10 1 1 39

Corporate social responsibility and economic performance in the top British companies: are they linked?
George Balabanis Hugh C. Phillips and Jonathan Lyall

The authors George Balabanis is a Lecturer at the European Business Management School, University of Wales Swansea, Swansea, UK. Hugh C. Phillips and Jonathan Lyall are at the Leicester Business School, DeMontfort University, Leicester, UK. Abstract This paper investigates the relationship between corporate social responsibility (CSR) and the economic performance of corporations. It rst examines the theories that suggest a relationship between the two. To test these theories, measures of CSR performance and disclosure developed by the New Consumer Group were analysed against the (past, concurrent and subsequent to CSR performance period) economic performance of 56 large UK companies. Economic performance included: nancial (return on capital employed, return on equity and gross prot to sales ratios); and capital market performance (systematic risk and excess market valuation). The results supported the conclusion that (past, concurrent and subsequent) economic performance is related to both CSR performance and disclosure. However, the relationships were weak and lacked an overall consistency. For example, past economic performance was found to partly explain variations in rms involvement in philanthropic activities. CSR disclosure was affected (positively) by both a rms CSR performance and its concurrent nancial performance. Involvement in environmental protection activities was found to be negatively correlated with subsequent nancial performance. Whereas, a rms policies regarding womens positions seem to be more rewarding in terms of positive capital market responses (performance) in the subsequent period. Donations to the Conservative Party were found not to be related to companies (past, concurrent or subsequent) nancial and/or capital performance.
In the modern commercial era, companies and their managers are subjected to well publicised pressure to play an increasingly active role in society so called Corporate social responsibility. It has been argued that an element in this development is simply enlightened self-interest in that social responsibility enhances corporate image and nancial performance. To date the evidence to support this thesis derives from North America. Outside this continent evidence for any relationship is sparse. This study investigates the claims that social responsibility and economic performance are linked and attempts to test the relationship within a UK context. This study will initially attempt to dene the concept of corporate social responsibility and to examine its guiding principles. Subsequently, the available empirical research into the link between corporate social responsibility and economic performance will be evaluated and research hypotheses will be formulated. Finally, the research method, measures used, ndings and conclusions will be presented.
Denitions
Corporate social responsibility (CSR) has recently been the subject of increased academic attention. While social responsibility has gured in commercial life over the centuries, in the modern era increasing pressure has been placed on corporations to play a more explicit role in the welfare of society. Although the topic rose to prominence in the 1970s (Carroll, 1979; Wartick and Cochran, 1985), the rst publication specically on the eld dates back to 1953, with Bowens Social responsibilities of the businessman. In this work Bowen argues that industry has an obligation to pursue those policies, to make those decisions, or to follow those lines of actions which are desirable in terms of the objectives and values of society (Bowen, 1953, p. 6). Epstein (1987), however, argues that the concept of specic business ethics can be traced further back to certain academics and businessmen in the nineteenth century who promulgated the belief that private business is a public trust. Bowen (1953) sets the scene in this eld by suggesting that the concept of specically corporate social responsibility emphasises that: businesses exist at the pleasure of society and that their behaviour and methods of 25
European Business Review Volume 98 Number 1 1998 pp. 2544 MCB University Press ISSN 0955-534X
Corporate social responsibility and economic performance
European Business Review Volume 98 Number 1 1998 2544
George Balabanis, Hugh C. Phillips and Jonathan Lyall
operation must fall within the guidelines set by society; and businesses act as moral agents within society. Wood (1991) expanded these ideas, encapsulating them into three driving principles of social responsibility, which are that: (1) business is a social institution and thus obliged to use its power responsibly; (2) businesses are responsible for the outcomes relating to their areas of involvement with society; and (3) individual managers are moral agents who are obliged to exercise discretion in their decision making. In general, the social responsibilities of a rm seem to arise from the intersection (and compatibility) of the political and cultural systems with the economic system (Jones, 1983). However, Friedman (1970) argued that the successful functioning of our society depends on the role specialisation of its institutions (or systems). According to him the corporation is an economic institution and thus should specialise in the economic sphere; socially responsible behaviour will be rectied by the market through prots. In Friedmans (1970) view business has only one social responsibility and that is to maximise the prots of its owners (to protect their property rights). Organisations are seen purely as legal entities incapable of value decisions. A manager who uses a rms resources for non-prot social purposes is thought to be diverting economic efciency and levying an illegal tax on the organisation. Opponents (Frederick et al., 1992) of this view, challenge the very foundations of Friedmans thesis the economic model. They claim that the economic model and role specialisation of institutions (or systems) are not working as suggested. This comes as a result of the rise of oligopolies in certain sectors; the separation of ownership and management; governments involvement in the economy and conversely industrys involvement in the political process through lobbying. In addition, if corporations do not adopt social responsibility, government with its potential for inefciency and insensitive bureaucratic methods may be forced to step in. With respect to Friedmans argument that the legal conception of corporations articles and memorandums of associations limits a rms involvement solely to economic roles, it can be claimed that they are broad 26
enough to allow departures from this narrow path. Social responsibility is also seen as a consequence of and an obligation following from the unprecedented increase of rms social power (as tax payers, recruiters, etc.) (Davis, 1975). Failure to balance social power with social responsibility may ultimately result in the loss of this power and a subsequent decline of the rm (Davis, 1975). Another school of thought sees social responsibility as a contractual obligation rms have towards society (Donaldson, 1983). It is society in the rst place that has permitted rms to use both natural and human resources and has given them the right to perform their productive functions and to attain their power status (Donaldson, 1983). As a result, society has an implicit social contract with the rm. Thus, in return for the right to exploit resources in the production process, society has a claim on the rm and the right to control it. The specics of this contract may change as social conditions change but this contract in general always remains the basis of the legitimacy of the demand for or assertion of the need for CSR (Epstein, 1987). A growing number of scholars take the view that rms can no longer be seen purely as private institutions but as social institutions instead (Frederick et al., 1992; Freeman, 1984; Lodge, 1977). The benets owing from rms need to be shared collectively. This thesis is similar to the stakeholders model (Freeman, 1984) and claims that a rm is responsible not only to its shareholders (owners) but to all stakeholders (consumers, employees, creditors, etc.) whose contribution is necessary for a rms success. Thus, CSR means that a corporation should be held accountable for any of its actions that affect people, communities and the environment in which those people or communities live (Frederick et al., 1992). Carroll (1979) suggests that CSR is dened as the economic, legal, ethical and discretionary demands that society places on business. Similarly, Zanies conceptualised CSR as the degree of t between societys expectations of business and the ethics of business. He argues that CSR is really nothing more than another layer of managerial responsibility resulting from the evolution of capitalism. An interesting twist to the argument is provided by Tuzzolino and Armandi (1981) who provide a motivational theory of
organisational social response based on Maslows hierarchy of needs. CSR is the fullment of a rms internal and external self-actualisation needs which are located on the top of their organisational needs pyramid. According to this view, rms adopt CSR after they have satised three earlier layers of needs (which include: physiological or survival needs fullled by corporate prots; safety needs such as dividend policy, conglomeration and competitive position; and afliative needs such as participation in trade association, lobby groups, etc.). Epstein (1987) attempted to differentiate business ethics and CSR and to incorporate them into a strategic process. According to him business ethics refer to issues and dilemmas related to the morality of organisational actions or decisions. CSR focuses more on the consequences of organisational actions. He dened CSR as the discernment of issues, expectations and claims on business organisations regarding the consequences of policies and behaviour on internal and external stakeholders (Epstein 1987, p. 101). Angelidis and Ibrahim (1993) dened CSR as corporate social actions whose purpose is to satisfy social needs. They developed an equilibrium theory based on social demand and supply, identifying a set of factors that affects them (social supply and demand). Thus, opinions differ in terms of the basis or scope of CSR and even the very denition of the term. As a consequence different aspects of a rms operations can be seen to come under its sway depending on the stance one adopts. As has been shown, what can be conceived as social responsibility can range from simply maximisation of prots, to satisfaction of stakeholders social needs, or fullment of social contractual obligations, fullment of a rms needs, achievement of a social equilibrium, etc. depending on the stance taken. While academic debate abounds at the theoretical level, at the operational level insights are more sparse. Schwarts and Dahl observed that socially acceptable behaviour of North American rms at the time of writing the 1970s included: disclosure of information to shareholders; disclosure of the board of directors; monopolistic behaviour (predatory pricing, etc.); equality of treatment for minorities; prot sharing; environmental protection; 27
ethics in advertising; and social impact of technology. However, according to Vyarkarnam (1992), many of these have now been regulated by statute. Present day concerns have changed focus. He found that current CSR concerns, which are in substance the same for both North American and the UK rms, encompass such areas as: environmental protection (e.g. reduction of emissions and waste and the recycling of materials); philanthropy (donating to charities, etc.); involvement in social causes (involving anything from human rights to AIDS education); urban investment (working with local government to regenerate small businesses and the inner city environment generally); and employee schemes (higher standards of occupational health and safety, good standard of staff treatment, job-sharing, exitime, etc.).
The relationship between corporate social responsibility and economic performance

