The Payoff to Corporate Well-Doing

(1)

The Payoff to Corporate Well-Doing

The Effect of Corporate Boards on the Link between

Social and Financial Performance

(2)

(3)

THE PAYOFF TO CORPORATE WELL-DOING: THE

EFFECT OF CORPORATE BOARDS ON THE LINK

BETWEEN SOCIAL AND FINANCIAL

PERFORMANCE

Master Thesis

Gerwin van der Laan

Groningen, August 2005

Supervisors:

Prof. Dr. Hans van Ees

Prof. Dr. Arjen van Witteloostuijn

University of Groningen, Faculty of Economics

Master of Science in Economics and Business (Research Master)

(4)

(5)

PREFACE

In 167 C.E. Marcus Aurelius published his Meditaties in which his philosophical reflections are preceded by a list of people to whom he said to owe his accomplishments. Aurelius, a Roman emperor, thanks his friends, his family, his teachers and others for showing him the importance of certain virtues and for shaping his personality. Although writing down every single person who has had an influence on what you achieved seems a time-consuming endeavour, Aurelius’ list includes only sixteen individuals (and ‘the gods’). A Dutch author thus made the observation that learning in the early days probably was a well-organised effort.

To Marcus Aurelius, I owe the first words of this Master thesis. I would not have been aware of Aurelius’ work, however, if I would not have read the Dutch authors’ book, which was recommended to me by someone else. I am afraid that listing all the people to whom I am indebted is a cumbersome activity. I refrain from undertaking that activity here, although some people deserve a special word of thanks for their respective contributions.

Writing this thesis on the payoff to corporate well-doing is the closing part of a five-year journey through the Economics curriculum. From my second year until today, Hans van Ees has been my most important tour guide, who has introduced me to science and constantly challenged me. Under his supervision I discovered corporate governance. Arjen van Witteloostuijn came to the fore two years after Hans did and showed me the beauty of team research. Together they allowed me to practice virtually every stage of the research process in real-world settings. I am honoured that they have provided all these opportunities to me and were willing to supervise me in writing this thesis.

I am also indebted to Gregory Maassen from Erasmus University (Rotterdam) and Stephanie Sliwicki from Spencer Stuart for sharing their (board) data with me.

Finally, I would like to express my gratitude to my friends and family who have contributed in many small and larger ways to this thesis. Maybe, I will once publish my own meditations and devote a section to you.

Gerwin van der Laan

(6)

(7)

ABSTRACT

The relationship between corporate social performance and corporate financial performance is investigated using panel data on S&P 500 firms in the period 1997-2002. Furthermore, the presence of interaction effects of various board variables is tested for.

The results show that a reputation for positive social performance is not the mirror image of a reputation for negative social performance. The two types of variables therefore enter the models separately, a feature which previous studies have not accounted for.

A bad reputation for corporate social performance is convincingly shown to hurt financial performance, controlling for firm size, industry and solvency. An outstanding positive reputation for social performance does not affect financial performance. The reverse hypothesis – which holds that financial performance causes social performance – does not receive much support. Tests of non-linear effects show some support for a minimum for negative social performance beyond which profitability does not decrease further.

Interaction effects show that firms with larger boards derive higher profits from good CSP or diminish the negative financial effects of bad CSP. Also, the striking conclusion can be drawn that if board members meet often, the potential benefits of investing in CSP decrease.

(8)

CHAPTER 1: INTRODUCTION

As of the last quarter of the 20th century the Western world is said to have entered the post-industrial era. This period is characterised by a concentration of economic production in the service industries, increasing globalisation, more pressing ecological limits to economic growth, and a tendency of democratisation of both countries and economic exchange (Shrivastava, 1995). These four characteristics together have led to a change in the power balance between corporations and governments. Increasingly, people are aware of the influence large corporations have on individuals’ rights and duties and the extent to which governments can safeguard these rights (Matten and Crane, 2005). As a consequence, corporations are asked to integrate the effects of their economic decisions on the environment and people into their investment decisions: companies are held responsible for the advancement of societies. Because of this expectation the evaluation of a firm’s activities is necessarily a social issue, whether businessmen assume a social responsibility or not (Elbing, 1970), and the societal impact of business as well as the strategies firms employ to deal with social issues become a branch of business economics.

In the business world, the profit maximisation mantra seems to be more popular than ever before and societal effects have become externalities. However, numerous corporate scandals, which are considered a consequence of too much emphasis on short-term financial gains, have gone public. These have initiated a discussion in business schools on the education of future managers and the ethical perspective with which they enter the labour market (Ghoshal, 2005). Also, an explosion of the number of corporate governance codes, which are to regulate the decisions managers make, can be discerned after 1995 (Aguilera and Cuervo-Cazurra, 2004). This rebirth of the interest in ethical issues in corporate decision making increases the relevance of the academic research into corporate social performance. It is to the relationship between corporate social and financial performance that this study intends to make its greatest contribution.

1.1 Problem Statement

The main question in this research is twofold: what is the relationship between corporate social performance and corporate financial performance, and what are the effects of corporate boards on this relationship?

(12)

studies in which multiple types of social performance are simultaneously assessed is small. One aim of this research is to assess the relationship using a larger (panel) data set in which these problems are less salient. The second feature that makes this research relevant to science is the assessment of processes that affect the relationship between social and financial performance, thus gaining insight into the potential mechanisms which cause the association.

For society, and specifically governments, the relevance is to be found in the insight this study creates into the extent to which government intervention is required to attain social goals. Types of corporate social performance which are shown to be positively related to corporate financial performance need not be subsidised. The good as such provides the incentives to invest; the only role for government is to provide firms with the appropriate information on the profitability of achieving the specific social good. The social good is in this case a positive externality or the absence of a negative externality which arises automatically as a result of profit maximising behaviour. The results of this research might therefore lead to guidelines for the development of corporate law: only those social goods which do not lead to financial performance need to be part of government regulation for these are externalities that are not automatically undone. The insights into the effect of board variables on the relationship between social and financial performance can be an input for the development of corporate governance codes.

Lastly, the relevance of this study from a managerial perspective is straightforward. The study shows which types of social performance have a positive net present value. Furthermore, given the board composition, the managers can determine which social goods have the largest marginal benefits.

