1. Introduction
Unethical business practices are nowadays under the spotlight, attracting the attention not only of the justice system but also of the media, due to the increasing awareness of the need to protect Planet Earth. Therefore, firms make significant efforts not only to take actions to protect the environment, the quality of life of their workers, and local communities but also to communicate those initiatives to a wide range of stakeholders, including, of course, the shareholders, as many of these firms have trade shares in international equity markets. All these actions are under the umbrella of the so-called corporate social responsibility (CSR) of companies.
Indeed, CSR must be understood under the “triple bottom line” approach explained by Elkington (1998), which expands on traditional corporate reporting to consider social and environmental performance in addition to financial performance. In addition, the acronym “Environmental, Social and Governance” (ESG) is commonly used to refer to CSR activities because these are the three most important factors when measuring corporate sustainable behaviour.
Many studies have analysed the relationship between the CSR of the companies listed in capital markets using a wide range of financial dimensions (for instance, information asymmetry, earnings management, audit fees, cost of debt, debt ratios or equity ratios). Benlemlih (2017) presents a broad literature review in all these different areas, referring to only six recent articles until 2014 on the specific topic of cost of equity (CoE). Note that CSR disclosure is a potential mediator of the CSR–CoE relationship, but not an indispensable one, and there may be other types of impact in corporate finance. For instance, the positive link between CSR and corporate financial performance (CFP) is also evidenced when companies with higher CSR scores have greater CFP reflected on return over assets (Waddock and Graves, 1997). CSR disclosure increments resources to finance profitable projects, thereby reducing capital constraints (Cheng et al., 2014) and increasing the firm’s market value, especially in weak equity and debt markets with higher transaction costs (El Ghoul et al., 2017). Hence, and especially for listed companies, CoE is a very powerful indicator in financing and strategical decision making. It represents the return or percentage gain capital providers will obtain by investing on the firm’s shares, which can be measured according to different definitions or proxies. These expected returns are valuable for capital budgeting, portfolio allocation, performance assessment, active risk control, and even firm valuation.
Our study is timely and valuable as this is a growing area of research and the latest literature review only revised six articles on CSR and CoE covering up to 2014 (Benlemlih, 2017). Note that the European Union requires greater disclosure of nonfinancial and diversity information by large companies (Directive 2014/95/EU, 2014). European countries have transposed this directive such that companies are required to include some nonfinancial information in their annual report or separate reports from 2018 onwards. Therefore, the new requirements imply a new setting where future research will try to assess the impact of increased CSR engagement on financial aspects such as CoE. In addition, companies might be even more willing to make new disclosures on sustainability if new evidence is found for the benefits of CSR regarding CoE reduction. Therefore, the contribution of this literature review study relies on its extension (22 articles), covering a wide time span from its origin in research (2001) up to the latest publications in 2018, at a time when more requirements for sustainability disclosure are being introduced in the European Union, and when there is also an increasing trend to inform about nonfinancial information around the world. Hence, the topic of CSR and CoE becomes even more attractive for future researchers, managers, or society as a whole.
The definitions and methodologies used to measure CoE are diverse, and there is also a wide range of variables measuring the sustainability profile of a company. Henceforth, the main purpose of this article is to present a literature review of the relationship between CSR and CoE on an international scale, focusing on the theoretical frameworks referred to, the methodologies employed, the different approaches to measure the cost of equity together with the sustainability measures and control variables, and the regional settings examined, with the aim of providing a base for future research.
The remainder of the paper is organized as follows. Section 2 explains the methodology employed to define the scope and sample of the studied articles. Section 3 describes different theoretical frameworks, the models implemented by authors to calculate the cost of equity, the independent variables required to measure the sustainable profile of firms, and control variables. Additionally, we analyse different regional contexts where the research between CoE and CSR has been examined through the years. The paper closes with the conclusions section, with new ideas for studies in this field in order to open future avenues for research within this specific topic—i.e., suggestions related to innovative studies on CSR and CoE.
3. Results and Discussion
3.1. Theoretical Frameworks
Existing research on CSR and CoE is framed mostly within three theories, as follows.
