Varieties of Capitalism and Environmental Sustainability: Institutional Explanations for differences in Firms’ Corporate Environmental Responsibility Reporting across 21 OECD Economies
Prepared for presentation at the APSA 2011 Annual Meeting, Seattle, Washington, September 1-4 Holger Meyer Ph.D. Candidate Department of International Affairs University of Georgia
[email protected]
Working draft - Not for citation without the author's permission Comments welcome!
Abstract: What factors explain cross-national differences in multinational corporations‘ (MNCs) environmental responsibility efforts? The logic of economic globalization suggests that multinational corporations operating in countries of comparable economic development and socio-cultural values orientation would pursue similar corporate environmental responsibility (CER) activities and reporting strategies in all of these markets. In reality, however, MNCs‘ efforts to reduce and report on their ecological footprint exhibit remarkable qualitative and quantitative cross-national differences. Rooted in the varieties of capitalism literature, this paper argues that cross-national institutional differences are a central and hitherto neglected factor in explaining the relationship between popular demands for environmental sustainability and corporate responses to these demands. This argument is tested by comparing quality and quantity of CER information reported by 20 Fortune Global 500 companies on their national websites in 21 OECD countries. Pattern recognition and classification methods are employed in order to construct a new measure of MNCs‘ reported CER activities. The regression of this measure of CER on Pryor‘s (2005) categorization of economic systems provides strong evidence that firms disclose more and better information on their corporate environmental responsibility efforts in less coordinated market economies than in more coordinated market economies. These findings challenge the common perception of a trend towards more homogeneous global corporate responsibility efforts. For now, MNCs‘ corporate environmental responsibility strategies remain embedded in national institutional frameworks.
Introduction Environmental degradation has become one of the central socio-political issues of the beginning 21st century. Growing concern for the environment across almost all countries (Esty and Porter 2001) has spurred a global debate on how best to achieve sustainability. In recent years, this discourse has become dominated by a fundamental dispute about the compatibility of capitalism and environmental protection (Inglehart 1995; 2003; Dalton, Recchia, and Rohrschneider 2003). On the one hand, supporters of economic models that emphasize governmental regulation and control accentuate shortcomings of market mechanisms with regard to managing public and common goods. They advocate strong regulatory regimes to preserve the environment (Zahrnt and Zahrnt 2011). On the other hand, proponents of free market capitalism point out that customer demand for sustainably produced goods and services will generate sufficient momentum to bring about environmentally sustainable economic systems. They caution that overregulation is detrimental to this process (Vorholz 2011). One of the most salient points of contention in the debate about how best to achieve ―green capitalism‖ (Harris, 2010) is the growing concentration of economic, financial, and not lastly political power in the hands of multinational corporations (MNCs) (Bracken 2004; Bock and Fuccillo 1975, 51; Scruggs 1999)). The growing transnational awareness of this development triggers societal demands for greater corporate responsibility and accountability (Cutler 2006; Utting 2002; Winston 2002; Levy and Newell 2005). In the words of Margolis and Walsh, ―the sheer magnitude of problems […] inspires a turn toward all available sources of aid, most notably corporations. Especially when those problems are juxtaposed to the wealth-creation capabilities of firms – or to the ills that firms may have helped to create – firms become an understandable target of appeals‖ (Margolis and Walsh 2003, 270). Even though comparative studies of corporate social responsibility (CSR) in general and corporate environmental responsibility (CER) in particular are relatively rare, a number of works have emerged in recent years (Albareda et al. 2006; Brammer and Pavelin 2005; Maignan and Ralston 2002; Williams and Aguilera 2008). The majority of these analyses of MNCs‘ responses to growing global demands for more corporate responsibility assert that the continuing globalization (or even ―Americanization‖ (Djelic 1998)) of management concepts, ideologies, and technologies results in a gradual harmonization of corporate response strategies throughout the world (Guler, Guillen, and Macpherson 2002). Doh and Guay, for example, argue