1.2 Theoretical Approach

Garriga and Melé (2004) distinguish among four classes of theories. In this research two of these will predominantly be used. In instrumental theories, corporate social responsiveness – the process which leads to corporate social performance (CSP) – is considered to be a variable which creates wealth. In this study the principle of corporate social responsibility is discussed to some extent, but since this issue belongs more to the area of philosophy than to economics or business, the main emphasis is on which types of social performance enhance corporate financial performance (CFP) and how this wealth creation takes place. In this thesis’ view, CSP is essentially an instrument for the enhancement of welfare. It follows immediately that ethical theories, one of the groups in Garriga and Melé’s (2004) overview, will not be dealt with in this research.

(13)

reasoning, but empirical estimation is cumbersome. This group, called integrative theories, claim that business should integrate social demands into its decision processes. It is obvious that for the types of social performance which are shown to enhance financial performance, such integration would indeed be desirable, but this is more a conclusion than a premise on which the theorising efforts are built. Donaldson (2003), among others, does call for the development of an integrative measure of firm performance including financial and social aspects but – as will be discussed – such a measure is not available and it is not very likely that it will become available.

In the derivation of the CSP-CFP linkages and the explanation of the CSP model, a variety of theories which are all in some way part of what is called the resource dependence theory are used. In this theory organisations are viewed as “being embedded in networks of interdependencies and social relationships (…). The need for resources (…) made organizations potentially dependent on the external sources of these resources” (Pfeffer and Salancik, 2003: xii). The analysis leading to the development of the hypothesis (# 1) that social performance enhances firm performance is justified by several arguments drawing on the notion of a competitive advantage.

FIGURE 1.1: Hypotheses in this Study

Board Variables - board size - number of outsiders - average age - number of meetings H3

Corporate Social Performance Corporate Financial Performance - 7 indicators of good CSP - accounting-based measures - 7 indicators of bad CSP H1 - equity market-based measures

(14)

Hypothesis 2, which claims that financial performance causes social performance, rests in the concept of slack resources or organisational slack (see, e.g., Cyert and March, 1963). Slack resources are defined as those resources that remain after the claims of all constituents of the firm have been met. No explicit claims on slack resources can be made, and managers can thus spend these on seemingly unprofitable activities in the realm of corporate social responsiveness (although that does not imply that the activities are unprofitable). In traditional economic models of the firm (see e.g. Chapter 3 of Milgrom and Roberts, 1992), slack resources are zero because competition drives down excess rents. However, imperfect competition typically results in non-zero organisational slack (Cyert and March, 1963).

The introduction of corporate boards into the CSP-CFP model cannot be complete without some treatment of agency theory, which is the main theoretical basis on which the large empirical body of research in this area has been built (Daily et al., 2003). The hypotheses on the moderation effects (Hypotheses 3a – 3d) are essentially grounded in resource dependency reasoning, though, and view the board as a device to link the organisation to its environment.

Figure 1.1 shows the relationships that will be tested in this thesis. It also shows – through the two dashed arrows – two relationships that are not part of this research. The first is the direct influence of board variables on CSP, the second is the direct influence of boards on CFP. The inclusion of main effects into the regression analyses that estimate the significance of the interaction effects does provide insights into these relationships, but a proper testing would also imply the inclusion of boards as instruments for corporate social performance if the relationship between CSP and CFP is estimated. This and the many research strategies which are appear when these two arrows are allowed for are left for future research. The same holds for feedback and other dynamic effects.

1.3 Organisation of the Thesis

In Chapter 2, the theory will be outlined. Chapters 3 and 4, respectively, introduce the data and methods and the findings of applying the methods to the data. Chapter 5 summarises this thesis. In Chapter 2, a first section deals with the renowned corporate social performance framework that has been developed in the past thirty years. In the second section hypotheses are formulated regarding the relationships between social and financial performance: both causal directions are subject of hypothesis development. Section 2.3 concludes Chapter 2 and deals with the moderating effect of board variables on the relationship between social and financial performance.

(15)

(16)

(17)

CHAPTER 2: THEORY

Every analysis of the relationship between business and society starts with an overview of arguments for and against the involvement of corporations in the reduction of human misery. This chapter starts with an overview of the arguments and then considers the renowned corporate social performance framework which was developed in the past thirty years (Section 2.1). In Section 2.2 the extensive research on the relationship between social and financial performance will be reviewed. The main theoretical contribution of the present research lies in the introduction of board variables to the aforementioned relationship. In Section 2.3 an introduction to board research and a first attempt to connect this literature to that on social performance is offered.

Margolis and Walsh (2003) summarise the economic arguments against business’ involvement in social issues under two headings: misappropriation and misallocation. “When companies engage in social initiatives, the first concern is that managers will misappropriate corporate resources by diverting them from their rightful claimants (…). Managers also misallocate resources by diverting those best used for one purpose to advance purposes for which those resources are poorly suited” (Margolis and Walsh, 2003: 272). They oppose misappropriation and misallocation to three duties: the duty to contribute to the conditions a firm has created, the duty to contribute to the conditions a firm benefits from and the duty of beneficence. Davis (1973) also lists many pros and cons of corporate socially responsible behaviour.

(18)

also be interpreted as imposing a tax on people’s wellbeing if the investment cannot be made by the people themselves. In the case of pollution, for example, the community may not be able to counter the discharge of toxic substances whereas the corporation could divert resources to the prevention of the emissions.

To these arguments, Davis (1973) adds that businessmen need not have the appropriate skills to tackle social problems, that spending attention to an assumed social responsibility weakens the attention paid to the core function of a business organisation, that if there is cross-national diversity in social demands socially responsive behaviour weakens the relative competitiveness of firms, that business has no formal line of accountability to the people and that business already has much social power without it being involved in social activities. These arguments are not all as convincing as those of Jensen and Friedman can be. For example, the lack of accountability and the governance of the social power of organisations are only valid arguments in certain institutional settings, which happen to be predominant but which are therefore not unalterable

Proponents of a socially responsive business world might use three duties to support their case (Margolis & Walsh, 2003). Firstly, business activities certainly have an impact upon society. As an employer, a firm affects the conditions under which its employees live; as a producer, the production process and the usage of resources affect the physical environment; and as a vendor, it affects the quality of the lives of its customers. Consequently, a firm might be held responsible for the harm it does to others, or – a term which is used in further sections of this thesis – the firm has a public responsibility. Secondly, a firm benefits from society as it is. Examples of this are a high educational level which yields highly productive workers and a good health system which guarantees short sickness leaves. Society might expect firms to contribute to these conditions from which it benefits just as individuals do. Thirdly, the duty of beneficence might be referenced to, “the duty to promote the well-being of others, in particular to provide aid to prevent or relieve suffering or dire conditions” (Margolis & Walsh, 2003: 292). The obvious problem with this third duty is the seemingly endless reach it has, a problem which might also be associated with the second duty.

