Corporate Social Responsibility Disclosure and Firm
Financial Performance: The Effect of
Multinationality and the 2008/2009 Financial Crisis
By C. Eltingh1
Supervisor: Dr. J.V. Tinang Nzesseu Submitted for the degree of
MSc International Financial Management Faculty of Economics and Business
University of Groningen
1
Abstract
This paper analyses the effect of corporate social responsibility disclosure on firm financial performance using a sample consisting of 32.311 firm-year observations over the period 2004 to 2018. Similar to the previous literature, a positive relationship exists between corporate social responsibility disclosure and firm financial performance. Besides this, the initial findings also show that both multinationality and the 2008/2009 financial crisis do have an effect on the relationship between CSR disclosure and firm financial performance. However, conclusions concerning the effect on the financial performance of firms being multinationals or the impact of the 2008/2009 financial crisis cannot be drawn easily. When different measures of both firm financial performance and corporate social responsibility disclosure are used, different results are found. This finding is important for future research on both topics. Conclusions should not only be drawn based on one of the measures, but should also be tested using other measures of firm financial performance and/or corporate social responsibility disclosure. Since the models show a positive effect of corporate social responsibility disclosure on firm financial performance, it should be taken into consideration by managers as an investment option.
2
1. Introduction
Global developments have increased public awareness on issues related to corporate social responsibility (CSR) which caused pressure on firms to disclose their CSR activities to inform their stakeholders of their actions taken on CSR related issues. This increased interest by the public and pressure on firms is recognised by, for example, Nekhili et al. (2017), Bagnoli and Watts (2017), and Chen, Hung, and Wang (2018) but as pointed out by Cho et al. (2015) researchers are also getting increasingly more interested in this area. As described by Castelló and Lozano (2011), the 2008/2009 financial crisis at the end of the previous decade has also contributed to the increased focus on CSR. They pointed out that during the financial crisis, public trust had decreased, and that firms’ activities are increasingly scrutinised by other stakeholders such as the public and non-governmental organisations. This increase in public scrutiny potentially impacts multinational corporations (MNC) more than domestic corporations (DMC). Park and Ghauri (2015) point out that especially for MNCs meeting local expectations is important. Marano and Kostova (2016) also point out that MNCs, in general, face more different pressures compared to DMCs as they are active in different countries and are for example sometimes seen as exploiters of foreign resources.
Because of these developments, this paper analyses whether disclosing CSR information influences firm financial performance and thus might be seen as an investment option for managers. However, as described above, MNCs and DMCs face different pressures and demands. As a consequence, a distinction is made between these two groups of corporations to compare the impact of CSR disclosure on their financial performance. Finally, as described before, the financial crisis caused the firms’ activities to come under increased scrutiny. Therefore, this study performs a closer examination of this sudden changing point in the level of scrutiny. What is important in this analysis is that the effect of CSR disclosure is lagged because only after the publication of the annual report of a firm, this information becomes public. Hence, an analysis is made to investigate the effect of disclosing CSR information in the year before the financial crisis on a firm’s financial performance during the financial crisis. This effect of CSR disclosure on firm financial performance during the financial crisis is compared to the effect of CSR disclosure on firm financial performance in the other years. The main questions that are answered in this paper are the following: Does multinationality
3
disclosure and firm financial performance and are used to develop the hypotheses that are tested in this paper. The overall implication is that CSR information disclosure by firms positively impacts their financial performance. One of the important papers in this area that is discussed in this research is the study by Waddock and Graves (1997). They did not only found a positive relationship between CSR disclosure and firm financial performance but also found that CSR disclosure affects next year’s firm financial performance positively.
With this insight, this paper constructs three models to test the three hypotheses that are developed in this paper. The reason is that, first, only the overall effect of CSR information disclosure on firm financial performance needs to be measured, but then also separately the influence of multinationality of firms and the effect of the financial crisis have to be examined. These hypotheses are tested using a sample over the period from 2004 to 2018, consisting of 32.311 firm-year observations of which 8.024 MNC firm-year observations and 24.287 DMC firm-year observations across 38 different countries. The initial results are in line with previous literature and indicate that firms that disclose CSR information are expected to have a higher financial performance compared to firms not disclosing CSR information. Also, a difference between MNCs and DMCs is found, and the results show that the firm financial performance of MNCs who disclose CSR increases less compared to DMCs who disclose CSR. Finally, the results also show that the 2008/2009 financial crisis did affect firms significantly and that firms that disclosed CSR in the year before the crisis benefitted from this with a higher financial performance during the crisis.
4
results, and in section 5 the robustness tests are presented. Finally, in section 6, the conclusion of the results of the paper and its implications for future research can be found.
2. Literature
2.1. Definition CSR disclosure
Before developing hypotheses concerning the effect of CSR disclosure on firm financial performance, the concept of CSR and CSR disclosure is elaborated. From the paper by De Bakker, Groenewegen, and Den Hond (2005), it becomes clear that the concept of CSR has been present in academic literature for some decades now. Besides the increased academic interest, McWilliams, Siegel, and Wright (2006) state that managers also have an increased interest in CSR. Despite the large quantity of academic literature and interest of the public, there is not yet a clear established definition of CSR. The description of CSR by Votaw (1973, pp 11) is widely used by other academics and describes the current situation of CSR, “The term
[CSR] is a brilliant one; it means something, but not always the same thing, to everybody”.
Having different definitions of CSR creates problems not only for reporting meaningful CSR practices, but also in the development, implementation, and academic analyses of these CSR practices as put forward by Sarkar and Searcy (2016). Sheehy (2015) also emphasises the importance of settling on one definition of CSR because the current situation with many definitions imposes a threat to the value and meaning of CSR.
Dahlsrud (2008) has examined the frequency of the used definitions, and the results are clearly in favour of the definition given by the Commission of the European Communities (2001). The Commission of European Communities (2001, pp 6) defines CSR as “a concept whereby
companies integrate social and environmental concerns in their business operations and in their interaction with their stakeholder on a voluntarily basis”. Michelon, Pilonato, and Ricceri
5
In this research, five different theories are used to analyse the relation between CSR disclosure and firm financial performance of which a one-sentence summary can be found in table 1. The signalling and transparency characteristics of CSR disclosure described by Lys, Naughton, and Wang (2015) and Chen, Hung, and Wang (2018) are important aspects of the theories that describe the relationship between CSR disclosure and firm financial performance.
2.2. Theories
6
Another theory that explains the CSR activity and CSR disclosure of firms is the stakeholder theory. As described by Freeman (1984), stakeholders are all individuals or groups that can affect the performance of a firm. The paper by Chiu and Wang (2015) gives a good insight into the influence of the stakeholder theory on CSR disclosure of firms. Managers should not only focus on the best interest of the shareholders but should instead take the interest of all stakeholders into account. The success of the firm in the long run extensively dependent on the support of the stakeholders and need to be appropriately addressed by management. What stands out from the paper by Dierkes and Antal (1985), is that both the financial reports, as well as the non-financial reports, are a way to monitor and control a company. If stakeholders want to see if their interest is taken into account by the managers, they can consult these reports, and for their interest in CSR activities, the CSR disclosure should be consulted. Managers take the power of each stakeholder group into account, and both engage and disclose CSR to maintain a good relationship with the stakeholders. This relation is also seen by different researchers, starting with Sweeney and Coughlan (2008) who report that the CSR disclosure of firms is in line with the expectations of the stakeholders of the firms. Also, Thijssens, Bollen, and Hassink (2015) have traced CSR disclosure back to the interest of the stakeholders and the extent to which each stakeholder can influence the firm. As theorised, the influence of each stakeholder on CSR disclosure is affected by their respective power on the firm. Yusoff, Mohamad, and Darus (2013) have also analysed the effect of the stakeholder theory and its effect on the financial performance of firms. They found evidence that when firms disclose their CSR activities which are in line with the expectation of the stakeholders, this would have a positive effect on the financial performance of the firm.