that companies responding to the demands of the increasingly conscience-focused transnational marketplaces of the 21st century deemphasize national contexts (Doh and Guay 2006). However, a few influential studies have emerged that challenge the validity of the global convergence thesis. Emphasizing the institutional embeddedness of firms‘ CSR and CER efforts, they expose significant cross-national variation in multinational corporations‘ response and reporting strategies (e.g. Tsalikis and Seaton 2006). Surprisingly, several of these studies provide counterintuitive and contradictory results. One cross-regional study found, for example, that European and Japanese companies are far more advanced with respect to managing environmental impacts than their North American or Asian (ex-Japan) counterparts‖ (EIRIS 2007, 3). These findings conflict with the prevailing assumption that companies headquartered in the US - the birthplace of the concept of corporate responsibility - would outperform competitors with regard to responsibility efforts. Even more surprisingly, there appears to be a mismatch between the extent of firms‘ efforts and strategic necessity. For instance, Tsalikis and Seaton find that German consumers are among the most pessimistic with regard to the future ethical behavior of businesses, while customers in the UK are among the most optimistic (Tsalikis and Seaton 2007). A number of studies find, however, that responsible behavior as part of companies‘ business strategy is more developed in the UK than it is in Germany (e.g. Habisch et al. 2005). 2
Aside from these notable exceptions, research in the environmental realm continues to rely heavily on anecdotal evidence and case studies. Supporting this observation, Esty and Porter criticize that ―there are precious little systemic data on which to base environmental judgments at both the public policy and corporate levels. This may explain why environmental fields remain mired in deep controversies over the best path forward, with debate often dominated by emotional claims and heated rhetoric‖ (Esty and Porter 2001, 78). Intended to ameliorate this significant shortcoming, this paper answer the following research question: What factors explain the discrepancies in Multinational Corporations’ environmental responsibility strategies across countries of similar socio-economic development? The analysis starts with a critical evaluation of the most important alternative explanations for why corporations invest in corporate environmental responsibility. Subsequently, a novel explanatory framework is introduced that incorporates variance in national institutional frameworks - a central and hitherto neglected predictor variable for MNCs‘ CER efforts. This framework explains why MNCs operating in more coordinated market economies (emphasizing the use of formal institutions to regulate the market and coordinate interactions between firms and other actors) are less likely to invest in CER activities than MNCs operating in more liberal market economies (primarily relying on market mechanisms to solve problems of coordination). Next a new, statistically derived, systematic method to measure CER is developed. This measure is utilized to test the effects of different institutional frameworks on MNCs‘ CER efforts in a comparative study of CER information publicly disclosed by 20 Fortune Global 500 (FG500) companies operating in 21 OECD economies. Finally, substantial findings of the analysis as well as shortcomings and potential issues are discussed and further research is suggested.
The Origins of Corporate Environmental Responsibility Corporate Environmental Responsibility (CER), like the umbrella terms Corporate Responsibility (CR) and Corporate Social Responsibility (CSR), remains an essentially contested phenomenon (Moon, Crane, and Matten 2005). Three decades ago, Abbott and Monsen identified two of the most frequently cited reasons for the prevailing skepticism: First, detailed quantitative information, consistently measured and reported across a large number of firms continues to be largely unavailable. Second, the field continues to struggle over the establishment of comprehensive methodologies to measure the full impact of known corporate activities on society at large. (Abbott and Monsen 1979): Despite continuing substantial scholarly criticism of the concept itself (Oosterhout and Heugens 2008), an increasing number of works on corporate environmental responsibility have been published in recent years. These works attempt to conceptualize, measure, and explain motivations behind corporations‘ responses to calls for taking on greater environmental responsibility, show transparency, and demonstrate accountability (Levy and Newell 2005; Levy and Newell 2006; Kolk and van Tulder 2005; Cashore, Auld, and Newsom 2004; Cutler, Haufler, and Porter 1999).