According to the legitimacy theory, companies face different political and social pressures from various stakeholders. Hence, firms might feel pushed to adopt a “social” behaviour to legitimise themselves (De Villiers and Van Staden, 2006) or disclose CSR information by facilitating media coverage so that society can accept the company’s business practices (Li et al., 2017). These disclosures may be manipulated by companies to maintain a socially responsible image (De Villiers and Marques, 2016). Controversial (e.g., mining, tobacco, gambling, and sexual activities) or sensitive industries often create negative externalities (Hong and Kacperczyk, 2009; El Ghoul et al., 2011), so these firms are expected to disclose more detailed ESG information in order to legitimise themselves and reduce their CoE by compliance with social norms to establish business operations and good relations with stakeholders (De Villiers and Marques, 2016).
Out of the 22 articles under analysis of CSR and CoE, 4 articles (18%) refer to legitimacy theory (Li et al., 2017; Martínez-Ferrero and García-Sánchez, 2017; Michaels and Gruning, 2017; Weber, 2018). In line with promoting a good image among stakeholders, CSR represents an important variable to improve firm’s legitimacy by receiving favourable news coverage (Li et al., 2017) and increasing its credibility with external assurance (Martínez-Ferrero and García-Sánchez, 2017).
Agency theory is based on the “stewardship problem” between principals (shareholders) and agents (management), since the latter may act according to their “economic self-interest”, possibly expropriating the firm’s assets against the shareholders’ interest because of misuse of private information. Furthermore, shareholders must incur costs to monitor agent’s activities and bear risk (Jensen and Meckling, 1976). Voluntary disclosure (such as CSR information, for instance) reduces information asymmetries by reducing transaction costs, increasing the demand of shares, and consequently decreasing CoE (Diamond and Verrecchia, 1991). Moreover, shareholder’s future risk estimations decrease and, finally, non-diversified risk also diminishes (Coles et al., 1995; Clarkson et al., 1996). The agency theory also provides a framework in the sense that management can be encouraged to pursue and disclose CSR activities in order to maximise salary bonus, secure job positions, and avoid future conflicts with stockholders (Healy and Palepu, 2001), reducing information asymmetry and the company’s financial uncertainty (Richardson and Welker, 2001; Verrecchia, 2001). In addition, voluntary external assurance of nonfinancial information (Sierra et al., 2014; Zorio et al., 2014) reduces agency conflict and their CoE (Martínez-Ferrero and García-Sánchez, 2017). In spite of all the literature above, Breuer et al. (2018) cite agency theory in anticipation of a positive relationship between CSR and CoE as they argue that CSR can be a costly diversion of scarce resources.
Out of the 22 articles reviewed for this literature review, 6 articles (27%) refer to agency theory (Gupta, 2015; Feng et al., 2015; Martínez-Ferrero and García-Sánchez, 2017; Suto and Takehara, 2017; El Ghoul et al., 2018; Breuer et al., 2018).
Finally, stakeholder theory posits that companies need to have good relationships with different agents—including employees, customers, suppliers, government institutions, and shareholders—as well as by practising and disclosing socially responsible activities. Regarding CSR nonfinancial disclosures, the firm’s objective is to inform stakeholders about the different socially responsible activities with the purpose of maintaining business operations, reducing financial risk and, consequently, the CoE (Li and Foo, 2015). Social arrangements determine the importance of different stakeholders. In fact, an empirical study related to CSR disclosures with companies operating in liberal market economies (United States and United Kingdom) and economies where the state leads a more important supervisory role in the economy (Spain, Portugal, France) concluded that firms with headquarters in state-led economies disclose more detailed stakeholder information (labour, environment, community) than firms in liberal market economies (Gallego-Alvarez and Quina-Custodio, 2017). As regards code of law countries, the state or government plays a major role in economic decisions and different social groups (e.g., unions, clients, providers of capital, suppliers) exert pressure to regulate political actions in favour of their interests (La Porta et al., 2008; Liang and Renneboog, 2017).
Finally, as regards our specific literature review on CSR and CoE, 10 out of the 22 articles (45%) are framed within the stakeholder theory (Richardson and Welker, 2001; El Ghoul et al., 2011; Reverte, 2012; Dhaliwal et al., 2014; Harjoto and Jo, 2015; Li and Foo, 2015; Martínez-Ferrero and García-Sánchez, 2017; Michaels and Gruning, 2017; El Ghoul et al., 2018; Breuer et al., 2018).
