1
The effects of Corporate Social Responsibility on the
capital structure of U.S. firms.
Erik Zuidema*
January 11, 2018
Abstract
There is an increasing demand for sustainability which stresses the importance of corporate social responsibility (CSR) and how this affects financial markets. Increased CSR performance results in firms to be perceived as less risky, which in turn increases credit ratings. In combination with tax advantages, this increases the attractiveness of debt financing. On the other hand, a high CSR performance increases demand for the associated firm’s stocks, which in turn result in a lower cost of equity which makes equity financing more attractive. This paper uses 494 U.S. firms in the period 2006 – 2016 to see how CSR performance affects the capital structure of those firms. We find that CSR scores are positively and significantly related to the equity ratio’s, implying that we find support for the equity financing hypothesis. A unit increase in the overall CSR score increases BookEquity by 9 basis points and MarketEquity by 26 basis points. CSR performance does not only affect a firm’s financial performance and cost of capital, but also the capital structure of the firm.
Keywords: Social Responsible Investment, Corporate Social Responsibility, Financial Performance, Cost of Equity Capital, Cost of Debt Capital, Capital Structure, Leverage
2
1. Introduction
Corporate Social Responsibility (CSR) has become increasingly important in the last decades. Increased publicity and media attention put more pressure on firms to operate and behave in a way which is perceived as sustainable. This pressure and increased attention affects the way in which corporations behave and report about their activities, but also affects the investment choice of large institutional investors and individuals. This gave rise to the term Social Responsible Investing (SRI), where investors take into account CSR related activities to decide whether or not to invest in a firm. These activities may be costly for firms, but go beyond the scope of profit maximization. The activities include social, economic and environmental activities that benefit all the stakeholders and hence the society as a whole.1 While cost of
related activities are often measurable directly, the benefits are indirect and appear in the long run. The activities may benefit firms in terms of reputation, increased sales, establishing loyal customer relationships, media attention and a lower cost of capital. CSR related activities are means of communication containing signals for stakeholders and hence are important for management in determining firm strategies. Numerous researchers have examined the relationship between CSR and financial performance, where some added cost of capital implications to this relationship. However, only a few papers examined how these relations affect the capital structure of a firm. In this paper we examine the effects of CSR performance on the capital structure of U.S. firms. We collaborate on theoretical mechanisms at work that cause CSR to affect a firm’s capital structure and empirically test how CSR performance affects the equity ratios of U.S. firms.
Cost of capital implications are important since the cost of equity is used as the discount rate that the market applies to a firms future cash flows and it is a key input in a firm’s long term investment decisions. Cost of capital implications, in turn, have implications on a firm’s capital structure. Modigliani and Miller’s (1953, 1968) irrelevance theorem taught us that a trade-off exists between tax advantages and bankruptcy risk. Whilst the theorem still has large implications today, a lot of research has been conducted that stresses the importance of the capital structure choice of a firm, which in turn affects firm strategy and strategic choices the firm makes (Jensen, 1986). The capital structure implications of CSR performance are important in several ways. First, they affect a firm’s risk profile: increased leverage results in increased bankruptcy risks. Second, it affects the ability to obtain funding and how expensive that funding is. Therefore, it is important to examine how CSR affects a firm’s capital structure.
1 The term Corporate Social Responsibility (CSR) has been defined differently by authors over the years. The
3
The implications affect manager’s decisions and hence strategic planning. In addition, it is interesting to see what the effects of the individual dimensions of CSR on capital structure are. Researchers that analyzed the effects of CSR on firm performance find both positive, negative as well as no relationship. First, CSR commitments can be seen as investments in intangible assets like a firm’s reputation that increase firm value. Second, CSR commitments can be thought of a costly representation of private benefits that actually destroy firm value. Third, CSR performance may not be priced by capital markets at all. The effect of CSR performance on financial performance may also occur through different channels, namely the cost of capital. First, better CSR performance is related to a lower cost of equity capital. This occurs through the size of an investor base and through perceived riskiness of a firm (Ghoul et al., 2011). A higher demand for high scoring CSR firms in combination with increased media attention result in a larger investor base of the associated firms which results in a lower cost of equity through diversification arguments. (Heinkel et al., 2001). The lower cost of equity capital increases the attractiveness for a firm to finance itself through equity, which we call the equity financing hypothesis. Moreover, increased CSR performance decreases the perceived riskiness of firms as there are less uncertain future explicit claims. For example, a firm that scores well in the environmental pillar has a lower chance of lawsuits that result from environmental scandals. This may increase a firm’s credit rating which results in lower interest rates. In combination with tax advantages of debt financing, this makes debt financing more attractive for the firm, which we call the debt financing hypothesis. We argue that the theoretical mechanism increase a firms access to financial markets and try to whether a net effect dominates.
To examine and verify the relationship between CSR performance and capital structure we apply different multivariate regression methods. First, we perform pooled OLS regressions with equity ratios as the dependent variable, a CSR performance measure as the explanatory variable and we include a set of control variables that affect the link between CSR performance and leverage. In addition to an overall CSR score, we run our regressions with individual dimensions of CSR as the explanatory variable to see whether particular dimensions have different effects on a firm’s capital structure. Afterwards, we apply other panel regression methods to increase the reliability of our results.
4
In this paper we examine the effects of CSR on the capital structure of U.S. firms. We try to verify existing theoretical mechanisms and extend the traditional interrelations between CSR performance, financial performance and cost of equity by taking on a different perspective. We do so through a firm’s capital structure. If it is true that mechanisms exist that result in lower cost of equity or debt capital, this should also be reflect in a firm’s leverage ratios. We find a significant and positive relationship between the CSR score of a firm and its associated equity ratio. A unit increase in the overall CSR score increases BookEquity by 9 basis points and MarketEquity by 26 basis points. The same results hold for the individual pillar scores with respect to their signs. This would suggest that equity financing becomes more attractive relative to debt financing. Our results are robust after performing several sensitivity analyses. We apply White’s cross-section standard errors and employ random effects models.
We contribute to the existing literature by including capital structure to the traditional framework of relationships between CSR performance, financial performance and cost of capital. In addition, we do not only examine a single overall CSR measure, but also decompose this measure to see whether different aspects of CSR may have different effects on a firm’s capital structure. The results have implications for management and strategic planning as they should reconsider the values and weights they assign to CSR commitments. It affects the pricing of shares, cost of capital, but also a firm’s reputation and the access to financial markets. Investors should take into account that financial markets works as a mechanism that rewards high CSR firms with lower required returns. Governments may also indirectly influence the relationship between CSR performance and capital structure through regulation with regards to CSR.
The remainder of the paper is structured as follows. Section 2 provides a theoretical background on the relationship between CSR and firm performance, CSR and cost of capital implications and how it relates to capital structure. Section 3 describes the methodology used in more detail. In section 4, we describe the data used in our research and provide summary statistics. Section 5 presents the results of our regressions and section 6 discusses these results and concludes.
2. Literature Review
5
aspects of investors and find that from the existing CSR factors2, both social responsible
investors as non-social responsible investors consider environmental aspects to be the most important.