Commentators have argued both for and against the view that corporate social responsibility is enlightened economic self-interest. The controversy at the theoretical level will be considered here, while the empirical evidence for and against will be presented later. Similarly, in order to clarify what is a complex and at times convoluted debate, the discussion will be divided into the relationships suggested with: concurrent and subsequent (to CSR) economic performance; and past economic performance. Relationship with concurrent and subsequent (to CSR) economic performance Those who have theorised that a negative relation exists between social responsibility and economic performance have argued that a high investment in social responsibility results in additional costs. According to McGuire et al. (1988, p. 855) the added costs may result from actions such as making extensive charitable contributions, promoting community development plans, maintaining plants in economically depressed locations and
establishing environmental protection procedures. These costs might put a rm at an economic disadvantage compared to other, less socially responsible, rms. In contrast, others have argued the case for a positive association. McGuire et al. (1988) cite the argument that a rm perceived as high in social responsibility may face relatively fewer labour problems or perhaps customers may be more favourably disposed to its products. Alternatively, CSR activities might improve a rms reputation and relationship with bankers, investors and government ofcials. Improved relationships with them may well be translated to economic benets. According to Spicer (1978a,b), Rosen et al. (1991), Graves and Waddock (1994) and Pava and Krausz (1996), a rms CSR behaviour seems to be a factor that inuences banks and other institutional investors investment decisions. Thus, a high CSR prole may improve a rms access to sources of capital. Modern corporate stakeholder theory (Cornell and Shapiro, 1987; Freeman, 1984; Jones, 1995; McGuire et al., 1988) can also explain part of the CSR/economic performance relationship. According to stakeholder theory the value of a rm is related to the cost of both explicit claims and implicit claims on a rms resources. Claimants include not only the legal owners of the rm but other constituencies such as lenders, employees, consumers, banks, government, etc. Stakeholders who have explicit claims on the corporation include besides its owners lenders, employees, government, etc. In addition, there are others with whom the rm has made implicit contracts, which could include the quality of service and CSR. According to McGuire et al. (1988), if the rm does not honour these implicit contracts, then it is argued that the parties to these contracts may attempt to transform them from implicit to explicit agreements. The latter may be more costly for the rms involved. According to Freeman (1984) and McGuire et al. (1988) the implications of the conversion of implicit to explicit contracts may have broader effects than the direct costs resulting from the forced change in its behaviour (e.g. cost of installment of gas emission control equipment). For example, socially irresponsible actions in one area (e.g. gas emissions) may spill-over and affect the corporate image in other areas as well (e.g. unregulated issues on labour relationships). This could in turn result in other implicit 28
stakeholders (e.g. trade unions) striving to make their claims explicit. Thus, rms with an image of high CSR may nd that they face both fewer and lower-cost explicit claims than those with a less enlightened stance. Thus, from a theoretical perspective, arguments can and have been made both for and against a positive relationship between social responsibility and concurrent or subsequent (to CSR) economic performance. Relationship with past economic performance According to Parert and Eibert (1975), Ullmann (1985) and Roberts (1992), if corporate social responsibility is viewed as a signicant cost, rms with relatively high past nancial performance may be more willing to absorb these costs in the future. It is also expected that poor performers would seek more immediate results and consequently they may prefer short-term and high-yield investments to the uncertain and in general longer-term CSR investments. A similar view is that policies and expenditures in discretionary areas such as social programmes may be especially sensitive to the existence of slack resources in the rm (McGuire et al., 1988). Ullmann (1985) argued that corporations must reach an acceptable level of economic performance before devoting company resources to meet social demands. This is supported by the assertion that corporations with strong prior economic performance appear to be more likely to have high current levels of social disclosure. Ullmann (1985) also suggested that companies with less stable stock market patterns would be relatively less likely to commit resources to social activities.
Prior empirical research

