Incorporating the Concepts of Sharing-In and Sharing-Out in CSR: Australian Consumers’ Perspective

Abstract

Corporate social responsibility (CSR) is based on attending to concerns beyond a corporation’s shareholders. This concern embraces the concepts of sharing and caring for others. Logically then, the literature on sharing might inform theory and practice around CSR initiatives. To date, however, theory around the theme of sharing is absent from extant CSR literature, and, presumably, it is omitted as a perspective on CSR practice. This paper addresses this gap by empirically investigating consumers’ responses to different types of sharing involved in a range of CSR initiatives. We test the sharing theory to provide generalizable conclusions. Data was collected via an online panel of Australian consumers. Findings reveal two distinct types of sharing across different CSR initiatives: Sharing-in and sharing-out. Sharing-in CSR initiatives are those perceived as being adopted for the benefit of the firm’s direct stakeholders and aimed at providing a direct, reciprocal benefit to the firm. Conversely, sharing-out initiatives are perceived as providing wider and unconditional support to all, irrespective of their proximity to the firm. This research makes an original contribution to the CSR literature by embedding the notion of sharing within the management of CSR initiatives. The sharing perspective adds to knowledge and may assist managers when developing CSR initiatives, and how those initiatives might be perceived by a firm’s relevant stakeholders.

1. Introduction

Corporate Social Responsibility (CSR) approaches and initiatives of firms lead to benefits that might be shared so broadly and indiscriminately that the firm will not necessarily expect a direct return on its efforts, or so narrowly among the firm’s direct stakeholders such that the firm might hold an expectation of achieving a direct return from its efforts. Consumers have shown varied response to how a firm defines and adopts CSR activities. CSR definitions have been constructed on the fit of the activity with the firm’s operations, the location of the CSR activities, and its proximity to stakeholders, etc. [1]. Broadly targeted CSR activities may appeal to consumers’ altruistic affiliations or personal values and effectively position the corporation in consumers’ minds as “caring” [2,3]. This could then encourage consumers to feel favourably disposed towards that corporation. In this way, consumers feel a “congruence” between their own and the corporation values [4,5]. Conversely, more narrowly targeted CSR activities may appeal to consumers for such reasons as parochialism, egoism, patriotism, etc. [6]. Much of this research has focused on consumers’ responses to different CSR scenarios.

Extant research has examined conditions under which consumers might respond supportively to corporations CSR positions. When surveyed, consumers have often reported that they will consider such positions favourably [7,8]. For example, Marquina [9] found a positive relationship between a firm’s CSR and consumers’ purchase intention and behaviour. Consumers showed a willingness to pay a premium price for a product (athletic shoes in this case) if the firm was involved in CSR, such as their environmental commitment and corporate giving to worthy causes. Similarly, Kim [10] examined consumer reactions to the food industry’s environmental CSR and found a stronger favourable response and purchase intention for proactive CSR initiatives compared to passive (compliance driven) CSR programs.

Two questions often asked of CSR are, first, what is the responsibility? And second, to whom is it owed? One frame for answering these questions is by considering the concept of sharing [11,12]. Sharing goes to the heart of the first of these questions and is relevant to CSR since it is the very reason that corporations are expected to behave in a responsible manner and engage in activities that bring about benefits (e.g., sustainability) for society and the environment. Belk [11] defines sharing as “…the act and process of distributing what is ours to others for their use” (p. 126). Earlier CSR researchers were clear that the reason any business activity is allowed, is for the benefit of the whole community and that firms must fulfil a social obligation [13]. Society grants a “social legitimacy” and a certain power to corporations and in exchange it expects them to act responsibly and share their bounty with others [14,15]. Corporations not fulfilling their societal obligations lose their legitimacy and power to operate as their acceptance of both “socio-human and socio-economic responsibilities” go together [14] (p. 74).

