This paper discusses the formation process of CSR images from a customer perspective. It analyses the influence of company-CSR coherence, motivational attribution and corporate credibility in the way customers evaluate CSR images in the banking industry. It also describes the impact of customer gender, age and educational level on the formation of CSR images. Results show that CSR images are based on customer perceptions of the company-CSR coherence, the attribution of altruistic motivations and corporate credibility when developing CSR initiatives. The findings also demonstrate that gender, age and educational level do not allow identifying differences in the way customers construct CSR images. Thus, they are not useful in segmenting customers for the design of better CSR and communication strategies.
The Corporate Social Responsibility (CSR) image refers to stakeholder perceptions of corporate responses to the general social concerns of stakeholder groups (Lai et al., 2010; Pérez and Rodríguez del Bosque, 2013). It is a concept that is gaining attention gradually both in the academic and professional literature because it improves intangible corporate resources, such as brand equity and reputation (Sen and Bhattacharya, 2001). The CSR image as perceived by customers has begun to attract the interest of scholars and practitioners because it is believed to boost customer-related benefits, such as satisfaction, repurchase intentions and customer willingness to recommend the company to other consumers.
When studying this concept, marketing scholars have begun by focusing on the impact of the CSR image on customer behaviour (Brown and Dacin, 1997; Sen and Bhattacharya, 2001; García de los Salmones et al., 2005; Matute et al., 2010; Tian et al., 2011). However, few studies so far have attempted to understand the process by which customers construct CSR images (Rifon et al., 2004; Becker-Olsen et al., 2006; Ellen et al., 2006; Bigné et al., 2009), and scholars have yet to present an integrative conceptual model that tests the formation process of CSR images in the real context of companies marketing services. For example, Rifon et al. (2004) provide the first integrative model to understand the formation process of CSR image. However, they use a laboratory experiment and they only test the adequacy of their model for companies marketing tangible products. Subsequent scholars such as Becker-Olsen et al. (2006), Ellen et al. (2006) or Bigné et al. (2009) only test and confirm parts of the seminal paper proposed by Rifon et al. (2004). It is also especially significant that the few papers analysing the formation process of the CSR image have given no attention to the role of customers’ demographic characteristics when these stakeholders construct their perceptions. Nevertheless, in other streams of research, scholars have demonstrated that customers with different genders, ages and educational characteristics have different orientations towards CSR (Diamantopoulos et al., 2003; Sharma et al., 2012; Gonçalves and Sampaio, 2012; Samarasinghe, 2012) and that they process CSR information differently (Quazi, 2003). For example, women and better-educated customers seem to have a greater CSR orientation than men and less-educated customers, while older customers are more sceptical than younger people when evaluating CSR (Serwinek, 1992; Quazi, 2003). These ideas allow the authors to anticipate significant effects of demographic characteristics on the way customers construct CSR images.
The gaps identified in previous literature limit the currently available knowledge of CSR from a customer perspective and hamper the generation of effective managerial strategies based on economic, social and environmental concerns (Rifon et al., 2004). Based on these ideas, the authors of this paper describe and test a reliable conceptual model to understand the way customers construct their perceptions of the CSR image of services companies in Spain. The authors also test the model on six customer samples with diverse genders, ages and educational characteristics to identify whether significant differences exist in the way customers construct CSR images. These analyses will generate great value for managers and businesses that can use the findings to design more effective CSR and communication strategies based on customer segmentation.
The paper begins by proposing a conceptual model to analyse the formation process of the CSR image among customers. Second, the authors discuss the role of customer demographic characteristics (gender, age and educational level) in that process. Several research hypotheses are presented. The fifth section of the paper shows the methodology applied to the study, which is based on structural equation modelling and multisampling tests. The authors describe the findings and the paper ends with a discussion of the most significant conclusions, managerial implications, limitations and future lines of research derived from the study.
Literature reviewCSR and CSR image: definitions and relevance for companiesCorporate social responsibility (CSR) is conceptualised as the construct describing the relationship between companies and society (D’Aprile and Mannarini, 2012). More precisely, van Marrewijk (2003) defines it as all company activities demonstrating the inclusion of social, environmental and economic concerns in business operations, and in interactions with stakeholders. As presented by D’Aprile and Mannarini (2012), to overcome critical challenges, companies have to rethink their role in society and consider themselves as socio-economic agents contributing to the human, civic and social progress of the community as a whole. Thus, this concept has gained much attention in the last decades because CSR investments have been demonstrated to lead to the recovery of corporate credibility in product and company crises (Lin et al., 2011), the improvement of employee attraction and retention (Kim and Park, 2011) or the establishment of beneficial relationships with customers and other primary stakeholders (Matute et al., 2010; Peloza and Shang, 2011). Based on these ideas, many scholars have recently focused on the study of CSR, especially from a management perspective and trying to determine how CSR investments and performance imply larger benefits for companies (Melo and Garrido, 2012).
However, less attention has been given to the study of customer perceptions of CSR (García de los Salmones et al., 2005; Matute et al., 2010; Tian et al., 2011; Stanaland et al., 2011; Pérez and Rodríguez del Bosque, 2013). Nonetheless, customer's expectations and opinions are considered to directly influence the design and effectiveness of corporate strategies and as so their study is essential in the academic and business fields (Fukukawa et al., 2007). In this regard, CSR image refers to customer perceptions of corporate responses to the general social concerns of stakeholder groups (Lai et al., 2010; Pérez and Rodríguez del Bosque, 2013).
Aligning with this definition, scholars have considered that the stakeholder theory (Freeman, 1984) allows academics and practitioners to conceptualise CSR image from a comprehensive and more real perspective than other theories adopted by many researchers in marketing, such as the sustainable development or the pyramidal model proposed by Carroll (Maignan and Ferrell, 2004). Stakeholder theory is also useful to understand CSR image in almost any research context (Pérez and Rodríguez del Bosque, 2013). Freeman (1984) defines stakeholders as those groups or individuals who can affect or are affected by the achievement of the company's objectives or are those actors with a direct or indirect interest in the company. The stakeholder theory describes CSR as the obligations companies have towards these groups (Maignan and Ferrell, 2004) and CSR image as customer subjective perceptions of the performance of companies concerning these obligations (Pérez and Rodríguez del Bosque, 2013). Based on this idea, Pérez et al. (2013) suggest a dimensioning of the CSR image according to the most influential audiences of companies, named customers, shareholders, employees and the society. These scholars also include a fifth dimension that comprises general legal and ethical responsibilities of companies in their interactions with stakeholders. These ideas are supported by previous relevant scholars (e.g., Maignan and Ferrell, 2004; Turker, 2009) and they will be used in this paper to measure CSR image in the context of a service industry.
The formation process of CSR images among customersIn this section, the authors propose a conceptual model that will allow researchers and managers to anticipate the formation process of CSR images. The model considers three direct antecedents of the CSR image: company-CSR coherence, motivational attribution and corporate credibility. The authors draw on the seminal paper of Rifon et al. (2004), who propose a causal model that explains the effect of company-CSR coherence, the attribution of corporate motivations to engage in CSR and corporate credibility on customer attitudes towards the CSR implemented by companies. The results of their study confirm the value of a company's involvement in sponsoring CSR initiatives that are perceived by customers as: (1) consistent with core business activities and products (company-CSR coherence), (2) altruistic in nature (motivational attribution) and (3) credible (corporate credibility).