The significant increase in CSR has fed the research on the relation between CSR and financial performance of firms and their associated stocks. It leads to an extension of the traditional asset pricing model where there’s a tradeoff between risk and return. Many papers find so-called ‘CSR-effects’ on financial performance, including positive and negative effects. Other papers find that CSR has no significant effect on financial performance.
Cai and He (2014) create an innovative way to screen stocks and find evidence that high scoring CSR firms outperform their low scoring counterparts in the long run. Using different methodologies, including Carhart’s four factor model and Fama-Macbeth regressions, they examine whether CSR engagements can be seen as cost beyond compliance that shrinks profits or as an asset which is preferred by investors and hence increases financial performance. They find that an SRI portfolio earned 3% above industry benchmarks in the 4th
year after screening. Hence, they suggest that SRI is an intangible asset which is likely to be undervalued by the market. CSR generates shareholder value that is generally underpriced by the markets due to its intangible feature, thereby triggering excess returns.
In contradiction, some papers find significant negative relationships between CSR performance and financial performance, most importantly because of increased costs that arise when firms are subject to environmental and social restrictions. Renneboog et al. (2008) examine the performance of SRI stocks worldwide and find that SRI funds underperform their domestic benchmarks by -2.2% to -6.5%. Building on models of Fama and French and Carhart, they find that SRI investors derive non-financial utility from investing in companies that meet certain CSR standards and are willing to accept a lower rate of return. Ciciretti et al. (forthcoming) successfully isolate the price of taste for responsible assets by ruling out both systematic and residual risk components using the Fama-Macbeth two step cross-sectional framework. Demand for SRI can be driven by a risk effect, where investors perceive favorable risk characteristics such as a reduction in stakeholder risk. Furthermore, the demand can be driven by taste effects, where investor’s preferences are unrelated to the risk-return tradeoff of an asset. They find a significant and negative relation between social responsibility scores and risk adjusted returns and find that existence of the taste effect results in lower risk adjusted returns amounting to 4.8% annually.
Mollet and Ziegler (2014) examine the relation between SRI and stock performance and find insignificant abnormal stock returns for both U.S. and EU stock markets, therefore supporting the view that SRI stocks are correctly priced by market participants. They also use several asset pricing models to calculate risk adjusted returns. Bauer et al. (2005) find no significant
2 The existing literature on Corporate Social Responsibility includes 4 pillars of interest. These include
6
differences in risk-adjusted returns between ethical and conventional funds, using the Carhart multifactor model. They conclude that CSR is not priced by financial markets and hence does not add or destroy value in terms of risk-adjusted returns. Galema et al. (2008) find an insignificant relationship and examine the contraction of positive and negative results more closely. They find that misinterpretation of results takes place, because controlling for systematic risk in financial performance calculations not fully captures the trade-off between CSR and financial performance, and that aggregation of SRI measures may confound existing relationships between individual dimensions of SRI and returns.
The cost increase as a result of CSR practices in turn resulted in the public debate who bears, or who should bear those increased cost. Firms can increase prices for their products or services and thus let the consumer bear the cost for CSR practices, but employees may also accept lower wages or shareholders may accept lower dividends in return for being socially responsible. Thereby they participate to the sustainability of the environment.
Research has been done with regards to the stock returns of ‘sin’ stocks. Glushkov and Statman (2009) find that typical social responsible portfolio’s do not include stocks of companies associated with tobacco, alcohol, gambling, firearms etc. These so-called ‘sin’ stocks have higher expected returns compared to socially responsible stocks, since demand for these stocks is lower and a higher return is required to guarantee investments in those stocks. In addition, Hong and Kacperczyk (2009) find that firms operating in those ‘sin’ industries are held less by norm-constrained institutions like pension funds, receive less coverage from analysts and media, and have higher expected returns.
While most of the papers focus on the relation between CSR and financial performance, only a few studies have focused on the effects of CSR on the cost of equity capital and cost of debt capital of firms. To my knowledge, very little research is done on how these CSR effects on the cost of capital actually affect the capital structure of firms and the firm’s associated choice to finance itself through debt or through equity.
7
They use multiple methods to calculate their cost of capital, including CAPM. They apply multiple regression methods for their different hypothesis and find evidence for the debt financing hypothesis, suggesting that the higher level of environmental risk management increases the firms leverage. However, in contrast to their theoretical explanation they empirically find that increased environmental risk management increases the cost of debt capital. To explain this contrary result, they state that debt markets possibly continue to see risk management as beyond necessary in terms of compliance and punish firms who engage in such behavior. In addition, increased environmental performance increases the willingness of debt markets to provide financing, and increased leverage viciously increases the cost of debt capital because of increased bankruptcy risks.
Ghoul et al. (2011) examine the relation between CSR engagement and the cost of equity. In their paper they build on the theoretical frameworks of Merton (1987) and Heinkel et al. (2001) and hypothesize that higher CSR firms have lower cost of equity compared to lower CSR firms, and that low CSR firms are being associated with a smaller investor base and a higher perception of risks. The conscious inclusion of high CSR stocks by investors increase the investor base for high CSR firms, and decrease the investor base of low CSR firms. Cost of equity is higher for the low CSR firms to compensate investors for a lack of risk sharing. In their paper, they use the ex-ante cost of equity implied in analyst earnings forecasts and stock prices, which is also their contribution to the existing literature since many papers rely on the CAPM to estimate cost of equity. They apply pooled cross-sectional time series regressions with robust standard errors at the firm level, and find that firms with a better CSR score exhibit lower cost of equity after controlling for firm specific determinants as well as for industry and year fixed effects. In addition, they find that investment in improving responsible employee relations, environmental policies and product strategies substantially contribute to reducing the firms cost of equity. They also examine two sin business sectors, and find that in these sectors the cost of equity is higher, which is in line with existing literature. Trinks et al. (2017) add to the literature by introducing a key transparent, quantitive and relative measure of environmental performance, namely Greenhouse Gas emission intensity. Using observations of 1920 firms in the period 2002 – 2016, they find that industry-adjusted GHG emissions intensity positively impacts the cost of equity. Whilst the focus lies here on the environmental aspect of CSR, they find that lower intensity is associated with a lower cost of equity capital, which would support the equity financing hypothesis. Ghoul et al. (2016) reexamine the relationship between environmental performance and cost of equity and find that cost of equity capital is lower when firms have a higher responsibility after controlling for firm-level characteristics, as well as industry, year and country effects. They successfully deal with endogeneity problems by using instrumental variables and account for noise in analysts’ forecasts.
8
positive, negative or no relation between CSR and financial performance. Moreover, research conducted provides a negative relationship between CSR commitment and cost of equity capital, and mixed results about the effects of CSR on the cost of debt capital. Whilst the reduction in cost of equity capital makes equity financing more appealing, the possible reduction in cost of debt in addition to tax advantages also increases the attractiveness of debt financing.