The purpose of this paper is to, therefore, investigate consumers’ responses to firms’ CSR activities based on whether the benefits of these activities are perceived to be shared-in or shared-out. We contribute to the extant body of CSR-consumer response research in several ways. First, we incorporate the notions of sharing-in and sharing-out of CSR initiatives and consumer behaviour literature. Second, we report on an empirical investigation of Australian consumers’ responses to different types of CSR initiatives, based on the initiatives’ perceived scope of sharing.

The remainder of the paper is structured as follows. In the two next sections, we provide brief reviews of the sharing construct and the CSR-consumer literature. We then describe the research methodology, including the validation of measures. We then present the empirical findings of our study and discuss the theoretical and practical implications of these results, highlight the study’s limitations and conclude by proposing several critical directions for future research.

2. Sharing and CSR

Sharing-in and sharing-out are important in this discussion as firms have multiple stakeholders that potentially benefit from the firm’s CSR initiatives. Some of these groups are internal, such as senior management, shareholders and employees, and some external, such as consumers, suppliers, government and publics [16]. In a study investigating diverse corporate stakeholder groups, Fiedler and Kirchgeorg [17] found that different stakeholder-specific targeting and positioning is required by firms for effective branding to these groups. Customers, for example, will require a different communication approach to employees, shareholders, or the media. It stands to reason that CSR initiatives of different scope, shared-in or shared-out, will resonate with different stakeholders. In a CSR-specific study of stakeholders and their response to CSR activities [1], a firm’s stakeholders were divided into two groups: High- and low-proximity based on the level of involvement with the firm’s processes. Employees and management were categorised as high-proximity and customers and society as low-proximity stakeholders. It was found that low-proximity stakeholders, being outside the firm, were not in a position to differentiate between symbolic and actual, substantive CSR actions. However, focusing CSR actions towards only internal stakeholder groups seemed to be limiting, unrealistic and overly simplistic.

The concept of sharing is somewhat conceptually similar to the term used in social psychology literature of the in-group and out-group. Some CSR studies have made a connection between consumers’ preferences, behaviour and/or identity with the group [18,19]. According to Price [20] “sharing” is “the most universal form of human economic behavior, distinct from and more fundamental than reciprocity” (p. 3). This is made more relevant for the present times by Belk [11] who argues that the “issues of social justice, consumer welfare, environmentalism, materialism, commoditization, global food security, sustainable environments, and much more, all stand to be vitally informed by work on sharing” (p. 730).

Recent studies have applied the concept of sharing within different scenarios. In a study exploring how individuals value and form attachments to virtual (online) possessions, Odom, Zimmerman and Forlizzi [21] found that “these practices collectively model Belk’s notion of sharing in, where people share within a social group as a way of strengthening bonds” (p. 1497). Another study by Wiese, Kelley, Cranor, Dabbish, Hong and Zimmerman [22] incorporated the theory of “sharing in” in their investigation of the features of interpersonal relationships that influence sharing in an online environment. They found this concept to impact on individuals’ willingness to share as well as the extent to which they shared their location and personal information. Whereas, Baskentli, Sen, Du and Bhattacharya [4] found consumers with high binding moral foundations (collectivistic) to react more favourably to group-oriented CSR domain. They differentiate between the individual- and group-orientated CSR domains as “whether… it is perceived as focusing on protecting and enhancing the welfare of individuals or groups” (p. 503). It can argue that the foundation of sharing-out CSR is to spread benefit and welfare to a larger section of society, i.e., groups, instead of individuals.

We posit that when firms engage in sharing-in CSR, they are in effect concentrating their efforts on the direct stakeholder groups (similar to the “high-proximity group”) of the firm, whereas, by engaging in sharing-out CSR, a firm is providing benefit to indirect stakeholder groups (similar to “low-proximity group”). Sharing-in CSR activities are posited to be limited in scope, expected to bring direct economic gain to the firm and benefit only direct stakeholders of the firm. Sharing-out CSR activities are posited to be less discriminatory in scope and might bring economic and reputational gains to the firm but are not adopted solely for this reason.