Nonetheless, the authors identify three limitations in this paper that justify further study along this line of research. First, the study is based on a lab experiment, so the perceptions of customers in a real market context are not evaluated. Secondly, Rifon et al. (2004) only study the reliability of the model in relation to the marketing of tangible products (Avon, Philip Morris, Reebok and Ortho), so it is still unknown if their results are applicable to the marketing of CSR in service industries. Further research has aligned with the proposal of Rifon et al. (2004) and scholars have tried to confirm their ideas in other real research contexts (Becker-Olsen et al., 2006; Ellen et al., 2006; Bigné et al., 2009). All these scholars confirm the findings of Rifon et al. (2004), although they have only tested the original model partially. For example, neither Ellen et al. (2006) nor Becker-Olsen et al. (2006) include corporate credibility as an antecedent of the CSR image. Similarly, Bigné et al. (2009) do not test the role of the CSR image in the model as they only test the relationships between company-CSR coherence, motivational attribution and corporate credibility. Finally, the third limitation of the paper of Rifon et al. (2004) is the fact that feasible moderating effects on the formation process of the CSR image have been ignored. For example, scholars have demonstrated that customer demographic characteristics, such as gender, age or educational level, determine CSR knowledge, attitudes and behaviour (Diamantopoulos et al., 2003; Sharma et al., 2012; Gonçalves and Sampaio, 2012; Samarasinghe, 2012). Thus, it is logical to expect that demographic characteristics also influence the way customers form CSR images, although this role has not been defined yet.
Thus, these flaws that the authors have identified in previous literature justify further study on the way customers form CSR images. Based on both theoretical and empirical evidence, in this paper the aims of the authors are twofold: (1) to corroborate the results of Rifon et al. (2004) in the context of a Spanish service industry (and taking into consideration all the four variables initially considered by these scholars) and (2) to extend their findings by examining the role of three demographic characteristics of customers (gender, age and educational level) on the way they form CSR images.
Research hypothesesAs far as the first goal of the authors is concerned, the model proposed by Rifon et al. (2004) starts by proposing a direct relationship between company-CSR coherence and motivational attribution. In this regard, company-CSR coherence refers to customer perceptions of the similarity between the mission and goals of the company and the needs of its CSR initiatives and/or partners. The schema (Lynch and Schuler, 1994) and associative learning (Till and Nowak, 2000) theories traditionally explain the effects of company-CSR coherence on customer cognitive processes. According to these theories, learning is a mechanism by which customers establish relationships among concepts to produce an associative network in their memories. Through this network, customers learn about a concept through its relationship to other nodes in their memory. In the field of CSR, these stimuli are social causes, non-profit organisations or institutions related to the promotion of CSR in the business arena. Thus, these theories point to improved corporate images as companies develop CSR initiatives consistent with their personalities. Nonetheless, Rifon et al. (2004) indicate that important elements of the schema and associative learning theories remain unclear. The study of what motivates an organisation to engage in CSR can bring greater clarity to these approaches. Motivational attributions are the result of a cognitive process by which customers assign a cause to an observed situation. Following the virtue ethics perspective, van de Ven (2008) suggests that CSR initiatives can only be considered virtuous if they satisfy an altruistic motivation of the company rather than an egoistic interest. Thus, only the attribution of altruistic motivations generates positive responses from the customer.
According to these ideas, there is a need for scholars to consider both company-CSR coherence and motivational attribution together, thus providing greater explanatory power for models designed to explain the formation process of CSR images. First, there is the possibility that company-CSR coherence will have a direct effect on motivational attribution (Rifon et al., 2004; Becker-Olsen et al., 2006; Ellen et al., 2006). Following the ideas proposed by the schema and associative learning theories, Rifon et al. (2004) empirically confirm that company-CSR coherence positively influences customer perceptions of corporate altruistic motivations. Ellen et al. (2006) also suggest that a greater compatibility between the goals of a CSR initiative and the core business of a company can make the allocation of strategic, value-based motivations more feasible, while simultaneously decreasing the perception of corporate egoistic motivations. Based on these ideas, the first research hypothesis is as follows:H1 Company-CSR coherence directly and positively influences the motivations attributed to the company in the development of its CSR.
Furthermore, company-CSR coherence can influence corporate credibility. In this regard, credibility refers to the degree to which customers, investors and other constituents believe in corporate expertise and trustworthiness when developing CSR initiatives (Goldsmith et al., 2000). One can argue that if customers perceive that CSR initiatives are consistent with corporate personality, CSR initiatives will be perceived as credible because the company-CSR link is easily understood and integrated in their mental schemas. In contrast, if CSR initiatives are not consistent with the identity the company is trying to convey, the effort customers have to devote to understanding CSR will reduce corporate credibility (Bigné et al., 2009).
Along this line, attribution theory (Folkes, 1988) describes customers as rational information processors whose behaviour is influenced by the causal inferences they make about events they observe and experience. In terms of CSR, the theory proposes that altruistic motivations can form the basis of corporate credibility and customer attitudes towards companies. Rifon et al. (2004) show that the more altruistic motivations customers perceive in companies engaging in CSR, the greater their evaluation of corporate credibility. Similarly, Becker-Olsen et al. (2006) demonstrate that the allocation of altruistic motivations may counteract the negative effect of low company-CSR coherence by positively impacting on corporate credibility. Based on these ideas, two new research hypotheses propose that:H2 Company-CSR coherence directly and positively influences corporate credibility. The motivations attributed to the company in the development of its CSR directly and positively influence corporate credibility.
It is also noted that corporate credibility is an essential factor in building brand value through CSR (Hoeffler and Keller, 2002). Along this line of thought, recent literature points to the significant role played by companies (as sources of messages) when determining customer responses to business strategies and communication processes (Kim and Choi, 2007). The persuasion (Hovland et al., 1953), cognitive response (Greenwald, 1968) and reasoned action (Fishbein and Ajzen, 1975) theories propose that an expert and reliable source of information is more persuasive than a source perceived as less reliable (Lafferty and Goldsmith, 1999). For example, Lafferty and Goldsmith (1999) confirm that higher levels of corporate credibility cause customers to construct positive attitudes towards companies. Similarly, Goldsmith et al. (2000) confirm that customers exposed to an ad from a highly credible company develop more positive attitudes towards the brand. In the CSR field, Hoeffler and Keller (2002) state that brand credibility is a necessary antecedent to building brand value through social marketing. The theoretical and empirical evidence provided by these studies allows the authors to suggest the following hypothesis:H4 Corporate credibility directly and positively influences CSR image.
For the following hypotheses, the authors also drew from the associative learning and attribution theories. First, the associative learning theory assumes that high company-CSR coherence improves customer attitudes towards companies becauseit leads them to believe that the corporate behaviour is adequate (Till and Nowak, 2000). Therefore, a good fit between the company and its CSR initiatives will be more easily integrated into customer associative networks, strengthening the perception of a positive relationship between the company and its CSR initiatives (Fiske and Taylor, 2008). In turn, company-CSR coherence strengthens the market position of the company, which is important in helping customers understand the adequacy of the company within its competitive environment, providing brand differentiation, reducing uncertainty and increasing purchase intentions (Becker-Olsen et al., 2006).