3. Methods
To analyze the effects of CSR scores on the capital structure of the firms in our dataset, we apply different multivariate regression methods. Predictions about the effects of CSR on capital structure and its sign are ambiguous. A higher CSR score may reflect that the firm is perceived as safer, and hence obtains a higher credit rating. Decreased interest rates would increase the attractiveness of debt financing for the associated firm. In addition, the firm may be able to reap higher profits because of the associated tax shield with debt financing. Consequently, we expect that the firm increases its leverage. However, we do not necessarily expect a reduction in the cost of debt capital, since increased leverage increases the cost of debt because of bankruptcy risks. On the other hand, there’s an increasing demand for sustainability, suggesting that demand for the high CSR firm’s stocks increases. This results in higher share prices and, ceteris paribus, decreases the cost of equity, which makes equity financing more attractive. We therefore expect that the firm will decrease its leverage. One of the main goals of this paper is see which effect on leverage dominates. Moreover, we try to take a different perspective in the field of interest by not solely focusing on the relationship between CSR and financial performance and the cost of equity. Instead, we try to verify the existence of these relationships through a different channel, namely a firm’s capital structure. First, we run several pooled OLS regressions using our book value and market value capital structure proxies as the dependent variables. Panel regression methods are preferred over alternative approaches because we are able to address a broader range of issues and the ability to remove the impact of omitted variable bias in our regressions. An important and relevant omitted variable bias could be unobserved firm heterogeneity (Horváthová, 2010). We adopt the following pooled OLS model:
𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒 𝑖,𝑡 = 𝛼 + 𝛽1∗ 𝐶𝑆𝑅𝑖,𝑡 + 𝛽2∗ 𝐵𝐸𝑇𝐴𝑖,𝑡+ 𝛽3∗ 𝑆𝐼𝑍𝐸𝑖,𝑡 + 𝛽3∗ 𝑃𝑇𝐵𝑖,𝑡+ 𝛽4∗
𝑆𝐺𝑖,𝑡+ 𝑢𝑖,𝑡 (1)
where 𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒 𝑖,𝑡 represents our proxy for capital structure (𝐵𝑜𝑜𝑘𝐸𝑞𝑢𝑖𝑡𝑦 and
𝑀𝑎𝑟𝑘𝑒𝑡𝐸𝑞𝑢𝑖𝑡𝑦) for firm 𝑖 at time 𝑡, 𝐶𝑆𝑅𝑖,𝑡 the CSR score for firm 𝑖 at time 𝑡. To increase the
9
respectively. Moreover, we follow previous literature by adding a set of important control variables to increase the robustness of our results. Explanations and definitions of the variables and its expected signs and impact are found in Section 4.
The pooled OLS regressions implicitly assume that average values of variables and relationships are constant over time and across all cross-sections. With the presence of firm heterogeneity or trends over time, this assumption may not be realistic. Therefore, we also introduce a fixed effects model, including year-fixed effects. The model captures time variation to see whether leverage changes over time but not cross-sectionally. Note that we remove the intercept from the equation to avoid the dummy variable trap, where there’s perfect multicollinearity between included dummy variables and the intercept. The model is given as follows:
𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒 𝑖,𝑡 = 𝛽1 ∗ 𝐶𝑆𝑅𝑖,𝑡−1+ 𝛽2∗ 𝑋𝑖,𝑡+ 𝑦𝑒𝑎𝑟𝑖+ 𝑢𝑖,𝑡 (2)
where 𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒 𝑖,𝑡 represents our proxy for capital structure (𝐵𝑜𝑜𝑘𝐸𝑞𝑢𝑖𝑡𝑦 and
𝑀𝑎𝑟𝑘𝑒𝑡𝐸𝑞𝑢𝑖𝑡𝑦) for firm 𝑖 at time 𝑡, 𝐶𝑆𝑅𝑖,𝑡 the CSR score for firm 𝑖 at time 𝑡. Again, White’s
cross-sectional standard errors are used. In addition to the weighted average score, we run the regressions for the individual pillar scores as well. 𝑋𝑖,𝑡 reflects our set of control variables
shown in Equation (1) and described in Section 3. 𝑦𝑒𝑎𝑟𝑖 denotes a year dummy variable that
takes the value of 1 for the concerning year and 0 otherwise and 𝑢𝑖,𝑡 captures the error term.
Additionally, we run a cross-section random effects model where we assume that the individual intercepts per firm arise from a common intercept α (which is the same for all the firms and over time) plus a random variable that varies cross-sectionally but is constant overtime. This random variable Є𝑖 measures the random deviation of each firm’s intercept
from the common intercept α. Note that there are no dummy variables anymore to capture firm heterogeneity but that this occurs via Є𝑖. The random effects model is given as follows:
𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒 𝑖,𝑡 = α + 𝛽0𝑖 + 𝛽1∗ 𝐶𝑆𝑅𝑖,𝑡+ 𝛽2∗ 𝑋𝑖,𝑡+ +ѡ𝑖,𝑡, ѡ𝑖,𝑡 = Є𝑖+ 𝑣𝑖,𝑡 (3)
where 𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒 𝑖,𝑡 represents our proxy for capital structure (𝐵𝑜𝑜𝑘𝐸𝑞𝑢𝑖𝑡𝑦 and 𝑀𝑎𝑟𝑘𝑒𝑡𝐸𝑞𝑢𝑖𝑡𝑦) for firm 𝑖 at time 𝑡, 𝐶𝑆𝑅𝑖,𝑡 the CSR score for firm 𝑖 at time 𝑡. 𝑋𝑖,𝑡 reflects our
set of control variables again and ѡ𝑖,𝑡 the new error term. It includes the cross-sectional error
term Є𝑖 and the individual observation error term 𝑣𝑖,𝑡. Again, White’s cross-sectional standard
errors are used. The model assumes that Є𝑖 has zero mean, is independent of 𝑣𝑖,𝑡 and 𝑋𝑖,𝑡, and
has a constant variance.
10
4. Data and descriptive statistics
We use the Asset4 database to obtain a set of companies and their associated company codes in North America. To be included in our dataset, the firms have to be large enough to be publicly traded, and there should be CSR scores available. Our final data set ends up as an unbalanced panel of 494 firms, with yearly data of 11 main variables from 2006 to 2016, an 11-year period. This results in a total number of 5242 firm-year observations. Moreover, we classify the industries the firms operate in using Standard Industry Classification (SIC) codes. SIC codes were developed by the U.S. government to provide a code that covers all economic activities. The codes contain multiple digits, where the first digit represent the business segment which provides the most revenue for the company. We classify the firms to the industries they operate in to see whether the effects of corporate social responsibility on capital structure differs across industries. Table A1 in the appendix shows the SIC codes and their associated industries. We divide our variables in three different categories, including corporate social responsibility scores, capital structure proxies and financial control variables. They are described in the following subsections, as well as the descriptive statistics and correlations between the variables.
4.1 CSR scores
To examine the effects of social responsible investments on capital structure, we need some proxies to reflect how firms are performing in the area of CSR. We use Thomson Reuters’ Asset4 database to obtain A4IR ratings for the companies in the period 2006-2016 on a yearly basis. The A4IR score reflects how a company's financial and extra-financial health can be equally weighted based on the information in Asset4's economic, environmental, social and corporate governance pillars. It reflects a balanced view of a company's performance in these four areas. The overall score lies between the boundaries of 0 and 100, where a higher score indicates a better overall performance in the mentioned areas. We use this score as our main indicator for CSR performance (CSR).In addition, we obtain the four pillar scores on which the overall CSR score is based individually, to see whether some pillars have a more significant impact on the capital structure of U.S. firms.
