Differences in corporate social responsibility disclosure between Japan and the USA

Purpose – The purpose of this paper is to examine how and why disclosure of corporate social responsibility (CSR) information was influenced by independent directors in Japan and the USA. Design/methodology/approach – The author used a pooled cross-sectional data set of 498 Fortune Japanese and American firms between 2006 and 2011 and fixed effects estimation method. The author analysed the results by employing a comparative approach between the two national contexts. Findings – This study found that independent directors in Japanese firms had a significant positive effect on CSR disclosure whilst no evidence was found in the US firms, although the proportion of independent directors on American boards traditionally and largely outnumbers that of the Japanese counterparts. Originality/value – The study results offer an insight that independent directors could be evaluated in terms of effectiveness and efficiency in CSR disclosure. The findings support the stakeholder theory in Japanese globalised companies while challenging the theory in the US context, thereby calling for further research into the stakeholder engagement models, particularly in the USA.


Introduction
Stakeholder theory (Freeman, 1984) advocates that companies have responsibility to a range of stakeholders, internally and externally, rather than merely the firm owner. This study uses the stakeholder view to examine the influence of independent directors on firm's corporate social responsibility (CSR) disclosure in the context of Japan and the US large companies. This paper empirically contributes to the literature by showing the evidence that the stakeholder approach is significant with Japanese independent directors, while not evident in American independent directors, thus challenging the global stakeholder theory.
To date, only very few studies (e.g., Gul and Leung, 2004;Patelli and Prencipe, 2007;Jizi et al., 2014) have examined how the factor related to independent directors affects CSR disclosure. These studies used the context of a single country (Hong Kong, Italy and the USA, respectively), not the context of two specific countries that allows the findings to be compared. Prado-Lorenzo and Garcia-Sanchez (2010) studied how the level of independence and diversity of board of directors affect disseminating information related to greenhouse gasesa type of CSR disclosurein multiple countries; however, their study does not explore effect on the disclosure of social performance of the firms. Bear et al. (2010) and Zhang et al. (2013) investigated how the factor related to independent directors of the US and non-US firms influences the firms' CSR reputation, not CSR disclosure.
This paper goes a step further by observing the two comparative national settings using pooled cross-sectional data related to multiple industries for an investigation of how independent directors impact CSR disclosure, in both environmental and social aspects. The paper is, to some extent, the response to the call for more research on the importance of informal governance mechanisms (Stafsudd, 2009;Filatotchev and Nakajima, 2010) and for cross-cultural studies to capture the context sensitivities of corporate governance (CG) and CSR ( Jain and Jamali, 2016).
The remaining sections are the literature review followed by the research method and the empirical results. Finally, the discussions and conclusion highlight the implications in theory and practice and suggest further research.

Literature review
The business and society literature focusses on the interdependence of businesses with other societal elements, viewing firms as corporate citizens (Matten and Crane, 2006;Jones and Haigh, 2007). Drawing the framework from the business and society literature, the stakeholder theory puts CSR into the perspective of social and environmental responsiveness which companies should adopt for the benefit of the stakeholders (Freeman and Velamuri, 2006).
Although there are vigorous debates on the possibility of a universal concept of CSR (Freeman and Hasnaoui, 2011) and how and why CSR differs among nations (Welford, 2005), there is a general consensus that stakeholder groups wholly or partly share similar social and environmental reversal impacts caused by firms' activities. Therefore, stakeholders evaluate corporate social and environmental performance, demand managerial responsiveness to CSR movements and monitor CSR activities (Wood and Jones, 1995). CSR-related experiences and beliefs can drive stakeholder trust and positive intent (Hillenbrand et al., 2013); however, the impact of modern economic activities on the quality of life has led to a growing concern about CSR among stakeholders (Raelin and Bondy, 2013).
There might be conflicts of interests between shareholders who expect financial gain and stakeholders who seek long-term social values. Thus, the reduction of conflicts of interests between shareholders and the stakeholders in order to push positive social change is one of the central issues of CSR management (Aguilera et al., 2007). The enlightened stakeholder perspective ( Jensen, 2001) suggests a possibility that stakeholders and shareholders compromise on their shared values (Porter and Kramer, 2006). More recently, stakeholder theory has been extended further into the strategy realm of value creation from the stakeholder synergy perspective (Tantalo and Priem, 2016). The point is to find out who in a company is able to maintain existence of the shared value and how to catalyse creation of the shared value in a specific legal, economic and cultural context.