Some of the analysed articles refer to a multifaceted theoretical approach (as Martínez-Ferrero and García-Sánchez, 2017, who build on the three theories mentioned above). On the other hand, other articles in our specific review on CSR and CoE do not specifically refer to any particular theory. Note for instance Xu et al. (2014), who simply mention stakeholder theory as a keyword but do not refer to it in the text, or Li and Liu (2018), who refer to stakeholders and information asymmetry reduction yet do not properly build the conceptual theories behind this.
Other articles in our review (e.g., Dhaliwal et al., 2011; Reverte, 2012; Martínez-Ferrero and García-Sánchez, 2017; Weber, 2018) also refer to voluntary disclosure theory (Botosan and Plumlee, 2005; Botosan, 2006), as sustainability engagement and CSR reports are used as a means to decrease information asymmetries and increase confidence in how social and environmental risks are managed (Li and Foo, 2015). From our point of view, this theory is very much in line with the postulates of the three theories above, being somehow a summary of the consequences of the other three explained theories.
Finally, Li et al. (2017) also refer to signalling theory, which investigates the implications of signals in information asymmetry.
3.2. Research Methodologies
Next, we explain the different methodological approaches of the 22 articles under study.
In most of the studies in our literature review, authors use regression analysis (Richardson and Welker, 2001; Sharfman and Fernando, 2008; Reverte, 2012; Feng et al., 2015; Kim et al., 2015; Li and Foo, 2015; Gupta, 2015; Eom and Nam, 2017; Michaels and Gruning, 2017; Weber, 2018; Li and Liu, 2018). Li et al. (2017) uses both ordinary least squares (OLS) and generalised least squares (GLS), while Weber (2018) uses a matched sample of GRI CSR with a propensity score as well as regression analysis.
Some studies address potential endogeneity and self-selection bias by carrying out two-stage regressions (2SLS) (as in Dhaliwal et al., 2011, 2014; Xu et al., 2014; Harjoto and Jo, 2015; Ng and Rezaee, 2015; Suto and Takehara, 2017; Breuer et al., 2018; Li and Liu, 2018) or the generalised methods of moments (GMM), following the technique developed by Blundell and Bond (1998), as in El Ghoul et al. (2011); Gupta (2015) or by Arellano and Bond (1991) in Martínez-Ferrero and García-Sánchez (2017). El Ghoul et al. (2018) uses both two-stage least squares (2SLSs) estimation and a dynamic system with GMM.
3.3. Dependent, Independent, and Control Variables
In the 22 empirical studies of our sample, the dependent variable is the implied CoE, also called cost of capital in some studies. It can be defined as the percentage rate of return to obtain the market value of an asset by discounting future cash flows dividends (El Ghoul et al., 2011); or, in other words, the internal rate of return that investors expect to gain by maintaining their capital investment on the firm (Reverte, 2012; Suto and Takehara, 2017).
The models used to obtain the different proxies for CoE are shown in Figure 1.
In the majority of the 22 analysed articles, ex-ante models are used to calculate CoE considering forecasted earnings based on an analyst’s consensus and current prices. Reverte (2012) argues that the CoE obtained with ex-ante models is a better proxy than ex-post realised stock returns (by definition, a backward-looking measure). CoE is more reliant on cross-sectional variation amongst the companies, and it does not need long-time series to be robust and is not dependant on a specific asset pricing model. Gupta (2015) also supports the use of ex-ante models because they can account for unexpected news on cash flow or a firm’s fundamentals, which obviously cannot be not depicted in ex-post measures, and provides many examples of the growing volume of literature using this approach. Hence, the ex-ante models to calculate CoE in the articles in the sample follow mostly Claus and Thomas (2001), Gebhardt et al. (2001), Easton (2004), and Ohlson and Juettner-Narouth (2005). Note that Claus and Thomas (2001) assume the market price expressed as projected residual and book value earnings for a time horizon of 5 years and constant dividend payouts. Gebhardt et al. (2001) consider stock prices as return over equity projections with a time horizon from 2 to 12 years. The price earnings growth (PEG) model considers a time horizon of 2 years, projected earnings per share for year 2, year 1, current market price, and zero dividend payments (Easton, 2004). The previous model is recommended for CoE estimations by Botosan and Plumlee (2005) and Botosan et al. (2011), and implemented by Reverte (2012), Martínez-Ferrero and García-Sánchez (2017), and Michaels and Gruning (2017). Out of the 22 studies on CSR and CoE, 19 (86%) implement this type of ex-ante model.