11
The environmental pillar (Environmental) measures a company's impact on living and non-living natural systems, including the air, land and water, as well as complete ecosystems. It reflects how well a company uses best management practices to avoid environmental risks and capitalize on environmental opportunities in order to generate long term shareholder value.
The social pillar (Social) measures a company's capacity to generate trust and loyalty with its workforce, customers and society, through its use of best management practices. It is a reflection of the company's reputation and the health of its license to operate, which are key factors in determining its ability to generate long term shareholder value.
4.2 Capital Structure proxies
We use Thomson Reuters’ Datastream to obtain yearly data on several financial variables that we need for our capital structure proxies. We use multiple proxies to reflect the capital structure of a firm to increase the robustness of our regressions. We define the book value of equity divided by total assets as BookEquity, which is our first proxy for a firm’s capital structure. In addition, we multiply the amount of shares outstanding times the official closing share price of the associated firm to calculate the market value of equity of a company. Subsequently, we divide this amount by our proxy for the market value of total assets to obtain MarketEquity, which is our second proxy for a firm’s capital structure. We follow previous literature with regards to our proxy for the market value of total assets, which is calculated by subtracting the book value of equity from total assets and adding the market value of equity. This procedure is used since it is very difficult to determine a market value of total assets.
4.3 Control variables
Consistent with previous literature, we include several financial control variables in our regression. Those variables are obtained from Thomson Reuters’ Datastream, where some variables are determined by Datastream self and others were obtained through the Worldscope database. First, we control for the beta of the associated firm (Beta), which reflects the relationship between the volatility of a stock price and the volatility of the market. The coefficient is based on between 23 and 35 consecutive month end price percent changes and their relativity to a local market index. Secondly, we control for firm size (Size), which is measured as the natural logarithm of total assets. Furthermore, we control for the price-to-book ratio of the firm (Price to Book), which is measured as the share price divided by the price-to-book value per share. Lastly, we include a measure for sales growth (Sales Growth) as a control variable, which is calculated as follows: the current year’s net revenues are divided by the net revenues five years ago, reduced to a compound annual rate. This number is deducted by 1 and multiplied by 100 to obtain our measure for sales growth.
12
(Feldman, Soyka, and Ameer, 1997). A higher CSR score may reflect a firm’s ability to better deal with sustainability changes and other events, which results in a lower beta and hence a lower cost of equity capital (Sharman and Fernando, 2008). Thus, we expect that an increase in a firm’s beta increases cost of equity capital and vice versa, and therefore we expect a negative relationship between Beta and our equity capital structure proxies. In addition, larger firms are considered to be less risky as they benefit from lower operating and financial risks (Fama and French, 1993), which supports the debt financing hypothesis through higher credit ratings and the increased ability to reap tax benefits. In contradiction,Merton (1987) develops a capital market model with incomplete information. Larger firms receive more attention from financial analysts and the media, which reduces information asymmetries. This in turn decreases the cost of equity and hence supports the equity financing hypothesis. The effect of size is therefore ambiguous. A high price-to-book ratio reflects that the firm is perceived as safe in contraction to low price-to-book ratio’s, who may indicate a financially distressed position and therefore are expected to earn higher ex post returns (Fama and French, 1992). We expect that financially distressed firms to be highly leveraged in relative terms (and therefore have lower equity ratios). Therefore, low price-to-book firms will have relatively low equity ratios and high price-to-book firms will have higher equity ratios. Thus, we expect a positive relationship between a firm’s price-to-book ratio and its equity ratios. Lastly, we find that sales growth signals solvency (Bradley and Chen, 2015), which will positively affect share prices and in turn decrease cost of equity capital. This supports the equity financing hypothesis and therefore we expect a positive relationship between sales growth and our capital structure proxies.
4.4 Descriptive statistics and correlations
Table 1 provides descriptive statistics for the key variables described in the above sections. Panel A reports the statistical properties for our capital structure proxies, namely BookEquity and MarketEquity. The average value of BookEquity at the firm level over our sample period is 0.36, which is lower than the median level of 0.37. This suggests that less than fifty percent of the firms have a BookEquity lower than the average (Panel A, column 2 and 3). The same holds for MarketEquity, where the mean is 0.55 compared to a median of 0.59. The average equity ratios for market values are higher than for book values, suggesting that firms are more leveraged in book values than in market values. The higher equity ratio’s may indicate that a firm’s principal and interest payments make up a less significant part of the firms cash flows and hence carry a smaller burden and can be perceived as more safe. On the other hand, the higher equity ratios also may indicate that a firm fails to engage in debt opportunities that result in increased profits from financial leverage and associated tax advantages.
13
a higher value corresponds with a higher level of responsibility. The average total responsibility score is 64.73, which is lower than the median of 71.83, suggesting that more than fifty percent of the firms have a score higher than the average (Panel B, column 2 and 3). When analyzing the different dimensions, we find out that the highest mean responsibility score is 77.72 for corporate governance, whilst the lowest mean responsibility score is 54.24 for the environmental dimension. This may be a surprising observation when we stress the importance of environmental behavior in the last decade. Societal and regulatory pressures may force firms to behave environmentally responsible which may result in increased environmental scores in the future.
Panel C gives the properties of the main financial control variables used in our regression. The average value of Beta is 1.06, implying that on average the firm’s stocks have a higher volatility than the market. The average price-to-book ratio is 2.65 which tells us that on average the market values of shares exceed the associated book values. The average 5 year sales growth is 6.80, suggesting that on average firms have a positive sales growth.
Table 1: Descriptive Statistics for Key Variables at Firm Level
This table reports the summary statistics of the yearly data about the three main categories of variables. Panel A shows statistical properties for our Capital Structure proxies. Panel B shows statistics for the Corporate Social Responsibility scores, including the overall score as the 4 individual dimensions. Panel C shows the statistical properties for the key financial control variables.
Panel A: Capital Structure proxies
N Mean Median St.Dev Min Max
BookEquity 5240 0.36 0.37 0.22 -1.88 1.03
MarketEquity 5240 0.55 0.59 0.24 0.00 1.00
Panel B: Corporate Social Responsibility scores
N Mean Median St.Dev Min Max
CSR 5240 64.73 71.83 27.11 4.34 98.22
Corporate Governance 5240 77.72 80.96 14.46 6.22 97.85
Economic 5240 61.27 66.05 26.78 1.70 98.97
Environmental 5240 54.24 59.55 32.61 8.26 96.86
Social 5240 56.93 60.42 27.67 3.53 98.59
Panel C: Financial variables
N Mean Median St.Dev Min Max
Beta 5240 1.063 1.020 0.66 -2.01 4.38
Size 5240 16.47 16.32 1.38 12.14 21.67
Price to Book 5240 2.65 2.33 41.58 -2211.22 1107.56
14
Table 2 describes the correlations between our key variables for our three categories of variables. We see a strong positive correlation between BookEquity and MarketEquity (Panel A). This is because both proxies have equity in the denominator.