Economic Explanations The most frequently cited explanations for why firms engage in CER are economic in nature. These accounts are informed by a variety of different assumptions and observations. While some scholars advance the argument that responsible behavior can lower the costs of complying with present and future environmental regulations (Hart 1995; Dechant and Altman 1994), others highlight that it can enhance firm efficiency, and drive down operating costs by reducing costs of energy, materials, and waste disposal (Russo and Fouts 1997). In addition, CER efforts can create a competitive advantage through the development of appealing ―eco-friendly‖ products 3
(Shrivastava 1995). Moreover, they can help avoiding costs of negative stakeholder reactions (Blacconiere and Patten 1994), for example by improving a firm‘s image and thus enhancing stakeholder loyalty (Berman et al. 1999). Not surprisingly, management texts conventionally assert that responsiveness to the demands of a broad range of stakeholders is in a firm‘s best long-term financial interest (Post, Lawrence, and Weber 2002). Despite plenty of anecdotal evidence supporting the existence of a positive link between environmental responsibility and financial performance, economic arguments for firms‘ CER efforts outside of niche markets remain highly disputed. More rigorous empirical studies find a weak or insignificant relationship between various measures of corporate responsibility and financial performance (Griffen and Mahon 1997; Guerard 1997; Waddock and Graves 2000). Vogel (2005) argues that it is difficult to realize monetary value derived from ‗supplying‘ responsibility to stakeholders other than consumers. Examining discrepancies between the public‘s perception of selected companies and their actual behavior, another recent work finds no correlation between consumer perception of corporate environmental performance and verifiable data (Aldhous and McKenna 2010). An earlier comprehensive review of 167 studies on the relationship between corporate responsible behavior and financial performance, conducted over a 35 year period concludes that while doing good doesn‘t appear to destroy shareholder value, there is only a minor correlation between responsible corporate behavior and good financial results (Margolis and Walsh 2001). In a follow up article, Margolis and Elfenbein postulate that ―the minor correlation that does exist could well be explained by deep pockets—a history of strong financial performance may simply give a company the wherewithal to contribute to society.‖ (Margolis and Elfenbein 2008, 20). Succinctly summarizing the ambiguities of the relationship between corporate responsibility and financial performance, a former executive of a large oil company remarked at a U.N.-sponsored corporate responsibility workshop that ―[if] the ‗win-win‘ argument were so compelling, then we wouldn‘t be sitting around this table‖ (Utting 2000). In short, the financial benefits of CER (and consequently arguments explaining CER as a purely business-driven phenomenon) remain controversial and have so far hardly been tested in large-n cross-national, cross-company comparative studies.
Organizational and Stakeholder Theory Explanations A number of organization theorists have tried to conceptualize the firm in a way that incorporates shareholders‘ interest in wealth creation as well as non-financial functions. Advancing theoretical models of the firm that rivals the contractarian perception, they develop arguments for the triple bottom line of ―people, planet, and profit‖ (Henriques and Richardson 2004, 186). Accordingly, stakeholder theory predicts that firms demonstrate responsibility to achieve legitimacy among stakeholders and the license to operate. The more concerned the relevant stakeholders are about the environment, the more extensive CER efforts can be expected from a company. Improved relationships with stakeholders create valuable intangible resources, which may in turn become sources of competitive advantages. Patten, for example, uses this theory to explain the extent of disclosures in the environmental reports of oil companies following the Exxon Valdez oil spill (Patten 1992). Several related studies attempt to quantify information needs of stakeholder groups such as investors, environmental organizations, governments and the press (Azzone et al. 1997; Rikhardsson et al. 2002). Their findings suggest that stakeholders have considerable interest in social and environmental performance information. There seems to be an increased motivation to report social and environmental information in an accountability context, and more companies are producing either obligatory or voluntary social or environmental reports (Rikhardsson et al. 2002). 4
Institutional Explanations In light of the unresolved academic debates concerning causes and consequences of CER efforts and the conflicting empirical evidence generated, political-institutional explanations promise to provide an alternative to accounts relying exclusively on competitive market pressures (Scott 1987). In his influential work, Institutions, institutional change, and economic performance, North defines formal and informal institutions as ―the rules of the game in a society‖ (North 1990). In a later work, he adds that ―[i]f institutions are the rules of the game, organizations and their entrepreneurs are the players. Organizations are made up of groups of individuals bound together by some common purpose to achieve certain objectives. Organizations include political bodies (e.g., political parties, the Senate, a city council, regulatory bodies), economic bodies (e.g., firms, trade unions, family farms, and cooperatives), social bodies (e.g., churches, clubs, and athletic associations) and educational bodies (e.g., schools, universities, vocational centers)‖ North (1998, p. 81). Institutions endure because they become ‗taken-for-granted‘ through repeated use and interaction or ‗legitimate‘ through the endorsement of some authoritative or powerful individual or organization. Institutions are not static, but their evolution generally processes in an incremental manner. This allows for a relatively clear-cut distinction between different national systems according to the characteristics of their respective long-lasting institutional frameworks (Clemens and Cook 1999, 445). North‘s argument