Empirical research into the effects of corporate responsibility has produced mixed results. Some studies have suggested a positive relation, whereas others have concluded that the effects are negative or inconsequential. For example, Belkaoui (1976) investigated the information content of pollution control disclosures. His results suggested a positive relationship between economic performance and social responsibility, at least in this area. Other studies produced results consistent with the notion that corporate social responsibility activities impact on the nancial markets (Anderson and Frankle, 1980; Shane and Spicer, 1983; Spicer, 1978a,b).
However, certain studies have replicated earlier research and found conicting results. Frankle and Anderson (1978) rejected Belkaouis (1976) interpretation and argued that non-disclosing rms had consistently performed better in the market. In a similar manner, Chen and Metcalf (1980) disagreed with Spicers (1978a,b) conclusions, arguing that his results were driven by spurious correlations. In response Spicer (1980) stated that Chen and Metcalf (1980) misinterpreted the purpose of his study, emphasising that associations not causal relationships were being investigated. Ingram (1978) concluded that the information content of social responsibility disclosures was conditional on the market segment with which a rm is identied. Alexander and Bulcholz (1978) and Abbott and Monsen (1979) found no signicant relationship between a corporations level of social responsibility activities and stock market performance. In addition, Chugh (1978), Trotman and Bradley (1981) and Mahapatra (1984) concluded that corporate social responsibility activities may lead to increased systematic risk. Cochran and Wood (1984) used corporate social responsibility rankings developed by Moskowitz (1972) to test the relationship between corporate social responsibility activities and rms performance. After controlling for industry classication and corporate age, a weak positive association between corporate social responsibility activities and economic performance was found. Mills and Gardner (1984) concluded in their analysis of the relationship between social disclosure and economic performance, that companies are more likely to disclose social responsibility expenditures when their nancial statements indicate favourable economic performance. One drawback of the above empirical studies is that they failed to distinguish between past, concurrent and subsequent to CSR economic performance, and thus to make possible reliable inferences about direction of causation. In most of the previous studies, economic performance covered a (commonly ve year) period surrounding the CSR performance and/or social disclosure periods. Routinely, the CSR performance and/or social disclosure periods were the midpoints of that period. However, in Mahapatra (1984) and Mills and Gardiner (1984) studies, economic performance periods were concurrent to the CSR performance period. 29
Only Shane and Spicer (1983) looked at economic performance subsequent to CSR disclosure period, nding a positive association. Practically, McGuire et al. (1988) were the rst to break this tradition and to separate economic performance into past, concurrent and subsequent to CSR performance. They used Fortune magazines ratings of corporate reputations to analyse the relationship between perceived corporate social responsibility and economic performance. Prior economic performance of the rms, as measured by both stock market returns and accounting based measures, were found to be more closely related to corporate social responsibility than was subsequent economic performance. McGuire et al. (1988) suggested that economic performance may be a variable inuencing social responsibility activities. Thus, the empirical research into the relationship between corporate social responsibility and economic performance is confusing and far from conclusive. According to Ullmann (1985) this may be attributed to the use of varying and questionable measures of CSR, differences in the research methodologies and the nancial performance measures used. To overcome these limitations, this study will use a more comprehensive measurement of CSR performance (admittedly within the context of the UK social and business environment), a combination of economic performance measures and including the necessary intervening variables in the research design.
Research methodology
The objective of the current study is to test the hypothesis that there is a relationship between CSR and economic performance of rms in terms of their: contemporaneous or subsequent economic performance; and past economic performance. As mentioned earlier, prior research has focused on the North American experience, whereas this study focuses on corporations that operate in the UK environment. In total 56 rms were included in the study. The main criteria for their selection were quotation on the London Stock Exchange and the availability of CSR ratings by the New Consumer Group (NCG) a UK public-interest research organisation (Adams et al., 1991).
The study covered more than 20 industry sectors, though there was a deliberate limitation in terms of industry coverage. The main focus of the NCG organisation is the consumer sector. Sectors such as nancial services and media related products were not included due to the difculties associated in the assessment of CSR performance (Adams et al., 1991, p. 55). Measurement of corporate social responsibility Previous studies in this area have primarily used three criteria to measure corporate social responsibility (McGuire et al., 1988): expert evaluations; content analysis of annual reports and other corporate documents; and performance in controlling pollution as a proxy measure. Each of these measures is subject to limitations. The validity of independent experts ratings rests on the expertise of the assessors and the accuracy of the information available to them. Commonly used expert-based rankings in the US studies are the Council of Concerned Businessmen Index and Business and Societys rankings (Moskowitz, 1972, 1975). The latter ones are based on the authors (Moskowitz) own evaluations and a panel of businessmen and MBA students evaluations. The informational value of annual reports and other public corporate documents as a source of hard data can be queried, being the product of the companys public relations programme. The pollution control proxy measure is limited in value to those industries where pollution is a signicant issue. In this study, as has already been mentioned, (conned by a general absence of relevant UK sources) the ratings of corporate social responsibility were obtained from the publication Changing Corporate Values (Adams et al., 1991), produced by the New Consumer Group (NCG). The assessment of companies in the NCG book covers the years 1988 and 1989. Evidence for the validity of the ratings comes from the number and diversity of sources used to produce a relatively comprehensive index of CSR performance. The sources include: commercial cutting services, trade associations, trade unions, business directories and databases, research and public relations rms and national and international public interest groups. The ratings were complemented by primary data collected via a 30
multi-wave mail survey directed to companies (for details of the questionnaire see Adams et al., 1991, pp. 531-58). Apart from the questionnaire, additional information for each company came from several (internal) informants (for each rm) who were contacted by NCG. As the authors emphasise, in some cases they had to get information (for a single company) from 20 different informants. This effort resulted in the development of company CSR proles which were sent to the rms for comment and renement. It seems that Adams et al. (1991, p. 531) took all reasonable precautions to minimise bias and systematic error in their measurement. The laborious triangulation of both information sources and data collection methods, and the use of multi-dimensional measurements of CSR, seem to have signicantly decreased the possibility of both the validity and reliability problems in their ratings. In addition, measurement limitations of prior studies such as: use of only one dimension (pollution proxies, etc.), reliance on a single information source (e.g. annual reports) or a single data collection method (panel of experts, content analysis, etc.) have been addressed and avoided. Each company was rated on 13 different aspects related to CSR (listed below). The last ve categories, in the list, apply only to a small number of industries or sectors. Therefore, they were not incorporated into the current study (i.e. only the rst eight were used). The 13 ratings produced by NCG were: (1) CSR disclosure. Disclosure was assessed on whether the company completed the questionnaires sent by NCG and the Council of Economic Priorities, the information provided by the company over and above its annual report, the extent to which the company went beyond minimum statutory disclosure requirements and the comments provided on the rst version of NCG CSR proles. Similar to other studies in this area, content analysis that included the quantity (the number of pages) and quality of information related to CSR in their annual reports was also used (Adams et al., 1991, p. 58). (A ve point scale was produced: 2 = well below average; 1 = below average; 0 = average; 1 = above average; 2 = well above average.) (2) The extent to which a company encourages the advancement of women. This rating was based on extent to which a company provides maternity leave,
career breaks, crche facilities and job share schemes. Other factors taken into account were: the recruitment, training and promotion of women and the proportion of women at different levels of management especially at senior management level. (Four point scale: 1 = below average, 0 = average, 1 = above average, 2 = well above average.) (3) The extent to which a company encourages the placement and advancement of ethnic minorities. This rating was based on the existence of well-designed ethnic monitoring and equal opportunities policies, membership of the Fullemploy organisation, specic initiatives and the number of ethnic minorities found at senior managerial positions. Fullemploy is a UK, London-based organisation founded in the 1970s which aims to help ethnic minorities, especially young blacks, in career advancement. (Three point scale 1 = below average, 0 = average, 1 = above average.) (4) Philanthropy or charitable giving and involvement to community projects. As well as charitable donations as a proportion of pre-tax prots, this rating included factors such as secondment of staff, gifts in kind and facilities provided to staff to engage in charitable activities. In addition, membership of Business in the community and Per cent club were also taken into account. Business in the community is a partnership of 250 companies and 30 (governmental, trade unions and/or voluntary) organisations. Its main aim is to promote community action by the private sector. Similarly Per cent club is another UK organisation that promotes charitable donations. To qualify for membership a rm should contribute more than 0.5 per cent of its pre-tax prots to the community. (Three point scale 1 = below average 0 = average, 1 = above average.) (5) Environmental action refers to initiatives the company is undertaking to reduce its environmental impact or improve its environmental protection performance. In addition, whether or not a rm has a written environmental policy and/or an environmental ofce/ofcer was taken into account. (Three point scale: 0 = none, 1 = some environmental action, 2 = concerted environmental action.) 31
(6) Whether the company has made a donation to the British political parties in the 1986-1990 period. As none of the examined rms has donated to the Labour or other parties, only donations to the Conservative Party were examined. (A yes/no scale was used.) Information on the size of donations provided was not complete for all rms and thus not usable. (7) Subscription to the Economic League, a blacklisting organisation that sells information to employers on individuals regarded as potential subversives. (A yes/no scale was used.) (8) The extent to which the companys activities have a signicant effect on the environment. This was more of an industry-based classication of a rms impact on the environment, disregarding the rm-specic efforts to alleviate it. A four point scale was used, where: 3 = industries with major environmental impact (chemical, oil and mining industries), 2 = industries with signicant environmental impact (clothing, pesticides, electrical goods, pharmaceuticals, agricultural goods and car manufacturing rms), 1 = industries with above average impact (tobacco, fast food, soft drinks and brewing companies), and 0 = industries with average impact (all the other industries and sectors). As this rating refers to the industrys rather than actual rms activities, it was used in the research as intervening or classicatory variable rather than a measure of CSR. (9) Respect for life (animal cruelty) refers to animal testing, conditions under which animals are reared for food, shing techniques used and genetic engineering. As ratings were available for only half of the sampled rms, this index was excluded from the study. (10) Respect for people refers to a companys involvement in the manufacturing and sales of alcohol and tobacco products as well as gambling. As this indicator is conned to specic businesses (e.g. brewery and tobacco industries), it was excluded from the study. (11) Doing business with oppressive regimes. The degree (depth) of a rms business involvement with South Africa was used as a proxy. However, this measure could not be used across the whole range of rms. Some of them had already