Panel B shows the correlations between our overall responsibility score and its four individual dimensions, as well as the correlations between the dimensions self. The overall responsibility score has the strongest correlation with the social dimension, namely a positive correlation of 0.90. In contrast, the overall score has the weakest correlation with the corporate governance dimension, which is still positive and has a value of 0.68. The main take away here is that all the four individual dimensions are strongly and significantly correlated to each other, all with a positive sign. The correlations between the main control variables beta, size, price-to-book and sales growth respectively are very weak or even zero and often insignificant, as can be seen in panel C of table 2. This suggests that the main control variables do not really influence each other.
Table 2: Correlations between Key Variables
This table reports the main pairwise correlations between our main key variables. Panel A shows the correlations between our Capital Structure proxies. Panel B shows the correlations between the responsibility scores and Panel C represents the correlations between the key financial control variables. * indicates a p-value below 0.1, ** a p-value below 0.05 and *** a p-value below 0.01
Panel A: Capital Structure proxies
1) 2)
BookEquity (1) 1.00
MarketEquity (2) 0.68*** 1.00
Panel B: Corporate Social Responsibility scores
1) 2) 3) 4) 5) CSR (1) 1.00 Corporate Governance (2) 0.68*** 1.00 Economic (3) 0.75*** 0.39*** 1.00 Environmental (4) 0.89*** 0.55*** 0.49*** 1.00 Social (5) 0.90*** 0.55*** 0.60*** 0.78*** 1.00
Panel C: Financial Variables
1) 2) 3) 4)
Beta (1) 1.00
Size (2) 0.00 1.00
Price to Book (3) -0.02 0.00 1.00
Sales Growth (4) 0.06*** -0.03** 0.00 1.00
15
pillars. This would not necessarily lead to biased estimates, but the separate effect of the individual scores will have higher standard errors. To deal with this multicollinearity, we run a regression using the total pillar score as independent variable. Afterwards, we run separate regressions using the individual pillar scores only as independent variable in the associated regression.
In addition, we classify the firms to their associated industries to see whether the leverage ratios differs across industries. Table 3 reflects the summary statistics for both BookEquity in panel A and MarketEquity in panel B across different industries. The mean value of BookEquity varies between 0.24 and 0.47, indicating that capital structures may vary across industries. The Finance, Insurance and Real Estate (FIRE) industry has the lowest value of BookEquity of 0.24. This indicates that this sector mostly finances itself through debt instead of equity. This may indicate that firms in this industry are able to finance itself through debt, and enjoy the benefits it brings in terms of profitability through tax shields. However, higher debt also increases the burden on the firm and hence the chance for insolvency. The sector can be categorized as relatively unstable but with high growth opportunities. On the other hand, the Construction sector has the highest BookEquity of 0.47. Firms in this industry finance itself through equity more relative to the other industries. This may indicate that they are less able to successfully apply for loans, less able to benefit from tax shields but a lower chance for insolvency. The sector can be categorized as stable with lower growth opportunities relative to the other industries. With regards to MarketEquity, the mean varies between 0.33 and 0.68, where the FIRE industry still has the lowest value. The highest value, however, is now achieved by the Retail Trade sector. They can be interpreted in the same way as described for BookEquity. An important remark has to be made with regards to the sample sizes of the industries. For example, there are only 77 observations for the Construction sector. When the sample size of this industry would be as high as for FIRE and Manufacturing, the BookEquity and MarketEquity would be more reliable.
16
Table 3: Industry Summary Statistics for Capital Structure proxies
This table reports the most important descriptive statistics for both BookEquity and MarketEquity per industry, allocated using the SIC classification. Panel A reports the statistics for BookEquity, Panel B for MarketEquity.
Panel A: BookEquity Panel B: MarketEquity
N Mean St.Dev N Mean St.Dev
Mining 319 0.41 0.14 319 0.60 0.17
Construction 77 0.47 0.15 77 0.56 0.19
Manufacturing 2033 0.41 0.22 2033 0.63 0.20
Transportation and Public
Utilities 615 0.30 0.14 615 0.46 0.17
Wholesale Trade 121 0.40 0.16 121 0.63 0.17
Retail Trade 483 0.41 0.23 483 0.68 0.18
Finance, Insurance and Real
Estate 1071 0.24 0.18 1071 0.33 0.25
Services 681 0.36 0.29 681 0.63 0.22
5. Results
First, we perform our pooled OLS regressions for both BookEquity and MarketEquity. We show these results using both the total CSR score (CSR) as explanatory variable, as well as for the individual pillar scores separately. We run those different regressions separately to see an individual effect per pillar and interpret its associated sign. All the pillars are highly, positively and significantly correlated with each other. This implies that often when a firm scores high in one pillar, it also scores high in other several pillars. When we would include all the pillars in one sole model, this would give rise to the problem of multicollinearity. To deal with this problem, we run regressions for the pillars separately and use White’s cross-section standard errors.
17
The regressions with MarketEquity also have a higher adjusted 𝑅2, which can be seen as the fraction of variance in the dependent variable that is explained by our explanatory variables. Therefore, our market value capital structure proxy will be more reliable. The values of the coefficients are very close to each other after subdividing the CSR scores into their individual pillar levels. Corporate governance has the highest impact on the market equity ratio and environmental the lowest. This is in contradiction with Berry and Junkus (2010), who found that the environmental pillar is actually considered the most influential pillar. We note that the coefficients are very close to each other. The control variables coefficients all have the expected sign and are mostly statistically significant, except for the price-to-book ratio. Beta and the equity ratios are negatively related at the 1% significance level. A higher beta results in a higher cost of equity, decreasing the attractiveness of equity financing as was expected. We find a significant and negative relationship between Size and our equity ratios, implying that larger firms are more leveraged. The firm being perceived more safely results in better credit ratings and lower cost of debt, and the ability to reap benefits from a tax shield. This effect is stronger than the resulting lower cost of equity because of lesser information asymmetries. We find that Price to Book is only significant for the corporate governance and the economic pillar at the 10% level (panel B and C). They do have the expected positive sign in these pillars, but it should be noted that they are very close to zero which implies that Price to Book would only have a very small impact on the capital structure of a firm. This is a surprising outcome, as it contradicts the existing literature where they find that the price-to-book ratio has a positive and significant impact on the equity ratio of a firm. Lastly, there is a positive relationship between sales growth and the equity ratios at the 1% level, which is in line with previous literature. The increased sales growth indicates solvency and profitable opportunities, resulting in a lower cost of equity capital making equity financing more attractive.
18
Table 4: Pooled OLS Capital Structure Regressions
This table presents results from regressing our capital structure proxies on the CSR scores and controls over the period 2006-2016. Panel A reflects the regressions using the weighted average CSR score as main explanatory variable for both BookEquity and MarketEquity as dependent variables. Panel B, C, D and E do the same using Corporate Governance, Economic, Environmental and Social as main
explanatory variables for both BookEquity and MarketEquity as dependent variables respectively. BookEquity is the book value of equity divided by total assets, MarketEquity the market value of equity divided by the market value of total assets. Beta reflects the relationship between a stock’s volatility with the market, Size reflects the natural logarithm of total assets, Price to Book the price-to-book-ratio at the end of the fiscal year and Sales Growth the sales growth of the associated firm over a 5 year period. * reflects statistical significance at the 10% level, ** at the 5% level, *** at the 1% level.