(19)

somewhat peculiar arguments are called ‘sociocultural norms’ and ‘let business try’. The first means that it is only human to be willing to do what is socially desirable, whereas the second means that for some problems – where many other institutions have failed to reach a solution – business, as a last resort, might be able to find a cure. Davis (1973) also claims that business has the resources to tackle social issues. Friedman’s answer – that the abundance of resources should lead to extra dividend payouts which the stockholders can spend on social problems if they desire to do so – is more convincing than Davis’ argument. A final argument, which is logically sound, is that if stockholders hold a diversified portfolio, their interest coincides with that of society which makes business’ assumption of a social responsibility justifiable. This argument agrees with Keim’s (1978) exposition of the enlightened self-interest model in which corporations make some investments which have a negative present value for the firm but a positive NPV for investors who hold a diversified portfolio. The firm need not even be the cause of the social problem, if the firm considers itself capable of solving the problem, the enlightened self-interest criterion prescribes its involvement (Fitch, 1976).

2.1 Corporate Social Performance

Several authors in the corporate social performance field have worked on the development of the framework that is depicted in Figure 2.1. This figure condenses the seminal articles of Carroll (1979), who classified nine definitions of corporate social performance along three dimensions; Wartick and Cochran (1985), who defined social responsiveness and social issues management in terms of Carroll’s categorisation; Wood (1991a, 1991b) who acknowledged that principles are a multi-level phenomenon, opened the black box of the processes somewhat, and added the outcomes of corporate behaviour to the model; and Swanson (1995, 1999) who calls for an integration of principles and processes, and offers a model in terms of value-based decision processes and corporate culture.

(20)

2.1.1 Principles of Social Responsibility

The first block in Figure 2.1 concerns principles of corporate social responsibility. Carroll (1979) claims that a first step in understanding social performance is defining what responsibilities business has to society. He distinguishes among four classes. First and foremost, business has an economic function in society: it provides consumers with products and services they desire. Secondly, this economic responsibility is to be fulfilled within the legal framework in which every agent in a country operates. Thirdly, society expects business to act along ethical norms, which go beyond the verbatim text of laws. Lastly, a firm has discretionary or volitional responsibilities: the moral obligation to do more than is expected. In sum: “the social responsibility of business encompasses the economic, legal, ethical, and discretionary expectations that society has of organizations at a given point in time” (Carroll, 1979:500).

FIGURE 2.1:

Corporate Social Performance Framework Principles Processes Outcomes Institutional level Environmental Social impacts Legitimacy assessment

Organisational level Stakeholder Social Public Responsibility management programmes Managerial level Issues management Social policies Managerial

discretion

Source: based on Wood (1991a:694)

This classification of responsibilities has shown to be far from perfect. Friedman (1970) writes that “a corporation is an artificial person and in this sense may have artificial responsibilities, but ‘business’ as a whole cannot be said to have responsibilities, even in this vague sense.” Thus it is necessary to specify what the unit of analysis is which is said to have a responsibility. Also, Wood (1991a) notes that the classification by Carroll refers to domains of principles but not to the principles themselves. Principles are supposed to enact processes, but the classification of Carroll does not easily lend itself for linking principles to processes (Wartick & Cochran, 1985).

(21)

Although Friedman (1970) argues that ‘business’ as such is not an entity which can assume whatever type of obligations, the firm being a business organisation can. In 1973 Keith Davis wrote that “[t]hough the long run may require decades, or even centuries in some instances, history seems to confirm that society ultimately acts to reduce the power of those who have not used it responsibly” (Davis, 1973: 314). Thus at the institutional level, the social responsibility of business is to maintain its legitimacy as a provider of products and services (Wood, 1991b). At the organisational level, business has a public responsibility. By taking resources from the environment and transforming these resources into (intermediate) goods, the corporation changes its environment. The public responsibility of the firm implies that it counters its harmful effects on the environment in which it operates. This responsibility holds for both the primary and the secondary areas of a firm’s involvement, that is: the consequences which are intrinsic to the activities of the corporation (primary) and the impact these activities have (secondary), a terminology which is due to Preston and Post (in Wood, 1991a). For example, a fast-food restaurant holds responsibility for the decomposability of the packaging materials it uses (primary area of involvement) and for the long-term effects which the regular consumption of fast-food might have on consumers’ health condition (secondary area of involvement). Finally, at the managerial level, corporate executives can be assumed to have some leeway as to the choice of means with which corporate targets can be pursued (a discussion of the extent to which this discretion exists can be found in Section 2.3). Principles of corporate social responsibility at this level of analysis imply that managers should use their discretion in a socially responsible way (Wood, 1991b).

2.1.2 Processes of Social Responsiveness

If a firm and its constituents assume a social responsibility, the next step is to respond to demands for socially responsible behaviour. In Figure 2.1 this step is shown in the block in the centre of the figure. In Carroll’s (1979) model, the philosophy of responsiveness is presented as the third element, next to principles of responsibility and social issues management. Carroll (1979) draws a continuum from ‘do nothing’ to ‘do much’ on which he localises three earlier contributions to the field. Although the terminology differs, the three models basically distinguish between strategies of reacting to explicit demands, doing only what is required, accommodating the inclusion of social issues in the decision procedures of the corporation and scanning the environment for nascent demands and acting proactively (see also Zenisek, 1979).

(22)

criterion against which the multitude of social demands can be evaluated. Furthermore, Davis’ (1973) iron law of responsibility – the above-quoted line saying that society will ultimately take away power from those who abuse it – is ignored by social responsiveness theorists: the fact that organisations are responsive to social demands does not imply that they are responsible, since the concept of responsibility relates to outcomes, whereas responsiveness has to do with processes (Wartick & Cochran, 1985). For example, a firm that free-rides on an industry standard can still achieve responsible outcomes without acting socially responsive. Wartick and Cochran continue by stating that principles and processes are interrelated and define the tasks of a firm as “(a) to identify and analyze society’s changing expectations, (b) to determine an overall approach for being responsive to society’s changing demands, and (c) to implement appropriate responses to relevant social issues.” (1985: 763). Note that this definition of a corporation’s tasks comes close to the ‘do much’-end of Carroll’s range of philosophies of responsiveness.