Furthermore, Barone et al. (2000) demonstrate that customers positively perceive brands that show altruistic motivations in the support of CSR initiatives. For some CSR initiatives, however, companies might be perceived as the main beneficiaries and, therefore, exploitative. Along this line, Becker-Olsen et al. (2006) show that when a company's operations are guided by egoistic motivations, their CSR initiatives will generate a greater number of unfavourable perceptions in the minds of customers. These thoughts lead customers to question corporate motivations, and these negative attributions ultimately reduce purchase intentions. The opposite occurs when customers perceive altruistic corporate motivations. In line with these ideas, two new research hypotheses propose that:H5 Company-CSR coherence directly and positively influences CSR image. The motivations attributed to the company in the development of its CSR directly and positively influences CSR image.
The second goal of the authors in this paper is to extend the findings of Rifon et al. (2004) by examining the role of three demographic characteristics of customers (gender, age and educational level) on the way they form CSR images. In this regard, customer demographic characteristics have been extensively applied to the study of CSR knowledge, attitudes and behaviour (Diamantopoulos et al., 2003; Sharma et al., 2012; Gonçalves and Sampaio, 2012; Samarasinghe, 2012); therefore, demographic characteristics are commonly used for customer segmentation (Roberts, 1996; Dietz et al., 2002). Among these features, customer gender (Arlow, 1991; Burton and Hegarty, 1999; Quazi, 2003; Samarasinghe, 2012), age (Serwinek, 1992; Quazi, 2003; Samarasinghe, 2012) and educational level (Anderson and Cunningham, 1972; Quazi, 2003; Samarasinghe, 2012) have been identified as the most significant characteristics influencing customer perceptions of CSR. Nonetheless, their study has yet to be applied to the formation process of the CSR image, leaving a lack of information regarding how customer CSR knowledge, attitudes and behaviour might influence the way in which customers construct CSR images.
As far as gender is concerned, scholars have demonstrated that men and women process information differently and that they use different informational clues to understand the stimuli they receive from companies or the media. For example, men appear as more analytical and logical, whereas women are characterised by their subjectivity and intuition. Furthermore, the so-called selectivity model (Meyers-Levy, 1989) proposes that men tend to be more selective when evaluating the available information about companies, products and services, while women tend to follow a more comprehensive strategy, attempting to assimilate all available information. Therefore, where men tend to focus their attention on the most accessible information about the tangible features of a specific topic, women appear to be more subjective because they use a greater number of heuristics that are, on occasion, more subtle, more tangential and less accessible than those used by men.
Scholars have also demonstrated that women have a greater CSR orientation in decision making (Arlow, 1991). For example, in their study of college students, Burton and Hegarty (1999) observe that women give special value to non-economic corporate endeavours such as philanthropy. In a study of customers, Roberts (1996) show that women also exhibit a greater CSR orientation than men. Dietz et al. (2002) confirm that women give greater importance to personal values, especially those related to the environment, when making purchase and consumption decisions. Along this line, Diamantopoulos et al. (2003) demonstrate that most studies investigating the link between gender and environmental knowledge, attitudes and behaviour conclude that males tend to have greater knowledge about green issues than females. However, they also conclude that females exhibit both greater concern and participate more frequently in various types of green behaviour (e.g., recycling or political action).
The greater CSR consciousness of women and the ideas proposed by the selectivity model are closely related to the formation process of the CSR image. Specifically, a greater CSR concern might make females more thoughtful when evaluating CSR initiatives. Thus, company-CSR coherence, motivational attribution and corporate credibility will have a greater impact on women's perceptions of the CSR image than men's perceptions. Based on these ideas,the authors propose the following hypothesis:H7 Customer gender moderates the formation process of CSR image. When compared to men, for women there is a stronger relationship between (a) company-CSR coherence and motivational attribution, (b) company-CSR coherence and corporate credibility, (c) motivational attribution and corporate credibility, (d) corporate credibility and CSR image, (e) company-CSR coherence and CSR image, (f) motivational attribution and CSR image.
Furthermore, an extensive line of research in marketing centres on studying the changes produced in cognitive processes as customers go through different chronological stages of their lives (Morris and Venkatesh, 2000). According to Morris and Venkatesh (2000), several scholars show that ageing is accompanied by a loss of intellectual ability to process complex and non-frequent information such as CSR-related data. Thus, younger customers are expected to evaluate more heuristics when constructing CSR images than older customers.
Scholars have also demonstrated that age is a clear predictor of CSR knowledge and the attitudes and behaviour of customers (Arlow, 1991). Most studies have shown that younger customers are more concerned about CSR than older customers and that they make greater efforts to behave accordingly (Anderson and Cunningham, 1972; Aldag and Jackson, 1977; Arlow, 1991; Diamantopoulos et al., 2003). For example, Arlow (1991) considers that age plays the most important role in customer CSR orientation, with the youngest people showing the greatest interest in CSR. Similarly, the study of Anderson and Cunningham (1972) shows that both younger customers and managers present clearer CSR orientations. Aldag and Jackson (1977) also confirm these results, contrasting the inverse relationship between age and CSR orientations: older customers are less concerned about CSR than younger customers.
Following the same reasoning proposed for customer gender, the authors believe the greater CSR consciousness of younger customers might result in their being more thoughtful than more mature or older customers when evaluating CSR initiatives. Thus, younger customers will place greater importance on company-CSR coherence, motivational attribution and corporate credibility when perceiving CSR images. That customers lose the intellectual ability to process information as they get older supports this belief. According to these ideas, it is proposed that all the relationships included in the conceptual model presented in this paper will be stronger among younger customers than among more mature customers. Consequently, the following hypothesis is proposed:H8 Customer age moderates the formation process of CSR image. The younger the customer, the stronger the relationship between (a) company-CSR coherence and motivational attribution, (b) company-CSR coherence and corporate credibility, (c) motivational attribution and corporate credibility, (d) corporate credibility and CSR image, (e) company-CSR coherence and CSR image, (f) motivational attribution and CSR image.
Finally, the authors analyse the effect of customers’ educational level on the formation process of the CSR image. Scholars attempting to characterise socially responsible customers based on their educational level have shown that these customers are better-educated than the mean (Roberts, 1996; Diamantopoulos et al., 2003). Roberts (1996) observe that CSR-conscious customers have prestigious job positions, high incomes and high educational levels. In addition, people with a higher degree, especially college degrees, develop a more elaborate perception of the implications of CSR for both companies and society (Quazi, 2003). They also exhibit a greater social orientation. In the particular CSR domain of environmental awareness, studies reporting a significant relationship between customer educational level and ecological knowledge, attitudes and behaviour have consistently reported that better-educated customers tend to score higher on all components of the environmental domain (Diamantopoulos et al., 2003). It is therefore suggested that customers with a higher educational level will understand the issues involved in CSR more fully. Therefore, these customers will be more concerned about CSR and will show a greater willingness to participate in more thoughtful processes to evaluate CSR images. Along this line of thought, the following hypothesis is proposed:H9 Customer educational level moderates the formation process of CSR image. When compared to less educated customers, for the better-educated there is a stronger relationship between (a) company-CSR coherence and motivational attribution, (b) company-CSR coherence and corporate credibility, (c) motivational attribution and corporate credibility, (d) corporate credibility and CSR image, (e) company-CSR coherence and CSR image, (f) motivational attribution and CSR image.