Gordon’s constant growth model is generalised by Ohlson and Juettner-Nauroth (2005), which defines the cost of equity using stock market price, one-year projected earnings, and perpetual growth rate. Ng and Rezaee (2015) proxy CoE with a variation of the price multiple—the industry-adjusted earnings–price ratio—as well as with the finite horizon expected return model (Gordon and Gordon, 1997).
In some studies (e.g., Dhaliwal et al., 2014; Li and Foo, 2015; Gupta, 2015; Breuer et al., 2018), CoE is calculated as an average of the three proxies developed by Gebhardt et al. (2001), Claus and Thomas (2001), and Easton (2004) and/or their further refinements, or even use the average plus some of the proxies, as in Xu et al. (2014). Some authors deduce risk free government note yields from the average result (Eom and Nam, 2017).
Ex-post models are also used to examine the relationship of CSR and CoE. For instance, the capital asset price model (CAPM) states that the expected market return is measured by implementing a single-variable stochastic model (Sharpe, 1964; Lintner, 1965). The model takes into account the risk-free data composed of treasury note debt yields; beta, which encloses the systematic relationship between the market and the firm’s individual risk; and, finally, the risk-free premium obtained by the difference between market return and the risk-Free rate. The CAPM approach is implemented to obtain CoE and analyse its relationship with environmental risk management disclosures in Sharfman and Fernando (2008).
Finally, other studies apply the three-factor model to obtain (or even explain) CoE. This model proposed by Fama and French (1993) considers additional variables, i.e., beta, the market-to-book ratio, and size (as in Suto and Takehara, 2017, following Fama and French, 1997). Note that some of the analysed articles used ex-ante models to estimate CoE and the variables in the three factor-model as control variables, as we explain in the next subsection (Dhaliwal et al., 2011; El Ghoul et al., 2011; Reverte, 2012; Li and Foo, 2015; Feng et al., 2015).
Some authors in our sample not only explore the relationship of CSR with CoE, but also with the cost of debt (e.g., Suto and Takehara, 2017), information asymmetry (Michaels and Gruning, 2017), corporate value through Tobin’s Q (Eom and Nam, 2017), or the firm risk and its investor base (Breuer et al., 2018).
The relationship between CoE and sustainability has been examined through many different lenses using a wide range of independent variables to contrast different hypotheses, controlling for different aspects somehow determined by the sample and, of course, including some kind of measure for CSR reporting or for the sustainable behaviour of the company.
Therefore, regarding independent variables, different authors provide several proxies. For instance, El Ghoul et al. (2011) uses a community score (based on volunteer programs, charity donations, relation with indigenous people, and taxes, among others), as well as an employee score and environmental score based on many other items. Reverte (2012) uses a CSR score quintile rank. Xu et al. (2014) use a CSR score, also taking into consideration partial scores as CSR investors (stockholder’s participation on firm’s decision making), CSR employees (health, salary, wages, social benefits, equal recruitment, etc.), CSR customers (R&D activities, consumer satisfaction, etc.), CSR suppliers (fair trade, protection of suppliers’ know-how), CSR community involvement (taxes, no child labour, anti-corruption policy, charities, etc.), and CSR environment (good environmental policies and management of pollutants, waste). Harjoto and Jo (2015) create a CSR index based on the disclosure items required by norms yet not by laws. Gupta (2015) further subdivides its main variable, environmental sustainability index (ESI), into three components, namely emission reduction, product innovation, and resource reduction. Michaels and Gruning (2017) develop a disclosure index based on an artificial intelligence model run on the actual CSR reports. Ng and Rezaee (2015) use an economic sustainability measure with three components—growth opportunities, operational efficiency, and research effort. El Ghoul et al. (2011), Dhaliwal et al. (2011), and Weber (2018) also consider if the company is a strong CSR performer or not. El Ghoul et al. (2018) use the ratio of environmental costs to total assets as a proxy for corporate environmental responsibility.