signicant investments in South Africa before its international condemnation and thus divestment could have been more difcult for them. In addition, if this indicator was used as a CSR measure, those rms who by chance or choice do not trade internationally would be appear misleadingly to be gratuitously credited with social responsibility. (12) Production and/or sales of military equipment. As this rating was only relevant to defence related industries, it could not be applied to any representative sample of rms. (13) Business relationships with the least developed countries. In this instance, NCG rated rms on the basis of their depth of business involvement with third world countries, disregarding their nature and thereby whether their actions could be regarded as socially responsible or not. For more information on what factors were included to produce the ratings of each category see Adams et al. (1991, pp. 5-54). A further advantage of the CSR measurement used is that it provides different ratings for CSR performance and disclosure. As NCG did not rely exclusively on information (disclosed or) emanating from rms themselves to assess CSR performance, a differentiation between CSR disclosure and CSR performance is operationally possible. Ullmann (1985) emphasised that a mistake common to studies of this kind is that researchers do not differentiate between the two. There is an obvious and important difference between what companies do and what they say they do (or did). Firms have a variety of reasons to misrepresent or even conceal their actual CSR performance. In particular, there is a tendency to overstate their CSR performance in order to create a positive impression as a part of their PR efforts. Inversely, according to Ullmann (1985), rms with poor CSR record (or with high CSR expenses) tend to under-report this information. This is because some of the CSR activities might be at the expense of other investments which were more protable and closer to shareholders interests. Thus, an interference between CSR disclosure and CSR performance seems to exist and needs to be checked. In this study, CSR disclosure is measured by the rst dimension in the list. CSR 32
performance is captured mainly by the following four indicators: womens position, ethnic minorities position, philanthropy and environmental actions. The content of these indicators covers all the CSR dimensions of the modern rms as identied by Vyarkarnam (1992); Trotman and Bradley (1981); Anderson and Frankle (1980) and Abbott and Monsen (1979) . Donations to the Conservative Party and participation in the Economic League due to their dichotomous nature will be used as complementary measures. Finally, the effect on the environment rating, as mentioned earlier, is used as an intervening variable. Periods of analysis The set of corporate social responsibility measures available related to the period 19881989. Based on the above period of time, economic performance data were divided into three periods: 1984-1987 (pre-assessment period); 1988-1989 (concurrent period); and 1990-1994 (post-assessment period). Economic performance measures were averaged for each period. However, information on donations to the Conservative Party and subscription to the Economic League are related to the 1986-1990 period (this was to cover party donations during the 1987 election period). Therefore, these are related to economic performance data for the periods: 1984-1985 (pre-assessment period); 19861990 (concurrent period); and 1991-1994 (post-assessment period). Measures of economic performance and hypotheses This study followed the precedent of most of the previous studies on the topic and used accounting data to measure nancial performance and capital market data to measure systematic risk and excess market valuation (capital market performance). For the purpose of this study, the term economic performance will be used to describe both nancial and capital market performance. The data were obtained from the Datastream on-line database. The variables used were: (1) three accounting based measures (Return on Capital Employed, ROCE; Return on Equity, ROE; and the ratio of gross prot to sales, GPS); and (2) two capital market based (Excess market valuation, EMV and Beta a measure of a rms systematic risk).
The ratios of return on capital employed and the return on equity measure the relative efciency of asset utilisation. A major strength of the ROCE ratio is that it is free from the effects of bias that can result from differences in capital structure between rms. However, both ROCE and ROE can be distorted by the effect of ination on the book value of the assets. For example, two rms may have identical physical assets that they purchased at different times. If so, the book value of these assets, in all likelihood, will be different, reecting the effect of ination on nominal asset prices. The third accounting-based measure to be used is the gross prot to sales ratio. Its main advantage is that it is free from leverage differences but it is handicapped in the sense that it is subject to ination distortions and, unlike the other measures, fails to capture the relative effectiveness of the use of assets. All three measures will be used in a complementary fashion so as to compensate for their disadvantages. As sustained growth in nancial performance is a primary goal for most managers, trends in accounting-based measures are frequently used in evaluating the performance of management. Given that in periods of low protability economic demands may have priority over discretionary social responsibility expenditures, satisfactory nancial performance may have a denite inuence on the level of support top corporate decision makers can commit to future social responsibility activities (Ullmann, 1985). Based on this argument, it can be suggested that a positive association should exist between past nancial performance and the level of CSR performance (Cornell and Shapiro, 1987; McGuire et al., 1988). H1a: The higher the level of past nancial performance, the higher the level of CSR performance. Conversely, in cases of low nancial performance, rms may be less inclined to disclose information on expenditures for CSR activities, as it might seem to be difcult to justify this redirection of prots towards CSR activities to shareholders (Ullmann, 1985). Thus: H1b: The higher the level of past nancial performance, the higher the level of CSR disclosure. As has been mentioned earlier, literature on the effects of CSR on (concurrent and subsequent) nancial performance (FP) is inconclusive. Both theoretical argument and prior 33
empirical work can be used to support the argument that CSR can have either a negative or a positive impact on nancial performance. The main arguments for a positive relationship are that CSR is an indicator of managements ability to deal effectively and proactively with stakeholders demands (and/or possible threats which if they materialised would have an adverse impact on rms future FP). Satisfying stakeholders explicit and implicit demands, gaining their support and/or averting potential threats is thought to improve rms FP. CSRs impact on FP is realised by either: increasing employee productivity and morale or reduction of potential labour problems; gaining consumers goodwill and support for its products; stopping possible actions by lobby groups against the rm and the resulting bad publicity; preventing government ofcials imposing costly regulation; and/or facilitating the ow of capital to the rm by improving its standing with bankers and investors (McGuire et al., 1988; Spicer, 1978a,b). H2a: The higher the level of a rms CSR performance, the higher its concurrent and subsequent nancial performance. Communication or disclosure of CSR performance information to stakeholders is necessary for these effects to take place and for stakeholders to fully appreciate a rms social contribution. Conversely, withholding of CSR information and secrecy may cultivate suspicion and/or distrust to stakeholders which can be translated to lower nancial performance. Put formally as a hypothesis this is: H2b: The higher the level of a rms CSR disclosure, the higher its concurrent and subsequent nancial performance. The argument against this view is that CSR consumes resources and can be costly, putting the rm at an economic disadvantage compared to its competitors. CSR activities such as charitable donations, support to community projects, installation of environmental protection equipment, etc. can be seen as carrying a signicant cost element. Also, emphasis on CSR may distract resources from more economically protable uses (e.g. investing in South Africa, before its democratisation) and reduce a rms range of strategic alternatives. Based on the above, hypothesis H3 is formulated: H3: The higher the level of a rms involvement in costly CSR activities, the lower
its concurrent and subsequent nancial performance. Information on the exact costs of each CSR performance variable and disclosure is not available. In addition, one has to recognise the difculty of estimating and attributing costs to each of the CSR activities. However, based on the information available from NCG, philanthropy and environmental action seem to be comparatively more costly than the other CSR activities examined in this study. CSR performance in hypothesis H2a includes both costly and other CSR performance indicators. The conict between hypotheses H2a and H3 with reference to the direction of relationship of the cost-adding CSR activities, is a reection of their contradictory theoretical underpinnings (Ullmann, 1985). Accounting-based measures are not free of limitations. They tap only historical aspects of a companys nancial performance. They can also be subject to managerial manipulation and differences in accounting procedures make comparability across rms and studies problematic (McGuire et al., 1986). To overcome these difculties, two stock-marketbased measures of performance were also used. Ullmann (1985) argues that these are more appropriate to studies of this type, as they capture investors evaluation of a rm. Given this debate about the proper measure of a companys nancial performance, the use of capital-market performance measures along with nancial performance measures compensates for any possible measurement weaknesses. The capital-market measures used were beta (a measure of a rms systematic risk) and excess market valuation. Systematic risk (beta) is dened as the covariance between returns on a risky asset (e.g. a corporations common stock) and market portfolio, divided by the variance of the market portfolio (Copeland and Weston, 1983). Corporations that have low measures of systematic risk are expected to have higher levels of CSR activities. Corporations with low systematic risk have a more stable pattern of stock market returns. Given that such economic considerations are central to corporate decision makers, stable market performance should enhance their discretion to commit resources in CSR activities in the subsequent periods (Roberts, 1992). H4a: The lower a rms past systematic risk or beta, the higher its CSR performance. 34
Similarly, rms with stable market performance would feel less tempted to conceal information about investments in CSR activities from their shareholders and investors, even if these activities were not evidently prot-generating or economically justied. Thus: H4b: The lower a rms past systematic risk or beta, the higher its CSR disclosure. Research also suggests that CSR activities may improve a rms access to capital and increase employee morale and productivity (McGuire et al., 1988; Moskowitz, 1972). Based on this information, market participants may view socially responsible rms as being better managed and less risky. A high degree of CSR may also permit a rm to have lower variance in its economic performance as a function of more stable relations with the government and the nancial community in general (McGuire at al., 1988; Spicer, 1978a,b). Compared to other rms, rms with high CSR may have lower total market risk as they should be less sensitive to certain external events, like governmental action in the eld, and thereby have a lower liability. However, it can be argued that the impact of CSR on a rms systematic risk may be minimal. Most events affecting one rms commitment to social responsibility do not systematically affect all other rms in the marketplace (Cornell and Shapiro, 1987). Based on this argument rms with high CSR performance are postulated to exhibit more stable market performance patterns (lower betas). H5a: The higher a rms level of CSR performance, the lower the concurrent and/or subsequent systematic risk or beta. The same arguments as above hold for the relationship between disclosure and beta. Disclosure of social responsibility activities is expected to have similar results as in many cases it is used to improve market evaluations (Ullmann, 1985). Thus, H5b: The higher a rms level of CSR disclosure, the lower the concurrent and/or subsequent systematic risk or beta. Excess market valuation (EMV) is dened as the difference between total rm market value and the book value of assets divided by sales. This measure captures the value premiums or discounts accorded by the market to various companies(Cochran and Wood, 1984, p. 50). If corporate social responsibility creates positive or negative market expectations about