7
Koparan, and Okan (2015) have found evidence for a positive relation between CSR isomorphism and firm financial performance using CSR disclosure. Martinez-Ferrero and Frias-Aceituno (2015) link the institutional theory to firm financial performance and found a positive relationship between a firm’s CSR practices, measured using CSR disclosure, and the firm’s financial performance.
Also, the legitimacy theory, which is closely related to normative isomorphism, can be used to explain CSR disclosure of firms. Legitimacy is defined by Suchman (1995, pp 574) as “a
generalized perception or assumption that the actions of an entity are desirable, proper, or appropriate within some socially constructed system of norms, values, beliefs, and definitions.”
Guthrie et al. (2004) describe legitimacy more simple and state that firms are bound to a social contract which they have to follow. This social contract states how society expects the firm to operate which, over time, may be subject to change. Although no legal documents are signed by the firm, it is obliged to follow this contract to be accepted by the society to do business. Suchman (1995) emphasises the importance of firms communicating with the public to manage their legitimacy. As mentioned above, firms can communicate to the wider public, which are the other parties involved in the social contract by using disclosures. Cho and Patten (2007) find that firms use disclosures to increase their legitimacy and by doing so, escape sanctions of the public. Lanis and Richardson (2013) find evidence of this acceptance of the public in the context of tax aggressiveness. They find that firms who are more tax aggressive also disclose more CSR as a way of compensating for its legitimacy loss. Podnar and Golob (2007) have found evidence that the firms’ engagement in CSR is a way to get a “license” to operate and gain acceptance of the public. This is also what Bachmann and Ingenhoff (2016) found, which is that firms can gain legitimacy by engaging in CSR. They found that even when stakeholders are sceptic, the advantages of CSR disclosure on the legitimacy of the firm outweighs the disadvantages. Even in controversial industries, CSR initiatives can have a positive effect on the legitimacy of those firms. Asmeri, Alvionita, and Gunardi (2017) confirm that society appreciates firms that show that they are socially concerned. As mentioned by Jayachandran, Kalaignanam, and Eilert (2013), increased legitimacy positively affects a firm’s financial performance.
8
more in CSR activities. Furthermore, Seifert, Morris, and Bartkus (2004) focus on corporate philanthropy, which can be considered as a form of CSR, their finding is also in line with the theory of slack resources. They found that firms which had slack resources engaged more in corporate philanthropy. The results of Waddock and Graves (1997) also confirm the theory of slack resources as they found a positive relationship between a firm’s resource availability and their CSR disclosure.
Agency theory By disclosing CSR, agency costs will be lower and thus should be positively related to firm financial performance.
Stakeholder theory A firm's success depends on the support of all stakeholders who monitor and control the firm by consulting not only their
financial disclosures but also their non-financial disclosures.
Institutional theory Changes in CSR disclosure are related to isomorphism and are found to be positively related to firm financial performance.
Legitimacy theory By disclosing CSR firms want to increase their legitimacy and in turn increase their financial performance.
Theory of slack resources Firms who have abundant resources tend to disclose more CSR which is positively related to next year’s financial performance.
Table 1: Summary of the five theories that are used to describe the relation between CSR disclosure and firm financial performance.
2.3. Hypotheses development
2.3.1. CSR disclosure and firm financial performance
9
firm financial performance. This implies that CSR disclosure can be related to a lower return on the firms’ assets. On the other hand, there are also some articles providing evidence for a positive relationship between CSR disclosure and firm financial performance. This positive relationship has been found by, for example, Bayoud, Kavanagh, and Slaughter (2012) for firms from Libya and Mohammed, Saheed, and, Oladele (2016) for manufacturing firms in Nigeria. Gutsche, Schultz, and Grathwohl (2017) have looked at the relationship between CSR disclosure and firm financial performance for S&P 500 firms for the period 2011 to 2014. They provide evidence for a positive relationship between CSR disclosure and firm financial performance. They have also compared the effect of CSR disclosure and CSR performance and found that the effect of CSR disclosure is larger than the effect of CSR performance.
Reference Relationship
CSR disclosure/ firm financial
performance
Period Selected firms
Selcuk and Kiymaz (2017) Negative 2009-2011 Firms listed on the Borsa Istanbul Bayoud et al. (2012) Positive 2007-2009 Forty selected Libyan firms Mohammed et al. (2016) Positive 2001-2012 Manufacturing firms in Nigeria Gutsche et al. (2017) Positive 2011-2014 Firms listed on the S&P 500
Table 2: Summary of the empirical findings of previous research.
As mentioned before, all discussed theories concerning the reasons why firms disclose their CSR activities and its effect on a firm’s financial performance suggest a positive relationship between CSR disclosure and firm financial performance. Besides this, also the majority of the discussed empirical analyses find a positive relationship between CSR disclosure and firm financial performance. Therefore, this paper tests the following hypothesis:
H1: CSR disclosure positively affects firm financial performance.
2.3.2. MNCs versus DMCs
10
countries, are hard to deny. The article by Tixier (2003) also points to a similar conclusion concerning the objective of MNCs, which have grown in both number and size. In their pursuit of generating profit, their success in foreign markets is in part affected by the ethical standards of the MNC. Polonsky and Jevons (2009) have developed this conclusion even further. They state that although MNCs are finding ways to maximise their profit in foreign countries, CSR communication has also become an important strategic tool. MNCs are recognising the value of CSR and are using CSR as a strategic tool in which benefits of the firm are dependent on the communication of their CSR activities.
Both Rathert (2016) and Park and Ghauri (2015) point out the importance of meeting the CSR demands from the foreign local stakeholders in the strategy of MNCs. MNCs face different influences and pressures on their CSR disclosure compared to DMCs, as is described by Marano and Kostova (2016). As MNCs are active in a number of different countries, they also face different expectations and pressures from the different countries in which they are active, which does not apply for DMCs. MNCs weigh the power of every institution, and the more power a foreign country has, the more an MNC will try to meet their expectation. In a transnational context, CSR is more complex, and the effect of a firm’s transnational organisational field can be more demanding for MNCs. Transnational fields can put more pressure on an MNC to engage in more CSR than is required in its home country. However, it also offers MNCs opportunities to become a more progressive and global organisation by engaging in practices from other transnational fields. It is thus clear that due to its international character MNCs face more and different pressures regarding CSR. Although the decision-making regarding their CSR is more complex, it also offers them opportunities which do not apply for DMCs. DMCs do not face these pressures, which makes their CSR decision-making less complex. However, this also implies that they do not have the corresponding opportunities of the MNCs.
11
From the literature, it does not immediately become clear whether MNCs financial performance is more affected by CSR disclosure than DMCs. Due to the international character of MNCs, their CSR decision-making process is more complex compared to DMCs, but it also provides them opportunities. Besides this difference, it has also become clear that MNCs are subject to great scrutiny from society. It is clear that differences exist between MNCs and DMCs with regard to their CSR disclosure, but not its implication for the relationship between CSR disclosure and firm financial performance. Therefore the following hypothesis is tested:
H2: The effect of CSR disclosure on firm financial performance is strengthened by multinationality.
2.3.3. Financial crisis
12
had lost during the financial crisis. In light of previously discussed literature, this should positively affect firm financial performance.
As shown and pointed out by Bansal, Jiang, and Jung (2015), the financial crisis of 2008/2009 provide a good natural situation which can function as a proper test case for the effect of CSR disclosure on firm financial performance. Castellé and Lozano (2011) point out that firms legitimacy were criticised during the 2008/2009 financial crisis. Public trust in firms was dropping fast during the 2008/2009 financial crisis and demand for greater scrutiny of the firms increased. Not only non-governmental organisations but also other stakeholders of the firms scrutinised and questioned firms’ activities more. In light of this sudden increase in scrutiny, and as CSR disclosure affects next year’s firm financial performance, firms which already had disclosed CSR in the year prior to the 2008/2009 financial crisis would be expected to do better during the financial crisis compared to the other firms. Therefore, the following hypothesis concerning CSR disclosure and the 2008/2009 financial crisis is tested:
H3: Firms which have disclosed CSR in the year prior to the 2008/2009 financial crisis will outperform the other firms in the year of the financial crisis.