Panel A: CSR Panel B: Corporate Governance
BookEquity MarketEquity BookEquity MarketEquity
CSR 0.0009*** 0.0026*** Corporate Governance 0.0008*** 0.0029*** (0.0000) (0.0000) (0.0000) (0.0000) Beta -0.0099*** -0.0365*** Beta -0.0125*** -0.0421*** (0.0071) (0.0000) (0.0007) (0.0000) Size -0.0522*** -0.1006*** Size -0.0476*** -0.0873*** (0.0000) (0.0000) (0.0000) (0.0000)
Price to Book 0.0001 0.0000 Price to Book 0.0001* 0.0001
(0.1312) (0.4623) (0.0923) (0.1901)
Sales Growth 0.0017*** 0.0026*** Sales Growth 0.0016*** 0.0024***
(0.0001) (0.0000) (0.0003) (0.0000) Intercept 1.1631*** 2.0644*** Intercept 1.0874*** 1.7961*** (0.0000) (0.0000) (0.0000) (0.0000) N 5242 5242 N 5242 5242 Adj. 𝑅2 0.1143 0.3499 Adj. 𝑅2 0.1080 0.3050 Panel C: Economic Panel D: Environmental
BookEquity MarketEquity BookEquity MarketEquity
Economic 0.0010*** 0.0023*** Environmental 0.0008*** 0.0020*** (0.0000) (0.0000) (0.0000) (0.0000) Beta -0.0090** -0.0348*** Beta -0.0114*** -0.0400*** (0.0357) (0.0000) (0.0016) (0.0000) Size -0.0517*** -0.0953*** Size -0.0528*** -0.0996*** (0.0000) (0.0000) (0.0000) (0.0000)
Price to Book 0.0001* 0.0001 Price to Book 0.0001 0.0000
(0.0912) (0.3552) (0.1652) (0.4960)
Sales Growth 0.0016*** 0.0023*** Sales Growth 0.0017*** 0.0026***
19 Panel E: Social BookEquity MarketEquity Social 0.0006*** 0.0022*** (0.0000) (0.0000) Beta -0.0113*** -0.0381*** (0.0035) (0.0000) Size -0.0503*** -0.0964*** (0.0000) (0.0000) Price to Book 0.0001 0.0000 (0.1342) (0.3795) Sales Growth 0.0017*** 0.0026*** (0.0002) (0.0000) Intercept 1.1553*** 2.0387*** (0.0000) (0.0000) N 5242 5242 Adj. 𝑅2 0.1110 0.3329
Pooled OLS regressions may have its limitations, maybe most importantly ignoring firm heterogeneity. As shown in the Data section (section 3), the mean value of the capital structure proxies varies notably between the different industries. This implies that firm heterogeneity will be relevant in our sample. Therefore, we also employ a period-fixed effects model (Eq.2) to test the effects of CSR on capital structure. The results are shown in table 5. The adjusted 𝑅2 has increased slightly. The coefficients do not differ a lot with our pooled OLS results in terms of values, signs and significance and therefore the same interpretations with regards to table 4 hold.
20
Table 5: Capital Structure Regressions including Period-Fixed Effects
This table presents results from regressing our capital structure proxies on the CSR scores and controls including period-fixed effects over the period 2006-2016. Panel A reflects the regressions using the weighted average CSR score as main explanatory variable for both BookEquity and MarketEquity as dependent variables. Panel B, C, D and E do the same using Corporate Governance, Economic, Environmental and Social as main explanatory variables for both BookEquity and MarketEquity as dependent variables respectively. BookEquity is the book value of equity divided by total assets, MarketEquity the market value of equity divided by the market value of total assets. . Beta reflects the relationship between a stock’s volatility with the market, Size reflects the natural logarithm of total assets, Price to Book the price-to-book-ratio at the end of the fiscal year and Sales Growth the sales growth of the associated firm over a 5 year period. * reflects statistical significance at the 10% level, ** at the 5% level, *** at the 1% level.
Panel A: CSR Corporate Governance
BookEquity MarketEquity BookEquity MarketEquity
CSR 0.0010*** 0.0026*** Corporate Governace 0.0010*** 0.0028*** (0.0000) (0.0000) (0.0000) (0.0000) Beta -0.0092*** -0.0363*** Beta -0.0121*** -0.0423*** (0.0080) (0.0000) (0.0007) (0.0000) Size -0.0522*** -0.1012*** Size -0.0471*** -0.0880*** (0.0000) (0.0000) (0.0000) (0.0000)
Price to Book 0.0001 0.0000 Price to Book 0.0001* 0.0001
(0.1480) (0.6291) (0.0973) (0.2960)
Sales Growth 0.0016*** 0.0025*** Sales Growth 0.0015*** 0.0024***
(0.0003) (0.0000) (0.0006) (0.0000) Intercept 1.1532*** 2.0714*** Intercept 1.0663*** 1.8113*** (0.0000) (0.0000) (0.0000) (0.0000) N 5242 5242 N 5242 5242 Adj. R2 0.1229 0.3623 Adj. R2 0.1136 0.3166
Panel C: Economic Panel D: Environmental
BookEquity MarketEquity BookEquity MarketEquity
Economic 0.0011*** 0.0023*** Environmental 0.0009*** 0.0020*** (0.0000) (0.0000) (0.0000) (0.0000) Beta -0.0081** -0.0347*** Beta -0.0109*** -0.0399*** (0.0381) 0.0000 (0.0017) (0.0000) Size -0.0515*** -0.0958*** Size -0.0526*** -0.1003*** (0.0000) (0.0000) (0.0000) (0.0000)
Price to Book 0.0001* 0.0000 Price to Book 0.0001 0.0000
21 Panel E: SOC BookEquity MarketEquity SOC 0.0008*** 0.0022*** (0.0000) (0.0000) Beta -0.0107*** -0.0380*** (0.0035) (0.0000) Size -0.0502*** -0.0971*** (0.0000) (0.0000) PTB 0.0001 0.0000 (0.1507) (0.5634) SG 0.0016*** 0.0026*** (0.0004) (0.0000) Intercept 1.1465*** 2.0483*** (0.0000) (0.0000) N 5242 5242 Adj. R2 0.1174 0.3462
Potential variables include other firm specific characteristics such as liquidity and commitment to research and development, and we follow literature that suggests these variables do not significantly affect the base regressions. (Trinks et al., 2017). Furthermore, the choice to engage in CSR and hence obtain higher CSR scores may not be independent of the firm’s capital structure, and hence the problem of reverse causality may arise.3 We admit that such
concerns also hold for the relation between CSR performance and capital structure, and that in addition to the firm-specific characteristics we control for in our regressions there may also be other channels that affect the capital structure of a firm.4 The main take-away after our
performed robustness checks is that the results are not materially different from our primary analysiss.