the rms prospects, then it will be related to a high or low excess value. However, a limitation of this measure is that it is subject to ination distortion of asset values. Similar to beta, this measure is a reection of market evaluation for a rm and its activities. The arguments put forward for the relation between systematic risk and CSR apply to EMV as well, as the underlying logic is the same. Favourable past market responses, as manifested in high EMV, will relieve some of the pressure on corporate decision makers to invest solely in activities that evidently generate prots in shorter periods of time than CSR investments. Thus, H6a: The higher a rms past EMV the higher its CSR performance. Similarly, decision makers under such circumstances (of favourable EMV) will be less inhibited to disclose information about potentially non-economically justiable and costly CSR activities to their investors and shareholders. Thus, H6b: The higher a rms past EMV the higher its CSR disclosure. The same factors affecting the beta CSR performance/disclosure relationship (i.e. improved access to capital, higher employee morale and productivity and more stable relationships with government and nancial institutions) inuence positively the capital markets reactions to rms investing in CSR activities. As a result such rms are postulated to exhibit higher EMVs in the concurrent and subsequent periods. Thus, H7a: The higher a rms level of CSR performance, the lower the concurrent and/or subsequent EMV. H7b: The higher a rms level of CSR disclosure, the lower the concurrent and/or subsequent EMV. Controlling for spurious relationships Before testing the hypothesis, it is necessary to control for intervening variables which might produce spurious relationships and/or obscure any key effects. The nexus of the postulated relationships between the main and intervening variables is given in Figure 1. The main focus of the study is on link 2 (hypotheses H1a, H1b, H4a, H4b, H6a and H6b), and links 3 and 4 (hypotheses H2a, H2b, H3, H5a, H5b, H7a and H7b). Another link of interest is link 1, that depicts the relationship between CSR performance and disclosure. As these variables 35
affect each other, their effects on each other will be isolated and controlled. An additional control was on the carry-over effects of economic performance from one period to the next. Links 13 and 14 depict these relationships. It was important to isolate these effects to control for the possibility of spurious relationships resulting from this link. A rms size and industry classication are commonly used as control variables in studies of this type. However, as rms in the sample are dispersed over more than 20 (2-digit SIC coded) industries and given the small size of the sample, it was difcult to statistically control for such effects. To overcome this problem given this constraint, the NCG classication of industries in terms of their impact on the environment was used instead. Industries with a signicant and probably more visible impact on the physical environment are less likely to avoid the attention and scrutiny of the public. Thus, it is reasonable to assume that rms in these sectors will feel more pressure to improve their CSR standing. In addition, visibility of their damage to the environment may affect negatively stakeholders behaviour towards the rm and as a consequence their nancial and capitalmarket performance. The relationship of environmental impact with CSR and economic performance is depicted with links 9, 10, 11 and 12 in Figure 1. Past empirical research has shown that a rms size affects economic performance and
Figure 1 The nexus of relationships between economic performance, CSR disclosure and CSR performance
CSR Disclosure
(1) Past financial performance (2) Subsequent financial (4) performance
CSR Performance
womens position ethnic minority philanthropy environmental action
(13)
(3) Concurrent financial performance (5) (7) (10)
(14)
(8)
(12)
(6) (9)
(11)
Firms Size
Environmental impact of industry
CSR performance and disclosure (Chen and Metcalf, 1980; Spicer, 1978b). According to Ullmann (1985, p. 548) larger companies are subject to more public scrutiny and are more likely to have the necessary nancial, managerial and technical know-how to undertake costly CSR activities than their smaller counterparts. An additional or alternative argument is that larger companies, being signicant taxpayers and employers, have a higher bargaining power that allows them to obtain concessions from legislators and/or enforcement agencies regarding the imposition of and compliance to CSR standards. A rms size then might inuence both voluntary and mandatory requirements for CSR behaviour. As a rms size is also postulated to affect both economic performance (links 6, 7 and 8 in Figure 1) and CSR (link 5 in Figure 1) it needs to be controlled. Turnover was used to measure a rms size.
T-test analyses failed to uncover any statistically signicant (p 0.05) differences between Conservative Party donors and nondonors, in terms of CSR disclosure or any of the CSR performance variables. However, the same was not true for membership of the Economic League. While the sample was small, a t-test showed that members of the EL rated signicantly (at p 0.05) lower than EL non-members in both womens position (mean difference = 0.583, at p 0.025) and ethnic minorities position (mean difference = 0.354, at p = 0.035) in their organisation. ANOVA was also used to check for joint effects (of Conservative Party donation and EL membership) on CSR performance and disclosure. No such effects were identied at p 0.05. The relationship between rms size, environmental impact of industry, corporate social responsibility performance and disclosure As can be seen in Table I, the two control variables, rms size and environmental impact, are inter-correlated. This can be explained by the fact that the industries with signicant environmental impact (chemicals, mineral extraction, etc.) all require a high level of investment and generally are large in size. Firms size also was found to be signicantly (at p 0.001) correlated to disclosure and two CSR performance measures, philanthropy and environmental action (see Table I). Surprisingly, correlation analysis shows that environmental impact of a rms industry is not statistically related to the rms activities to alleviate it (i.e. the environmental action measure). One would have expected that rms in industries with higher levels of environmental damage would be more oriented towards activities designed to alleviate this, than companies in other sectors, apparently, this was not so. T-test analysis also showed that membership of the Economic League and donations to the Conservative Party are not statistically related either to the rms size or the environmental impact of the industry. Similarly, either environmental impact or rms size was found to be statistically correlated (at p 0.05) to any of the economic performance measures employed in all three periods of analysis. 36
Findings
Interrelationships among corporate social responsibility measures Principal component analysis (roots criterion) of the four CSR performance variables (environmental action, womens position, ethnic minorities position and philanthropy) identied one factor (explaining 42.3 per cent of the total variance). That conrms the unidimensionality of all four variables which seem to be manifestations of the same underlying factor, that is CSR performance as dened here. Factor scores will be used as a proxy of the overall CSR performance which will be examined together with the four CSR performance variables. Examining the relationship between disclosure and CSR performance (link 1 in Figure 1), regression analysis showed a quite strong relationship. In particular, CSR performance variables can explain 36 per cent (R2 = 0.363) of CSR disclosure. As expected, rms with good performance in terms of CSR are more inclined to disclose more. Twenty-one out of the 56 (37.5 per cent) sampled rms had donated to the Conservative Party, whereas only eight (14.3 per cent) were members of the Economic League (EL), of which only four were contributors to the Conservative Party as well. A chi square test showed that contributions to the Conservative Party were not statistically related to EL membership.
Table I (Two-tail) correlation between CSR disclosures, CSR performance, environmental impact and rms size
1 1 2 3 4 5 6 Disclosure Womens position Ethnic minorities position Philanthropy Environmental action CSR performance (factor scores) 7 Impact on the environment 8 Firms size
0.512**1 0.283*2 0.264*4 0.178 0.188 0.105 0.464**1 0.404**5 0.134
0.230
0.579**1 0.773**1 0.520**1 0.534**1 0.732**1 0.240 0.155 0.139 0.127 0.228 0.343*3 0.084 0.089 0.318*6 0.354**7
0.111 0.237
0.476**1
Note: N = 56, *p 0.05, **p 0.01, [1p = 0.000, 2p = 0.034, 3p = 0.012, 4p = 0.049, 5p = 0.002, 6p = 0.018, 7p = 0.008] Relationships between past, concurrent and subsequent economic performance and the impact of recession Before we embark on to examining the impact of CSR on nancial and capital-market performance, it might be useful to examine how nancial performance measures are interrelated at different periods under study. As expected, correlation analysis showed that economic performance is positively related to the next period of time performance. As nancial and capital-market performance were found to be strongly inter-correlated, principal component analysis (PCA) was used to identify the underlying factors. PCA (roots criterion) extracted a single factor for each set of variables for the past, concurrent and subsequent periods of study. Factor scores were used as overall measures of a rms nancial and capital market performance respectively. The uctuation of the nancial and capitalmarket performance of the sampled rms during the 1984-1994 period was also examined, so as to assess the impact of the business cycle (recession). Figure 2 illustrates the changes in the annual average (arithmetic mean) of all economic ( nancial and capitalmarket) performance variables during this period. As can be seen, there is no clear pattern of a cyclical movement for any of the measures. In order to statistically assess the impact of recession in the British economy during the period 1987-1994, pairwise t-test analysis was used to check for any signicant (at p 0.05) differences between the 19841986 (pre-recession period) and 1987-1994 (recession period). Only one statistically signicant difference (p 0.02) was detected in regard to the return on equity (ROE) measures. Surprisingly, the average ROE in the 37 1984-1986 (pre-recession) period was 16.11 (standard deviation: 8.18) whereas during the 1987-1994 (recession) period it was 20.82 (standard deviation: 13.08) higher than that of the pre-recession period. No other statistically signicant differences were identied. It seems that recession had no statistically identiable effects on the economic performance indicators of the sampled rms. The relationship between past economic performance, CSR performance and disclosure Table II displays the (two-tail) semi-partial correlation matrix of CSR disclosure, CSR performance and past economic performance measures. Possible effects resulting from a rms size and its industry environmental impact (links 5 and 11 in Figure 1) were controlled for (partialled out). Similarly disclosures effects on CSR performance (link 1, in Figure 1) and vice versa were partialled out. In both instances, the unstandardised regression residuals of the main variable against the control ones were used. Philanthropy was found to be positively related to both excess market valuation and gross prot to sales ratio. It seems that rms with higher pre-tax (gross) prots and gains in market value feel more comfortable in investing in community projects and charitable donations. This partly supports hypotheses H1a and H6a. However, ndings (in Table II) do not support hypotheses H4a and H4b (relationship between beta and CSR performance and disclosure, respectively). Of interest was the relation of past nancial performance and donations to the Conservative Party and Economic League. Logistic regression analysis was used to identify a potential relationship. Forward stepwise
Figure 2 Average (mean) performance of sampled rms in the period 1984-1994 25
20
performance fail to predict the probability of a company donating to the Conservative Party or being a member of the Economic League. Similarly, t-test analysis conrmed that there were no signicant differences in any of the past economic performance indicators between contributors and non-contributors to the Conservative Party and members or nonmembers of the EL, respectively. The relationship between CSR performance and concurrent and subsequent economic performance Semi-partial correlation analysis similar to the one used earlier (see Table III) revealed only one statistically signicant relationship between contemporaneous economic performance and CSR disclosure. Gross prot to sales ratio was found to be positively related to CSR disclosure (at p 0.004). The controls applied in this case included: rms size, environmental impact and past economic performance (links 7, 10 and 13 in Figure 1). The interaction effects between CSR performance and CSR disclosure were also partialled out of each other. No other statistically signicant relationships (at p 0.05) were identied. In addition, hierarchical regression analysis (see Table IV) was used to conrm and assess the CSR performance and disclosure (after controlling for rms size and environmental impact) effects on concurrent economic performance (free of the effects of past economic performance). Variables entered the regression equation at three consecutive steps. At the rst step the control variables, rms size and environmental impact, at the second step CSR disclosure and at the third step the four CSR performance indicators were entered. At each step changes between the