Who forms the "catalyst"?
Executives are advised and monitored by a board of directors to protect the owner's rights ( Jensen and Meckling, 1976) while the company activities related to social and environmental issues are externally overseen by a wide range of stakeholder groups (Clarkson, 1995) since their benefits are impacted by the firm's activities. Board of directors basically focus on the traditional support of shareholder value creation instead of dealing with broader responsibilities including CSR (Prado-Lorenzo and Garcia-Sanchez, 2010). Between the internal and external directors on the board, the external directors are more likely to take a neutral view of the concerns of external stakeholders due to the role of independent monitoring, advising and resource provision (Pfeffer and Salancik, 1978) stated on their employment contracts. Further, boards are often expanded for political reasons to include the non-executive directors with the task of stakeholder representatives and with concerns other than shareholders' value (Agrawal and Knoeber, 2001).
The concept of independent directors remains controversial in the literature (Tricker, 2015). Independent directors have been loosely defined as people who have never been employees of the firms (Dore, 2005) or its subsidiaries in both Japan and the USA (Aguilera, 2005). Whilst the structure, composition and independence of the board create the conditions of board effectiveness, it is actually the conduct of the non-executive vis-à-vis the executive that determines board effectiveness (Roberts et al., 2005). Given the probability of executive dominance over the board, non-executive directors should bring independent judgements to bear on issues of strategy, performance and standards of conduct (Cadbury, 1992). Independent directors are subject to the terms and conditions as the agent for shareholders; moreover, by having networks with stakeholders (Powell, 1990;Kim and Cannella, 2008), independent directors can provide information, advice and connections to other organisations, and access to external resources that can be of added value for shareholders, leading to strategic change (Haynes and Hillman, 2010).
Above all, two main arguments have been advanced in the support of the impact of independent directors on using the stakeholder approach in the board rooms during periods of economic recession. First, the inclusion of independent directors is related to better supervision of executive decisions and activities ( Jensen and Meckling, 1976). Independent directors are expected to focus less on short-term financial performance targets and more on the measures that enhance long-term corporate sustainability, such as engaging in CSR (Hung, 2011) and reporting on CSR (Ibrahim et al., 2003;Jamali et al., 2008;Arora and Dharwadkar, 2011). Second, independent directors offer objective advice to corporate boards on strategic decisions (Tricker, 2015); thus, board resource diversity could be gained from having more independent members (Bear et al., 2010). The greater the diversity of board resource, the greater the potential for problem understanding and problem solving with a positive engagement in CSR performance and CSR disclosure tends to be (Hafsi and Turgut, 2013). Research shows that a separation of the two leadership roles, in line with these two arguments, can positively impact future performance of a firm when current performance is poor (Krause and Semadeni, 2013).

How to "catalyse"?
On the one hand, independent directors should offer their opinions on a possibility of reducing costs, for example, on CSR activities, due to a concern about CEO opportunism, such as building ethical reputation (Trevino et al., 2000) and concerns about shareholders' costs (Fama and Jensen, 1983). As a result, CSR might not potentially be of shareholder interest because CSR activities can prove costly to them. Thus, independent directors might argue for low CSR but high disclosure of CSR activities to placate stakeholders.
On the other hand, in favour of enlightened shareholder value ( Jensen, 2001), independent directors might argue that CSR would actually serve shareholder long-term interests by controlling agency costs related to social and environmental conflicts between firms and a wide range of stakeholder groups. Meanwhile, CSR is likely to constitute a resource that leads to a wide range of long-term benefits, namely a sustained competitive advantage (Hart, 1995;Mcwilliams et al., 2002;Porter and Kramer, 2006), attracting more socially responsible customers (Baron, 2001), reducing negative events that would otherwise harm the firm's financial performance (Peloza, 2006), experiencing a significant stock price increase (Gao et al., 2011), or giving job seekers the signals that ultimately affect organisational attractiveness ( Jones et al., 2014). Thus, independent directors might push for CSR disclosure to improve corporate values.
Higher proportion of independent directors on a corporate board would limit managerial opportunism because managers are pushed to be more accountable to shareholders who are increasingly disconnected from management, given the vertical and horizontal expansions of corporations (Fama and Jensen, 1983). Cadbury's (1992) report recommends that the calibre and number of non-executive directors on a board should be such that their views will carry significant weight in the board's decisions. It is expected that independent directors would be more able to exert greater influence on management decisions to disclose information when the proportion of independent board members is higher. This will, in turn, enhance the comprehensiveness and quality of disclosures (Forker, 1992;Beasley, 1996) and hence contribute to better governance following Williamson's (1985) theoretical framework which links disclosure quality with CG.