13
3. Methodology and Data
3.1. Methodology
To test the above-stated hypotheses, appropriate and proper regression models need to be developed. To test the first hypothesis, a regression model needs to be developed, which tests the relationship between firm financial performance and CSR disclosure. The dependent variable in this model is firm financial performance, and the independent variable is CSR disclosure. Besides these two variables, also control variables need to be taken into account. These control variables are variables that should be added to the model as they have been proven to have an influence on the relationship between CSR disclosure and firm financial performance. First, the control variables will be introduced after which a description is given on how the analysis takes into account the bidirectional relationship that is described earlier.
The first firm control variable that is implemented into the model is a measure of firm size. As described by Kang, Lee, and Huh (2010), the implementation of this variable eliminate the effect on the firm financial performances that is due to their size. Larger firms, among other things, enjoy economies of scale, thus positively influencing its financial performance. This relation between firm size and firm financial performance is also described by Waddock and Graves (1997). Furthermore, it has been taken into account in various studies in the context of CSR, for example by Barnett and Salomon (2012). Besides controlling for firm size, Hirsch (1991) also uses firm sales growth in the context of firm financial performance. As stated by Maury (2006), the firm’s sales growth reflects the value of the firm’s growth opportunities and is also emphasised by Schmalensee (1989). From Zhang (2005) it becomes clear that the cash holdings of firms should also be included in the analysis of firm financial performance and Lu, Shailer, and Yu (2017) point out the importance of a firm’s cash holdings in the context of CSR disclosure. Therefore, a measure of cash holdings is implemented in the model to cover its effect on the relationship between CSR disclosure and firm financial performance. The next firm control variable that needs to be taken into account in the model is the firm’s leverage. Ibhagui and Olokoyo (2018) provide a good overview of the literature and show with the use of signalling theory and agency theory that there is evidence for a relation between the firm’s leverage and financial performance. They also find a mitigating role of firm size on the relationship between leverage and firm financial performance.
14
behind this is that firms invest in positive net present value CSR projects which cause an increase in the firms’ financial performance and hence increased dividend pay-outs. This relation is also shown by Benjamin, Biswas, and Yang (2018) and as dividend has an influence on the relationship between CSR disclosure and firm financial performance, this is implemented in the model. Also, some board characteristics have been shown to have an influence on the relationship between CSR disclosure and firm financial performance. Alabdullah, Ahmed, and Muneerali (2019) and Alqatan, Chbib, and Hussainey (2019) found that both the board size of firms as well as CEO duality, which means that the CEO of the company is also acting as the chairman of the board, influences the relationship. Therefore, these board characteristics need to be taken into account for the analysis. Furthermore, country control variables are added to the model to control for country differences that affect the relationship between CSR disclosure and firm financial performance. Based on the research by Poddi and Vergalli (2009) and Konchitchki (2011), the country moderators are measures for economic development and the inflation rate of the countries.
A cross-country panel data regression analysis is used to test the relationship between CSR disclosure and firm financial performance. As pointed out by Waddock and Graves (1997) does current CSR disclosure affect the firm’s financial performance in the future. Intuitively, only after the annual report has been published does this information become public, and will thus have a delayed effect on the firm. The disclosures in the annual report of year t will not affect the firm in year t, but in year t+1. Furthermore, they also show that the relationship between CSR disclosure and firm financial performance faces a potential reverse causality problem. That is that the firms disclosing their CSR see an increase in their next year’s financial performance which in turn leads to an increase in CSR disclosure. To take this into account and to overcome the potential reverse causality problem, the model is lagged by one year. The regression equation is as follows:
𝐹𝑖𝑟𝑚 𝐹𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒𝑖,𝑗,𝑡 = 𝛽0 + 𝛽1(𝐶𝑆𝑅 𝐷𝑖𝑠𝑐𝑙𝑜𝑠𝑢𝑟𝑒)𝑖,𝑡−1 +
15
interaction term between the variables CSR disclosure and multinationality. The model that is used to test the second hypothesis is as follows:
𝐹𝑖𝑟𝑚 𝐹𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒𝑖,𝑗,𝑡 = 𝛽0 + 𝛽1(𝐶𝑆𝑅 𝐷𝑖𝑠𝑐𝑙𝑜𝑠𝑢𝑟𝑒)𝑖,𝑡−1 +
𝛽2(𝑀𝑢𝑙𝑡𝑖𝑛𝑎𝑡𝑖𝑜𝑛𝑎𝑙𝑖𝑡𝑦)𝑖,𝑡−1 + 𝛽3(𝐶𝑆𝑅 𝐷𝑖𝑠𝑐𝑙𝑜𝑠𝑢𝑟𝑒)𝑖,𝑡−1∗ (𝑀𝑢𝑙𝑡𝑖𝑛𝑎𝑡𝑖𝑜𝑛𝑎𝑙𝑖𝑡𝑦)𝑖,𝑡−1 +
𝛽𝑥(𝐹𝑖𝑟𝑚 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑠𝑥)𝑖,𝑡−1 + 𝛽𝑦(𝐶𝑜𝑢𝑛𝑡𝑟𝑦 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑠𝑦)
𝑗,𝑡−1 + 𝜀𝑖,𝑗,𝑡 (2
The third hypothesis focuses on the 2008/2009 financial crisis. As stated by Von Hagen, Schuknecht, and Wolswijk (2011), the default of the Lehman Brothers in September 2008 marks the start of the financial crisis. They describe the situation in the period before the default as turmoil and only after the Lehman Brothers default the situation is described as a crisis. To test the third hypothesis, a new model must be made that includes a variable to indicate the financial crisis. This is done by implementing a dummy variable that indicates the first full financial crisis year of 2009 (value 1) or whether the analysis is performed for another year (value 0). Although in previous models, all independent variables were legged by one year, this is not the case in this model. In this model, the variable that indicates the crisis year of 2009 will not be lagged as it is intended to indicate the crisis year. Contrary to the effect of CSR disclosure does the financial crisis affect the firms immediately and will thus not be lagged by one year. Furthermore, in this model an interaction term is added, which is similar to the interaction term in the second model, between the variables CSR disclosure and crisis. The model that is used to test the third hypothesis is as follows:
𝐹𝑖𝑟𝑚 𝐹𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒𝑖,𝑗,𝑡 = 𝛽0 + 𝛽1(𝐶𝑆𝑅 𝐷𝑖𝑠𝑐𝑙𝑜𝑠𝑢𝑟𝑒)𝑖,𝑡−1 + 𝛽2(𝐶𝑟𝑖𝑠𝑖𝑠)𝑖,𝑡 + 𝛽3(𝐶𝑆𝑅 𝐷𝑖𝑠𝑐𝑙𝑜𝑠𝑢𝑟𝑒)𝑖,𝑡−1∗ (𝐶𝑟𝑖𝑠𝑖𝑠)𝑖,𝑡 + 𝛽𝑥(𝐹𝑖𝑟𝑚 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑠𝑥)𝑖,𝑡−1 +
𝛽𝑦(𝐶𝑜𝑢𝑛𝑡𝑟𝑦 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑠𝑦)𝑗,𝑡−1 + 𝜀𝑖,𝑗,𝑡 (3
16
robust standard errors and clustered standard errors at the firm level are used to correct for heteroskedasticity and autocorrelation.