that variation in longstanding, historically entrenched institutions must be incorporated into neo-classical theory, because it facilitates the creation of different organizations and strategies by the actors in each system, lends itself to extrapolation to the study of MNCs‘ corporate responsibility efforts. One of the more recent institutionalist theoretical developments is the ―varieties of capitalism‖ literature. Spearheaded by Hall and Soskice, it focuses on how state, market and civil society relations are organized differently across capitalist systems and how divergent capitalist models impact business strategy and behavior (Hall and Soskice 2001; Amable 2006; Crouch 2005; Whitley 1998). A number of researchers in this tradition focus on dissimilar economic developments in countries with different institutional frameworks. Pauly and Reich find that ―durable national institutions and distinctive ideological traditions still seem to shape and channel crucial corporate decisions‖ (Pauly and Reich 1997, 1). Aguilera and Jackson add that stakeholder identities and interests vary cross-nationally and that therefore some of the assumptions of agency-oriented analysis are too simplistic (Aguilera and Jackson 2003). According to Gjolberg, ―the global features of corporate responsibility efforts might lead to the assumption that national dynamics are secondary or even irrelevant. However, while CSR might be of a global nature, recent research suggests that it is applied differently across different social, economic, cultural, legal and political contexts‖ (Gjølberg 2009, 10). Despite these insights, very few scholars explicitly root their analysis of corporate responsibility patterns in the established literature on political systems. Even those scholarly works that do so (e.g. Habisch et al. 2005; Matten and Moon 2008; Midttun, Gautesen, and Gjølberg 2006) do not provide large-n statistical analysis of the assumed causal mechanisms at work. As a case in point, Matten and Moon make a powerful theoretical argument that companies from liberal, laissez-faire economies choose more explicit activities than those operating in coordinated market economies (Matten and Moon 2004). However, their study remains limited to a small number of countries, does not address CER, and provides little empirical evidence. In short: There is substantial evidence that institutional frameworks impact the formation of firms‘ national corporate responsibility strategies. However, we are far from a scholarly consensus on the extent of this impact and on how to classify institutional frameworks.
5
Theory The theoretical argument advanced in this paper follows the institutionalist school of thought and is rooted in the varieties of capitalism approach (Hall & Soskice, 2001): Cross-national differences in MNCs‘ CER strategies can to a significant part be explained by analyzing dissimilarities in historically grown, durable institutional frameworks that shape ―national business systems‖ (Whitley, 1997). The proposed framework explains why we should expect higher corporate CER efforts in less coordinated economies than in more coordinated economies. In doing so, it opposes arguments interpreting contemporary CER efforts as a revival of a socially embedded economy, where one would expect companies operating in traditionally most embedded welfare states with old neo-corporative relations and coordinated market economies to be the strongest performers (Scruggs 1999). At the same time it rejects explanations of corporate responsibility efforts as purely business-driven and detached from political initiative (Matten and Moon 2008) even though it recognizes a significant degree of MNC agency.
Level of Coordination of Market Economies In Varieties of Capitalism, Hall and Soskice argue that the capitalist economy does not assume a single, universal form but varies across states (Hall and Soskice 2001). According to their view, market economies can be situated on a scale with the polar opposites Liberal Market Economies (LME) and Coordinated Market Economies (CME). Even though the framework has been criticized on various grounds (e.g. privileging institutional continuity over discontinuity, internal coherence and equilibrium over internal contradiction and crisis, coordination and mutual accommodation over conflict and contention, policy over politics, and business over the state) it remains the ―state of the art of institutional analysis‖(Howell 2003). The two ideal types - LME and CME - are distinguished by their ways of solving problems of ‗coordination‘ between interest groups. In more liberal market economies coordination occurs primarily through market mechanisms. Industrial compliance with a minimalist set of basic market rules provides a basis for otherwise free pursuit of industrial interests with a focus on efficiency, competitiveness and profitability (Kikeri and Nellis 2004). In contrast, more coordinated market economies tend to rely much more on formal institutions in governing the economy and regulating firm relations with stakeholders. They are characterized by an interest group system in which groups are organized into national, specialized, hierarchical and monopolistic peak organizations. These interest groups are incorporated into the process of policy formation and implementation through collective bargaining processes. Operating in these dissimilar political environments, firms ―develop corporate strategies to take advantage of the institutional support available in any economy for particular modes of coordination, deriving from this a new perspective on issues in strategic management‖ (Hall and Soskice 2001). In short: Different societies employ different mechanisms to hold companies accountable and grant stakeholders opportunities to influence the accountability process. Consequently, it can be assumed that firms‘ decision to invest in CER activities is influenced by divergences in political institutions. In more coordinated economies institutions help a large number of actors to decide upon and harmonize environmental protection strategies. In more liberal market economies these institutions are less developed and firms have more liberty in designing individual CER strategies. The expected utility of implementing individual CER strategies is consequently higher in less coordinated market economies than in more coordinated ones.