CSR disclosure is defined as the communication of the social and environmental impacts resulting from an organisation's economic actions on particular interest groups and on society at large (Gray et al., 1996). It is necessary for executives to make efforts to facilitate the proper counselling and resource provision functions of the independent directorate system by disseminating information in a timely and appropriate manner to all concerned. Besides that, the establishment of relationship with external stakeholders enables the independent board members to access information sources and capture hands-on information. Knowledge sharing between the external and internal board members can be valuable to the board, and thus, enhances the decision-making process ( Johansen and Pettersson, 2013).
Stakeholders tend to expect that independent directors, as representatives of external stakeholders, will drive greater corporate responsiveness to societal needs ( Johnson and Greening, 1999). Independent directors are supposed to have an impact on managerial accountability to internal and external stakeholders, which leads to an assumption that independent directors, if adequate numbers exist on board membership, would be able to influence management decisions on CSR disclosure. Therefore, the following hypothesis is proposed: H1. A higher proportion of independent directors on a board is likely to result in increased level of CSR disclosure.
To the best of the author's knowledge, only a few recent studies looked into independent directors' direct impact on CSR-relevant strategy. In a study using longitudinal data covering 98 observations for 2009, 97 observations for 2010 and 96 observations for 2011, Jizi et al. (2014) indicated that board independence is positively associated with CSR disclosure in the US banking sector. Although this study employs longitudinal data, it only looks at a single industry, banking, in a single country, the USA. Zhang et al. (2013) found a positive link between greater presence of independent directors on the board and better CSR reputational rankings of the firms from the Fortune World's Most Admired (FWMA) survey conducted in 2007, the rankings released in 2008. This study used the data of 516 US and non-US companies in 2007 for the independent and control variables. Due to the limitation of using cross-sectional data for the study, Zhang et al. (2013) called for the future research utilising other CSR measures in a longitudinal study design that assesses CSR performance. Bear et al. (2010) did not find that diversity of director resourceshaving outside directors on the board as one of the criteria of the diversitysignificantly affects CSR reputational ranking. Their study used the data of 51 US and non-US healthcare firms on CSR reputational rankings released by Fortune 2009 for the dependent variable. For the independent and control variables, the study used the 2007 data of the studied firms, which was one year behind that of the Fortune survey conducted, i.e. 2008. However, their study only looks at a single industry.
Overall, the significance and magnitude of the impact of independent directors on CSR strategy remain empirically unanswered; this is a challenge against the practicality of stakeholder theory, given the pressure of stakeholders heightened on the socio-economic agendas worldwide. The limitations of those empirical papers, either using cross-sectional data or exploring a single industry, point to the need for further research on the impact of having outside directors in board membership on CSR strategy in various contexts in terms of industry, time and country.
This study is of importance for the four further reasons. First, integrating CSR-driven principles into CG has generated higher demand among stakeholders (Devinney et al., 2013), particularly since the early 2000s. As external stakeholders are exposed to the environmental and social impacts of daily business activities, they expect corporations to use the stakeholder approach in monitoring and advising the management on the issues related to stakeholders' interests, such as CSR commitments (Mason and Simmons, 2014). However, the validity of stakeholder theory in governance practices demands further research (Aguilera et al., 2016), since different configurations of governance characteristics influence directors' engagement in governance behaviours (Yoshikawa et al., 2014).
Second, scholars are often sceptical about investigating the role of outside directors as a governance mechanism in Asian firms because, traditionally, Asian boards consist of mostly insiders (Peng, 2004) or close contacts of the firm's founder (Kim, 2005). Independent directors in Asian firms, though, play important roles in providing supplementary resources, advice and counsel to the executives to make better decisions in order to enhance firm capacity and performance (Chen, 2014). To this extent, the independent directors are as important as those in western firms. Independent directors are expected to develop and maintain good relationships with external stakeholders, meaning that shareholders can benefit from the resources supplied by external networks ( Johanson and Østergren, 2010).