3.2. Data and sample
The analysis contains firm-level data collected from Thomson Reuters Datastream and country-level data obtained from The World Bank and are summarised in Table 3. The dependent variable, firm financial performance, is measured by the firms’ Tobin’s q, as was also done by Bharadwaj, Bharadwaj, and Konsynski (1999). The definition of Tobin’s q used in this paper is the sum of the market value of total equity, the book value of preferred stock, and the book value of debt divided by the book value of total assets. As described by Majeed, Aziz, and Saleem (2015), different measures for CSR disclosure have been used in research that all have their shortcomings. Similar to the research by Harjoto and Jo (2015), CSR disclosure is measured using a dummy variable indicating whether a firm discloses CSR in either a separate report or in their annual report (value 1) or do not disclose CSR information at all (value 0). To separate the dataset between MNCs and DMCs, a dummy variable is constructed indicating whether a firm has international operating income (value 1) or not (value 0) as done in the analysis of Erel, Jang, and Weisbach (2020). The crisis variable in the third model is a dummy variable that indicates the first full financial crisis year of 2009, as stated by Von Hagen, Schuknecht, and Wolswijk (2011).
17
Ward (2011), and accounts for the size of the countries. The second country moderator is the inflation rate of the countries, measured by the consumer price index. The consumer price index is a common proxy of inflation rates and is also used by, for example, Khan and Vieito (2013).
Firm Financial Performance Tobin's q defined as the sum of the market value of total equity, book value of preferred stock, and book value of debt divided by the book value of total assets.
CSR Disclosure Equals 1 if the firm discloses CSR in their annual report or publish a separate CSR report, otherwise 0.
Multinationality Equals 1 if the firm has generated operating income in foreign countries before adjustments and eliminations, otherwise 0.
Crisis Equals 1 for the 2008 financial crisis year, otherwise 0.
Firm Size Natural logarithm of total assets in US dollar.
Sales Growth Equals the change in net sales compared to the previous year.
Cash Holdings Represents the natural logarithm of money available for the use in the normal operations of the company in US dollar.
Leverage Equals total debt divided by total assets.
Dividend Equals dividend per share divided by the share price.
Board Size The total number of board members at the end of the fiscal year.
CEO Duality Equals 1 if the CEO is also chairman of the board, otherwise 0.
Economic Development Natural logarithm of GDP per capita in US dollar.
Inflation Annual consumer price index.
Table 3: Description of the variables that are used in the regression analyses.
18
being an MNC (value 0). Following Hyland and Diltz (2002) are the firms operating in the financial sector (SIC-code 6000-6999) and the firms operating in the utility sector (SIC-code 4900-4999) excluded from the sample. The firms operating in the financial sector are excluded because their financial ratios are harder to compare with the financial ratios of non-financial firms. Firms operating in the utility sector are excluded from the model because the utility sector is a highly regulated sector. Also, all firms with missing variables are removed from the dataset. Furthermore, similar to Falato, Kadyrzhanova, and Sim (2013) and Brown and Higgins (2005) are firms with less than five firm year-observations and the firms in countries which have less than 100 firm year-observations are excluded. Lastly, the continuous firm-level variables are winsorized at the top and bottom 1% of the distribution to mitigate the influence of outliers. This results in a final sample of 32.311 observations consisting of 8.024 MNC year-observations and 24.287 DMC year-year-observations from 38 different countries.
3.3. Descriptive statistics
19
Mean Median Std. Dev.
Tobin's q 1,506 1,135 1,160 CSR Disclosure 0,562 1,000 0,496 Multinationality 0,248 0,000 0,432 Firm Size 15,362 15,354 1,471 Sales Growth 0,096 0,061 0,234 Cash Holdings 12,429 12,633 1,920 Leverage 0,244 0,231 0,176 Dividend 0,021 0,016 0,021 Board Size 10,184 10,000 3,489 CEO Duality 0,385 0,000 0,487 GDP per capita 10,477 10,695 0,742 Inflation 0,020 0,019 0,019 Nr. of obs. 32.311
Table 4: Descriptive statistics (mean, median, and standard deviation) of the variables that are used in the regression analyses.
20 Nr. of obs. Tobin's q CSR Disclosure Multinationality Firm Size Sales Growth Cash Holdings
Leverage Dividend Board Size CEO Duality GDP per capita Inflation Australia 2.433 1,579 0,337 0,170 13,652 0,150 10,505 0,200 0,029 6,791 0,112 10,924 0,022 Austria 134 1,128 0,672 0,172 15,395 0,093 11,901 0,223 0,029 11,701 0,045 10,753 0,019 Belgium 265 1,177 0,592 0,147 15,492 0,075 12,214 0,257 0,023 11,479 0,234 10,683 0,019 Brazil 473 1,518 0,689 0,078 15,522 0,139 11,775 0,346 0,025 9,810 0,410 9,262 0,060 Canada 1.902 1,369 0,382 0,225 14,639 0,138 10,961 0,225 0,020 9,455 0,362 10,748 0,017 Chile 118 1,312 0,669 0,000 15,674 0,085 11,733 0,304 0,024 9,076 0,076 9,541 0,030 China 645 1,180 0,470 0,031 16,341 0,145 13,622 0,255 0,017 11,023 0,293 8,820 0,024 Denmark 261 2,345 0,747 0,084 14,865 0,096 12,101 0,187 0,015 9,253 0,069 10,960 0,015 Finland 314 1,378 0,771 0,048 15,155 0,051 12,102 0,219 0,038 8,041 0,162 10,748 0,014 France 1.010 1,174 0,827 0,250 16,451 0,068 13,543 0,264 0,025 12,878 0,670 10,599 0,013 Germany 921 1,232 0,689 0,188 16,046 0,073 13,234 0,236 0,021 14,165 0,109 10,662 0,014 Greece 154 1,120 0,429 0,123 14,804 0,071 10,826 0,319 0,027 12,019 0,513 10,064 0,016 Hong Kong 1.414 1,300 0,446 0,166 15,588 0,154 13,127 0,237 0,024 10,763 0,465 10,544 0,030 India 581 2,484 0,745 0,072 15,457 0,133 11,590 0,232 0,012 11,528 0,358 7,362 0,072 Indonesia 214 2,256 0,752 0,005 14,831 0,100 11,487 0,205 0,026 6,860 0,065 8,161 0,049 Ireland 136 1,664 0,507 0,132 14,575 0,111 11,949 0,285 0,017 11,809 0,081 10,941 0,011 Israel 106 1,348 0,283 0,132 15,580 0,035 12,288 0,340 0,041 10,500 0,217 10,451 0,013 Italy 303 1,038 0,647 0,158 15,927 0,052 12,148 0,325 0,024 12,677 0,152 10,456 0,016 Japan 4.599 1,059 0,638 0,306 15,773 0,048 13,412 0,205 0,016 11,262 0,479 10,590 0,003
21 Nr. of obs. Tobin's q CSR Disclosure Multinationality Firm Size Sales Growth Cash
Holdings Leverage Dividend