6
Citizen Demands vs. Consumer Demands Having theorized that companies adapt to institutional incentives, it is important to explain how different institutions convert popular demand for environmental protection into motivations for firms to provide CER. For this, it is useful to split popular demand for environmental sustainability into two distinctive categories: consumer demand and citizen demand. Consumer Demand encompasses demands that concern the environmental sustainability of specific goods and services (cf. Vogel 2005). They are primarily directed at firms whom the people in their function as consumers perceive as being responsible for certain aspects of environmental protection. By choosing products based on the size of their environmental footprint consumers can change the incentive for companies to invest in CER efforts. Firms address these demands in order to secure or expand their customer base. A function of variables such as economic prosperity, socio-cultural values orientation, education and individual preferences consumer demand develops relatively independently from the level of coordination of a particular market economy (Manget, Roche, and Münnich 2009). Consequently, the level of coordination in an economy is expected to have no significant influence on consumer demand. Citizen Demand on the other hand, encompasses demands that reflect the population‘s opinion about individual business’ general responsibility in ensuring environmental sustainability vis-à-vis government’s and society’s responsibility. Comparatively higher public support for coordination among interest groups is expected to correlate with less demand for individual firms‘ CER efforts as sustainability is to be achieved through coordinated action. Consequently, more coordinated, less pluralist societies in which firms are more tightly embedded in and regulated by institutional and legal frameworks leave smaller shares of responsibility issues to the discretion of individual companies. Companies operating in these societies have less incentive to develop, implement, and communicate individual CER efforts as these activities are likely to yield smaller benefits (in terms of reputation and prevention of regulation) than in less coordinated economies. The lack of authority of peak associations in less coordinated economies compared to more coordinated economies exemplifies this distinction. In the latter, these associations ―have the authority and inclination to ensure that there is close monitoring and general compliance with environmental laws, lest some firms cheat to the detriment of others under their authority. In addition, those actors comprising the associations have a greater ability and incentive to pursue common solutions to industry pollution problems, thus diffusing ‗best practices‘ more readily throughout the economy‖ (Scruggs 2001, 687). In short, this theory (cf. Figure 1) explains crossnational differences in MNCs CER activities by differentiating between consumer demand for environmental sustainability (having similar effects on similarly developed economies, notwithstanding their level of coordination) and citizen demands (emphasizing state responsibility in more coordinated economies and firm responsibility in less coordinated ones).
Hypotheses H1 (Central Hypothesis): The level of multinational corporations‘ reported environmental responsibility activities depends on the level of coordination within a national economy: The less coordinated the economy the more substantial the CER efforts of individual companies. H2 (Auxiliary Hypothesis): Consumer and citizen demand for firms‘ CER efforts depends on the perceived size of their respective industries‘ environmental footprint. Consequently, demands for CER are more pronounced with regard to companies operating in industries with a comparatively larger environmental footprint than with regard to companies operating in industries with a comparatively smaller environmental footprint. 7
Customer Demand for CER
High
Citizen Demand for CER
High
Low
Low Level of Coordination
Level of Coordination
Aggregate Demand for CER High
Citizen Demand for CER Customer Demand for CER Low
High
Low
Overall Demand for CER
Level of Coordination
Figure 1: Theoretical Framework
Research Design The hypotheses about the determinants of firms‘ reported CER activities are tested with recoded data self-reported by companies on their nation-specific websites.