Wood (1991a), who defines principles at different levels of analysis, also suggests a link between these different principles and types of responsiveness. She divides the continuum of Carroll in three parts – reactive, responsive, and interactive firms – thereby picking a somewhat inconvenient name for the middle category since all are actually types of responsiveness. Assuming that the principles of legitimacy, public responsibility and managerial discretion are hierarchical, she claims that reactive firms are motivated by the principle of legitimacy only, whereas responsive firms are motivated by legitimacy and public responsibility concerns and interactive firms respond to all the principles. She also echoed Wartick and Cochran’s definition of a firm’s tasks by proposing three activities that make up the process of responsiveness: environmental assessment, stakeholder management, and issues management (Wood, 1991b), as displayed in Figure 2.1. If uncertainty about societal expectations exists and the environmental assessment is local, researchers have to consider imitative behaviour within industries, which is observed frequently especially for firms which satisfy instead of exceed societal expectations (Bansal & Roth, 2000; Christmann, 2004). A clear distinction between the latter two is not provided, but one could argue that stakeholder management involves satisfying the (social) demands of those entities in the environment that can immediately affect the production process of the firm, such as suppliers and customers, whereas social issues management refers to the engagement of a firm in social issues which do not have a direct link with the production process of the firm, for example the development of training programmes for minorities or the reduction of pollution caused by the firm’s activities. This distinction is in line with that of Hillman and Keim (2001).

(23)

hand, executives in a value-attuned organisation acknowledge the influence of personal values on their decisions and are aware of the variety of values that are present in the environment. As a consequence, the environment will not be as narrowly defined as in a value-neglecting organisation and the corporation will be more capable of integrating social demands into its decision processes (Swanson, 1999). Obviously, one might reason that the value neglecting organisation will already have faced bankruptcy and that therefore Swanson’s typology is somewhat odd. The key research strategy then becomes to analyse which values enhance a community’s utility. Based on a classification by Frederick, Swanson (1995) proposes to classify the values along three key areas of corporate activity: economising, ecologising, and power-seeking values. Economising values refer to the ability of a corporation to convert input into output. Ecologising values are those that determine the ability of a corporation to engage into collaborative linkages with other organisations. Thus, where economising values are of a competitive nature, ecologising values are cooperative. Power-seeking values are those that enhance the position of a firm in a network. In sum, neglected decision processes and value-attuned decision processes can be empirically compared as to the extent to which they affect the economising, ecologising and power-seeking values of the corporation.

2.1.3 Outcomes and Measurement of Social Performance

The six main articles that have been used to describe the development of the CSP-model in Figure 2.1 – Carroll (1979), Wartick and Cochran (1985), Wood (1991a, 1991b), and Swanson (1995,1999) – largely leave aside the issue of the outcomes of socially responsible and/or responsive behaviour, with the notable exception of the two articles by Donna Wood. Carroll nor Wartick and Cochran consider the outcomes as part of their models (these contain two processes blocks: stakeholder management and issues management, using the labels Wood attached to them). Swanson intends to improve the model by integrating principles and processes, and therefore does not consider outcomes either.

Wood (1991a, 1991b) claims that the outcomes of corporate socially responsible and/or socially responsive behaviour – policies, programmes, and social impacts – form the only part of the model which is actually measurable. Principles cannot be objectively assessed and processes can at best be inferred from outcomes. She proposes two interesting research strategies, the first being to assess whether the behaviour is institutionalised – thus implying that social responsiveness has become a principle-based decision process – or not, in which case socially responsible behaviour and socially responsive behaviour need not coincide. Secondly, she proposes to research the perspectives of the various stakeholders in a firm on the extent to which socially responsive behaviour of a specific type is desired (Wood, 1991a).

(24)

advantage lies in the possibility to objectively score the releases, for example by counting words. However, the quality of these documents may be subject to a bias towards socially correct formulations. A corporation is after all not very likely to admit its lack of investment in an area of CSP. Furthermore, one measures what a company says to do and that need not be what it actually does. The second type of measures are direct impact measures, or – using the terminology of Abbott and Monsen (1979) – social accounting. Although these can be considered the most objective measures available, the scope of one specific measure generally is limited to one dimension of corporate social performance, often concerning environmental issues. The third type of measures is the reputation survey. The advantage is that a relatively broad set of indicators of overall social performance can be assessed, but the subjectivity involved can be substantial. Interestingly, as will be mentioned in Section 2.2, the results obtained in empirical works do not seem to be a function of the method used to measure social performance (Orlitzky et al., 2003).

Policies and programmes are among the outcomes Wood (1991a) identifies. One class of measures of CSP thus involves content analysis of these policies. Robertson and Nicholson (1996) research the disclosing behaviour of public companies in the UK. They reason that if codes of ethics are to have a meaning, the phrasing of corporate disclosures in this area should be tailored to the corporation. In this sense, it is worthwhile remarking that Schreuder (1981) found that 70 per cent of the employees of five large Dutch corporations consider the corporate social report a useful means of corporate communications. Obviously, “we have six on-site child development centres with a capacity of more than 900” (Snider et al., 2003: 182) is a more explicit link between text and behaviour than “we value and respect our customers, for without them we would not exist” (Snider et al., 2003: 181). A content analysis which makes no classification of the specificity of a claim, like Ingram and Frazier (1980), may find spurious relationships or might not find a strong relationship between disclosure and performance. In the article of Ingram and Frazier’s (1980, the lack of significant findings could be the consequence of the absence of a classification in this sense.

(25)

to which the CFO personally finds the code useful in making decisions in five areas. Weaver et al. (1999) show that this type of integration is easily decoupled from daily operations. True integration implies that every employee is constantly aware of the ethical dimension of his work. Easily decoupled social performance is found to be a function of external pressures – as Stevens et al. (2005) claim –, but truly integrated social performance is a function of management commitment, not of external pressures (Weaver et al., 1999).

2.2 Corporate Social and Financial Performance

The model of corporate social performance outlined in Section 2.1 holds the premise that if the environmental scanning technique and the management of stakeholders are effective and efficient the supply of solutions to social demands enhances corporate financial performance.

