | 53 | 53 |
Notes
-
The table reports the testing results from the following regression models:
where CSR is overall corporate social responsibility index, GOVERN is corporate governance index, SOCIAL is corporate social index, BOOK is measured as the sum of shareholders’ equity over the sum of total assets for five years from 2005 to 2009, and ROA is measured as the sum of net earnings over the sum of total assets for five years from 2005 to 2009.
- ***, **, and * denote significance at the 0.01, 0.05, and 0.10 levels, respectively, based on one-tailed t-test.
I use White's (1980) test to find that heteroskedasticity is not a problem.
The CSR ranking from Corporate Knights Research Group reveals that tax payment is a factor that determines social and final scores (i.e., tax payment contributes 20–25 percent to social scores and 6–10 percent to final scores). The tax-payment factor is calculated based on the average percent of statutory taxes paid over the last four fiscal years (via annual reports), that is, cash tax over taxes owed at the statutory rate.
In a robust test, I adjust the social and final CSR scores by excluding tax payment. I calculate the average cash effective tax rate over four years before CSR ranking is disclosed. For those firms with an average cash effective tax rate higher than the statutory tax rate, their social scores are reduced by 20–25 percent and the final scores are reduced by 6–10 percent. The results, not presented in the paper, are similar to those in Table 5 to the extent that ETR is positively associated with the social, governance, and final scores.
It is desirable to directly compare the results in this paper with those for the U.S. firms presented in prior literature (Watson, 2011a,b; and Davis et al., 2013). However, the CSR measurements used in prior literature are different from those used in this study. For example, Watson (2011a,b) and Davis et al. (2013) use CSP ratings produced by Kinder, Lydenberg and Domini Research & Analysis. The data are only for the U.S. firms, which provide integer scores of strength and concerns in several categories including corporate governance, community, diversity, environment, human rights, and so on. The ratings can be positive or negative.
To make a comparison between the U.S. firms and the Canadian firms, in a supplementary test I have collected the overall CSR rankings for 2012 and 2013 for the global 100 most sustainable corporations from Corporate Knights Research Group. There are 18 U.S. firms and 13 Canadian firms in 2013 among the global top 100 sustainable corporations. In 2012, there are 10 U.S. firms and 10 Canadian firms among the global top 100 sustainable corporations. The database from Corporate Knights Research Group provides a uniform measurement of overall CSR for both the U.S. firms and Canadian firms. Initially there are 51 firm-year observations. After deleting 6 observations with unavailable financial reports, negative pre-tax earnings, or effective tax rates falling outside [0, 1], there are 45 observations left. Financial data for the U.S. firms have been collected from the SEC 10-K database. Financial data for Canadian firms have been collected from the SEDAR database. A Canadian indicator variable is included in model 10 to test whether or not there is a difference in the Canadian context. Therefore, the regression model is designed as follows:
where
CAN = 1 for Canadian firms and 0 for U.S. firms.
Table 7 presents the results from model 16. It shows that the coefficient on CAD×ETR is positive and statistically significant at the 0.05 level. It suggests that, in Canada, paying more taxes contributes to a higher overall CSR ranking.
| Exp. sign | Model 16 | |
|---|---|---|
| Intercept | ? | 0.428* |
| (1.643) | ||
| CAN | ? | −0.037 |
| (−0.857) | ||
| ETR | + | −0.246* |
| (−1.693) | ||
| CAN×ETR | + | 0.297** |
| (1.878) | ||
| ROA | + | 0.71** |
| (2.34) | ||
| LEV | ? | −0.126** |
| (−1.781) | ||
| INV | ? | 0.244 |
| (1.098) | ||
| FIX | ? | −0.016 |
| (−0.214) | ||
| LOSS | ? | 0.17** |
| (2.387) | ||
| SIZE | + | 0.014 |
| (0.675) | ||
| Energy | + | −0.048* |
| (−1.469) | ||
| Technologies | + | −0.015 |
| (−0.35) | ||
| Financial | + | −0.067* |
| (−1.358) | ||
| Biotech | + | 0.003 |
| (−0.081) | ||
| R2 (adj.) | 0.09 | |
| Observations | 45 |
Notes
-
The table reports the testing results from the following regression models:
where ETR is annual effective tax rate, CSR is overall corporate social responsible index for 2012 and 2013, CAN is indicator variable, equal to 1 for Canadian firms. Firm-specific control variables are defined as follows. SIZE: measured as log total assets; ROA: measured as net earnings over total assets; LOSS: indicator variable, equal to 1 with net loss in last year; LEV: measured as the sum of short- and long-term debts over total assets; FIX: capital intensity, the ratio of fixed assets to total assets; INV: inventory intensity, the ratio of inventory to total assets.
- ***, **, and * denote significance at the 0.01, 0.05, and 0.10 levels, respectively, based on one-tailed t-test.
Conclusion and Summary
This paper examines the relationship of CSR, tax aggressiveness and firm market value. An economic model is developed based on the idea by Barboza and Trejos (2011) to show that profit-maximization firms would be willing to incur additional costs in CSR (such as paying more taxes) as long as they can differentiate their products from those of non-CSR firms, and that socially conscious consumers will buy products from CSR firms at prices higher than those of non-CSR firms. The empirical study in this paper indicates that socially responsible firms are less likely to undertake aggressive tax activities, while those firms that are less interested in being socially responsible are more likely to undertake aggressive tax activities. It also indicates that a reputation of higher CSR will enhance a firm's market value. Using Canadian companies listed in the S&P/TSX 60 index, I find that both firms’ five-year effective tax rates and annual effective tax rates are positively associated with their overall CSR scores as well as their social index. Firms’ five-year effective tax rates are also positively associated with their governance index. However, the annual effective tax rates are not associated with the governance index. Based on Ohlson's (1995) residual income model, I also find that firms’ overall CSR ranking and governance index are positively associated with their market value. However, the social index is not associated with firm market value. Overall, the empirical results show that paying more taxes leads to higher CSR rankings and thus enhances firm market value. In a supplementary test for cross-country comparison, I find that paying more taxes contributes to a higher overall CSR ranking in the Canadian context.
It should be noted that firms implementing CSR may enjoy favorable tax treatments; for example, charitable donations are tax deductible as are R&D expenditures incurred in environmental and sustainability activities. These tax treatments will reduce the tax liability of CSR firms and hence will be biased against the results in this paper.
This study provides important evidence that increases our understanding of the relation between CSR and tax aggressiveness in countries beyond the United States that have a different institutional environment and industry mix.
This study is of interest to policymakers, corporate managements, and academics who wish to examine the relationship between CSR and other firm characteristics. Given that CSR has recently attracted increased attention from academics, businesses, governments, and other bodies, future studies should be conducted to explore the reason why CSR differs considerably across firms. Future studies may also extend the model in this paper to include other business goals besides maximizing profits.
Finally, the results from this study must be interpreted with caution, since the sample is a small portion of the large population of Canadian public firms. The small sample size may reduce the strength of the tests and generalizability of some of the results.