The authors conducted a study based on personal surveys administered to customers of banking services in Spain. Fifty interviewers were hired and subjected to a training programme on how to carry out personal surveys. The interviews were undertaken over one month in spring 2010. In the questionnaire, each customer had to evaluate his/her main banking service provider on CSR-related issues, such as company-CSR coherence, motivational attribution, corporate credibility and customer perceptions of the CSR initiatives implemented by banking institutions. Because all the banking institutions analysed by customers regularly undertook CSR initiatives, no fictitious information had to be generated for the study. This fact adds value to the research results because the findings in this paper refer to the real CSR images of Spanish banking institutions.
The Spanish banking industry was selected as the context for the research because of the difficulties it is facing in emerging from the international economic crisis. It is also an interesting research scenario because since the beginning of the crisis in 2007, society has devoted increasing attention to CSR in this sector (Pérez and Rodríguez del Bosque, 2012a). The crisis has led to the loss of society's confidence in banking institutions and an increased social conscience of stakeholders who now demand better tools for evaluating business practices. Thus, the authors of the paper anticipate that concepts such as company-CSR coherence, motivational attribution and corporate credibility might play a determining role in the formation of customer perceptions in the banking industry, different from their impact in other more robust industries. Nevertheless, CSR in the banking industry has been little analysed by scholars thus far (Pérez and Rodríguez del Bosque, 2012a), and the authors believe new and useful implications could be derived from a specific study of Spanish banking institutions.
SampleA non-probabilistic sampling procedure was used to design the research sample. With the purpose of guaranteeing a more accurate representation of the data, a multi-stage sampling by quotas was used based on customer gender and age. After the collection and processing of the information, a total of 1124 valid surveys remained (response rate=93.67%). The customers who did not answer the survey (76) claimed that they did not have enough knowledge about some aspects of the CSR of banking institutions to answer the questionnaire. They were mostly concerned about the CSR oriented to shareholders and employees. The remaining 1124 customers admitted to having enough information about their main banking institutions to give a confident answer to all the sentences in the questionnaire.
Table 1 shows the profile of the sample, which was segmented according to the categories of the moderating characteristics in this paper. Thus, customers were grouped by gender (men and women), age (18–45 and >45) and educational level (non-college and college). Following the ideas of Samarasinghe (2012), the authors chose 45 as the age cutting for the analyses because it has been demonstrated that this age allows researchers to identify the greatest differences in the CSR knowledge and awareness of customers. Customers who are in the 18–45 age category show more CSR knowledge and awareness than customers above 45, who do not usually check CSR information when taking purchase decisions (Samarasinghe, 2012). As far as the educational level is concerned, the authors chose to differentiate between customers with college or without college degrees because this is the most common classification used by previous scholars when evaluating CSR attitudes and knowledge (Roberts, 1996; Diamantopoulos et al., 2003). The sample was 48.67% male and 51.33% female, which was comparable to the population in the country (50.97% male and 49.03% female). Regarding age, 46.54% were customers between 18 and 45 (46.84% in the national population), while 53.46% were customers over 45(53.16% in the national population). The sample was slightly more educated than the population (36.57% vs. 13.55% of customers with college degrees, respectively).
Sample profile.
| Variable | Spanish population | Sample | ||
|---|---|---|---|---|
| N | % | N | % | |
| Gender | ||||
| Male | 18,903,405 | 49.03 | 547 | 48.67 |
| Female | 19,650,236 | 50.97 | 577 | 51.33 |
| 38,553,641 | 100 | 1124 | 100 | |
| Age | ||||
| 18–45 | 18,058,247 | 46.84 | 523 | 46.54 |
| >45 | 20,495,394 | 53.16 | 601 | 53.46 |
| 38,553,641 | 100 | 1124 | 100 | |
| Educational level | ||||
| Non-college | n.a. | 48.04 | 713 | 39.86 |
| College | 13.55 | 411 | 36.57 | |
| 100 | 1124 | 100 | ||
To address the research hypotheses, first the global data were used to test a Structural Equation Model (SEM) to determine how customers constructed CSR images based on company-CSR coherence, motivational attribution and corporate credibility. Fig. 1 displays the conceptual model tested in this paper. The SEM was examined using the statistical software EQS 6.1.
Seven-point Likert-type scales were used to measure all the concepts in the model (Table 2). First, the CSR image was measured using a scale specifically developed by the authors for the banking services context. This scale has already been tested in previous studies (Pérez and Rodríguez del Bosque, 2012b). The scale was based on stakeholder theory, which has been demonstrated to perfectly fit the banking industry approach to CSR (Pérez and Rodríguez del Bosque, 2012a). Twenty-two items were generated and gathered in five reflective dimensions: customers (items 1–5), shareholders and supervising boards (items 6–8), employees (items 9–13), society (items 14–19) and a general dimension concerning legal and ethical issues, which included corporate responsibilities towards a broad array of stakeholders (items 20–22). A detailed explanation of the scale's development can be found in (Pérez and Rodríguez del Bosque, 2012b). Corporate-CSR coherence, which refers to the symbolic similarity between corporate personalities and CSR initiatives, was measured using a four-item scale based on previous proposals by Lafferty et al. (2004) and Bigné et al. (2009). Motivational attribution was measured using a four-item scale oriented towards identifying customer perceptions of the egoistic or altruistic nature of banking institutions in developing CSR initiatives. The works of Becker-Olsen et al. (2006) and Bigné et al. (2009) were used as references when designing the scale. Finally, to measure corporate credibility the authors followed Newell and Goldsmith's (2001) proposal. These are the only researchers to date who have tested the content, construct, convergent and discriminant validity of their metric scale. Thus, the authors applied a four-item scale to evaluate corporate expertise and trustworthiness in developing CSR initiatives.
Measurement scales.
| Latent construct | Items |
|---|---|
| Company-CSR coherence | (1) Carrying out CSR activities is compatible with this institution's core business; (2) It makes sense that this institution carries out CSR activities; (3) Carrying out CSR activities is complementary to this institution's core business; (4) There is a logical fit between the core business of this institution and the CSR activities that it carries out |
| Motivational attribution | This company… |
| (1) Acts unselfishly; (2) Is altruistic; (3) Acts guided by the global benefit of its stakeholders instead of by its self-interest; (4) Is generous | |
| Corporate credibility | This company… |
| (1) Has a great expertise in corporate social responsibility; (2) Is competent in the implementation of its responsibilities towards its stakeholders; (3) Its commitment to its stakeholders is credible; (4) Is honest about its commitment to its stakeholders | |
| CSR image | This company… |
| CUSTOMERS: (1) Establishes procedures to comply with customers’ complaints; (2) Treats its customers honestly; (3) Has employees who offer complete information about corporate products/services to customers; (4) Uses customers’ satisfaction as an indicator to improve the product/service marketing; (5) Make an effort to know customers’ needs. | |
| SHAREHOLDERS AND SUPERVISING BOARDS: (6) Tries to maximise its profits; (7) Keep a strict control over its costs; (8) Tries to insure its survivals and long-term success. | |
| EMPLOYEES: (9) Pays fair salaries to its employees; (10) Offers safety at work to its employees; (11) Treats its employees fairly (without discrimination or abuses); (12) Offers training and career opportunities to its employees; (13) Offers a pleasant work environment (e.g., flexible hours, conciliation) | |
| SOCIETY: (14) Helps solving social problems; (15) Uses part of its budget for donations and social projects to advance the situation of the most unprivileged groups of the society; (16) Contributes money to cultural and social events (e.g., music, sports); (17) Plays a role in the society beyond the economic benefits generation; (18) Is concerned with improving the general well-being of society; (19) Is concerned with respecting and protecting the natural environment | |
| GENERAL: (20) Always respects rules and regulations defined by law; (21) Is concerned with fulfilling its obligations vis-à-vis its shareholders, suppliers, distributors and other agents with whom it deals; (22) Is committed to well established ethic principles |
Before proceeding to the analysis of the SEM, the reliability, validity and goodness of fit of the scales were globally tested to ensure that the model properly fitted the data. For this purpose, first the authors implemented two first- and second-order confirmatory factor analyses (CFA) to validate the multidimensionality of the CSR image scale. The results of these analyses are presented in Tables 3 and 4. The findings confirmed the reliability, convergent and discriminant validity of the scale. They also confirmed the multidimensionality theoretically proposed in this paper. Thus, it was demonstrated that customers perceived CSR as the collection of corporate actions as they related to customers, shareholders, employees, the society and a general dimension including legal and ethical concerns. Based on these results, the authors incorporated CSR image as a second-order construct in the global model proposed in this paper.