Harrison and Freeman (1999) note that a methodological problem in defining CFP-CSP relationships exists. Originally, they claim, stakeholder theory was intended to find an integrative measure of firm performance that includes both the financial and the social aspect. Economic science has, however, not developed into a state in which this integrative measure is readily available. Graafland (2002), as an example, models the trade-off between profits and principles, but in order to arrive at a maximisation problem with only one maximiser, the assumption has to be made that a curvilinear relationship between profit and corporate social performance exists. Others, in an attempt to find an integrative measure like the concept of sustainable corporate performance as advocated by Steg et al. (2004), do make a convincing plea for the integration of various performance dimensions into one variable, but do not provide the weights that have to be assigned to the various performance elements. Graafland and Eijffinger (2004) intend to solve this problem by means of a questionnaire which was sent to corporations. Although this certainly is a step ahead, it still does not define the universal measure, since it is only the corporation’s perspective that counts here. Theoretically, one could send questionnaires to every stakeholder – if all stakeholders could be identified –, but then the question would arise which stakeholders are the most important. Therefore, although Harrison and Freeman (1999) have a point, a solution is not yet available.

(26)

productivity (Waddock and Graves, 1997; Zahra et al., 1993). Also, consumers might have a higher willingness to pay for the products of socially responsive corporations (Waddock and Graves, 1997; Zahra et al., 1993). Social responsiveness might also be a signal of good management (Alexander and Buchholz, 1978; Pava and Krausz, 1996; Waddock and Graves, 1997). McWilliams and Siegel (2001), finally, argue that a reputation for corporate social responsibility is a means of differentiating products from those of competitors. A higher degree of differentiation lowers consumers’ price elasticity and thereby increases the profit the firm can charge. In line with these arguments, it is hypothesised that:

Hypothesis 1a: corporate financial performance is a positive function of corporate social performance

Pava and Krausz (1996) argue that if corporate social responsibility is universally beneficial, all firms should be ‘doing it’. Since although the benefits of social responsiveness are open for questioning, costs are obviously present, corporations might face a trade-off between the benefits and costs of engaging in social responsiveness: improving upon the reputation for good CSP might first lead to an improved financial position, but as the social status of the corporation is higher, it might be more costly to add further to that reputation. Therefore, the returns to social performance can be diminishing and it is suggested that:

Hypothesis 1b: corporate financial performance is a curvilinear function of corporate social performance

Since the exact shape of the curve that will be tested is partly the consequence of the configuration of the data, more specifically the record for positive CSP will be analysed independently from the record for negative CSP, a discussion of the shape will be taken up in Chapter 4 after the presentation of the data has been completed (see Figure 4.10).

Many studies have researched the link between social and financial performance. The methodological problems are significant, generally stemming from the difficulty of measuring corporate social performance (Aupperle et al., 1985). Generally, the social performance data that happened to be available, defined the methodology which was applied. Consequently, sample sizes were often small, control groups ill-defined or absent and the time periods studied were too short to asses the long-term effects of superior social performance (Cochran and Wood, 1984). Despite these methodological problems, three meta-analyses show convergence on the issue.

(27)

since it controls for measurement and sampling errors. The conclusion is that the correlation between environmental and financial performance is lower than the correlation between other CSP variables and financial performance, but the relationships are all positive. Also, a relationship is not only found between concurrent CSP and CFP, but subsequent CFP is found to be a function of CSP as well. Studies that use reputation indices exhibit the strongest correlations. Furthermore, accounting-based measures of financial performance are generally better predicted by CSP than equity market-based measures of financial performance. The largest meta-analysis, containing 127 studies that appeared between 1972 and 2002, has been performed by Margolis and Walsh (2003). They do not correct for the errors Orlitzky et al. (2003) do control for but apply simple vote-counting. Although the method is simple, the results are again striking: out of the 109 studies that dealt with CSP as the independent variable, 54 proved a positive relationship to be existent. In only seven cases a negative relation was found. The other articles either reported non-significant (28) or mixed (20) results.

A relatively small stream of empirical research has argued for a reversed causality: the more resources a firm has at its disposal, id est the larger the organisational slack, the more a firm is free to engage in corporate social responsiveness (Pava and Krausz, 1996; Waddock and Graves, 1996). Thus, the categories of Carroll (1979) are considered hierarchical: firms first satisfy economic and legal responsibilities before turning to ethical and discretionary responsibilities. This so-called slack resources argument gives:

Hypothesis 2: corporate social performance is a positive function of corporate financial performance

Only 22 studies in the sample of Margolis and Walsh (2003) take social performance to be the dependent variable. Sixteen out of these studies found a positive correlation. Orlitzky et al. (2003), in an attempt to directly measure which of the two hypotheses fits the data best, find that the correlation between CSP and concurrent and prior CFP is slightly higher than that between CSP and CFP. They conclude that a bidirectional relationship exists.

(28)

other strategic choice variables that cause differentiation. McGuire et al. (1988) show that there is a strong negative relationship between the debt-to-assets ratio and CSR. Thereby, they support Pava and Krausz’s call for the inclusion of risk in the equation.

It is also necessary to specify industry or firm effects. In the earlier paragraphs of this Chapter, it has been argued that the extent to which a firm behaves socially responsive is different from one industry to another. Industries differ in the importance of a specific stakeholder to the firms in that industry too. If firm effects are specified, more degrees of freedom will be lost. However, if there is much heterogeneity inside an industry, the fit of a model is better if firm effects instead of industry effects are included. The discussion on whether the firm or the industry accounts for the largest share of the variance in profitability has been a prominent topic in the structure-conduct-performance literature. Research in this area has relied on datasets which only include manufacturing industries and has used methodologies that falsely inferred statements on causality (McGahan and Porter, 2002). Recent studies, nonetheless, have confirmed the results of the earlier research whilst finding solutions to the aforementioned issues. Firm effects are generally found to be more important than industry effects (see e.g. McGahan and Porter, 1999; Waring, 1996).

2.3 Board Behaviour

Although researching the relationship between corporate social and financial performance can yield interesting insights into what types of social investments work or, conversely, how much financial slack is sufficient for socially useful investments to take place, the managerial implications need not be straightforward. For example, if science would agree upon education of youngsters in the area in which a corporation is located being financially advantageous, no information as to how those financial benefits can be attained is available. Thus, the question on the relationship between financial and social performance is immediately followed by the question on what policy instruments are relevant to attain the advantages. In this section, a plea for the analysis of a possible influence of the corporate board on the relationship between CSP and CFP will be made.