Board Size CEO Duality GDP per capita Inflation Korea 740 1,117 0,582 0,176 15,954 0,098 12,943 0,269 0,014 8,980 0,441 10,156 0,020 Malaysia 328 1,709 0,799 0,079 15,134 0,065 11,778 0,267 0,028 9,207 0,149 9,237 0,023 Mexico 220 1,642 0,677 0,355 15,832 0,114 12,556 0,292 0,016 15,182 0,495 9,184 0,041 Netherlands 374 1,213 0,845 0,209 16,219 0,065 13,156 0,270 0,026 7,939 0,051 10,816 0,015 New Zealand 119 2,024 0,370 0,328 14,118 0,086 10,072 0,295 0,044 7,160 0,118 10,497 0,020 Norway 260 1,207 0,588 0,115 15,363 0,086 12,463 0,296 0,028 8,154 0,185 11,330 0,020 Philippines 100 1,573 0,610 0,030 15,632 0,136 12,273 0,343 0,021 9,920 0,580 7,898 0,031 Poland 121 1,188 0,529 0,107 14,802 0,102 11,677 0,258 0,027 7,950 0,025 9,506 0,016 Portugal 102 1,230 0,725 0,343 15,391 0,058 10,415 0,403 0,031 13,373 0,402 9,981 0,015 Russian Federation 241 1,259 0,697 0,066 16,438 0,146 13,051 0,303 0,033 10,598 0,178 9,355 0,081 Singapore 478 1,249 0,400 0,123 15,335 0,102 12,120 0,242 0,030 10,031 0,213 10,779 0,019 South Africa 724 1,473 0,949 0,247 14,235 0,102 11,391 0,199 0,027 11,275 0,130 8,768 0,055 Spain 316 1,832 0,845 0,120 15,551 0,076 12,608 0,305 0,027 13,472 0,522 10,279 0,017 Sweden 548 1,548 0,743 0,155 15,372 0,093 12,159 0,259 0,029 10,504 0,161 10,873 0,011 Switzerland 599 2,002 0,673 0,192 15,182 0,069 12,798 0,189 0,018 8,504 0,451 11,226 0,003 Thailand 196 1,949 0,872 0,041 15,409 0,099 12,425 0,330 0,029 14,439 0,056 8,691 0,014 Turkey 163 1,300 0,693 0,104 15,528 0,171 12,669 0,290 0,032 12,712 0,037 9,306 0,090 United Kingdom 3.262 1,503 0,706 0,288 14,816 0,100 11,840 0,233 0,027 8,965 0,127 10,670 0,022 United States 7.437 1,883 0,398 0,394 15,740 0,088 12,935 0,269 0,013 10,154 0,682 10,871 0,019 Total 32.311 1,506 0,562 0,248 15,362 0,096 12,429 0,244 0,021 10,184 0,385 10,477 0,020
22
23
Tobin's q CSR Disclosure
Multinationality Crisis Firm Size
Sales Growth
Cash Holdings
Leverage Dividend Board Size CEO Duality GDP per capita Inflation Tobin's q 1,000 CSR Disclosure -0,109 1,000 Multinationality 0,039 -0,051 1,000 Crisis -0,093 -0,033 0,084 1,000 Firm Size -0,293 0,337 0,070 0,035 1,000 Sales Growth 0,142 -0,111 0,000 0,042 -0,068 1,000 Cash Holdings -0,134 0,246 0,111 0,000 0,712 -0,041 1,000 Leverage -0,137 0,044 -0,060 0,002 0,234 -0,042 0,046 1,000 Dividend -0,114 0,125 -0,076 0,006 0,044 -0,171 -0,058 0,081 1,000 Board Size -0,116 0,204 0,046 0,016 0,473 -0,065 0,345 0,108 0,034 1,000 CEO Duality 0,047 -0,057 0,125 0,030 0,162 -0,020 0,171 0,021 -0,119 0,089 1,000 GDP per capita -0,039 -0,114 0,106 0,038 -0,032 -0,046 0,045 -0,023 -0,018 -0,137 0,092 1,000 Inflation 0,073 0,003 -0,061 0,181 -0,053 0,157 -0,134 0,017 0,023 -0,010 -0,100 -0,556 1,000
24
The second hypothesis focuses on the effect of the variable multinationality on the relation between CSR disclosure and firm financial performance. Therefore, the descriptive statistics for the two groups, MNCs and DMCs, are examined separately and can be found in table 7. It can be seen that, on average, Tobin’s q of MNCs are higher than DMCs. On the other hand, it can also be seen that on average more DMCs disclose CSR compared to MNCs. Besides these differences, MNCs have on average a larger firm size, more cash holdings, are less leveraged, pay out less dividend, have a larger board, operate more often under CEO duality, are more often active in a country with higher economic development, and are more often active in countries with lower inflation rates compared to DMCs. What is further noticeable is the fact that, on average, both MNCs and DMCs have almost the same average sales growth rates. For MNCs, the average sales growth rate is 0,095, whereas for DMCs it is 0,096, and for the total sample this is 0,096. Of the total sample, there are 8.024 year-observations for MNCs and 24.287 year-observations for DMCs.
MNC DMC
Mean Median Std. Dev. Mean Median Std. Dev.
Tobin's q 1,594 1,222 1,158 1,477 1,109 1,160 CSR Disclosure 0,513 1,000 0,500 0,578 1,000 0,494 Multinationality 1,000 1,000 0,000 0,000 0,000 0,000 Firm Size 15,539 15,485 1,439 15,303 15,317 1,477 Sales Growth 0,095 0,066 0,223 0,096 0,059 0,238 Cash Holdings 12,791 12,919 1,697 12,309 12,518 1,974 Leverage 0,227 0,216 0,168 0,250 0,236 0,178 Dividend 0,018 0,014 0,018 0,022 0,017 0,022 Board Size 10,486 10,000 3,549 10,085 10,000 3,463 CEO Duality 0,491 0,000 0,500 0,350 0,000 0,477 GDP per capita 10,608 10,711 0,493 10,434 10,680 0,803 Inflation 0,018 0,019 0,017 0,021 0,019 0,020 Nr. of obs. 8.024 24.287
Table 7: Descriptive statistics (mean, median, and standard deviation) separated for MNCs and DMCs.
25
large impact in the financial performance of the firms. During the financial crisis, the average Tobin’s q is “just” 1,077, while in the other years, the average is 1,534. Also, the number of firms disclosing CSR is lower during the financial crisis as the average of the CSR disclosure dummy variable is during the financial crisis year 0,505 while for the other years it is 0,566. Furthermore, the average sales growth and country inflation rates are much lower in the financial crisis year compared to the other years. What is noticeable is that the average firm size, cash holdings, leverage, and GDP per capita are not so much lower during the financial crisis year compared to the other years. The average leverage, for example, is in the crisis year 0,240, while during the other years, this ratio is 0,244. Another noteworthy variable is dividend as the average dividend yield is even higher during the financial crisis year (0,037) compared to the other years (0,020). Also, the average board size and the number of firms that have CEO duality in place during the financial crisis year is higher than during the other years.
Crisis Other years
Mean Median Std. Dev. Mean Median Std. Dev.
Tobin's q 1,077 0,834 0,817 1,534 1,160 1,174 CSR Disclosure 0,505 1,000 0,500 0,566 1,000 0,496 Crisis 1,000 1,000 0,000 0,000 0,000 0,000 Firm Size 15,274 15,236 1,430 15,367 15,361 1,474 Sales Growth -0,016 -0,030 0,244 0,103 0,064 0,232 Cash Holdings 12,328 12,514 1,886 12,436 12,641 1,922 Leverage 0,240 0,226 0,173 0,244 0,231 0,176 Dividend 0,037 0,030 0,034 0,020 0,016 0,020 Board Size 10,237 10,000 3,590 10,181 10,000 3,482 CEO Duality 0,438 0,000 0,496 0,382 0,000 0,486 GDP per capita 10,463 10,618 0,667 10,478 10,703 0,746 Inflation 0,006 0,001 0,022 0,021 0,019 0,018
Table 8: Descriptive statistics (mean, median, and standard deviation) separated for the financial crisis year (2009) and the other years.
4. Regression results
To test and analyse the above-stated hypotheses, a total of six regression models have been tested. All regression models are tested including year fixed effects, country fixed effects, and industry fixed effects, and the estimation results of these models can be found in table 9.