Third, it is believed that a national system of CG evolves in a manner consistent with a country's history, legal environment and socio-cultural traditions (Kim, 2005); thus, it is important to investigate how independent directors in a unique national CG system influence CSR activities. Globally, there are two mechanisms for the implementation of governance codesmandatory or voluntary regulations of compliance (Aguilera and Cazurra, 2009). The classic examples of the mandatory approach are legislations, e.g., the US Sarbanes-Oxley Act of 2002, and the Japan Financial Instruments and Exchange Law ( J-SOX in short) of 2006, both of which provide strict rules for the internal control of financial reporting in order to protect investors by improving the accuracy and reliability of disclosures. Given the unique characteristics of the culture and history in each country (Hooghiemstra et al., 2015), Japan and the USA provide interesting contexts for comparing the effect of independent directors on stakeholder engaging strategies in each institutional environment that shape CSR (Wang et al., 2016).
Fourth, in the implementation of governance principles, "the ways by which the game is actually played" should be of vital concern (Aoki et al., 2007). Whether there are actual changes or solely at face value in CG reform remains open because directors might be largely influenced by management (Zhu and Westphal, 2014) due to their vested interests. Given controversial debate on how to embed the stakeholder model (Carroll and Buchholtz, 2012) into governance systems and the volume of related rhetoric reporting recently in the press, there is a pressing need for objective investigation of how independent directors affect CSR to develop the insights of the underlying mechanisms in the governance reform and of the applicability of stakeholder theory in this reform.

Research method
3.1 Context Following Jo and Harjoto's (2012) finding of a causal effect of governance on CSR, I tested the hypothesis on impact of independent directors on CSR disclosure in top companies in the two comparative contexts of Japan and the USA. I was also interested to see the change this impact has within each studied firm across the years of the recent financial crisis. I chose Japan and the USA as the contexts of study because I am partially motivated by the statement that Japan might lag behind Anglo-Saxon countries in its understanding and adoption of CSR, a Western-led concept (Fukukawa and Teramoto, 2009) and that the global financial crisis could be a test for CSR-related theories (Kemper and Martin, 2010).
The US Sarbanes-Oxley Act enacted in 2002 as the response to the need for a sustainable financial market in the USA has been the global sweeping changes affecting CG in the early 2000s (Zhang et al., 2013). Although the act did not address the issue of board member independence outside the audit committee from which its members must be independent, the overall board composition tends to be influenced by the act whereby more independent members are likely to be nominated and elected to the corporate board (Valenti, 2008).
The globalisation of the stock market has increased the call for financial sustainability from Anglo-Saxon systems and disseminated CSR movements into Japan (Dore, 2005). The J-SOX legislation requires companies to have a subcommittee on internal controls established for the evaluation of the effectiveness of internal control over financial reports, for which the requirements of J-SOX are similar to those under the US Sarbanes-Oxley Act. Compliance with J-SOX requirements is effective for fiscal years beginning on or after April 1, 2008. Under this new legislation, the Tokyo Stock Exchange issued the new rules in 2009, which require a listed company to secure among its outside directors at least one person who is unlikely to have any conflicts of interests with general shareholders and to notify the issuer of their appointment as independent director(s)/auditor(s).
The selection of the US and Japan contexts for this study is, therefore, driven by the similarities that both countries have in terms of the mandatory CG mechanisms, which enables the relative comparisons of the hypothesis testing results. The selection is also due to the strong differences in the percentage of independent board members in Japan and the US companies.
The period between 2006 and 2011 was chosen since the recent global financial crisis spanned these years, in which 2008 was the most unprofitable year recorded by the companies in the data set where the lowest negative means of annual return on equity ROE were shown (data available upon request). Consequently, the hypothesis testing using the context of the observed years would reflect the time factor related to this credit crisis (Stein, 2015).
There are four main reasons for the selection of the globally known FWMA Japanese and American firms for testing the hypothesis. First, FWMA ranking exercises are conducted on the firms selected based on the similar worldwide applied criteria of Fortune. The exercises are based on the surveys with a large number of executives, directors and security analysts who rated the companies in their own industry for the selection of the companies they admired the most, in which CSR was considered as one of the key areas of leadership of a company in the relevant industry. Thus, given accelerated global competition, FWMA companies seem to be the leaders in adopting CSR principles and practices from the Anglo-Saxon nations (Flammer, 2015). Second, previous studies have positively reported that FWMA firms of these countries provide more CSR information (Chan et al., 2014). It is also empirically evident that preserving the already established reputation requires a firm to deliver consistent performance over time (Petkova et al., 2014). Third, out of all firms that appeared on the FWMA ranking result from 2006 to 2011, the number of US firms is always the largest followed by the number of Japanese firms. This is relatively proportionate to the size of these two global leading economies in the early 2000s. Fourth, with a global reputation and total assets between over USD12 billion and 1,900 billion for Japanese firms in 2011, between USD800 million and 2,300 billion for the US firms in the same year, each of FWMA firms has had a profound impact on their global value chain and the world economy.