First-order CFA of the CSR image.
| Latent variable | Measured variable | Standardised lambda | R2 | Cronbach's α | AVE | Goodness of fit |
|---|---|---|---|---|---|---|
| Customers | CSR1 | 0.72 | 0.52 | 0.85 | 0.54 | S-Bχ2(199)=982.40(p=0.00)NFI=0.93NNFI=0.93CFI=0.94IFI=0.94 |
| CSR2 | 0.75 | 0.56 | ||||
| CSR3 | 0.71 | 0.50 | ||||
| CSR4 | 0.73 | 0.53 | ||||
| CSR5 | 0.75 | 0.56 | ||||
| Shareholders | CSR6 | 0.69 | 0.47 | 0.79 | 0.55 | |
| CSR7 | 0.72 | 0.52 | ||||
| CSR8 | 0.81 | 0.65 | ||||
| Employees | CSR9 | 0.74 | 0.55 | 0.88 | 0.59 | |
| CSR10 | 0.82 | 0.67 | ||||
| CSR11 | 0.82 | 0.68 | ||||
| CSR12 | 0.77 | 0.60 | ||||
| CSR13 | 0.67 | 0.45 | ||||
| Society | CSR14 | 0.77 | 0.59 | 0.89 | 0.57 | |
| CSR15 | 0.76 | 0.58 | ||||
| CSR16 | 0.73 | 0.54 | ||||
| CSR17 | 0.80 | 0.64 | ||||
| CSR18 | 0.79 | 0.63 | ||||
| CSR19 | 0.68 | 0.47 | ||||
| General | CSR20 | 0.75 | 0.56 | 0.82 | 0.60 | |
| CSR21 | 0.80 | 0.63 | ||||
| CSR22 | 0.77 | 0.60 |
| Discriminant validity | |||||
|---|---|---|---|---|---|
| Relationships | Correlation | Confidence interval | Relationships | Correlation | Confidence interval |
| Customers–shareholders | 0.52 | (0.44–0.60) | Shareholders–society | 0.32 | (0.24–0.40) |
| Customers–employees | 0.57 | (0.51–0.63) | Shareholders–general | 0.48 | (0.40–0.56) |
| Customers–society | 0.57 | (0.51–0.63) | Employees–society | 0.62 | (0.56–0.38) |
| Customers–general | 0.67 | (0.61–0.73) | Employees–general | 0.68 | (0.62–0.74) |
| Shareholders–employees | 0.44 | (0.36–0.52) | Society–general | 0.69 | (0.63–0.75) |
In this regard, the authors implemented a first-order CFA with all the latent constructs of the conceptual model and taking into consideration the global sample in the study. The CFA results are presented in Table 5. Although the Satorra-Bentler χ2 was significant (S-Bχ2(113)=820.14, p-value <0.05), because of the large sample used in the analysis (over 200 cases), all the Comparative Fit Indexes – NFI, NNFI, CFI and IFI – exceeded the minimum recommended value of 0.90, demonstrating an adequate model fit. In addition, the standardised lambdas obtained for company-CSR coherence, motivational attribution, corporate credibility and the CSR image were significant and greater than 0.50, ensuring the convergent validity of the model. Finally, the authors evaluated the discriminant validity of the factorial structure, estimating the confidence intervals for the correlation between pairs of constructs. The results verified the discriminant validity of the model because no confidence interval included the digit 1.
First-order CFA of the conceptual model.
| Latent variable | Measured variable | Standardised lambda | R2 | Cronbach's α | AVE | Goodness of fit |
|---|---|---|---|---|---|---|
| Company-CSR coherence | Coher1 | 0.84 | 0.71 | 0.89 | 0.67 | S-Bχ2(113)=820.14(p=0.00)NFI=0.93NNFI=0.93CFI=0.94IFI=0.94 |
| Coher2 | 0.88 | 0.77 | ||||
| Coher3 | 0.77 | 0.59 | ||||
| Coher4 | 0.77 | 0.60 | ||||
| Motivational attribution | Motiv1 | 0.84 | 0.71 | 0.82 | 0.54 | |
| Motiv2 | 0.75 | 0.56 | ||||
| Motiv3 | 0.53 | 0.28 | ||||
| Motiv4 | 0.78 | 0.61 | ||||
| Corporate credibility | Credi1 | 0.73 | 0.54 | 0.87 | 0.63 | |
| Credi2 | 0.82 | 0.68 | ||||
| Credi3 | 0.85 | 0.73 | ||||
| Credi4 | 0.82 | 0.67 | ||||
| CSR image | Customers | 0.70 | 0.49 | 0.84 | 0.51 | |
| Shareholders | 0.56 | 0.32 | ||||
| Employees | 0.72 | 0.52 | ||||
| Society | 0.76 | 0.57 | ||||
| General | 0.79 | 0.63 |
| Discriminant validity | |||||
|---|---|---|---|---|---|
| Relationships | Correlation | Confidence interval | Relationships | Correlation | Confidence interval |
| Coherence–attribution | 0.27 | (0.19–0.34) | Attribution–credibility | 0.57 | (0.52–0.63) |
| Coherence–credibility | 0.56 | (0.50–0.62) | Attribution–CSRImg | 0.46 | (0.40–0.52) |
| Coherence–CSRImg | 0.44 | (0.37–0.51) | Credibility–CSRImg | 0.54 | (0.49–0.60) |
In a second step of the analysis, the authors tested the moderating role of customer demographic characteristics in the formation process of the CSR image. For this purpose, three multisampling analyses were implemented to test the proposed SEM for each of the six subsamples of banking service customers previously described. First, the authors implemented the three multisampling analyses (one for gender, one for age and one for educational level) to obtain a multi-group solution of causal relationships. The purpose of this step was to determine the standardised coefficients of the six relationships of the model in each subsample. A further step was testing the factorial invariance of the SEM among the categories of customers in each multisampling analysis (men vs. women; 18–45 vs. >45; non-college vs. college). This step ensured that the latent constructs were understood in the same way among customers with different demographic characteristics so the model would be comparable among them. The factorial invariance was studied using the Lagrange Multiplier (LM) test, which allowed the authors to compare the chi-square differences of the relationships when eliminating the restriction of equality among the factorial lambdas in the diverse customer categories in each multisampling analysis. When the analysis results showed non-significant chi-square improvement values (p-value>0.05), the factorial invariance was confirmed. Finally, the last step consisted of estimating the structural invariance of the model among all the customer categories in each multisampling test. This property was evaluated by recalculating the proposed SEM to include the restriction that the standardised betas of the relationships among all the latent constructs were equal among customer categories. Again, the suitability of this restriction was determined using the LM test. This time, it was necessary that the chi-square differences be significant (p-value<0.05) to confirm that each customer demographic characteristic (gender, age and educational level) was a moderator of the relationship under scrutiny.