15
10
0 84 Key Return on Equity Beta-systematic risk ROCE Excess Market valuation Gross Profit to sales 85 86 87 88 89 90 91 92 93 94
variable selection and the likelihood-ratio test were used to select the predictor variables for the regression equations (at p 0.05). Firms size and industrys environmental impact were also included in the equation as control variables. None of the past economic performance or control variables entered any of the regression equations. Thus it would seem, that past nancial and capital-market
Table II (Two-tail) semi-partial correlation analysis between past economic performance and social corporate responsibility controlling for rms size and industrys environmental impact
Women positiona Beta systematic risk Excess market value Capital market performance (factor) Gross prot to sales ratio Return on capital employed Return to equity Financial performance (factor) 0.022 0.231 0.130 0.022 0.190 0.088 0.111
Ethnic minoritya 0.067 0.187 0.074 0.250 0.079 0.086 0.151
Philanthropya 0.003 0.332*1 0.209 0.384*2 0.176 0.145 0.260
Environmental CSR actiona performancea Disclosureb 0.007 0.155 0.092 0.164 0.174 0.278 0.137 0.009 0.246 0.159 0.037 0.215 0.206 0.166 0.017 0.109 0.057 0.149 0.184 0.128 0.181
Note: N = 56, *p 0.05; a: variables controlled for disclosure; b: controlled for CSR performance [1p = 0.031, 2p = 0.012] 38
consecutive R2s (R2s) were estimated. R2s together with standardised regression coefcients (betas) were used to check the size of the independent variables effects on the concurrent economic performance. Variance ination factors and condition indices detected no collinearity effects among the independent variables. The condition index for the lowest eigenvalue was 5.08, well below the conventional standard (10). As can be seen in Table IV, the analysis conrms the semipartial correlation analysis results. CSR disclosure was found to explain around 5 per cent of the variance in the concurrent GPS ratio (R2 = 0.05). This provides partial 2 support for hypothesis H2b. However, hypotheses H2a and H3 were not supported by the ndings. In order to check the interactive (joint) effect of disclosure and CSR performance on
concurrent economic performance, moderated (or multiplicative) multiple regression analysis is used. In this analysis an additional multiplicative term (disclosure CSR performance) is created to encompass the interaction effect. The formed regression equation is as follows: Concurrent economic performance = a + b1 disclosure + b2 CSR performance + b3 (disclosure CSR performance). where: a is the intercept (constant) and b1, b2 and b3 the regression coefcients. Past economic performance, rms size and environmental impact effects were all partialled out from each of the concurrent economic performance measures used. Standardised scores of disclosure were formed before the formation of the multiplicative term so as to be measured in the same units as
Table III (Two-tail) semi-partial correlation analysis between concurrent economic performance and social corporate responsibility controlling for rms size, industrys environmental impact and past economic performance
Women positiona Beta systematic risk Excess market value Capital market performance (factor) Gross prot to sales ratio Return on capital employed Return to equity Financial performance (factor) 0.066 0.034 0.063 0.318 0.242 0.134 0.007
Philanthropya 0.026 0.026 0.040 0.169 0.107 0.020 0.044
Environmental CSR actiona performancea Disclosureb 0.070 0.070 0.124 0.318 0.210 0.009 0.068 0.079 0.012 0.055 0.263 0.135 0.099 0.039 0.039 0.213 0.1814 0.340**1 0.104 0.130 0.045
Note: N = 56, **p 0.01; a: variables controlled for disclosure; b: controlled for CSR performance [1p = 0.004]
Table IV Hierarchical regression analysis of concurrent economic performance against disclosure and CSR performance
Independent variables Size of the rm Environmental impact 2 R1 Disclosure 2 R2 Women position Ethnic minority Philanthropy Environmental action 2 R3 Final R2
Systematic risk, betaa 0.064 0.3681 0.073 0.012 0.002 0.229 0.154 0.076 0.173 0.045 0.120
EMVa 0.275 0.035 0.109 0.245 0.058 0.061 0.045 0.022 0.002 0.003 0.170
Dependent variables (concurrent period) Market GPS responsibilitya ratioa ROCEa 0.157 0.188 0.073 0.192 0.042 0.212 0.133 0.055 0.091 0.025 0.140 0.4042 0.278 0.100 0.448*3 0.050 0.4134 0.051 0.230 0.039 0.125 0.275*5 0.298 0.017 0.052 0.065 0.001 0.338 0.234 0.150 0.124 0.120 0.173
ROEa 0.370 0.082 0.061 0.087 0.040 0.037 0.128 0.104 0.016 0.022 0.123
Financial performancea 0.242 0.153 0.041 0.004 0.005 0.040 0.184 0.048 0.029 0.025 0.072
Note: a: past economic performance effects have been partialled out N = 56, *p 0.05, **p 0.01 [1p = 0.062, 2p = 0.061, 3p = 0.049, 4p = 0.076, 5p = 0.05] 39
the CSR performance. For parsimony reasons, the overall CSR performance factor scores for each rm were used instead of the raw scores for each CSR performance indicator. According to Cohen and Cohen (1983) the existence of an interaction effect can be identied by checking for (statistically signicant) changes of the R2 (R2) when the multiplicative term enters the equation. Thus, a two step hierarchical regression analysis was undertaken, CSR performance and disclosure enter the equation at the rst step and the multiplicative term at the second. If an interaction effect is present then the difference between the two R2s values would be statistically signicant (p 0.05) (Cohen and Cohen, 1983). The size of the interaction effect is determined by the size of the R2. Following this procedure, an interaction effect was identied in two instances. First, the CSR performance disclosure term had a signicant effect on concurrent ROE (controlled for past economic performance, rms size and environmental impact of industry effects). The regression equation was as follows: Concurrent ROE = 0.42 + 0.40 disclosure 0.023 CSR performance + 0.62 interaction (nal R2 = 0.35, R2 = 0.34, signicant at p = 0.001). Second, a similar effect was revealed for concurrent nancial performance (factor scores) proxy (free of the intervening variable impact). The second equation was: Concurrent nancial performance = 0.32 + 0.22 disclosure 0.03 CSR performance + 0.47 interaction (nal R2 = 0.18, R2 = 0.17 signicant at p = 0.006). Condition indices showed very low collinearity effects (the condition index for the lowest eigenvalue was 3.4 for both equations). The size of the interaction effect (R2) in the rst equation is larger than that in the second one. Jaccard et al.s (1991) instructions for interpretation of equations of this type were followed. Accordingly, both equations showed that rms that combine high disclosure and high CSR performance (and rms that combine low disclosure and low CSR performance) perform better in terms of ROE (and nancial performance in the concurrent period) than others, namely, rms that combine low CSR performance with high disclosure and the ones with high CSR performance and low disclosure. Interpretation of the 40
equations revealed that, in general, rms that combine high CSR performance and high disclosure outperform the ones that combine low CSR performance and low disclosure (in terms of contemporaneous ROE and nancial performance factor scores). In the case of subsequent performance, some statistically signicant relationships were identied (at p 0.05). In line with the previous section, a semi-partial correlation analysis was used (controlling for rms size and environmental impact of the industry, links 8 and 12 in Figure 1). Additionally, possible concurrent economic performance effects on subsequent economic performance were partialled out (link 14 in Figure 1). Results (in Table V) indicate that environmental initiatives are negatively related to a rms subsequent ROCE, whereas CSR disclosure was found to be positively related to subsequent GPS. It seems that costly CSR activities such as environmental action have a negative carry-over effect on subsequent nancial performance. However, disclosure seems to have exactly the opposite results on subsequent nancial performance. This seems to provide some support for hypothesis H3. Results also showed (Table V) that improvements in womens position get noticed by the capital markets in a favourable way in the subsequent period. As can be seen in Table V, capital-market performance factor scores in the subsequent period were positively related to improvements in womens position. Hierarchical multiple regression analysis (the same procedure as before) was employed again to assess relationships of CSR performance and disclosure with subsequent economic performance (Table VI). The collinearity effects in the equation were very low, the recorded condition index for the lowest eigenvalue was 5.07. As can be seen in Table VI, disclosure was found to have a negative effect on subsequent EMV. However, as R2 shows 2 the size of this effect is very small, it explains only 2.1 per cent of the subsequent EMVs variance. On the contrary, capital markets seem to be more positively disposed to improvements in womens position (R2 = 3 0.153). Unlike capital-market performance, disclosure seems to have a signicant effect on subsequent nancial performance. In particular, disclosure was found to be positively related to both GPS and nancial performance (factor scores). The changes in R2 values (R2 s) were 0.186 and 0.136 respectively. It 2
Table V (Two-tail) semi-partial correlation analysis between subsequent economic performance, disclosure and CSR performance controlling for rms size and industrys environmental impact
Women positiona Beta systematic riskc Excess market valuec Capital market performance (factor)c Gross prot to sales ratioc Return on capital employedc Return to equityc Financial performance (factor)c 0.120 0.282 0.315*1 0.054 0.132 0.137 0.110
Philanthropya 0.019 0.0966 0.104 0.144 0.229 0.005 0.151
Environmental CSR actiona performancea Disclosureb 0.068 0.069 0.103 0.063 0.342*2 0.036 0.179 0.032 0.1855 0.176 0.123 0.124 0.156 0.016 0.015 0.233 0.206 0.305*3 0.262 0.087 0.2874
Note: N = 56, *p 0.05; a: variables controlled for disclosure; b: controlled for CSR performance; c: controlled for concurrent economic performance [1p = 0.037, 2p = 0.023, 3p = 0.044, 4p = 0.059] seems that disclosure has a favourable effect on the other stakeholders whose contributions are necessary for improved nancial performance (e.g. employees, customers, etc.) but to have adverse effects on the capital market participants. Hierarchical regression analysis (Table VI) also conrms the environmental actions negative relationship with subsequent ROCE. Overall, CSR performance variables seem to explain 17.2 per cent (R2 = 0.172) of subse3 quent ROCE variance. This provides additional support for hypothesis H3. Moderated or multiplicative multiple regression analysis (using the same procedure as in the case of concurrent economic performance) failed to identify the existence of any signicant (disclosure CSR performance) interaction effects on any of the subsequent economic performance variables. The impact of the two dichotomous variables (donations to the Conservative Party and participation in the Economic League) and their interaction (joint impact) on both concurrent and post-assessment period (subsequent) performance were also examined. Past economic performance carry-over effects on concurrent economic performance were partialled out. In a similar manner concurrent performance effects were partialled out of subsequent economic performance variables so as to avoid the possibility of any spurious relationships. Analysis of covariance (ANCOVA) was used to uncover any statistically signicant (p 0.05) relationships of the two dichotomous
Table VI Hierarchical regression analysis of subsequent economic performance against disclosure and CSR performance
Independent variables Size of the rm Environmental impact 2 R1 Disclosure 2 R2 Women position Ethnic minority Philanthropy Environmental action 2 R3 Final R2
Systematic risk, betaa 0.004 0.369*1 0.083 0.013 0.001 0.212 0.100 0.069 0.094 0.026 0.110
EMVa 0.160 0.043 0.132 0.426*2 0.021 0.3763 0.154 0.063 0.192 0.160 0.277*4
Dependent variables (concurrent period) Market GPS responsibilitya ratioa ROCEa 0.131 0.194 0.103 0.3725 0.012 0.482*6 0.067 0.011 0.255 0.153 0.2697 0.114 0.183 0.061 0.513*8 0.186 0.111 0.078 0.134 0.003 0.033 0.220 0.159 0.220 0.080 0.3139 0.090 0.205 0.153 0.241 0.390*10 0.172 0.348**11
ROEa 0.34812 0.415*13 0.151 0.117 0.055 0.044 0.185 0.006 0.045 0.033 0.239
Financial performancea 0.30914 0.348*15 0.122 0.363*16 0.136 0.101 0.134 0.170 0.172 0.061 0.320*17
Note: a: concurrent economic performance effects have been partialled out N = 56, *p 0.05, **p 0.01 [1p = 0.045, 2p = 0.031, 3p = 0.052, 4p = 0.049, 5p = 0.059, 6p = 0.014, 7p = 0.059, 8p = 0.012, 9p = 0.088, 10p = 0.014, 11p = 0.009, 12p = 0.077, 13p = 0.019, 14p = 0.098, 15p = 0.037, 16p = 0.049, 17p = 0.025] 41