Data collection process
The data collection process comprised three stages. First, the data from all of the FWMA US firms and Japanese firms that were released in at least one of the years between 2007 and 2012 from the Fortune website were captured. The FWMA surveys were conducted in the previous year of the releasing year, i.e. from 2006 to 2011. There were initially 586 US firms and 60 Japanese firms; each firm was provided an ISIN code or a Bloomberg ticker. Second, annual data on environmental disclosure score (E ), social disclosure score (S ) and the combined environmental-social-governance (ESG) disclosure score, the percentage of independent directors, return on equity, debt-to-equity ratio, sales growth, turnover, total assets, number of employees and industry sectors from 2006 to 2011 were collected automatically from Bloomberg with the Bloomberg template spread sheet. Third, the list of the above-mentioned companies was narrowed down to the firms that meet the criterion of being an active public company as of July 2012.  Table III.

Model
Following Jo and Harjoto's (2012) finding of a causal effect of governance on CSR and Chan et al. (2014) suggestion of a link between CG quality and CSR disclosure on company annual report, I tested the hypothesis on the impact of independent directors on CSR disclosure. A multivariate linear regression model was estimated using the OLS method. To observe this hypothesised impact over time, following Ntim and Soobaroyen (2013), fixed effect estimation was applied to control for unobserved heterogeneity and time-invariant firm-specific effects (Wooldridge, 2002).
Drawn on Bear et al.'s (2010) work using lagged data of the explanatory variables, I used one-year lagged data for the independent variables and control variables under the assumption that independent directors must be in their roles for some time to exert their influence on CSR information disclosed by the executives. In addition, I control the lagged dependent variable, since the disclosure level demonstrates its heritage manner. The proposed empirical model is characterised by: where, the dependent variable that reflects CSR disclosure is measured by three alternative proxies, ESG, E and S, used in separate regressions. ESG was calculated on the amount of ESG information that a company disclosed while the two individual components of ESG, i.e. E and S, were calculated on the basis of amount of the information on E and S, respectively. These scores were measured by Proprietary Bloomberg ESG group based on the extent of company disclosure of environmental, social and governance data. The scores were also tailored to different industries; in this way, each company was evaluated in terms of the data relevant to its industry sector. Companies that are not covered by the Proprietary Bloomberg ESG group and companies that do not disclose anything will have no score. The scores range from 0.1 for companies that disclosed a minimum amount of data to 100 for those that disclosed every data point. Each data point is weighted in terms of importance, with environmental data carrying a greater weight than other disclosures in ESG, greenhouse gas emission carrying greater weight than other environmental disclosures in E, and workforce data carrying greater weight than other social disclosures in S.
The key explanatory variable, indirector, is defined by the percentage of independent directors on board membership following Rashid's (2015) study that employed the percentage of outside directors as the proxy for board independence.
For the control variables, following Ntim and Soobaroyen (2013), return on equity (roe), debt-to-equity ratio (leverage), sales growth (salesgrowth) were fixed. Additionally, the extent of a firm's CSR disclosure might be subject to the firm's size, since several aspects of firm size may influence governance in a way that tempers the board's ability to effect change (Dalton et al., 1999). Previous studies employed turnover (Prior et al., 2008;Ammann et al., 2011), total assets (Frye et al., 2006;Lo and Sheu, 2007) or number of employees (Glavas and Piderit, 2009) to quantify firm size. In this study, the control variables for firm size are turnover (sales), total assets (assets) and number of employees (employee). Further, industry effect and year effect were controlled, since voluntary disclosure principles and practices likely vary from industry to industry (Campbell et al., 2006) and from year to year. industry is the dummy variable for each of the industries; and year is the dummy variable for each of the six years from 2006 to 2011. Table I demonstrates the two similarities in the attributes of the data of both countries. First, the standard deviations and the gaps between the minimum and maximum values of ESG, E and S are substantially large, which reveals that the amount of social and environmental information embedded in nonfinancial disclosure varies noticeably across the firms. Second, the mean values of E are considerably higher than those of S, suggesting that the amount of environmental disclosure is higher than the amount of social disclosure in both countries.