FindingsTest of the conceptual modelBefore discussing the SEM results, the authors present several descriptive statistics to understand how banking customers evaluated each of the constructs studied in this paper. The results are presented in Tables 6 and 7. These descriptive results showed that customer evaluated the company-CSR coherence (mean=5.22), corporate credibility (mean=4.62) and CSR image (mean=5.24) positively. They seemed to be more sceptical about the motives behind the implication of banking institutions with CSR goals (mean=3.71). As to what the dimensions of the CSR image were concerned, customers considered that the banking institutions were doing especially well in satisfying their shareholders and supervising boards (mean=5.42) and complying with legal and ethical responsibilities (mean=5.39). CSR initiatives oriented to the society were among the poorest evaluated dimensions (mean=4.92).
Descriptive statistics of the banking institutions under scrutiny.
| Banking institution | N | Company-CSR coherence | Motivational attribution | Corporate credibility | CSR image | ||||
|---|---|---|---|---|---|---|---|---|---|
| Mean | SD | Mean | SD | Mean | SD | Mean | SD | ||
| Santander | 275 | 5.22 | 1.18 | 3.58 | 1.37 | 4.63 | 1.11 | 5.30 | 0.79 |
| BBVA | 87 | 4.95 | 0.90 | 3.69 | 1.35 | 4.48 | 1.11 | 5.11 | 0.83 |
| CaixaBank | 82 | 5.54 | 1.20 | 3.97 | 1.28 | 5.18 | 1.04 | 5.49 | 0.81 |
| Bankia | 33 | 5.33 | 1.13 | 3.55 | 1.32 | 4.70 | 1.11 | 5.13 | 0.89 |
| Regional bank | 500 | 5.26 | 1.06 | 3.81 | 1.28 | 5.18 | 1.04 | 5.49 | 0.81 |
| Others | 147 | 5.03 | 1.08 | 3.50 | 1.26 | 4.27 | 1.12 | 5.20 | 0.86 |
The results were also disaggregated for each of the banking institutions evaluated by customers. In this regard, more than 85% of customers evaluated one of the following institutions: Santander, BBVA, CaixaBank, Bankia or the most relevant regional bank. The findings showed that CaixaBank and the regional bank were the best evaluated institutions by customers in terms of their coherence, altruism, credibility and CSR image. These results were in accordance with the public image of both institutions. On the one hand, CaixaBank has been traditionally linked to CSR through its Foundation and the social projects it sponsors. It is also one of the banking institutions that invest more resources in CSR communications nationally. On the other hand, the local character of the regional bank places its values and CSR initiatives very close to the community and this fact was also reflected in its evaluations. Big national banks such as Santander and BBVA were among the worst evaluated institutions in each of the constructs evaluated in this paper. These results are justified by the CSR tradition of banks and savings banks in Spain. More precisely, only recently banks (Santander, BBVA) have started to integrate CSR principles in their strategic plans while CSR has been part of the corporate philosophy of savings banks (CaixaBank, Bankia, regional bank) since their foundation (Pérez and Rodríguez del Bosque, 2012a).
Furthermore, Table 8 shows the estimationsof the causal relationships presented in Fig. 1. Again, the Comparative Fit Indexes were above 0.90, showing an adequate model fit. Furthermore, the results confirmed the proposed relationships among company-CSR coherence, motivational attribution and corporate credibility. First, there was a significant relationship between company-CSR coherence and motivational attribution (β=0.27, p-value <0.05) on the one hand and corporate credibility (β=0.44, p-value <0.05) on the other, so hypotheses H1 and H2 were supported. Furthermore, there was a significant relationship between motivational attribution and corporate credibility (β=0.46, p-value <0.05); therefore, hypothesis H3 was also supported. The results showed the importance of companies fostering credibility when developing their CSR initiatives, a sit allowed them to further improve their CSR image(β=0.29, p-value <0.05). Consequently, hypothesis H4 was supported. Furthermore, company-CSR coherence directly contributed to the generation of the CSR image(β=0.21, p-value <0.05); therefore, hypothesis H5was also supported. Finally, the attribution of altruistic or egoistic motivations automatically determined the CSR image of banking institutions(β=0.23, p-value <0.05); therefore, hypothesis H6 was supported.
Summary of results (H1–H6).
| Hypothesis | Causal relationship | Std. coefficient | T-value | Contrast |
|---|---|---|---|---|
| H1 | Coherence→Attribution | 0.27 | 7.07* | Supported |
| H2 | Coherence→Credibility | 0.44 | 10.91* | Supported |
| H3 | Attribution→Credibility | 0.46 | 12.85* | Supported |
| H4 | Credibility→CSR image | 0.29 | 5.10* | Supported |
| H5 | Coherence→CSR image | 0.21 | 4.48* | Supported |
| H6 | Attribution→CSR image | 0.23 | 5.66* | Supported |
| Goodness of fit measures | ||||
| S-Bχ2(113) | NFI | NNFI | CFI | IFI |
| 1,024.14 (p-value=0.00) | 0.93 | 0.93 | 0.95 | 0.96 |
*p<0.05
Three ANOVA analyses were implemented to determine whether significant differences existed in the way customers evaluated company-CSR coherence, motivational attribution, corporate credibility and the CSR image depending on their gender, age and educational level (Table 9).
Descriptive statistics (ANOVA results).