6. Conclusion and discussion
The participants in the financial markets have increasingly been paying attention to corporate social responsibility commitments of firms, which in turn affects the choice to invest in the associated firms. Existing literature focuses on the relationship between CSR performance and financial performance, and fewer papers examine the effects of CSR on cost of capital. To my knowledge, this paper is among the first to examine the effects of CSR on the capital structure of publicly listed firms. There are several theoretical mechanisms at work here which increase the attractiveness of both debt and equity financing. An increased
3 Waddock and Graves (1997) find that there are 2 hypothesis that may explain the causality in the relationship
between CSR commitment and cost of equity. According to the good management hypothesis, increased CSR performance enhances the relationship between the firm and its key stakeholders, leading to an increased financial performance. This in turn will decrease the cost of equity and hence supports the equity financing hypothesis. On the other hand, there’s the slack resources hypothesis, which states that an increased financial performance leaves room for additional resources to increase CSR performance.
4For example, social norms may influence the capital structure choice of a firm. Bae et al. (2011) find that
22
CSR performance signals the financial markets that this firm is committed to participate in sustainability changes which, in turn, may result in a reduced perceived riskiness of the firm. Consequently, this increases the firm’s credit rating, resulting in the fact that the firm is able to borrow at lower interest rates. The lower interest rates in combination with tax benefits increases the attractiveness of debt financing, which we call the debt financing hypothesis. In contradiction, with (large) investors consciously taking CSR performance of firms into account when selecting their portfolios, we have seen an increased demand for stocks of firms that perform well in the area of CSR. This, ceteris paribus, would decrease the firms cost of equity and makes equity financing more appealing. We call this the equity financing hypothesis. Our main finding is that CSR performance is positively related to our equity ratio’s, indicating that a higher CSR performance is associated with an increase in equity financing. We thus find support for the dominance of the equity financing hypothesis. This finding is robust to using White’s cross-section standard errors and other panel regression methods. In addition, we zoom in to the concept of CSR and its associated four dimensions, and examine the effects of the dimensions separately on capital structure. We find
significant and positive relationships between the dimensions and our equity ratio’s, where the economic dimension is the most influential and the social dimension the least influential. We try to extend the existing literature by focusing on the effects on capital structure and its implications, rather than solely focusing at the links of CSR with financial performance or cost of capital.
We contribute to the existing literature in several ways. We do not focus solely on the environmental aspect of CSR, but include four different dimensions associated with CSR. Furthermore, we do not focus on the traditional relationships between CSR, financial performance and cost of capital, but take on a different perspective on how these relations affect the capital structure of firms. Our results are in line with the literature that examine the relationship between CSR and equity cost of capital, where they find that an increased CSR performance decreases the cost of equity capital, except for ‘sin’ stocks. They are consistent with studies of Heinkel et al. (2001), El Ghoul et al. (2011) and Trinks et al. (2017). Whilst our paper is consistent with Sharfman and Fernando (2008) with respect to cost of equity reduction, it contradicts their hypothesis that an increased CSR performance increases a firm’s leverage. This difference may be due to the fact that they use well criticized data as CSR proxies and different methodologies. Moreover, they find that
increased environmental performance actually increases the cost of debt. This decreases the attractiveness of debt financing and may explain why the equity financing hypothesis
dominates.
23
turn may increase a firm’s ability to finance itself through both debt capital and equity capital markets. Moreover, investors that invest in high CSR firms may reduce exposure to sustainability risks, which comes at the cost of lower expected returns. The financial market works as a mechanism that rewards firms that have high CSR commitments in terms of a lower cost of capital. Regulators also play a role in the CSR process which affects capital structure. Their impact goes beyond just regulations that influence financial markets. To illustrate, Ghoul et al. (2016) find a positive relationship between CSR commitments and media freedom. Regulation with respect to the degree of media freedom thus affects CSR commitments, which affects the cost of capital and capital structure.
We feel that one of the main limitations of our study is the ability to effectively distinguish the debt financing effect from the equity financing effect. We use equity ratios as capital structure proxies and hence are only able to see a net effect. We do know that there is a positive significant relationship between CSR and capital structure. Whilst there are multiple theoretical mechanisms that explain this relationship, we do not successfully distinguish the counteracting effects empirically that result in this positive net effect. Moreover, our study possibly suffers from endogeneity problems and omitted variable bias. This may cause a spurious positive relationship between CSR commitments and equity ratio’s, and hence lead to causality problems. A higher CSR commitment may increase the financial performance of stocks and result in a reduction of cost of equity capital, which in turn increases the equity ratios of a firm. However, it is also possible that the cost of equity reduction and increased financial performance result in an abundance of resources, which allows firms to assign these resources to CSR commitments. Moreover, despite adding significant control variables in our regressions, there may be omitted variables which affect the relationship between CSR commitments and capital structure. Firm-specific policies with regards to capital structure, social norms, and other factors may possibly affect a firm’s capital structure. Subsequently, the results of this study leave suggestions for future research. The most important future research implication is the empirical examination of the multiple
24
References
Bae, K., Kang, J., Wang, J., 2011. Employee treatment and firm leverage: a test of the stakeholder theory of capital structure. Journal of Financial Economics, 100(1), 130-153. Bauer, R., Koedijk, K., Otten, R., 2005. International evidence on ethical mutual fund performance and investment style. Journal of Banking & Finance, 29(7), 1751-1767.
Berry, T.C., Junkus, J.C., 2010. Social responsible investing: An investor perspective. Journal of Business Ethics, 112(4), 707-720.
Brooks, C., 2008. Introductory econometrics for finance. Cambridge University Press, Cambridge.
Cai, L., He, C., 2014. Corporate environmental responsibility and equity prices. Journal of Business Ethics, 125(4), 617-635.
Chen, K., Chen, Z., Wie, K., 2011. Agency cost of free cash flow and the effect of shareholder rights on implied cost of equity capital. Journal of Quantitative Analysis, 46(1), 171-207. Ciciretti, R., Dalò, A., Dam, L., 2017. The price taste for socially responsible investment. Unpublished working paper. University of Rome, Italy.
Dhaliwal, D., Zhen, Li, O., Tsang, A., Yang, Y.G., 2014. Corporate social responsibility
disclosure and the cost of equity capital: The roles of stakeholder orientation and financial transparency. Journal of Accounting and Public Policy, 33(4), 328-355.
El Ghoul, S., Guedhami, O., Kwok, C., Mishra, D., 2011. Does corporate social responsibility affect the cost of capital? Journal of Banking & Finance, 35(9), 2388-2406.
El Ghoul, S., Guedhami, O., Hakkon, K., Park, K., 2016. Corporate environmental
responsibility and the cost of capital: International evidence. Journal of Business Ethics, 1-27. 10.1007/s10551-015-3005-6.
El Ghoul, S., Guedhami, O., Nash, R. Patel, A., 2016. New evidence on the role of the media in corporate social responsibility. Journal of Business Ethics, 1-29. 10.1007/s10551-016-3354-9.
Fama, E.F., French, K.R., 1997. Industry cost of equity. Journal of Financial Economics, 43, 153-194.
Galema, R., Plantinga, A., Scholtens, B., 2008. The stocks at stake: Return and risk in socially responsible investment. Journal of Banking & Finance, 32(12), 2646-2654.