(29)

extend the theory to the (European) two-tier system, although that involves complicated issues on the interaction between the two entities. A second limitation as to the theory development lies in the direction of causality that is assumed in the relationship between social and financial performance. It is assumed in this section that causality runs from social to financial performance. This assumption is made because of the emphasis in the empirical literature on this relationship. The reverse causality hypothesis will be tested empirically as an extension to Hypothesis 2.

2.3.1 The Nature of Corporate Boards

The effect of leaders on performance in corporations is widely studied. Pfeffer (1972a) notes that the power which is generally ascribed to a leader is constrained by various factors. Firstly, leaders form a group which is relatively homogeneous and consequently the influence of different leaders on corporate outcomes need not be different. Secondly, the leader is embedded within the organisation and is expected to behave in certain ways. The behavioural options that are open to him/her are therefore restricted. Thirdly, there are many intervening variables between the efforts of the leader and the corporate outcomes; variables which the leader cannot influence. Consequently, one might reason that leaders are not important to the organisation, their role would be symbolic: to serve as a scapegoat if things go wrong or to personalise the organisation if things go well. However, studies have shown that variables which proxy for top executives’ activities do explain corporate outcomes partially (Lieberson and O’Conner, 1972; Jackson, 1992). For example, Lieberson and O’Connor (1972) show that the firm’s upper echelons account for 14 to 69 percent of the variance in the profit margin in different industries corrected for year and company influences. Likewise, Richardson et al. (2003) show that director fixed effects are significant in a sample of 885 large US-firms. They argue that since director fixed effects are found to be uncorrelated with the director’s tenure, the evidence proves that directors match their policy choices with the firm for factors traditionally not accounted for. The stronger explanation that directors impose their preferences on the firm would require such a correlation. The literature on whether the industry or the firm contributes most to profitability, which was touched upon in Section 2.2, also hints at the importance of the upper echelons of a corporation, since firm effects have been proven to exist.

(30)

eight tot ten board meetings are held annually (Spencer Stuart, 2004). Sometimes, subgroups are installed who meet to discuss topics in a specific area. From the descriptive statistics in Hayes et al. (2004) one can derive that the number of committees that deals with charity, the environment, health issues, safety matters and the like is substantial. Furthermore, one committee generally seems to deal with various topics from the abovementioned list, thus hinting at the institutionalisation of social performance issues in board practice.

The tasks of boards are generally considered to fall under two headings: the monitoring role and the strategy and service role. The monitoring role has received by far the most attention in the literature (Daily et al., 2003). It rests in agency theory and is based on the assumption that managers maximise a utility function which does not only depend on variables which are beneficial to the firm. For example, a manager might be tempted to engage in prestigious investments on behalf of the firm, which add to the manager’s status but hurt company performance. Since the shareholders, dispersed as they are, face free-riding problems in monitoring the firm themselves – they would bear the full cost of monitoring but only receive a fraction of the benefits through a higher value of their shares, whereas non-monitoring shareholders would also receive that benefit without incurring the costs – mechanisms are erected to restrict the manager’s opportunistic behaviour (Moerland, 1995). One of these mechanisms is the board of directors and, conversely, one of the roles of the board of directors is to monitor the top management team.

The second role of the board of directors rests in resource-dependency theory. It holds that directors are a means of linking the firm to its environment; directors bring expertise to the fore which they can use to further corporate goals (Pfeffer, 1972b; Pfeffer and Salancik, 2003). Directors thus advice top executives on strategic issues. McNulty and Pettigrew (1999) note that the two roles are linked because managers know that in the end, directors will have to approve the proposals they make. Thus, a director’s advice might also be sought for to increase the probability that a plan will succeed. If this is true, the finding of McNulty and Pettigrew (1999) that 90 to 95 percent of the proposals that reach the (UK) boards in their sample are accepted, the strategic role of the board can be very substantial. Obviously, a rationale for this finding might also be that boards of directors are ineffective in distinguishing among good and bad management proposals.

(31)

In the next section, hypotheses will be derived on the moderating effect of board size, the number of outsiders that serve on the board, the average age of the board members and the number of meetings of the board. In doing so, the perspective of upper echelon theory will be adopted (Hambrick and Mason, 1984). In this strand of research, the assumption is made that objective characteristics of upper echelons (boards) and their constituents proxy for the processes that determine corporate outcomes. Hambrick and Mason (1984) reason that strategic choices are the result of the perception, interpretation and values of decision makers and that these can be measured by objective, demographic variables.

In their evaluation of twenty years of upper echelon research, Carpenter et al. (2004) mention two important shortcomings of the studies they evaluate: the upper echelon methodology is static in the sense that characteristics exert an influence on (unobserved) processes which affect corporate performance without taking into account the possible effect of performance on the selection of the top management team. Furthermore, the research has shown that demography matters, but the processes through which they matter remain unclear. In this area, Carpenter et al. (2004) echo Jackson’s (1992) call for the analysis of strategic issue processing groups. On the other hand, attempts to open the black box – of which Forbes and Milliken’s (1999) model is an excellent example – are generally difficult to test because access to the processes of top executives’ decision making is not easily gained.

(32)

2.3.2 Hypotheses on Corporate Boards Moderating the CSP-CFP relationship

In a review of the empirical evidence on the relationship between board size and financial performance, Hermalin and Weisbach (2001) found a negative relationship to be most common. For European firms, Conyon and Peck (1998) find similar evidence. Barnhart et al. (1994) note, however, that the specification of the estimated equation matters for the results that can be derived. Underlying this research is the theory that larger boards bring inefficiencies in the form of conflict which harms the effectiveness of the board (Ensley et al., 2002).