Descriptive statistics
There are clearly two main differences in the trends of the ESG, E and S data for each country in Table I. First, the US data on ESG, E and S are particularly extreme. In other words, there are larger standard deviations and wider gaps between the minimum and maximum values of EGS, E and S of the American firms compared to those of the Japanese firms. A considerable number of American companies reported only a minimum amount of E and S data, while a small number of the other US companies reported substantially on E and S (the data are available upon request). Second, the means and medians of ESG, E and S in the Japan panel are all higher than those in the US panel. Moreover, there is an increase in E and S in the Japanese data while those in the American counterparts fell during 2007-2009 when the global financial crisis was at its peak (see Figures 1 and 2).
Regarding the proportion of independent directors on board, there is an opposing tendency between Japan firms and the US firms. Quite a few Japanese firm-year observations report 0 per cent independent directors; in contrast, there are a considerable number of American firms reporting upto 100 per cent of independent board members (see Table I). For all firm-year observations, the overall standard deviation in Japan data is larger than that in the US data (17.33 vice versa 11.21). On the contrary, the overall mean of Japanese firms is remarkably smaller than that of US firms, 17.25 per cent of the former compared to 81.82 per cent of the latter, and so as the overall medians (13.81 per cent compared to 84.62 per cent, respectively). Likewise, in each of the observed years, the annual means and medians for the Japanese firms are all substantially smaller than those of the US counterparts, as shown in Figure 3. Table II displays the coefficients of bivariate correlation between each pair of the variables for each data panel to test for multicollinearity. The correlation coefficient between each pair of the independent variable and the dependent variable ESG (E ) (S ) is less than 0.31, which is considered small.

Regression results
The result of Breusch and Pagan's (1980) Lagrangian multiplier test indicates that there is a panel effect in both the Japan and the US data. The F-test results (p o0.05) suggest that fixed effects estimation could be used to estimate the parameters of the variables when Japanese data and US data were employed alternatively. Particularly, Hausman's (1978) specification test results suggest that fixed effects estimation is preferred for Japanese data ( po 0.05) rather than the random effects estimation. As seen in Table III, the models 1a, 2a and 3a demonstrate the fixed effects estimation results for the Japanese panel, while the models 1b, 2b and 3b for the US panel.

Result interpretations
The regression results can be interpreted as two main points. First, only the regression run on Japanese data shows that the percentage of independent directors in Japanese firms significantly and positively affects the level of ESG, E and S (Model 1a: β ¼ 0.12, p o0.001; Model 2a: β ¼ 0.14, p o0.001; Model 3a: β ¼ 0.17, p o0.01, respectively). This is interpreted   The preference of the panel estimation method rather than a simple OLS method is confirmed by performing Breusch and Pagan's Lagrangian multiplier test. The test results ( p o 0.05 in each of the model using ESG, E and S alternatively as the dependent variable) allow the rejection of the null hypothesis of variances across firms equal to zero (i.e. no panel effect hypothesis); while the R 2 between the studied firms and the overall R 2 are somewhat disappointing, this weakness has been minimised by the use of fixed-effect estimation where the unobserved year-invariant factors are fixed, leaving the year-variant factors to become more explanatory in the models. Thus, the reported magnitudes of R 2 within the studied firms are at the acceptable level in social science research. *po 0.05; **p o 0.01; ***p o 0.001 Table III. Results of fixed effects regressions on the level of CSR disclosure The descriptive statistics, correlation coefficients and regression outputs reveal the opposing trends between two countries. The effect of independent directors on ESG (E ) (S ) of the Japanese firms is significant, as illustrated by the fixed effects regression outputs despite the insignificant bivariate correlations in the Japanese data. In contrast, although at first the bivariate correlations between independent directors and ESG (E ) (S ) in the US firms is significant, the underlying effect shown in the regression results turns out insignificant on the US data. This evidence is of greater importance, since the means and medians of the proportion of independent directors on the board of Japanese firms are less than 20 per cent, while those of the US firms are higher than 80 per cent. This contrast points to the need for further investigation of the universal validity of stakeholder theory in shaping CG configurations and CSR. Second, each of 2007Second, each of , 2008Second, each of , 2009Second, each of and 2010 generates an interesting insight although the year dummy is not the key explanatory factor in the proposed model. Each of the years from 2007 to 2010 turns out to be more significantly, strongly and positively related to CSR disclosure on the US data than it does on the Japanese data. This could be inferred that American independent directors might have become interested in stakeholder engagement in the wake of the global financial crisis; interestingly, the year factor is more explanatory for the studied impact in the US ( p o0.01) than it is in Japan.