Dependent Variable: Firms’ CER activities reported on the national level Adopting core aspects from previous characterizations, this project defines the dependent variable CER as a firm‘s actions intended to further environmental goods beyond the direct financial interests of the company and that which is required by law. It is differentiated from the environmental responsibilities of the government by the fact that its precise manifestation and direction of responsibility lie at the discretion of the corporation. Empirically, CER consists of clearly articulated and communicated policies and practices that reflect business responsibility for the totality of their impact on people and the planet (Matten and Moon 2008, 405). As mentioned previously, substantial skepticism regarding the evaluation of the effectiveness of CER stems from the fact that detailed quantitative information consistently measured and reported across a large number of firms used to be largely unavailable (Abbott and Monsen 1979). Attempting to overcome this severe shortcoming, scholars employed a number of different methods that attempt to measure CER efforts by proxy. Reputation analysis (e.g. Apéria et al. 2004), for example, analyzes the public images of corporations through consumer surveys. The value of results derived through these methods remains questionable, as they are highly dependent on respondents‘ level of information and experience, age, and access to the mass media, the corporation's size, changes in corporate names and a variety of other variables. In recent years, however, changes in reporting strategies, requirements and the information revolution – notably the development of the internet – have helped to alleviate the problem of data scarcity. The disclosure of a significant amount of information on corporate 8
responsibility efforts on firm‘s websites as well as in comprehensive stand-alone sustainability reports has become the norm (Rikhardsson et al. 2002; Crane 2008). A structured analysis of such publicly available quantitative and qualitative information promises to allow for a more representative comparison of firms CER efforts and a more consistent way of measuring such activities than other methods of comparison. For instance, one pioneering study relying on such publically available information measured the social involvement of 82 food-processing firms by comparing the percentage of the space in annual reports pertaining to corporate responsibilities and activities (Bowman and Haire 1975). It goes without saying that using self-reported information to evaluate MNCs‘ CER efforts creates several challenges. The biggest issue is arguably the danger of mistaking ―greenwashing‖ - the deceptive use of green marketing to promote the perception that a company's operations are environmentally friendly - for genuine CER (Clegg 2009). As a case in point, in 2005, General Electric launched its multimillion-dollar "Ecomagination" campaign, highlighting activities in areas such as clean technology or renewable energy. An Earthsense survey run in May 2008 revealed that GE seemed to be reaping the benefits of Ecomagination, scoring first for consumer perception as an environmentally friendly company within its sector and seventh overall. At the same time, however, the company‘s emissions remained largely unchanged (Aldhous and McKenna 2010). Moreover, strong theoretical arguments can be made for why a company might under- or overreport its sustainability efforts. For instance, shareholders might consider a firm‘s substantial CER activities detrimental to the management‘s efforts to put highest priority on maximizing income which is to be distributed as dividends (Friedman 1970). At the same time, however, stockholders have a vested interest in the stability and legitimacy of the entrepreneurial institution and the autonomy of that institution from state control. CER engagement and reporting are useful tools in achieving these goals (Abbott and Monsen 1979). As managers have incentive to disclose good news and to withhold bad news (Verrecchia 1983) extensive environmental disclosures are posited to be a positive signal concerning the firm‘s exposure to environmental risk and an indicator for active environmental sustainability efforts. Public statements by top executives, for example, have become a frequently used indicator to rate the credibility of firms‘ responsibility efforts (Hull 1971). In short, it would be problematic at best to equate CER information, voluntarily disclosed by firms for communication purposes, with their actual aggregate impact on the environment. Despite these issues, voluntarily disclosed information does provide the researcher with verifiable, quantifiable and comparable data on corporate behavior. In fact, a majority of the multinational corporations under observation seek external assurance of the validity of their reported information. Validity and accuracy of subject matter, methodologies, and data presented in national, regional, and global CER reports are frequently assured by governmental agencies, environmental organizations or accountancy firms (e.g. Daimler 360: Facts on Sustainability 2011: assurance by PricewaterhouseCoopers; Toyota Australia 2010 Sustainability Report: assurance by Environmental Resources Management Australia Pty Limited (ERM)). Equally important, environmental NGOs and other stakeholder groups hold companies accountable for the accuracy of disclosed information by publicly denouncing firms that misrepresent their engagement (Sasser et al. 2006). Measuring CER: Even though raw data have become more available, they are not immediately usable for research purposes. It is necessary to formulate a set of categories and code the raw data in terms of the categories. Errors of two types are possible: (1) the formulation of categories that do not reflect all issues encompassed by the above definition of CER and (2) inaccuracy in coding the raw data in terms of the selected categories. These errors potentially affect validity and reliability of the resulting scale. 9