| Variable | Categories | N | Company-CSR coherence | Motivational attribution | Corporate credibility | CSR image | ||||
|---|---|---|---|---|---|---|---|---|---|---|
| Mean | SD | Mean | SD | Mean | SD | Mean | SD | |||
| Gender | Men | 547 | 5.14 | 1.10 | 3.62 | 1.31 | 4.54 | 1.07 | 5.17 | 0.84 |
| Women | 577 | 5.29 | 1.10 | 3.80 | 1.30 | 4.71 | 1.10 | 5.30 | 0.82 | |
| F-stat | – | 5.24* | – | 5.22* | – | 6.97** | – | 7.56** | – | |
| Age | 18–45 | 523 | 5.19 | 1.10 | 3.59 | 1.31 | 4.52 | 1.10 | 5.14 | 0.87 |
| >45 | 601 | 5.25 | 1.10 | 3.81 | 1.30 | 4.71 | 1.07 | 5.32 | 0.78 | |
| F-stat | – | 0.86 | – | 7.50** | – | 8.09** | – | 13.88** | – | |
| Education level | Non-college | 713 | 5.17 | 1.12 | 3.65 | 1.31 | 4.59 | 1.10 | 5.26 | 0.83 |
| College | 411 | 5.30 | 1.07 | 3.81 | 1.31 | 4.68 | 1.07 | 5.21 | 0.84 | |
| F-stat | – | 3.23 | – | 4.03* | – | 1.81 | – | 0.75 | – | |
The results showed that significant differences existed in the way men and women evaluated company-CSR coherence (meanmen=5.14 vs. meanwoman=5.29; F=5.24, p-value <0.05), motivational attribution (meanmen=3.62 vs. meanwoman=3.80; F=5.22, p-value <0.05), corporate credibility (meanmen=4.54 vs. meanwoman=4.71; F=6.97, p-value <0.01) and CSR image (meanmen=5.17 vs. meanwoman=5.30; F=7.56, p-value <0.01). Women rated companies significantly better than men in all the latent constructs in the conceptual model. In addition, customers with diverse age characteristics rated motivational attribution (mean18–45=3.59 vs. mean>45=3.81; F=7.50, p-value <0.01), corporate credibility (mean18–45=4.52 vs. mean>45=4.71; F=8.09, p-value <0.01) and the CSR image (mean18–45=5.14 vs. mean>45=5.32; F=13.88, p-value <0.01) differently. However, no significant differences existed in the way customers evaluated company-CSR coherence (mean18–45=5.19 vs. mean>45=5.25; F=0.86, p-value >0.05). According to these results, customers over 45 believed in the CSR expertise and trustworthiness of their banking institutions to a greater extent than younger customers. They also believed in the altruistic motivations of their banking institutions to a larger extent than customers below 45 and their CSR images were also better than those of the 18–45 cohort. Finally, educational level only allowed differentiating customer evaluations of motivational attribution (meannon-college=3.65 vs. meancollege=3.81; F=4.03, p-value <0.05) although significant differences were not perceived concerning how customers evaluated company-CSR coherence (meannon-college=5.17 vs. meancollege=5.30; F=3.23, p-value >0.05), corporate credibility (meannon-college=4.59 vs. meancollege=4.68; F=1.81, p-value >0.05) and CSR image (meannon-college=5.26 vs. meancollege=5.21; F=0.75, p-value >0.05). Thus, the only difference between customer categories was that more educated customers seemed to be less sceptical when evaluating banking institutions than less educated customers.
For the multisampling analyses, the model's factorial invariance was first confirmed in the three analyses performed (p >0.05 in 100% of the lambdas compared for gender and age and 71.4% for educational level), which demonstrated that the model was appropriate for understanding the CSR image formation process of the different types of customers independently consulted (men, women, 18–45, >45, non-college and college). Thus, the authors proceeded to test research hypotheses H7a–H9f.
Table 10 presents the multisampling analysis results for gender, age and educational level. The model results for each customer category, the LM test of the structural invariance and the goodness of fit of the final model (once the model restrictions have been eliminated) are included in the table. Based on the significance of the chi-square differences included in Table 10, the results showed that gender (Dif.χ2(1)=1.66; p-value >0.05), age (Dif.χ2(19)=21.85; p-value=0.29)and educational level (Dif.χ2(2)=3.95; p-value >0.05) did not moderate the relationships studied in this research. Thus, the hypotheses H7a to H9f were not supported. Some differences were observed in the way customers of different age cohorts constructed the CSR images of their banking institutions. In this regard, customers over 45 gave more importance to company-CSR coherence when evaluating the motivations of banking institutions to engage in CSR (β18–45=0.18, p-value <0.05; β>45=0.34, p-value <0.05; Dif.χ2=3.98, p-value <0.05). Nonetheless, the differences were not as significant as to allow the authors to consider age as a relevant moderator of thewhole CSR image formation process.
Summary of results (H7a–H9f).
| Moderating variable: Gender | |||||
|---|---|---|---|---|---|
| Hypothesis | Causal relationship | Standardised coefficients | Dif.χ2 (1) | Contrast | |
| Men | Women | ||||
| H7a–H7f | Coherence→Attribution | 0.28* | 0.25* | 0.18 | Not supported |
| Coherence→Credibility | 0.37* | 0.40* | 0.81 | ||
| Attribution→Credibility | 0.48* | 0.49* | 0.46 | ||
| Credibility→CSRImg | 0.29* | 0.27* | 1.60 | ||
| Coherence→CSRImg | 0.26* | 0.20* | 1.54 | ||
| Attribution→CSRImg | 0.25* | 0.22* | 1.28 | ||
| Fit | S-Bχ2(213)=541.38 (p<0.001); NFI=0.93; NNFI=0.95; CFI=0.96; IFI=0.96 Dif. χ2(1)=1.66 (p=0.20) | ||||
| Moderating variable: Age | |||||
|---|---|---|---|---|---|
| Hypothesis | Causal relationship | Standardised coefficients | Dif.χ2(1) | Contrast | |
| 18–45 | >45 | ||||
| H8a–H8f | Coherence→Attribution | 0.18* | 0.34* | 3.98* | Not supported |
| Coherence→Credibility | 0.51* | 0.54* | 2.23 | ||
| Attribution→Credibility | 0.39* | 0.40* | 0.87 | ||
| Credibility→CSRImg | 0.36* | 0.24* | 2.17 | ||
| Coherence→CSRImg | 0.19* | 0.23* | 0.65 | ||
| Attribution→CSRImg | 0.18* | 0.27* | 0.46 | ||
| Fit | S-Bχ2(241)=537.92 (p<0.001); NFI=0.93; NNFI=0.96; CFI=0.96; IFI=0.96 Dif. χ2(19)=21.85 (p=0.29) | ||||
| Moderating variable: Educational level | |||||
|---|---|---|---|---|---|
| Hypothesis | Causal relationship | Standardised coefficients | Dif.χ2(1) | Contrast | |
| Non-college | College | ||||
| H9a–H9f | Coherence→Attribution | 0.24* | 0.31* | 0.67 | Not supported |
| Coherence→Credibility | 0.41* | 0.33* | 1.76 | ||
| Attribution→Credibility | 0.48* | 0.51* | 0.05 | ||
| Credibility→CSRImg | 0.25* | 0.30* | 0.18 | ||
| Coherence→CSRImg | 0.25* | 0.23* | 0.14 | ||
| Attribution→CSRImg | 0.28* | 0.15* | 0.25 | ||
| Fit | S-Bχ2(213)=557.88 (p<0.001); NFI=0.93; NNFI=0.95; CFI=0.96; IFI=0.96 Dif. χ2(2)=3.95 (p=0.14) | ||||
The aim of this empirical study has been to clarify the process that leads customers to construct CSR images. For this purpose, the authors have designed a conceptual model that allows researchers and practitioners to anticipate the CSR image based on three constructs: the coherence between the company and its CSR initiatives, the attribution of motivations for the company to implement CSR initiatives and its corporate credibility in developing CSR. The results have demonstrated that coherence, motivational attribution and credibility play significant roles in the formation of CSR images. The findings have also demonstrated that customer demographic characteristics are no longer relevant moderators of this formation process.
First, there is a direct relationship between customer perceptions of corporate motivations, corporate credibility and CSR images. Along this line and in accordance with the proposals of attribution theory (Folkes, 1988), this study has shown that when customers perceive that companies have altruistic motivations for designing and implementing CSR initiatives, they are more credible and customers perceive more positive CSR images. In contrast, companies lose credibility when they are perceived as egoistic, such as when customers anticipate corporate intrinsic motivations for developing CSR initiatives. The loss of credibility contributes to the deterioration of the CSR image, which, as an essential component of corporate image, can have direct consequences for the company's reputation in the market as well as indirect effects in areas such as customer satisfaction, retention or identification with the company.