26
(1) (2) (3) (4) (5) (6)
Tobin's q Tobin's q Tobin's q Tobin's q Tobin's q Tobin’s q
CSR Disclosure -0.098*** 0.127*** 0.126*** 0.148*** 0.117*** 0.139*** (0.027) (0.026) (0.026) (0.029) (0.027) (0.029) Multinationality 0.051 0.055 (0.039) (0.039) Crisis -0.482*** -0.480*** (0.042) (0.042) CSR Disclosure * -0.084* -0.092** Multinationality (0.043) (0.043) CSR Disclosure * 0.145*** 0.160*** Crisis (0.034) (0.034) Firm Size -0.287*** -0.286*** -0.286*** -0.286*** -0.286*** (0.017) (0.017) (0.017) (0.017) (0.017) Sales Growth 0.474*** 0.473*** 0.473*** 0.472*** 0.472*** (0.040) (0.040) (0.040) (0.040) (0.040) Cash Holdings 0.058*** 0.058*** 0.057*** 0.058*** 0.057*** (0.008) (0.008) (0.008) (0.008) (0.008) Leverage -0.378*** -0.376*** -0.373*** -0.376*** -0.373*** (0.085) (0.085) (0.085) (0.085) (0.085) Dividend -0.446 -0.429 -0.427 -0.424 -0.422 (0.492) (0.492) (0.492) (0.492) (0.493) Board Size 0.020*** 0.020*** 0.020*** 0.020*** 0.020*** (0.004) (0.004) (0.004) (0.004) (0.004) CEO Duality 0.070*** 0.071*** 0.071*** 0.071*** 0.071*** (0.025) (0.025) (0.025) (0.025) (0.025) GDP per capita -0.102 -0.101 -0.109 -0.109 (0.074) (0.074) (0.074) (0.074) Inflation -0.487 -0.502 -0.482 -0.499 (0.656) (0.656) (0.657) (0.657) Constant 0.969*** 4.346*** 5.434*** 5.400*** 5.516*** 5.488*** (0.161) (0.232) (0.800) (0.800) (0.800) (0.800) Observations 28,560 28,560 28,560 28,560 28,560 28,560 R-squared 0.373 0.430 0.430 0.430 0.430 0.430 Year x x x x x x Country x x x x x x Industry x x x x x x
Significance: *** p<0.01, ** p<0.05, * p<0.1; robust standard errors in parentheses
Table 9: Ordinary least squares regression results.
27
Tobin’s q. This means that in this simple model, a firm’s Tobin’s q is expected to go down when they start to disclose CSR.
As discussed earlier, there are also other variables that influence the relationship between CSR disclosure and firm financial performance. Where in the first simple regression model none of the earlier mentioned control variables are considered, this is not the case for the second regression model shown in table 9. In the second regression model, not only the main variables of interest for the first hypothesis are considered, but also the firm-level control variables have been included. As can be seen in table 9, the introduction of the firm-level control variables does have a significant impact on the estimation results of the CSR disclosure variable. Where in the first simple regression model the estimation result was negative, in the second model the estimation result for CSR disclosure is positive at the 1% significance level. This means that, contrary to the first model, when a firm starts to disclose CSR, an increase in their Tobin’s q should be expected.
The control variables firm size and leverage are negatively related to a firm’s Tobin’s q at the 1% significance level. On the other hand sales growth, cash holdings, board size, and CEO duality have a positive relation with Tobin’s q at a 1% significance level. What is noticeable from the second model is that the estimation result for the dividend variable is not significant. In the research of, for example, Amidu (2007), Farrukh et al. (2017), and Ouma (2012) a positive relationship between dividend and firm financial performance is found. The estimations of the second (and the following) model of the relation between dividend and Tobin’s q is negative and insignificant and are thus contrary to the findings by Amidu (2007), Farrukh et al. (2017), and Ouma (2012). The difference in the estimation results for the constant in the two models also stands out. In the first simple model, the estimation result for constant is 0,969 and significant at 1%. However, in the second model the estimation result for the constant is 4,346 and also significant at 1%. The second (and the following) model can be considered to describe the relationship better as the goodness of fit of the model described by the R-squared is higher compared to the first model.
28
that firms that disclose CSR have a higher financial performance compared to the firms not disclosing CSR.
The estimations of the firm-level control variables sales growth, cash holdings, board size, and CEO duality are positively related to Tobin’s q and also significant at 1%. The estimation results of firm size and leverage are at 1% significance negatively related to Tobin’s q. Similar to the second model, also in the third model the estimation result for the dividend variable is not significant. Furthermore, also the two country-level control variable estimations that have been added to the model are not significant. In this regression model, the estimation results for GDP per capita and inflation rate are negatively related to Tobin’s q but are not significant (and is also not significant in any of the following models). These findings contradict the results of, for example, Konchitchki (2011) and Churchill and Valenzuela (2019).
Taken this all into account, based on the third model, there is no evidence to reject the first hypothesis. The first hypothesis states that CSR disclosure positively affects firm financial performance. This is also what is seen from the estimation results as CSR disclosure is positively related to Tobin’s q at 1% significance. This means that firms that disclose CSR expects to have a higher Tobin’s q, and thus is CSR disclosure positively related to firm financial performance. The five theories that have been discussed previously predicted a positive relationship, which is now also found in this empirical analysis. These results are contradictory to the previously presented literature by Selcuk and Kiymaz (2017) and in line with Bayoud, Kavanagh, and Slaughter (2012), Mohammed, Saheed, and Oladele (2016), and Gustsche, Schultz, and Gratwohl (2017).
The fourth model in table 9 is a further extension to the third model and includes a variable to measure whether a firm is either an MNC or DMC. This is needed to test the second hypothesis, which is stated in section 2. Similar to the previous two models, also in the fourth model the estimation results of sales growth, cash holdings, board size, and CEO duality are positively related to Tobin’s q at the 1% significance level. Firm size and leverage have a negative relation with Tobin’s q at the 1% significance, whereas dividend, GDP per capita, and inflation are not significant.
29
and the estimation result of the multinationality variable is not significant. From this model, DMCs that disclose CSR are expected to have a 0,148 higher Tobin’s q compared to DMCs that do not disclose CSR. For MNCs, this relation is also found, but the effect of MNCs disclosing CSR compared to MNCs that do not disclose CSR is less than the difference for DMCs. The difference in Tobin’s q between MNCs that disclose CSR compared to not disclosing CSR is estimated to be 0,064 (0,148-0,084). There is no expected difference between DMCs that do not disclose CSR and MNCs that do not disclose CSR.
Taken this together it means that DMCs that disclose CSR are expected to have a larger increase in Tobin’s q compared to MNCs that disclose CSR. This evidence is not in line with hypothesis 2 and thus does provide evidence to reject hypothesis 2. Although still a positive relation for MNCs is found, the results show a negative effect of multinationality on the relation between CSR disclosure and firm financial performance and thus weakens the relationship. MNCs that disclose CSR are expected to have a lower increase in Tobin’s q compared to DMCs that disclose CSR, and are thus expected to have a lower increase in firm financial performance. However, for both DMCs and MNCs, disclosing CSR is expected to affect Tobin’s q and thus firm financial performance positively, but the effect is less for MNCs.
The fifth model in table 9 is another extension of the third model and includes a variable to measure the financial crisis. This is needed to test the third hypothesis that is already elaborated above. Similar to the previous model, also in this model, the estimation results for sales growth, cash holdings, board size, and CEO duality are positively related to Tobin’s q. The estimation results for firm size and leverage are also negatively related to Tobin’s q, while the estimation results for dividend, GDP per capita, and inflation are not significant.
30
0,145). The firms that have disclosed CSR in the year before the financial crisis have thus a total expected benefit of 0,262 (0,117 + 0,145) from disclosing CSR in the year before the financial crisis compared to the firms that did not disclose CSR in the year before the financial crisis.
As the firms that had disclosed CSR in the year before the financial crisis are estimated to have a smaller reduction in the Tobin’s q and thus a smaller reduction in their financial performance, the evidence is in favour of the third hypothesis and is thus not rejected.
Model 6 in table 9 combines the previous three models that were used to test the three hypotheses and includes two interaction terms. This allows to see if the three hypotheses hold when they are combined in one regression. The results show that no major changes emerge compared to the previous three models and therefore also jointly the conclusions drawn up earlier holds.