Overall, the hypothesis was confirmed on the Japanese data but no statistical evidence found on the American part. While supporting the stakeholder theory in the Japanese firms, the hypothesis testing results challenge this theory by questioning whether independent directors in the US firms adopted the stakeholder approach in exercising their functions in relation to CSR. The empirical evidence shapes to the two ways of interpretation.
On the one hand, the unsupported hypothesis tested on the US data highlights the need for the stakeholder approach in the board rooms of the US firms with regard to their social and environmental footprints. This point of discussion is aligned with Rose's (2007) findings that the US Fortune firm directors sometimes made decisions that emphasise the legal defensibility of profit maximisation at the expense of long-term social responsibility and personal ethics. If being accountable to external stakeholders can partially build firm reputation, the results of this study challenge Musteen et al.'s (2010) findings that US Fortune firms with a greater proportion of outside directors attract a better reputation than those with a higher proportion of internal directors.
On the other hand, the empirical findings possibly indicate that the board of directors and the top management in Japanese global firms might have adopted the stakeholder approach into their business in the way that they disclosed the ESG (E ) (S ) information to their external stakeholders. The significant fixed effects regression results on the Japanese data confirm Bansal and Roth's (2000) findings that Japanese firms focussed their attention on their priorities articulated by the social norms, and they often operated with similar standards in a socially cohesive environment. This confirmation is further supported by the year factor in the regression outputs using Japanese data, suggesting that the global financial crisis spanning 2006-2011 might not be the main factor that triggered the Japanese FWMA companies to engage with the stakeholders.

Robustness analysis
The author conducted a number of analyses on the robustness of the regression outputs. First, given the inconstant variants of the variables, Table III only reports the regression results after the robust check was done. A visual inspection of the correlation coefficients would indicate concerns for multicollinearity; therefore, I used the test for variance inflation factor (VIF) to measure how much the variance of the coefficients is inflated by multicollinearity. For Japanese data, the mean VIF of the data used in the regression on ESG is 6.98, that on E is 6.27 and that on S is 6.49. For the US data, the mean VIF of the data used in the regression on ESG is 1.62, that on E is 1.82 and that on S is 1.62. All of the mean VIFs are generally well below the rule-of-thumb value of 10, which demonstrates that the models do not seriously violate the assumption of no perfect multicollinearity.
Second, the independent and dependent variables of the Japanese data set show that the absolute values of skewness range from 0.01 to 1.29, and the absolute values of kurtosis from 2.73 to 5.44. The US data for the independent and dependent variables show that the absolute values of skewness range from 0.61 to 1.50, and the absolute values of kurtosis from 2.48 to 5.46. This implies that the assumption of normal distribution of the variables is not seriously violated.
Third, there are intrinsically unobserved elements in the dependent variable that cannot be estimated in the error term in statistical models. The independent variable is endogenous when there is a correlation with the error term (Wooldridge, 2013). Generally, a loop of causality between the independent and dependent variables of a model leads to endogeneity ( Jia and Skaperdas, 2012). To examine this problem, ESG (E ) (S ) was alternatively regressed against indirector and the control variables to ensure that the direction of causality is from indirector to ESG (E ) (S ), and not the reverse (Chen, 2014).
The next two steps are to investigate if indirector is endogenous with the error term, that is, to investigate the likely association between indirector and the residual of the models. In the first step, the residuals R1, R2 and R3 were obtained from the pooled OLS models using ESG, E and S as separate dependent variables. The correlation between indirector and each of the residuals was examined. The test results show that there is no significant correlation between indirector and each of the residuals using the data set ( p o0.05). The second step further investigated the association between indirector and each of the residuals via regressions alternatively using fixed effects and random effects estimations. The regression results display that there are no significant associations between indirector and each of the residuals ( po0.05).
Finally, I investigated the reversal consequence of higher ESG, E and S on having more independent directors in the board membership to ascertain whether the reversal impact caused by endogeneity is insignificant. I also controlled ROE, leverage, sales growth, sales, total assets, employee number, industry effects and year effects. The regression coefficient of ESG (E ) (S ), either using Japanese data or US data, is not statistically significant, suggesting that endogeneity is not a serious problem in this study.