As a key element to facilitate the attribution of altruistic motivations and to improve corporate credibility, it is essential that a fit exist between corporate personality and CSR initiatives. When it makes sense to customers that their banking institutions perform CSR, they will have better perceptions of corporate motivations and credibility. Therefore, CSR initiatives consistent with corporate missions and goals will help improve the CSR image. This result confirms associative learning ideas (Till and Nowak, 2000), which argue that the association with CSR goals can transfer meaning from CSR initiatives to the organisation itself.
Furthermore, previous researchers have demonstrated that customer demographic characteristics can assist practitioners in segmenting the market to achieve greater efficacy in their corporate and commercial strategies. Based on this idea, the authors have also analysed how the formation process of the CSR image might be moderated by customer gender, age and educational level. Nonetheless, the results have demonstrated that these three demographic characteristics are not good moderators of the formation process of CSR images among customers. Thus, the market segmentation based on demographic characteristics is not useful for the design of CSR and communication strategies in the banking industry. In this regard, the findings are in line with the most recent studies in the CSR field that demonstrate that customer demographic characteristics are no longer sufficient to characterise customer behaviour (Roberts, 1996; Laroche et al., 2001). Thus, although in the past gender, age and educational level were determining factors in the perception of CSR, women, younger and highly-educated customers being more socially concerned (Arlow, 1991; Roberts, 1996), CSR principles are increasingly making their mark on society. Consequently, nowadays both men and women, younger and older customers and customers with or without a college education are more conscious of the importance of CSR activities. Therefore, there are no significant differences in the treatment they make of the diverse variables that contribute to better CSR images. For example, Diamantopoulos et al. (2003) demonstrate that although gender, age and educational level impact environmental attitude, they do not affect other relevant dimensions of environmental consciousness, such as knowledge and behaviour, which are also closely related to the conceptual model analysed in this paper.
Managerial implicationsThe findings presented in this paper have significant practical and managerial implications. First, companies devoted to CSR strategies must create initiatives that are consistent with their own corporate identities and that are perceived as altruistic programmes that enhance corporate credibility. Both coherence and motivational attributions are the main precursors of positive perceptions among customers and of strengthening the customer–company nexus.
The adequacy of developing coherent CSR initiatives in the banking industry is reflected in the success of certain institutions emerging from the latest financial crisis. For example, Spanish banks that operate internationally, such as Santander or BBVA, currently focus their CSR initiatives on microfinance and on customers’ financial education (Pérez and Rodríguez del Bosque, 2012a). These activities are highly congruent with their corporate personalities, which are oriented to the development of the society in all countries where they operate by means of education, which is the most effective way to contribute to social progress. For example, Santander's CSR programme is devoted to building strong alliances with universities through the Santander Universities programme. At the same time, these CSR orientations are properly integrated into the corporate strategy and contribute to corporate financial goals. For example, focusing on customer financial education can broaden commercial opportunities for companies by extending their business network through the bottom of the pyramid. However, other banking institutions in Spain, such as building societies or cooperative banks, are more focused on supporting society at large, which is a strategy that has little congruence with the personality of a banking services provider (Pérez and Rodríguez del Bosque, 2012a). These institutions have recently suffered many solvency problems; some have even collapsed, and a new governmental law has been promoted to restructure building and cooperative societies by allowing massive mergers that could help these companies rationalise their financial accounts. Authors such as Pérez and Rodríguez del Bosque (2012a) believe the differences between banks and building or cooperative societies are directly related to the management of CSR and stakeholders in both types of institutions, and they argue that the incongruence of CSR in building and cooperative societies makes their initiatives less effective than those of banks. Even though the findings of this paper have suggested that in 2010 the CSR image of banks was poorer than the image of savings banks, the problems faced by savings banks since 2010 have significantly changed the competitive context of banking institutions and CSR perceptions might be quite different today.
Meanwhile, the results concerning motivational attribution imply that unconditional CSR investments (e.g., unconditional donations or the sponsorship of social causes), where the company is not perceived to obtain an immediate economic compensation, will result in better customer perceptions than conditioned investments (e.g., CrM), where the company does not make the actual investment but rather acts as a mere agent connecting the social cause with customers and profits from the mediation. When customers do not perceive that the company benefits directly from the CSR initiative, they evaluate CSR more positively. However, when the collaboration of the company in a social cause is conditioned upon customers buying the company's products or services, customers may perceive egoistic motivations and may be sceptical of the whole CSR programme. Other risks of conditioned investments, such as CrM, include the loss of reputation, trust and appeal.
Finally, the role company-CSR coherence, motivational attribution and corporate credibility play in the CSR image is no longer dependant on customer demographic characteristics. Thus, scholars and practitioners have to keep searching for other moderating variables that could assist managers in designing better CSR and communication strategies. Along this line, recent scholars have considered that psychological characteristics, such as personal values or personality (Laroche et al., 2001; Bigné et al., 2009), are better predictors of customer behaviour than demographic factors. Thus, practitoners should not devote resources to communicate different things to men and women, customers between 18 and 45 and over 45 or customers with college and non-college degrees. They should better invest their budgets in determining if psychological characteristics can allow them to segment the CSR market to design more effective strategies.
Limitations and future lines of researchSome limitations of this paper must be admitted. In this regard, the fact that data was gathered from customers in a single country (Spain) can limit the generalisation of our results to other research contexts. In this regard, the Spanish banking industry has suffered the negative consequences of the latest financial crisis to a larger extent than more solid banks across Europe. This fact might especially determine the perceptions of customers who may react differently to CSR in this industry when compared to other international customers. A second limitation comes from the fact that the data gathered from customers refers to the CSR images of twenty one different companies in the banking industry. In this regard, the authors could not evaluate the information separately for each company because in some cases there was not enough amount of data to carry out independent SEM analyses (for some banking institutions there was just one or two valid surveys). Thirdly, recently scholars have started to consider additional variables that are closely related to some of the constructs of the conceptual model proposed in this paper. For example, Skarmeas et al. (2013) analyse CSR scepticism as a consequence of motivational attribution, while resilience to negative information and equity are consequences of customer scepticism. Thus, the authors consider that it would be necessary that future scholars took into consideration the new approaches to the study of the formation process of CSR images in order to complement and update the original model proposed by Rifon et al. (2004). Fourthly, the way the sample was segmented according to customers’ age might also be considered a limitation of the paper. Following the procedure implemented by many previous scholars, the authors of the present paper have segmented customers in two cohorts with 45 as the age cutting of the sample (Samarasinghe, 2012). Nonetheless, we can argue that 45 might be a subjective figure and that a different timeframe could have been chosen for the analyses. Future scholars using larger samples could segment customers in a larger number of clusters in order to achieve more robust findings.
Finally, some possible future lines of research are worth mentioning. Specifically, two propositions stand out. On the one hand, the authors suggest that future scholars implement additional tests of the moderating role of customer age in order to give external validity to our results. On the other hand, the authors align with recent proposals in marketing literature which identify psychological attributes as better predictors of customer behaviour (Laroche et al., 2001; Bigné et al., 2009). Thus, the analysis of personal values, lifestyles and attitudes (Laroche et al., 2001) should complement this study to determine whether these new variables are better criteria for market segmentation.