To minimize the risk of ignoring critical dimensions of CER, selecting as many CER indicators as possible is preferred to limiting the data collection to a few intuitively important indicators ex-ante. Consequently, quantitative information on 35 environmental sustainability indicators identified by the Global Reporting Initiative (GRI)1 was collected on each corporation‘s national website for all 21 countries.2 Divided into ten categories, these indicators measure (1) general commitment to CER as expressed in company guidelines and policies, (2) the availability of quantitative information on input materials, (3) energy consumption, (4) water consumption, (5) steps taken to protect biodiversity, (6) emissions, effluents, and waste, (7) initiatives to mitigate the environmental impact of products and services, (8) compliance with environmental regulations, (9) the environmental impact of transporting products and materials, and (10) a firms‘ total environmental protection expenditures. Even though some of these indicators overlap in content, all reflect crucial aspects of CER activities of the individual firms. The first column of Table 1 displays the ten categories, the second column contains the respective indicators, and the third column indicates the indicators‘ GRI classification number. Indicator Selection Pattern recognition and classification methods were used in order to reduce the number of indicators as much as possible, while retaining those capturing key differences among firms‘ reported CER efforts. In a first step, cluster detection algorithms were employed to identify patterns of CER performance. Initially, Support Vector Clustering (SVC) (Ben-Hur et al. 2001) a clustering method without any inherent explicit bias of either the number, or the shape of clusters - was used to project the data into 35 dimensions (the number of initial variables). In these dimensions it detects the planes that best split the points into separate clusters. SVC determined a total of 17 groups for the data at hand. However, the groups were quite inhomogeneous, with a small number of clusters encompassing the majority of the data. Therefore, the K-means clustering algorithm (MacQueen 1965) was used to calculate multidimensional distances among all observations based on the variables (35 indicators) provided. As K-means requires determining the number of clusters ex ante, the results from the previous SVC procedure were used to set the number of clusters to 15, 17, and 20 respectively. In a second step, decision trees were used to identify those indicators most influential in the creation of the recognized patterns. The purpose of this step is to eliminate indicators that only provide redundant or insignificant information. Decision trees are not only easily interpretable but they also arrange the variables into a clear hierarchy of progressive partitions, allowing the researcher to identify the most influential ones. However, as the method is sensitive to small changes in the training dataset, it constructs different trees from very similar datasets. In order to increase the robustness of the results, the algorithm was provided with three different training sets, each consisting of the original database and subjected to a K-means clustering set to detect 15, 17 and 20 clusters. (Figure 2 shows the decision tree output for 20 clusters). Only those indicators that showed up repeatedly were chosen for further analysis. 13 indicators showed up in at least two of the three decision trees, 8 indicators were present in all of them. For 1
The Global Reporting Initiative (GRI) is a registered not-for-profit organization that produces one of the most comprehensive and widely used standards for sustainability reporting. Its globally accepted sustainability reporting guidelines are applied by companies and organizations reporting on their environmental, social, and economic activities. The guidelines are intended to capture the quality of the revealed information, the information‘s reliability, and the firms expressed commitment to future sustainability efforts. The website of the organization can be found at: http://www.globalreporting.org/Home. All companies under observation for this paper have publically declared to adhere to these guidelines in their sustainability reporting efforts. 2 For the purpose of this paper, these indicators were treated as dichotomous measures. However, more detailed information (measured on an ordinal scale from 0-6) was also collected to be used in follow-up studies.