Reverte (2012) takes CSR into account through the ratings provided by the Observatory on Corporate Social Responsibility reports. Suto and Takehara (2017) use data on corporate social performance from an annual CSR questionnaire survey sent to all listed firms in Japanese capital markets. Richardson and Welker (2001) use social performance measures from Canadian society, that sponsored assessments of the annual reports.
Inclusion in a capital market index to indicate socially responsible investment has also been used as a proxy for sustainability (Eom and Nam, 2017).
Carbon information disclosure is obtained from CSR reports (Li et al., 2017, to calculate disclosure index) or from the National GHG Emission Information System of the Ministry of the Environment (Kim et al., 2015 to calculate carbon intensity as total GHG emission/sales).
Table 2 shows the CSR dimensions or measures studied by the revised articles, as well as the implemented methodology (e.g., OLS, GLS, and robustness tests).
Different control variables have been used by extant research (Table 3).
3.4. Sample Characteristics, Regional Settings, and Main Research Findings
In this section, we analyse the samples used by prior research, paying special attention to their size, covered period, and regional setting. Given the nature of the dependent variable (CoE), this type of study is obviously restricted to listed companies. Studies vary from having a very small sample size (114 observations in Reverte, 2012, in just one country, Spain) to having large samples in a multi-country setting (74,077 observations from 31 countries as in Dhaliwal et al., 2014, for instance).
The first study covered just three years (Richardson and Welker, 2001), whereas the latest data in the samples used correspond to 2002–2015 in Breuer et al. (2018). The more recent studies show many robustness checks, bearing in mind many different categories for some variables (for instance, the years in order to control for the effect of the economic crisis as in El Ghoul et al., 2018).
All of the studies confirm the negative relationship expected between CSR and CoE, except for Richardson and Welker (2001), and, in addition, Suto and Takehara (2017) and Eom and Nam (2017) do not find significant results.
The first study on CoE and sustainability covered 124 firms from Canada, with data from 1990–1992 (Richardson and Welker, 2001). Evidence was found of a significant positive relation between social disclosures and the cost of equity capita, even though companies had better financial performance.
However, most of the existing research demonstrates a negative relationship between CSR and CoE. In addition, the majority of studies are based on US samples because of the availability analyst forecast data. Sharfman and Fernando (2008) use a sample with 267 US firms, concluding that benefits from improved environmental risk management lead to a reduction in CoE. Dhaliwal et al. (2011) analyse 1190 CSR reports from 294 US firms, finding that companies with a high CoE in the previous year tend to initiate CSR disclosure in the current year, and that initiating firms with high CSR performance enjoy a subsequent reduction in CoE, attracting institutional investors and analyst coverage. El Ghoul et al. (2011) with a sample of 12,915 firm observations conclude that firms with better CSR scores have lower CoE, whereas belonging to “sin” industries (namely, nuclear power, and tobacco) increases CoE. Harjoto and Jo (2015) analyse 2034 US firms, demonstrating that, amongst other consequences, CSR intensity (more specifically, legal CSR) reduces CoE. Finally, Weber (2018) uses 878 CSR reports prepared under the G3 or G3.1 Guidelines from 2005 to 2013, finding that GRI disclosure level has not had an impact on CoE per se, yet a negative association for poor CSR performers reporting at high GRI level and CoE is observed, if the report is assured.
In Asia, and more particularly in China, Korea, and Japan, studies on this relationship have also been undertaken.