From this model, it becomes clear that firms that disclose CSR have a better financial performance compared to firms that do not disclose CSR. However, for MNCs the effect of disclosing CSR is lower than for DMCs but is still positive. It is also clear that during the financial crisis, the firms that had disclosed CSR in the previous year benefited from this as they have, on average, a higher average Tobin’s q.
These results do have some implications for managers and policy-makers of firms. It is clear that firms that disclose CSR have a higher financial performance compared to the firms that do not disclose CSR. Although it depends on the type of firm, it can be used to calculate whether spending money on CSR and CSR disclosure can benefit the firm. Managers can also use CSR disclosure as a hedging strategy as it can be seen that firms that had disclosed CSR in the year before the financial crisis benefitted from this. However, also these calculations are different for each firm as the managers need to make estimations on their own about the cost, the negative impact of a financial crisis, and the likelihood of such a crisis to emerge.
5. Robustness
31
the paper by Selcuk and Kiymaz (2017) found a negative relationship between CSR disclosure and firm financial performance using return on assets (ROA) as measure for firm financial performance. On the other hand, Bayoud, Kavanagh, and Slaughter (2012) found a positive relationship with both the firms’ ROA and return on equity (ROE) as measure of firm financial performance. Therefore, the regression models 3,4, and 5 which were used to test the three hypotheses individually are also computed using two other measures of firm financial performance of which the full formula can be found in table 10. Besides only examining the effect of different measures for firm financial performance, these models are also computed using another measure of CSR disclosure which is also described in table 10.
5.1. Robustness variables
32
ROA Return On Assets: (Net Income - Bottom Line +((Interest Expense on Debt-Interest Capitalized)*(1-Tax Rate)))/Average of Last Year's and Current Year's Total Assets.
ROE Return on Equity: (Net Income - Bottom Line - Preferred Dividend
Requirement)/Average of Last Year's and Current Year's Common Equity.
ESG score An overall company score based on the self-reported information in the environmental, social, and corporate governance pillars.
Table 10: Description of the additional variables used for the robustness tests.
5.2. Robustness hypothesis 1
In table 11, the regression results of the model used to test the first hypothesis can be found using the three different firm financial performance measures and two different measures for CSR disclosure. In the first model, Tobin’s q is used as the measure for firm financial performance to examine the difference between the models properly. Similar to the regression results in table 9, also in this model the regression estimation of the CSR disclosure dummy variable is positive. However, what stands out is that the estimations of the variables leverage and GDP per capita are different. The estimation result of leverage is more negative, and the estimation result for the GDP per capita variable has become significant at 5% significance. When looking at the variables CSR disclosure, sales growth, cash holdings, board size, and CEO duality are positively related to Tobin’s q and are all significant. The variables firm size, leverage, and GDP per capita are significant and negatively related to Tobin’s q. Finally, the estimations of the variables dividend and inflation are not significantly related to Tobin’s q.
33
significantly negative related to firm financial performance, and board size and inflation are insignificant. Comparing the control variables of the second and third model, the only differences are leverage, which is in the third model insignificant, and board size, which is in the third model positive at 10% significance. When comparing the goodness of fit of the three models, the first model which uses Tobin’s q as measure for firm financial performance stands out. Whereas the first model has an R-squared of 0,435, the other two models have a significant lower R-squared value. The second model using ROA as measure for firm financial performance has an R-squared of 0,247 and the model using ROE as measure for firm financial performance an R-squared of 0,206. Therefore the model which uses Tobin’s q as measure for firm financial performance fits the observations better compared to the other two models.
34
CSR dummy ESG Score
(1) (2) (3) (4) (5) (6)
Tobin's q ROA ROE Tobin’s q ROA ROE
CSR Disclosure 0.127*** 0.011*** 0.037*** 0.724*** 0.037*** 0.129*** (0.026) (0.002) (0.006) (0.093) (0.006) (0.019) Firm Size -0.259*** -0.003*** -0.015*** -0.292*** -0.005*** -0.019*** (0.016) (0.001) (0.004) (0.018) (0.001) (0.004) Sales Growth 0.464*** 0.039*** 0.082*** 0.472*** 0.039*** 0.082*** (0.040) (0.004) (0.008) (0.040) (0.004) (0.008) Cash Holdings 0.054*** 0.003*** 0.011*** 0.050*** 0.003*** 0.011*** (0.008) (0.001) (0.002) (0.008) (0.001) (0.002) Leverage -0.661*** -0.057*** 0.035 -0.632*** -0.056*** 0.039* (0.087) (0.006) (0.022) (0.087) (0.006) (0.022) Dividend -0.393 0.338*** 1.022*** -0.716 0.325*** 0.975*** (0.482) (0.045) (0.136) (0.480) (0.045) (0.136) Board Size 0.018*** 0.000 0.001* 0.018*** 0.000 0.002** (0.004) (0.000) (0.001) (0.004) (0.000) (0.001) CEO Duality 0.065*** 0.004** 0.013*** 0.074*** 0.005*** 0.015*** (0.024) (0.002) (0.005) (0.024) (0.002) (0.005) GDP per capita -0.146** -0.049*** -0.071*** -0.156** -0.050*** -0.075*** (0.072) (0.006) (0.017) (0.072) (0.006) (0.017) Inflation -0.741 -0.063 0.109 -0.634 -0.061 0.117 (0.648) (0.052) (0.155) (0.651) (0.052) (0.155) Constant 5.676*** 0.555*** 0.786*** 6.001*** 0.575*** 0.854*** (0.780) (0.067) (0.182) (0.780) (0.067) (0.183) Observations 27,736 27,736 27,736 27,736 27,736 27,736 R-squared 0.435 0.247 0.206 0.440 0.247 0.207 Year x x x x x x Country x x x x x x Industry x x x x x x
Significance: *** p<0.01, ** p<0.05, * p<0.1; robust standard errors in parentheses
Table 11: Ordinary least squares regression result for the robustness test of hypothesis 1.
5.3. Robustness hypothesis 2
35
but insignificant. For the interaction term, there are some differences between the models. Although in all models the relationship is negative, in the second and third model the significance level is higher. The conclusions that have been drawn previously still applies for both the second and third model, however, the strength of the evidence is higher due to the higher significance level of the interaction term. The differences in the control variables of the first three models presented in table 12 are the same as the differences in the same models presented in table 11. There are only some minor magnitude differences between estimates of the regression models in table 12 compared to table 11.