10
comparative purposes, two measures of CER were constructed. CER1 containing the 13 indicators (cf. Table 1, col. 4) was used as the dependent variable in Model 1. An alternatively specified dependent variable - CER2 - incorporating only those 8 indicators that showed up in all three decision trees (cf. Table 1, col. 5) was used as dependent variable in Model 2. To avoid biases resulting from misspecification in combining the different indicators, the dependent variable CER1 (CERII respectively) represents a composite measure of all 13 (8) indicators - treated as dichotomous, equally-weighted measures. The individual indicators take on the value ―1‖ when quantitative information is reported and ―0‖ otherwise. Consequently, the highest possible CER1 (CER2) score is 13(8) (performance reported for all indicators) and the lowest score is 0 (no performance reported for any indicator). Figure 3 illustrates the distribution of CER1 (CER2) indicators among the 420 units of analysis (20 companies in 21 countries). Table 1: CER Performance Indicators Category
General Commitment
Material
Energy
Water
Biodiversity
Emissions, Effluents, & Waste
Products & Services Compliance Transport Overall
3
Indicator3
Selected GRI Index
1. CER Report 2. Statement 3. External assurance 4. Subscription to environmental charters 5. Membership in env. Associations 6. Input material 7. Recycled input material 8. Direct energy consumption 9. Indirect energy consumption 10. Energy saved due to eff. improvements 11. Products/services energy efficiency 12. Reduction in ind. energy consumption 13. Water withdrawal by source 14. Water sources affected 15. Water recycled & reused 16. Property in/adjacent to protected areas 17. Impact on biodiversity in protected areas 18. Habitats protected/restored 19. Biodiversity management 20. IUCN red list species affected 21. Total greenhouse gas 22. Other indirect greenhouse gas 23. Reduction of greenhouse gas emissions 24. Ozone depleting substances 25. Significant air emissions 26. Water Discharge 27. Waste Disposal 28. Significant Spills 29. Transported Hazardous Waste 30. Habitats affected by water discharge 31. Mitigation of products‘ env. Impact 32. Reclaimed packaging material 33. Fines and sanctions for non-compliance 34. Environmental impact of transport
X 1.1 3.13 4.12 4.13 EN1 EN2 EN3 EN4 EN5 EN6 EN7 EN8 EN9 EN10 EN11 EN12 EN13 EN14 EN15 EN16 EN17 EN18 EN19 EN20 EN21 EN22 EN23 EN24 EN25 EN26 EN27 EN28 EN29
35. Total env. protection expenditures
EN30
Model 1
Model 2
(13 indicators)
(8 indicators)
Additional information on each indicator can be found at http://www.globalreporting.org/ReportingFramework/.
11
*for a close-up view, please see Appendix
Figure 2: Decision Tree output (20 clusters)* CER1
CER2
Figure 3: Distribution of reported corporate responsibility efforts among 420 observations
Unit of Analysis: 420 National Websites maintained by 20 FG500 companies For the purpose of this project, a national website is defined as (a) a company‘s primary official platform to present itself on the internet that (b) uses the internet country code top-level domain (ccTLD) of the relevant country (in cases of redirection to a country specific website located at a different top level domain (e.g. http://www.siemens.ie http://www.siemens.com/entry/ie/en/) the latter is considered the ―national website‖ for the purpose of this study). Companies: This project analyzes the CER efforts of the 20 largest FG 500 companies operating in all 21 countries under observation as of 20104. Companies that do not operate in all 21 countries and do not maintain country specific websites were removed from the analysis. If a MNC operates in several industries and/or sells a variety of different products or services, the analysis focuses on the primary industry and the primary product or service provided (in terms of revenue). Table 2 contains all companies whose national websites were analyzed for this study. Table 2: Companies (FG 500 rank) 1. Toyota Motor Corp. (5) 2. General Electric Company (13) 3. Volkswagen AG (16) 4. Allianz SE (20) 5. Ford Motor Company (23) 6. Hewlett-Packard Company (26) 7. Daimler AG (30) 8. Samsung Electronics (32) 9. General Motors Company (38) 10. Siemens AG (40) 4
11. American International Group, Inc. (41) 12. Nestlé S.A. (44) 13. Hitachi, Ltd. (45) 14. International Business Machines Corp. (48) 15. Honda Motor Company, Ltd. (51) 16. Nissan Motor Company, Ltd. (63) 17. LG Corp. (67) 18. Sony Corp. (69) 19. Hyundai Motor Company (78) 20. BASF SE (81)
Fortune Magazine‘s Global 500 ranking of the top 500 corporations worldwide as measured by revenue (2010).
12
Countries: Of the 34 members of the OECD, 21 countries were selected (cf. Table 3). A preliminary analysis of FG 500 companies‘ national websites revealed that a large majority of them either do not operate or do not maintain country specific websites in Iceland, Slovenia, and Luxembourg. Consequently, these three countries (population size ≤ 2 million citizens) were dropped from the analysis. In order to maintain a high degree of comparability, only countries with a comparatively high GDP were kept in the dataset. Slovakia, Hungary, Estonia, Poland, Turkey, Chile, and Mexico were dropped from the analysis due to their comparatively low GDP (PPP) per capita (