In China, Xu et al. (2014), using a sample of listed firms in Shanghai Stock Exchange, show that firms with higher CSR scores have significantly lower CoE, an effect which is more pronounced in times of economic recession. In addition, even though state-owned companies have better CSR and lower CoE than the others, the effect of CSR in reducing CoE is weaker in state-owned companies. Li and Foo (2015) use 1015 CSR report quality scores, finding evidence that CSR report quality is strongly and negatively related with CoE, with a much higher impact in lowering CoE for privately owned corporations, even though the distinction between mandatory versus voluntary CSR disclosures does not have a significant impact on CoE. Li et al. (2017) focus on 161 listed companies operating in heavily polluting industries, showing that media reporting improves carbon information disclosure (both financial and nonfinancial disclosures), which is negatively associated with CoE. Finally, also in China, Li and Liu (2018) analyse 1708 observations from 2008 to 2014, finding that the quality of the CSR disclosure is negatively related to CoE, especially amongst environmentally sensitive industries, state-owned enterprises, and those that are larger in size.
In the Korean setting, the sample of Kim et al. (2015) includes 379 firms from the period 2007 to 2011, concluding that carbon intensity is positively related to CoE, no matter whether the companies voluntarily published sustainability reports or not, with this effect being lower for firms belonging to industrial sectors with large carbon emissions. Eom and Nam (2017) use 86 companies listed in Korea, from 2009 to 2013. Their results cannot provide evidence of any significant relation between the incorporation of the SRI index and CoE or corporate value. Moreover, depending on the phase of introduction of the SRI index, the results revealed a negative or positive correlation with CoE, probably because of mixed optimistic and pessimistic expectations from investors about CSR activities, or because this index may not correctly reflect CSR performance.
Suto and Takehara (2017), using 3461 observations of Japanese firms, did not provide evidence of a significant relationship between corporate social performance and CoE, yet institutional ownership has a strongly negative influence on CoE.
In Spain, Reverte (2012), using only 114 firm-year observations from 26 firms covering from 2003 to 2008, finds a significant negative relationship between CSR disclosure ratings and CoE, which is even more pronounced for those firms operating in sensitive industries.
In Germany, Michaels and Gruning (2017) analyse the relation between CoE and CSR disclosures of 264 sensitive firms in 2013/2014. The disclosure variable was obtained through combining a series of coding schemes to score CSR disclosures. Results demonstrate that CSR-sensitive firms have greater equity costs but implementing voluntary disclosure decreases CoE.
In an international setting, several studies find evidence of the negative relationship between CSR and CoE. Using data from different countries obviously allows authors to explore the impact on national characteristics, such as stakeholder protection and governance or cultural variables.
Dhaliwal et al. (2014) using a sample with 5135 standalone CSR reports from 1093 companies find a negative link between CSR disclosures and CoE which is more pronounced in stakeholder-oriented countries. They also provide evidence that financial and ESG disclosures behave like substitutes in reducing CoE. The sample of Feng et al. (2015) includes 10,803 firm-year observations from 25 countries. They find evidence that, in general, firms with better CSR scores have a significantly reduced CoE in North America and Europe, yet these results do not hold in Asian countries (which is somehow contrary to the evidence obtained for China or Korea by Li and Foo (2015), Li et al. (2017), Li and Liu (2018), and Kim et al. (2015)) as explained above.
Ng and Rezaee (2015) use a sample with over 3000 firms (countries not explicit), from 1990 to 2013, finding a negative association between CoE and growth and research (environmental and governance) sustainability performance, with social performance being only marginally related to CoE. In general, the relationship is strengthened when ESG performance is strong.
Gupta (2015) uses 23,301 firm-year observations from 2002 to 2012. The obtained findings suggest that better environmental practices (mostly reduction of emissions and waste) lead to lower CoE, with this effect being more pronounced in countries where country-level governance is weak.
Martínez-Ferrero and García-Sánchez (2017) use a sample of 1410 companies from 17 countries between 2007 and 2014, and find evidence of lower CoE for companies that publish and assure their social and environmental reports, especially if the assurance is provided by a Big 4 firm. Breuer et al. (2018), with a sample of 19,183 firm-year observations from between 2002 and 2015, find that if investor protection is high (low), CoE decreases (increases) if the company invests in sustainability. El Ghoul et al. (2018) analysed manufacturing firms in 30 countries from 2002 to 2011 and find that CoE is lower if companies have higher environmental responsibility.
In most of the studies in an international setting, we noted that the samples tend to be dominated by Anglo-Saxon countries because of data availability (for example, see Breuer et al. 2018 and El Ghoul et al., 2018).