36
CSR dummy ESG score
(1) (2) (3) (4) (5) (6)
Tobin's q ROA ROE Tobin’s q ROA ROE
CSR Disclosure 0.148*** 0.013*** 0.041*** 0.729*** 0.036*** 0.123*** (0.029) (0.002) (0.006) (0.101) (0.007) (0.020) Multinationality 0.047 0.003 0.007 0.005 -0.004 -0.018 (0.039) (0.003) (0.007) (0.075) (0.006) (0.016) CSR Disclosure * -0.082* -0.008** -0.018** -0.016 0.004 0.027 Multinationality (0.043) (0.003) (0.009) (0.127) (0.009) (0.028) Firm Size -0.259*** -0.003*** -0.015*** -0.292*** -0.005*** -0.019*** (0.016) (0.001) (0.004) (0.018) (0.001) (0.004) Sales Growth 0.464*** 0.039*** 0.082*** 0.472*** 0.039*** 0.082*** (0.040) (0.004) (0.008) (0.040) (0.004) (0.008) Cash Holdings 0.053*** 0.003*** 0.011*** 0.050*** 0.003*** 0.011*** (0.008) (0.001) (0.002) (0.008) (0.001) (0.002) Leverage -0.658*** -0.057*** 0.036 -0.632*** -0.056*** 0.039* (0.087) (0.006) (0.022) (0.087) (0.006) (0.023) Dividend -0.391 0.338*** 1.021*** -0.717 0.324*** 0.973*** (0.482) (0.045) (0.136) (0.481) (0.045) (0.136) Board Size 0.018*** 0.000 0.001* 0.018*** 0.000 0.002** (0.004) (0.000) (0.001) (0.004) (0.000) (0.001) CEO Duality 0.065*** 0.004** 0.013*** 0.074*** 0.005*** 0.015*** (0.024) (0.002) (0.005) (0.024) (0.002) (0.005) GDP per capita -0.145** -0.049*** -0.071*** -0.156** -0.051*** -0.075*** (0.072) (0.006) (0.017) (0.072) (0.006) (0.017) Inflation -0.761 -0.068 0.100 -0.638 -0.064 0.109 (0.648) (0.052) (0.154) (0.652) (0.052) (0.155) Constant 5.646*** 0.553*** 0.781*** 6.000*** 0.577*** 0.860*** (0.780) (0.067) (0.181) (0.779) (0.067) (0.183) Observations 27,736 27,736 27,736 27,736 27,736 27,736 R-squared 0.436 0.247 0.206 0.440 0.247 0.207 Year x x x x x x Country x x x x x x Industry x x x x x x
Significance: *** p<0.01, ** p<0.05, * p<0.1; robust standard errors in parentheses
37
5.4. Robustness hypothesis 3
Finally, table 13 presents the regression results to the third hypothesis using Tobin’s q, ROA, and ROE respectively. Also, when comparing these first three models, the CSR disclosure variables as well as the crisis variable are similar. For the first three models, the CSR disclosure variable is positive and significant at 1%, and the crisis variable is negative and significant at 1%. Also, for this robustness test, differences exist in the estimates for the interaction term. Where the interaction term in the first model was positive and significant at 1%, for the second model, the sign has changed, and a negative relationship has been found at 5% significance. Also, for the third model, a negative but insignificant relation is found. This means that using Tobin’s q as a measure for firm financial performance gives different estimation results compared to using ROA or ROE as measure for firm financial performance. Where the first model estimates that firms disclosing CSR in the years before the financial crisis are expected to have a smaller reduction in their firm financial performance during the financial crisis, the opposite is found in the second model. In the second model, the estimations find that disclosing CSR in the year before the financial crisis would be expected to have a negative effect on firm financial performance during the financial crisis. As the estimation result for the interaction term in the third model is insignificant, it would be expected that the effect of the financial crisis on the financial performance of firms is not related to CSR disclosure in the year before the financial crisis.
38
CSR dummy ESG score
(1) (2) (3) (4) (5) (6)
Tobin's q ROA ROE Tobin’s q ROA ROE
CSR Disclosure 0.118*** 0.011*** 0.038*** 0.712*** 0.037*** 0.127*** (0.027) (0.002) (0.006) (0.001) (0.000) (0.000) Crisis -0.475*** -0.028*** -0.084*** -0.485*** -0.028*** -0.092*** (0.042) (0.004) (0.012) (0.062) (0.007) (0.020) CSR Disclosure * 0.136*** -0.008** -0.012 0.229** 0.001 0.030 Crisis (0.034) (0.004) (0.011) (0.094) (0.011) (0.032) Firm Size -0.259*** -0.003*** -0.015*** -0.292*** -0.005*** -0.019*** (0.016) (0.001) (0.004) (0.018) (0.001) (0.004) Sales Growth 0.464*** 0.039*** 0.082*** 0.472*** 0.039*** 0.082*** (0.040) (0.004) (0.008) (0.040) (0.004) (0.008) Cash Holdings 0.053*** 0.003*** 0.011*** 0.050*** 0.003*** 0.011*** (0.008) (0.001) (0.002) (0.008) (0.001) (0.002) Leverage -0.661*** -0.057*** 0.035 -0.631*** -0.056*** 0.039* (0.087) (0.006) (0.022) (0.087) (0.006) (0.022) Dividend -0.389 0.338*** 1.021*** -0.712 0.325*** 0.975*** (0.482) (0.045) (0.136) (0.480) (0.045) (0.136) Board Size 0.018*** 0.000 0.001* 0.018*** 0.000 0.002** (0.004) (0.000) (0.001) (0.004) (0.000) (0.001) CEO Duality 0.065*** 0.004** 0.013*** 0.074*** 0.005*** 0.015*** (0.024) (0.002) (0.005) (0.024) (0.002) (0.005) GDP per capita -0.152** -0.049*** -0.070*** -0.159** -0.050*** -0.076*** (0.072) (0.006) (0.017) (0.072) (0.006) (0.017) Inflation -0.736 -0.063 0.109 -0.633 -0.061 0.117 (0.648) (0.052) (0.155) (0.652) (0.052) (0.155) Constant 5.754*** 0.551*** 0.779*** 6.044*** 0.575*** 0.859*** (0.781) (0.067) (0.181) (0.781) (0.067) (0.183) Observations 27,736 27,736 27,736 27,736 27,736 27,736 R-squared 0.436 0.247 0.206 0.440 0.247 0.207 Year x x x x x x Country x x x x x x Industry x x x x x x
Significance: *** p<0.01, ** p<0.05, * p<0.1; robust standard errors in parentheses
39
To conclude, as shown in all robustness test that has been performed, the chosen measure for both firm financial performance and CSR disclosure matters. It is clear that using different measures for these variables will result in different conclusions. Therefore, these results should be taken into account when analysing CSR disclosure and/or firm financial performance. In future research, the results should not be based solely on one measure, but different measures for these variables should be taken into account.
6. Conclusion
In this study, the relationship between CSR disclosure and firm financial performance is analysed by first examining existing theories and previous literature and secondly by empirical analyses. In finding a theoretical relationship between CSR disclosure and firm financial performance, five different theories have been considered. The five theories are the agency theory, stakeholder theory, institutional theory, legitimacy theory, and the theory of slack resources. Although these theories are different in explaining why CSR disclosure and firm financial performance are related, the outcome of the theories is the same. All theories predict a positive relationship between CSR disclosure and firm financial performance. Another focus of the literature review has been on the differences between MNCs and DMCs, and its implications for the relation between CSR disclosure and firm financial performance. It has become clear that MNCs are under greater scrutiny, and due to its international character, face more different expectations compared to DMCs. What has not become clear is whether the characteristics of being an MNC has an effect on the relationship between CSR disclosure and firm financial performance. Also, the effect of the 2008/2009 financial crisis on the relationship between CSR disclosure and firm financial performance has been examined. From the presented theories, it would be expected that firms that were disclosing CSR before the financial crisis, would still benefit from this during the beginning of the financial crisis.
40
relation between CSR disclosure and firm financial performance. When analysing the relationship for the whole sample during the 2008/2009 financial crisis, it has become clear that the effect of disclosing CSR before the financial crisis has a positive effect on the firm financial performance during the financial crisis. The presented evidence shows that the firms that disclosed CSR in the year before the indicated financial crisis year of 2009 benefited from this during the 2009 financial crisis year. Although all firms show a relatively large reduction in the financial performance in the financial crisis year, the firms that had disclosed CSR in the year before the financial crisis were less affected by the financial crisis.
There are, however, some important notes that need to be made regarding the empirical analysis. When using different measures for firm financial performance, the conclusion regarding the relation between CSR disclosure and firm financial performance still holds. Small differences exist in the analyses of the effect of being an MNC when different measures for firm financial performance are used. However, in the analyses of the 2008/2009 financial crisis relative large differences exist between the models using different measures for firm financial performance. This is important as the difference in the outcomes of the models result in different conclusions. Also, when using ESG score as measure instead to proxy CSR disclosure, the regression results are different, and different conclusions are to be drawn compared to the models using the CSR disclosure dummy variable. This becomes especially clear from the regression results of the second hypothesis where the conclusion is that, regardless of the measurement of firm financial performance, there is no difference between MNCs and DMCs which disclose CSR. Also, in the other models, some different results are presented, which can also change again when different measures for firm financial performance are used.