variables on economic performance. Firms size and environmental impact were used as covariates. ANCOVA results showed that members of the EL have signicantly (at p = 0.028) lower subsequent ROCE (controlled for concurrent economic performance effects) (mean = 0.49) than non-members (mean = 0.11). However, the size of the effect attributed to (or the variability explained by) EL membership (partial eta-squared = 0.126) was relatively small (12.6 per cent). Similarly, membership of the EL had a signicant (at p 0.006) impact on subsequent excess market valuation. Members of EL displayed a lower average EMV (mean = 0.72) than nonmembers (mean = 0.14). However, the size of the EL membership effect on EMV (partial eta-squared = 0.165) was comparatively higher than that on ROCE. No signicant differences in the (concurrent and subsequent) economic performance of donors and non-donors to the Conservative Party were identied. A t-test analysis was also performed to check independently the differences in the means between donors and non-donors to the Conservative Party and between subscribers and non-subscribers to the Economic League. It was (t-test analysis) also conrmed that members of the Economic League have a signicantly (at p 0.02) lower EMV (concurrent economic performance effects have been partialled out) for the subsequent period 1991-1994 (mean = 0.716) compared to non-members (mean = 0.135). Overall, it appears that donations to the ruling (Conservative) party have no effect on any of the concurrent and subsequent economic performance variables. However, membership of the Economic League has a negative effect on subsequent ROCE and EMV.
Conclusions
The results of the empirical tests are of interest for a number of reasons. First, the major theories concerning social responsibility which had been developed in the North American context were tested within the distinct cultural and economic environment of the UK. Second, a multi-dimensional measure of social corporate responsibility performance was used, which allowed for a more comprehensive measurement to be undertaken. The results of the empirical research supported only a few of the postulated relationships between CSR disclosure and 42
CSR performance with past, concurrent or subsequent economic performance. It seems that past nancial performance can explain variation in certain elements of corporate social responsibility. Specically, philanthropic activity seems to be affected by gross prot to sales ratio and excess market valuation in the past. This seems to agree to a certain extent with the ndings of McGuire et al. (1988). Interestingly enough donations to the Conservative Party or membership of the Economic League were found not to be particularly related to companys past economic performance. However, CSR disclosure was found to be associated with concurrent nancial performance. In particular, gross prot to sales ratio was found to affect disclosure positively. A combination of high CSR performance and high disclosure was also found have positive effects on rms overall protability. In contrast, low CSR disclosure combined with good CSR performance or high CSR disclosure combined with poor CSR performance were found not to be economically-rewarding strategies. Even poor CSR performance accompanied by low level of disclosure was found to be a better strategy (than the other combinations) in the short term (concurrent period). However, this effect seems to fade away in the longer run (i.e. subsequent periods). Another important point is the impact of CSR activities with a signicant cost element. In particular environment care related activities assumed to have a higher cost were found to be negatively related to subsequent nancial performance (ROCE). On the other hand less costly CSR activities, like the enhancement of womens corporate position were found to have a positive but not instantly realisable (as it was detected only in the subsequent period) effect on capital markets. In contrast, the reaction of the capital markets in the subsequent period to companies with high CSR disclosure was found to be negative. Paradoxically, CSR disclosure was found to be more positively received by other stakeholders responsible for a rms nancial performance (GPS). Again, disclosure effects seem not to materialise immediately as they could only be detected in the subsequent period. Donations to the Conservative Party were found to have no effect on any of the concurrent and subsequent economic performance variables. However, this was not true for
membership of the Economic League. Members of the Economic League were found to have lower nancial (ROCE) and lower capital market performance (EMV) in the subsequent assessment period than non-members. The hypothesis of the ethical investor (that capital markets tend to reward socially responsible rms) is not necessarily empirically supported by this research. Quite the opposite, ndings suggested that the capital market seems to be rather indifferent to rms that undertake some CSR activities. Even more surprisingly the degree to which a rm discloses CSR information had a negative effect on capital market participants (postulated to be one of the main motives and targets of disclosure). Overall, the ndings seem to suggest that other factors than CSR concerns were more important determinants of investors behaviour. Obviously the relationship is more complex than has been hitherto suggested. Future research should examine the impact of CSR concerns in conjunction with other factors which might affect decision making and expectations formation of different segments of investors. Overall, the results of this study (weakness of the identied relationships) imply that additional factors need to be taken into account to explain higher proportions of the economic performance CSR performance/disclosure relationship. Further research should then concentrate on the stakeholder model of management (Clarkson, 1995; Freeman, 1984; Jones, 1995) and in addition to investors (capital-market participants), other specic groups such as customers and suppliers could be included directly in the empirical tests. The social context within which CSR behaviour is moulded and subsequently rewarded or punished by the interested parties (stakeholders) is very important. Development of proper operationalisations of the social pressures and universe surrounding different rms would improve future research designs and unravel new, richer contingencies affecting the main relationship. In addition, the lack of systematic CSR measurement for other groups of companies imposes certain limitations on the generalisability of the ndings (beyond the largest British rms). The inclusion of medium-sized rms in the future might improve the research design. 43
References
Abbott, W. and Monsen, R. (1979), On the measurement of corporate social responsibility: self-reported disclosure as a measure of corporate social involvement, Academy of Management Journal, Vol. 22, pp. 501-15. Adams R., Carruthers, J. and Hamil, S. (1991), Changing Corporate Values, Kogan Page, London. Alexander, G. and Bulcholz, R. (1978), Corporate social responsibility and stock market performance, Academy of Management Journal, Vol. 21, pp. 47986. Anderson, J. and Frankle, A. (1980), Voluntary social reporting: an iso-beta portfolio analysis, Accounting Review, Vol. 55, pp. 468-79. Angelidis, P. and Ibrahim, N. (1993), Social demand and corporate supply: a corporate social responsibility model, Review of Business, Vol. 15, Fall, pp. 7-10. Belkaoui, A. (1976), The impact of the disclosure of the environmental effects of organization behaviour on the market, Financial Management, Vol. 5 No. 4, pp. 6-31. Bowen, H.R. (1953), Social Responsibilities of the Businessman, Harper & Row, New York, NY. Carroll, A.B. (1979), A three-dimensional model of corporate performance, Academy of Management Review, Vol. 4 No. 4, pp. 497-505. Chen, K. and Metcalf, R. (1980), The relationship between pollution control record and nancial indicators revisited, Accounting Review, Vol. 55, pp. 168-77. Chugh, E. (1978), Impact of pollution control regulations on the market risk of securities in the US, Journal of Economic Studies, Vol. 16 No. 2, p. 67. Clarkson, M.B.E. (1995), A stakeholder framework for analysing and evaluating social corporate performance, Academy of Management Review, Vol. 20 No. 1, pp. 92-117. Cochran, P. and Wood, R. (1984), Corporate social responsibility and nancial performance, Academy of Management Journal, Vol. 27 No. 1, pp. 42-56. Cohen, J. and Cohen, P. (1983), Applied Multiple Regression for the Behavioral Sciences, Laurence Erlbaum, Hillsdale, NJ. Copeland, T. and Weston, J. (1983), Financial Theory and Corporate Policy, Addison-Wesley, Reading, MA. Cornell, B. and Shapiro, A. (1987), Corporate stakeholders and corporate nance, Financial Management, Vol. 16, pp. 5-14. Davis, K. (1975), Five propositions for social responsibility, Business Horizons, Vol. 18 No. 3, pp. 19-24. Donaldson, T. (1983), Constructing a social contract for business in Donaldson, T. and Werhane, P. (Eds), Ethical Issues In Business, Prentice-Hall, Englewood Cliffs, NJ, pp. 153-65. Epstein, E. (1987a), The corporate social policy process and the process of corporate governance, American Business Law Journal, Vol. 25, pp. 361-83.
Epstein, E. (1987b), The corporate social policy process: beyond business ethics, corporate social responsibility and corporate social responsiveness, California Management Review, Vol. 29 No. 3, pp. 99-114. Frankle, A. and Anderson, J. (1978), The impact of the disclosure of environmental effects of organizational behaviour on the market: comment, Financial Management, Vol. 21, Summer, pp. 92-107. Frederick, W.C. et al. (1992), Business and Society, McGraw-Hill International, New York, NY. Freeman, R.E. (1984), Strategic Management : A Stakeholder Approach, Pitman, Boston, MA. Friedman, M. (1970), The social responsibility of business is to increase its prots, New York Times Magazine, 13 September, pp. 33+. Graves, S.B. and Waddock, S.A. (1994), Institutional owners and corporate social performance, Academy of Management Journal, Vol. 37 No. 4, pp. 103446. Ingram, R. (1978), An investigation of the information content of (certain) social responsibility disclosures, Journal of Accounting Research, Vol. 16, pp. 270-85. Jaccard, J., Turrisi, R. and Wan, C.K. (1991), Interaction Effects in Multiple Regression, Sage, London. Jones, T. (1983), An integrating framework for research in business and society: a step toward the elusive paradigm?, Academy of Management Review, Vol. 8, pp. 559-64. Jones, T.M. (1995), Instrumental stakeholder theory a synthesis of ethics and economics, Academy of Management Review, Vol. 20 No. 2, pp. 404-37. Lodge, G.C. (1977), The connection between ethics and ideology in Hoffman, M.W. (Ed.), Proceedings of the First National Conference on Business Ethics, The Center for Business Ethics, Bentley College, Weltham, MA. McGuire, J., Schneeweis, T. and Hill, J. (1986), An analysis of alternative measures of strategic performance in Lamb, R. and Stravasta, P. (Eds), Advances In Strategic Management, Vol. 4, JAI Press, Greenwich, CO, pp. 107-53. McGuire, J., Sundgren, A. and Schneeweis, T. (1988), Corporate social responsibility and rm nancial performance, Academy of Management Journal, Vol. 31 No. 4, pp. 854-72. Mahapatra, S. (1984), Investor reaction to corporation social accounting, Journal of Business and Finance and Accounting, Vol. 11, pp. 29-40. Mills, D. and Gardner, M. (1984), Financial proles and the disclosure of expenditure for socially responsible purposes, Journal of Business Research, Vol. 14 No. 4, pp. 407-24. Moskowitz, M. (1972), Choosing socially responsible stocks, Business and Society, Vol. 1, pp. 71-5.
Moskowitz, M. (1975), Proles in corporate social responsibility, Business and Society, Vol. 13, pp. 29-42. Parert, I.R. and Eilbirt, H. (1975), Social responsibility: the underlying factors, Business Horizons, Vol. 18 No. 4, pp. 5-10. Pava, M.L. and Krausz, J. (1996), The association between corporate social-responsibility and nancial performance the paradox of social cost, Journal of Business Ethics, Vol. 15 No. 3, pp. 321-57. Roberts, R. (1992), Determinants of corporate social responsibility disclosure an application of stakeholder theory, Accounting Organizations and Society, Vol. 17 No. 6, pp. 595-612. Rosen, B.N., Sandler, D.M. and Shani, D. (1991), Socialissues and socially responsible investment behavior a preliminary empirical-investigation, Journal of Consumer Affairs, Vol. 25 No. 2, pp. 221-34. Schwartz, J. (1975), Reforming the corporation from within, in Heilbroner, R.L. and London, P. (Eds), Corporate Social Policy: Selections from Business and Society Review, Addison-Wesley, Reading, MA, pp. 12-18. Shane, P. and Spicer, B. (1983), Market response to environmental information produced outside the rm, Accounting Review, Vol. 58, pp. 521-38. Spicer, B. (1978a), Investors, corporate social performance and information disclosure: an empirical study, Accounting Review, Vol. 53, pp. 94-111. Spicer, B. (1978b), Market risk, accounting data and companies pollution control records, Journal of Business Finance and Accounting, Vol. 5, pp. 67-83. Spicer, B. (1980), The relationship between pollution control record and nancial indicators revisited: further comment, Accounting Review, Vol. 55, pp. 178-85. Trotman, K. and Bradley, G. (1981), Association between social responsibility disclosure and characteristics of companies, Accounting Organizations and Society, Vol. 6, pp. 355-62. Tuzzolino, F. and Armandi, B. (1981), A need hierarchy framework for assessing corporate social responsibility, Academy of Management Review, Vol. 6 No. 1, pp. 21-8. Ullmann, A. (1985), Data in search of a theory: a critical examination of the relationships among social performance, social disclosure, and economic performance of US rms, Academy of Management Review, Vol. 10 No. 3, pp. 540-57. Vyakarnam, S. (1992), Social responsibility: what leading companies do, Long Range Planning, Vol. 25 No. 5, pp. 59-67. Wartick, S. and Cochran, P. (1985), The evolution of the corporate social performance model, Academy of Management Review, Vol. 10 No. 4, pp. 758-69. Wood, D. (1991), Corporate social performance revisited, Academy of Management Review, Vol. 16 No. 4, pp. 758-69.
44