25
Heinkel, R., Kraus, A., Zechner, J., 2001. ‘The effect of green investment on corporate behaviour. Journal of financial and quantitative analysis, 36(4), 431-449.
Hong, H., Kacperczyk, M., 2009. The price of sin: the effects of social norms on markets. Journal of Financial Economics, 93, 15-36.
Horváthová, E., 2010. Does environmental performance affect financial performance? A meta-analysis. Ecological Economics, 70(1), 52-59.
Jensen, M.C., 1986. Agency cost of free cash flow, corporate finance and takeovers. American Economic Review, 76, 323-332.
Modigliani, F., Miller, M.H., 1958. The cost of capital, corporate finance and the theory of investment. American Economic Review, 48, 261-297.
Modigliani, F., Miller, M.H., 1963. Corporate income taxes and the cost of capital: A correction. American Economic Review, 53, 433-443.
Mollet, J.C., Ziegler, A., 2014. Socially responsible investing and stock performance: New empirical evidence for the US and European stock markets. Review of Financial Economics, 23(4), 208-216.
Perrini, F., Pogutz, S., & Tencati, A., 2006. Developing corporate social responsibility: A European perspective. Edward Elgar Publishing Ltd, Cheltenham.
Renneboog, L., Ter Horst, J., & Zhang, C., 2008. The price of ethics and stakeholder governance: The performance of socially responsible mutual funds. Journal of Corporate Finance, 14(3), 302-322.
Sharfman, M.P., Fernando, C.S., 2008. Environmental risk management and the cost of capital. Strategic Management Journal, 29(6), 569-592.
Simerly, R.L, Mingfang, L., 2000. Environmental dynamism, capital Structure and
performance: A theoretical integration and an empirical test. Strategic Management Journal, 21(1), 31-49.
Toffel, M.W., Marschall, J.D., 2004. Improving environmental performance assessment: a comparative analysis of weighting methods used to evaluate chemical release inventories. Journal of Industrial Ecology, 8(2), 143-172.
26
Appendix
1. SIC Codes
Standard Industrial Classification (SIC) codes are codes developed to classify firms according to the business they operate in. These SIC codes are assigned to both U.S. and non-U.S. companies. A company may have up to eight codes assigned to depending on the number of business segments that make up the company's revenue. If a sales breakdown for segments is available SIC Code 1 would represent the business segment which provided the most revenue. SIC Code 8 would represent the segment that provided the least revenue. If a sales breakdown is not available the SIC Code is assigned according to the best judgement of Worldscope.
Table A1: Standard industrial Classification codes
This table shows the industrial classifications in which the firms operate. The firms are allocated to an industry based on which business segment of the company provides the most revenue.
Code Industry
01_09 Agriculture, Forestry and Fishing
10_14 Mining
15_17 Construction
20_39 Manufacturing
40_49 Transportation and Public Utilities
50_51 Wholesale Trade
52_59 Retail Trade
60_67 Finance, Insurance and Real Estate
70_89 Services
91_99 Public Administration
2. Year summary statistics
27
Table A2: Year Summary Statistics for BookEquity and MarketEquity This table reports the most important descriptive statistics for both BookEquity and MarketEquity per year concerning the period 2006 – 2016. Panel A reports the statistics for BookEquity, Panel B for MarketEquity.
Panel A: BookEquity Panel B: MarketEquity
N Mean St.Dev N Mean St.Dev
2006 491 0.40 0.21 491 0.60 0.24 2007 493 0.39 0.21 493 0.60 0.24 2008 493 0.36 0.25 493 0.54 0.25 2009 493 0.37 0.23 493 0.52 0.25 2010 493 0.38 0.22 493 0.53 0.24 2011 493 0.37 0.22 493 0.51 0.24 2012 493 0.36 0.22 493 0.53 0.24 2013 493 0.36 0.20 493 0.56 0.23 2014 493 0.34 0.20 493 0.57 0.22 2015 493 0.33 0.21 493 0.55 0.23 2016 493 0.31 0.22 493 0.56 0.23
Figure A1: Equity Ratios per Year for U.S firms in the Period 2006-2016 This figure shows the mean values of the equity ratios per year. The bright bars and dark bars represent the mean value of BookEquity and MarketEquity respectively. The vertical axis represents the leverage ratio, the horizontal axis the associated year. TAPU is defined as Transportation and Public Utilities, and FIRE stands for Finance, Insurance and Real Estate.
28
3. Random effects regressions
The results of regressing our CSR scores and control variables on our capital structure proxies including year random effects are shown in table A3. The results do not much differ from our primary analysis in terms of coefficients, signs and interpretations. The same conclusion is drawn after including cross-section random effects.
Table A3: Capital Structure Regressions including Year-Random Effects
This table presents results from regressing our capital structure proxies on the CSR scores and controls over the period 2006-2016 including year random effects. Panel A reflects the regressions using the weighted average CSR score as main explanatory variable for both BookEquity and MarketEquity as dependent variables. Panel B, C, D and E do the same using CGV, ECN, ENV and SOC as main explanatory variables for both BookEquity and MarketEquity as dependent variables respectively. BookEquity is the book value of equity divided by total assets, MarketEquity the market value of equity divided by the market value of total assets. Beta reflects the relationship between a stock’s volatility with the market, Size reflects the natural logarithm of total assets, Price to Book the price-to-book-ratio at the end of the fiscal year and Sales Growth the sales growth of the associated firm over a 5 year period. * reflects statistical significance at the 10% level, ** at the 5% level, *** at the 1% level.
Panel A: CSR Panel B: Corporate Governance
BookEquity MarketEquity BookEquity MarketEquity
CSR 0.0009*** 0.0026*** Corporate Governance 0.0008*** 0.0028*** (0.0000) (0.0000) (0.0000) (0.0000) Beta -0.0099*** -0.0363*** Beta -0.0124*** -0.0423*** (0.0071) (0.0000) (0.0007) (0.0000) Size -0.0522*** -0.1011*** Size -0.0475*** -0.0879*** (0.0000) (0.0000) (0.0000) (0.0000)
Price to Book 0.0001 0.0000 Price to Book 0.0001* 0.0001
(0.1312) (0.6100) (0.0934) (0.2797)
Sales Growth 0.0017*** 0.0025*** Sales Growth 0.0016*** 0.0024***
29
Panel C: Economic Panel D: Environmental
BookEquity MarketEquity BookEquity MarketEquity
Economic 0.0010*** 0.0023*** Environmental 0.0008*** 0.0020*** (0.0000) (0.0000) (0.0000) (0.0000) Beta -0.0089** -0.0347*** Beta -0.0114*** -0.0399*** (0.0359) (0.0000) (0.0016) (0.0000) Size -0.0516*** -0.0958*** Size -0.0528*** -0.1001*** (0.0000) (0.0000) (0.0000) (0.0000)
Price to Book 0.0001* 0.0000 Price to Book 0.0001 0.0000
(0.0917) (0.4405) (0.1652) (0.6174)
Sales Growth 0.0015*** 0.0022*** Sales Growth 0.0017*** 0.0025***