Pfeffer (1973) finds that board members are selected from the perspective of the function a member is supposed to execute. For example, hospital boards in agricultural areas consist of more members with an agricultural background than hospital boards in other areas. This is explained by the need to link the hospital to the agricultural pressure groups in the area, thus providing a link between the organisation and its environment (Pfeffer, 1973; Pfeffer and Salancik, 2003). Pfeffer (1973) also finds that types of hospitals he considers to be more in need of these linkages – for example hospitals which are heavily funded by private institutions – have larger boards. If more directors hold a seat on the board, the firm might be assumed to have stronger linkages with key stakeholders, providing a fore for the communication on the socially responsive activities of the firm. Consequently, the stakeholders might be less reluctant to provide the firm with a license to operate (Pfeffer and Salancik, 2003) and the reputation for good social performance might pay off in terms of profitability. The partial derivative of corporate financial performance with respect to social performance might thus be a function of board size; consequently, board size is a moderator variable in the CSP-CFP relationship (Jaccard and Turrisi, 2003). This topic suggests:

Hypothesis 3a: In firms with larger boards, corporate social performance has a stronger effect on corporate financial performance than in firms with smaller boards.

A second variable which traditionally has been widely studied in the literature on board performance is the number of outsiders. Hermalin and Weisbach (2001) conclude that there is no evidence for a relationship between the number of outsiders and financial performance, a relationship for which the specification problem mentioned before also holds (Barnhart et al., 1994). Theoretically, boards with a relatively high number of outsiders are supposed to be more effective in monitoring the firm and consequently a positive relationship is usually postulated.

(33)

mixed. In the study of Zahra et al. (1993) the zero-order correlations between the number of outsiders and corporate social performance variables are generally negative, whereas the coefficients in a regression analysis – which might not control for the appropriate variables – are found to be positive. Luoma and Goodstein (1999) show that the number of stakeholders who hold a seat on the board has increased significantly between 1984 and 1994. However, some robustness checks evoke doubts on the reliability of this finding. Johnson and Greening (1994), finally, develop a covariance structure model, which indicates a positive relationship between the number of outsiders and two dimensions of corporate social performance.

Ibrahim et al. (2003) find that the corporate social responsibility orientation of outside directors is higher than that of inside directors, where this orientation is measured by the four categories distinguished by Carroll (1979). Insiders are found to be more concerned with the company’s economic responsibilities, whereas outsiders place more emphasis on discretionary responsibilities. The above reasoning is condensed in:

Hypothesis 3b: In firms with more outsiders serving on the board, corporate social performance has a stronger effect on corporate financial performance than in firms in which the number of outsiders on the board is lower.

A negative correlation between the age of executives and the amount of risk a corporation is exposed to has been found in the empirical literature (Jackson, 1992). Younger managers are said to be driven by a desire to build a reputation of high performance and therefore take risks to achieve such a reputation. Indeed, Grimm and Smith (1991) found that in a sample of railroad companies changes in environmental policies were associated with younger managers. Likewise, the hypothesis might be postulated that younger directors are more engaged in representing the corporation than older directors are. Having noted that, in this context, the alleged differences are not between young and old directors but more between old and older directors, it follows that:

Hypothesis 3c: In firms with a high average director age, corporate social performance has a stronger effect on corporate financial performance than in firms in which the average age of the directors is lower.

(34)

is said to diminish board and consequently firm performance by the same two studies. Since cohesiveness is positively affected by the number of encounters a group has and this same number diminishes the probability that affective conflict will arise, it is proposed that:

(35)

CHAPTER 3: DATA AND METHODS

3.1 Data

In the following sections the sources of the data and the transformation of data into variables are explained. Section 3.1.1 deals with social performance, Section 3.1.2 with financial performance and Section 3.1.3 with board data, whereas 3.1.4 derives the control variables. A list of all the variables used in this study is presented in Appendix A. In Appendix B the relevant transformations of the data are provided.

3.1.1 Social Performance Data

Various operationalisations of the concept of corporate social performance have been used in empirical studies. As is apparent from the review in Chapter 2, content analyses of annual reports and other corporate disclosures, reputation indexes and pollution ratings are commonly used. Since many studies run into problems of sample size, this research creates a panel dataset in which all S&P 500 firms are measured in six consecutive years. In doing so, the results are probably representative for a larger set of firms and thus the generalisability can be expected to be substantial.

Source. Corporate social performance is measured based on the Kinder Lydenberg & Domini Ratings Data. This database is compiled by KLD Research Analytics Inc. (KLD), an advisory service which provides social research products to institutional investors (see www.kld.com). The database used in this study is a derivative of the SOCRATES database which contains social records of many US and international companies. The Ratings Data – which are a summary of SOCRATES – are screened for firms which appear in the Standard & Poors 500 index in the interval 1997-2002, resulting in a sample of 734 firms with one to six observations in the aforementioned period. Reputation data, as has been argued in Chapter 2, is preferred above content analysis and pollution measures since it provides a means to analyse many CSP dimensions simultaneously. Furthermore, the KLD database is preferred above the other extensive database available, the Fortune Index, because this latter index has been shown to measure a reputation for financial performance more than one for social performance (Fryxell and Wang, 1994).

(36)

products to the economically disadvantaged. Hillman and Keim (2001) call the qualitative screens ‘stakeholder management’ and assign the term ‘social issues participation’ to the exclusionary screens. KLD undiscloses the method it uses to rate firms, but from Waddock and Graves (1997) and Hillman and Keim (2001) it is known that various corporate sources (e.g. annual reports and press releases) and external sources (business periodicals) are consulted; that – because these ratings involve a certain degree of subjectivity – KLD’s raters meet to discuss borderline cases and that “KLD is willing to respond to company concerns where accuracy is at issue” (Waddock and Graves, 1997:308). Not much is known regarding the construct validity of the Ratings Data. Sharfman (1996) has shown that the correlation between the KLD database and the other commonly used source, the Fortune Reputation Index, is significantly positive. The correlation coefficients he produces generally do not exceed 0.55. These coefficients do improve when only non-zero ratings – see below – are considered. This analytical procedure implies that the Fortune data are assumed to be a fair measure of corporate social performance, but according to Fryxell and Wang (1994) who factor analysed this dataset, a financial factor dominates the social performance factor (‘subordinate capabilities factor’) and thus the Fortune data may be a measure of financial reputation more than it is a measure of reputation for good social performance.