10 1 1 39

Hochgeladen von

Dokumentinformationen

Originalbeschreibung:

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

10 1 1 39

Hochgeladen von

Copyright:

Verfügbare Formate

Corporate social responsibility and economic performance in the top British companies: are they linked?

George Balabanis Hugh C. Phillips and Jonathan Lyall

Corporate social responsibility and economic performance

European Business Review Volume 98 Number 1 1998 2544

George Balabanis, Hugh C. Phillips and Jonathan Lyall

Corporate social responsibility and economic performance

European Business Review Volume 98 Number 1 1998 2544

George Balabanis, Hugh C. Phillips and Jonathan Lyall

The relationship between corporate social responsibility and economic performance

Corporate social responsibility and economic performance

European Business Review Volume 98 Number 1 1998 2544

George Balabanis, Hugh C. Phillips and Jonathan Lyall

Prior empirical research

Corporate social responsibility and economic performance

European Business Review Volume 98 Number 1 1998 2544

George Balabanis, Hugh C. Phillips and Jonathan Lyall

Corporate social responsibility and economic performance

European Business Review Volume 98 Number 1 1998 2544

George Balabanis, Hugh C. Phillips and Jonathan Lyall

Corporate social responsibility and economic performance

European Business Review Volume 98 Number 1 1998 2544

George Balabanis, Hugh C. Phillips and Jonathan Lyall

Corporate social responsibility and economic performance

European Business Review Volume 98 Number 1 1998 2544

George Balabanis, Hugh C. Phillips and Jonathan Lyall

Corporate social responsibility and economic performance

European Business Review Volume 98 Number 1 1998 2544

George Balabanis, Hugh C. Phillips and Jonathan Lyall

Corporate social responsibility and economic performance

European Business Review Volume 98 Number 1 1998 2544

George Balabanis, Hugh C. Phillips and Jonathan Lyall

Corporate social responsibility and economic performance

European Business Review Volume 98 Number 1 1998 2544

George Balabanis, Hugh C. Phillips and Jonathan Lyall

(3) Concurrent financial performance (5) (7) (10)

Environmental impact of industry

Corporate social responsibility and economic performance

European Business Review Volume 98 Number 1 1998 2544

George Balabanis, Hugh C. Phillips and Jonathan Lyall

Corporate social responsibility and economic performance

European Business Review Volume 98 Number 1 1998 2544

George Balabanis, Hugh C. Phillips and Jonathan Lyall

0.512**1 0.283*2 0.264*4 0.178 0.188 0.105 0.464**1 0.404**5 0.134

Corporate social responsibility and economic performance

European Business Review Volume 98 Number 1 1998 2544

George Balabanis, Hugh C. Phillips and Jonathan Lyall

Figure 2 Average (mean) performance of sampled rms in the period 1984-1994 25

Ethnic minoritya 0.067 0.187 0.074 0.250 0.079 0.086 0.151

Philanthropya 0.003 0.332*1 0.209 0.384*2 0.176 0.145 0.260

Corporate social responsibility and economic performance

European Business Review Volume 98 Number 1 1998 2544

George Balabanis, Hugh C. Phillips and Jonathan Lyall

Ethnic minoritya 0.045 0.045 0.023 0.054 0.121 0.099 0.156

Philanthropya 0.026 0.026 0.040 0.169 0.107 0.020 0.044

Corporate social responsibility and economic performance

European Business Review Volume 98 Number 1 1998 2544

George Balabanis, Hugh C. Phillips and Jonathan Lyall

Corporate social responsibility and economic performance

European Business Review Volume 98 Number 1 1998 2544

George Balabanis, Hugh C. Phillips and Jonathan Lyall

Ethnic minoritya 0.004 0.158 0.124 0.047 0.195 0.234 0.229

Philanthropya 0.019 0.0966 0.104 0.144 0.229 0.005 0.151

0.512**1 0.2832 0.2644 0.178 0.188 0.105 0.4641 0.4045 0.134

Philanthropya 0.003 0.3321 0.209 0.3842 0.176 0.145 0.260