Discussions and conclusion
Delving into the black box of the validity of stakeholder theory observed in the influence of independent directors on CSR disclosure in Japan and the USA, I would suggest the following two points that possibly explain my findings.
First, there might be risks of inflation of CSR information disclosed by the management who aim to please both stakeholders and shareholders when firms are facing difficulties in recession. The boundary between a true and greenwashing CSR disclosure is unclear due to information asymmetry (Akerlof, 1970). Therefore, more independent supervision on CSR disclosure by external directors are needed to maintain corporate legitimacy (Suchman, 1995), corporate reputation and risk reduction (Delgado-García et al., 2013).
Second, although the CSR concept has received much attention in theory and practice for over a half of a century in Anglo-Saxon society (Tricker, 2015), this study confirms Young and Thyil's (2014) findings that contextual factors are important to the incorporation of CSR within governance. Given the benefit of my proposed model in unpacking the effectiveness of the Japanese and American independent board members in their influence on disclosure of CSR activities, this study implies that it is impossible to apply a single CSR-driven governance model in different national governance bundles.
At the firm level, in line with the previous findings that organisational culture influences a firm's orientation towards responsible treatments to stakeholders (Galbreath, 2010) and that stakeholder's orientation of CG is positively associated with social and environmental disclosure (Mallin et al., 2013), my findings support the stakeholder theory in Japanese corporate context but challenge the theory in the US corporate context. Although previous studies confirm the positive and significant impact in the banking sector (Bear et al., 2010) and health sector ( Jizi et al., 2014) in the USA, this study takes a further step by challenging the cross-industry generalisation of this impact. My findings support Aguilera and Cazurra's (2009) statement that the outcomes are not as straightforward as one might think, and hence it is important to move the debate beyond the convergence/divergence dichotomy and to pay more attention to the dynamics of how firms apply certain aspects of the governance codes and not others.
At the country level, it is worth analysing why my results show a significant relationship between the proportion of independent board members and CSR disclosure in the Japanese context, but remain inconclusive in the USA one. According to García-Sánchez et al. (2015), the effect of having independent members on the board is conditioned by the shareholder orientation characteristics of the CG system in each country. While there had been no legal requirement for the inclusion of external members on boards in Japan until the governance reform in the early 2000s, it seems that when innovating governance practices, Japanese companies decoupled from the original context and customised their governance practices to their particular circumstances (Aguilera and Cazurra, 2009). Japan's CG system neither fully converges with, nor completely diverges from the Anglo-Saxon model (Yoshikawa et al., 2007). This study reveals that using independent directors is likely to be an effective mechanism for stakeholder management in the Japanese context. However, this mechanism is questioned in the US context, although CSR movements were initiated from western countries.
Country-level studies show that, increasingly, firms tend to adopt a higher percentage of code recommendations despite their voluntary nature (Aguilera and Cazurra, 2009). The USA was the first country to issue the code of good governance in 1978, while Japan issued their first code 22 years later, in 1997. Although the USA has had a well-developed national governance context creating a favourable environment for independent directors to do their tasks (Yoshikawa et al., 2014) for much longer than Japan has, voluntary integration of social and environmental sustainability into business remains a question for American firms as there are no statistical evidence found in this study. Therefore, the national context offers the setting for stakeholder theory to be realized in shaping independent directors' behaviours relating to stakeholder engagement.
CSR disclosure in this study is measured in terms of its level. One might suppose that independent directors could be influential on the annual change of CSR disclosure. There is an assumption that the US independent directors may actually play their roles by making change in the CSR disclosure while its levels are affected by a number of other significant efforts and/or governance models such as board model. Thus, this could be the avenue for future research. This paper has certain limitations, which may open up other significant avenues for further research. The first limitation is resulted from the objective paradigmatic approach of the research (Burrell and Morgan, 1979), since it attempts to quantify human attitudes and behaviours embedded in social constructs of CSR. This rests on the false assumption that managerial behaviour is observable and that observers have the time and ability to watch all human behaviours (Eisenhardt, 1989). Using either the environmental disclosure score or the social disclosure score, or the combined ESG disclosure score as the proxies for CSR disclosure is a necessary condition but not a sufficient one. The second limitation is that the firms which did not meet the selection criteria were not included in the data set, which hinders possibility for the generalisation of the findings to small-and medium-sized firms in the USA and Japan. Likewise, this paper neither looks into the management nor controls for the differences between institutional investors (Aguilera et al., 2006) in the USA and Japan. Therefore, these limitations turn to be the directions for future research.