For every screen, a number of criteria exists on which firms are rated (see KLD Research & Analytics Inc., 2003). All these criteria are dichotomous and can be positive (‘strengths’) or negative (‘concerns’). Since matched pairs of criteria can be found, the criteria are in fact scaled on a three-point scale: for example, a firm with strong union relations receives a score equal to ‘1’ on that employee relations strength criterion whereas a firm that has weak union relations receives the same score on the associated employee relations concern criterion. It is also possible that a firm receives no score on either of the criteria, thus creating three categories for the union relations criterion. For exclusionary screens, strengths are not assigned and thus the scale is binary. Since this yields a measure of social performance which is too crude, and since the elements of CSP that are measured by the exclusionary screens are not at the core of the argument made in the previous chapter, the advice of Sharfman (1996) to leave these screens out is accepted. In the terminology of Hillman and Keim (2001), stakeholder management is considered, but social issue participation is not.

Variables. As was mentioned above, several criteria exist for the seven qualitative screens in the KLD Ratings Data. Table 3.1provides an overview of the number of criteria per strength (+) and concern (-).

(37)

2001; Waddock & Graves, 1997) have created a range from -2 (major concern) to +2 (major concern) which implies that good social performance is seen as the mirror image of bad social performance. In this study, this assumption is not made: for it might well be that a bad reputation lasts longer than a good one (see also McGuire et al., 2003).

TABLE 3.1

Number of Items in KLD database

Area 1997 1998 1999 2000 2001 2002 + - + - + - + - + - + - Community 5 3 5 3 5 3 6 4 6 4 6 3 Diversity 8 3 8 3 8 3 8 3 8 3 8 3 Employee relations 4 5 5 5 5 5 5 5 5 5 5 5 Environment 6 6 6 6 6 7 6 7 6 7 6 7 Human rights 2 3 2 4 2 4 3 5 3 5 4 4 Product 4 4 4 4 4 4 4 4 4 4 4 4 Corporate governance 3 4 3 4 3 4 3 4 3 4 3 4

Note: + = strength; - = concern

Although this operation made the fourteen indicators comparable – i.e. seven qualitative screens with a strength and concern score for each – the number of indicators of corporate social performance is still inconveniently high for a quick overall view of a firm’s social performance to emerge. In the hypotheses in which CSP is not a dependent variable, the various indicators are able to show whether different types of social performance have a different effect on financial outcomes. However, in the other hypotheses one overall measure of CSP is to be preferred.

To integrate the various indicators, a weighting schedule has to be developed. An objective candidate could be derived from factor analysis. However, none of the criteria is normally distributed and – ignoring this finding for the time being – even if a factor analysis is performed, the first factor explains only 17% of the variance in the indicators and the four factors with an eigenvalue above 1 together account for less than 50%. Thus, the explanatory power of the factors is below conventional levels of acceptability.

Two corporate social performance variables are developed. The first involves the computation of a simple average of the indicators – thus implying a weighting scheme in which every dimension of CSP is deemed equally important – where positive and negative indicators are still treated separately. Thus an average of the positive indicators and one of the negative indicators is computed. The second set of variables is composed of all the separate indicators. This second measure is thus a refined version of the first.

(38)

performance, but it does provide insights into the relationships among CSP-screens. Running cluster analyses for every year and assuming that the variables are continuous, a two-cluster solution results for the years 1997-2000 and a three-cluster solution appears for 2001 and 2002, regardless of the method which was used to obtain the clusters, thus indicating robust results. Computing the means of the indicators in every cluster shows that there is always one cluster with both the extremely positively and the extremely negatively rated firms, whereas the firms with a mediocre score are in an other cluster. Therefore, none of these cluster analyses yields a measure which discriminates between good and bad social performance. Clustering does hint at the conclusion that positive and negative criterions are positively correlated, which is rather counterintuitive.

In short, two sets of variables are developed. The first set, referred to as the ‘composite measure’, consists of two variables each of which is the mean of seven indicators of CSP. The first variable in this set is the mean of the positive indicators, the second the mean of the negative indicators (see Table 3.1 for an overview of the indicators). The second set of variables comprises the fourteen indicators of corporate social performance. The sets are entered separately into the regression models to see whether a possible non-significant effect of the composite measure is due to the balancing out of effects of the indicators, for example: a negative effect of a strong reputation for diversity and a positive effect of a strong reputation for environmental issues might render the effect of the composite measure insignificant.

3.1.2 Financial Performance Data

Financial performance data was obtained from Thomson Financial’s database Datastream. This database contains company accounts information on 50,000 stocks. Generally, three measures of financial performance are distinguished: accounting-based measures, equity market-based measures and risk measures. The latter category is not considered here, for tbe data set does not allow for the computation of standard deviations without having to drop many cases.

(39)

flow. This variable conceptually fits the hypothesis that companies are more socially responsive if the amount of slack resources is higher better than ROA or ROS.

Equity market-based measures of financial performance. The company accounts files which were derived from Datastream do not provide information on share prices. There are three performance variables which can be characterised as market-based measures of financial performance, though: earnings per share (EPS), dividends per share and the market-to-book value of the firm. EPS is calculated as the after-tax profit over the number of outstanding common shares. This amount can be paid out to shareholders, as dividends, or retained in the firm to smoothen the periodical dividend payment or to make new investments. Note that the variable dividends per share thus only is a proper measure of corporate performance if the retention rate is constant across time. The market-to-book value, lastly, is the ratio of the market value of the firm – i.e. the number of outstanding shares times the current price per share – and the value of the firm according to the internal accounting system. Higher values of this ratio imply that investors are willing to pay more for the company than the value of the physical components. This means that investors envisage growth potential for the firm. In determining the book value, intangible assets are not taken into account.

3.1.3 Board Data

The Payoff to Corporate Well-Doing

The Payoff to Corporate Well-Doing

The Effect of Corporate Boards on the Link between

Social and Financial Performance

THE PAYOFF TO CORPORATE WELL-DOING: THE

EFFECT OF CORPORATE BOARDS ON THE LINK

BETWEEN SOCIAL AND FINANCIAL

PERFORMANCE

Master Thesis

Gerwin van der Laan

Groningen, August 2005

Supervisors:

Prof. Dr. Hans van Ees

Prof. Dr. Arjen van Witteloostuijn

University of Groningen, Faculty of Economics

Master of Science in Economics and Business (Research Master)

PREFACE

ABSTRACT

TABLE OF CONTENTS

CHAPTER 1: INTRODUCTION

1.1 Problem Statement

1.2

Theoretical Approach

1.3

Organisation of the Thesis

CHAPTER 2: THEORY

2.1 Corporate Social Performance

2.2 Corporate Social and Financial Performance

2.3 Board Behaviour

CHAPTER 3: DATA AND METHODS

3.1 Data