3. Corporate Social Responsibility (CSR) and Consumers

We adopt the European Commission’s recent definition of CSR, which states that “(it) refers to companies taking responsibility for their impact on society.” [23]. Although a substantial amount of research has been conducted on CSR definitions [24,25], studies’ examining the scope of benefits arising from CSR activities have been limited. Matten and Moon [26] identified implicit and explicit types of CSR, with the explicit CSR tending to voluntary pro-social activities while respecting various stakeholder groups’ expectations, whereas, implicit CSR implied complying with the law and regulations while also adhering to the societal norms. They argued that these societal norms differ due to the impact of different political, financial, education, labour systems and culture with the nature of the firm in different countries. Their study identified US firms pursued more explicit, while European firms pursued more implicit CSR. However, little research is extant on the different types of CSR initiatives based on their scope of perceived benefit, and consumer responses to this scope, that is, to share-in or shared-out CSR.

Much recent CSR research attention has focused on the value of corporate responsibility in the marketplace, and considered, inter alia, its effects on consumers. Broad benefits ascribed to firms being perceived as socially responsible include: Gaining favourable consumer response which might manifest in the form of positive word of mouth, increased patronage, brand loyalty [27], relative immunity from financial/reputational disasters [28,29], longevity, seen as caring [3], competitive advantage over other players in the industry [30], talent acquisition and retention [31] and global legitimacy and social licence to operate [32]. Research into the CSR-consumer relationship has revealed consumers’ willingness to pay more for a product where the firm engages in CSR activity [9,10] and consumers’ inclination to give the benefit of the doubt to firms with a long CSR history in the face of reputational crises [28]. Consumers have been seen to be lenient for transgressions of smaller firms compared to larger ones [33], bestow positive CSR judgement of a firm based on limited CSR knowledge [34], and punish irresponsible corporate behaviour severely [35].

This variation in consumers’ response to CSR suggests a considered and customised CSR approach may be warranted to achieve the firm’s image-related CSR goals [4]. Several CSR scholars have found that consumers tend to have different preferences for different types of CSR domains [4,36]. This highlights two important issues. First, that there is an abundance of CSR domains (such as employees, community, environment, etc.) with their associated activities, suggesting that there is a strategic decision to make as to which ones the firm will pursue. Second, this decision may be quite different given that consumers are unlikely to equally prefer, nor respond to, all CSR activities equally. Therefore, making the right decision as to which CSR domains and associated activities to adopt is crucial for any firm. For example, a firm can decide to share-in and focus on its direct stakeholders such as employees only, in an expectation of a potential economic gain. However, another firm may decide that prospective reputational benefits of adopting far-reaching CSR initiatives outweigh other rewards and thus engage in sharing-out CSR. Therefore, to gain the most out of their CSR initiatives, in terms of marketplace support from consumers, firms need to pay attention to the CSR expectations and preferences of the consumer segments they wish to target. However, in spite of extant research findings there remain gaps in our knowledge as to why some consumers prefer certain CSR activities over others. Investigating whether consumers prefer shared-in or shared-out CSR will provide a key insight into this dilemma.

4. Materials and Methods

To ensure that a proper scale development procedure is followed, the guidelines from DeVellis [37] were adopted. Six large firms operating in Australia from three key industries were chosen to generate a large item pool for both social and environmental CSR initiatives. Two of the three chosen industries were known for their exposure to retail consumers. These were supermarkets in the retail sector (Woolworths and Wesfarmers) and the automotive industry (Toyota and Honda). The third industry, well-known for its CSR programs that tend to be restricted to certain locations, was the mining industry (BHP Billiton and Woodside). All firms were known for their significant economic contribution to the Australian economy and were large employment providers [38,39,40,41].

Though the mining industry is business to business (B2B) and therefore, does not have direct contact with everyday consumers, it was decided to add this industry to generate a realistic pool of CSR initiatives for this study for several reasons. First, due to the nature of this industry, the potential to cause harm communities and the environment is significant. Second, mining companies have been in the media spotlight due to this very reason and consumers, although not directly in contact with these companies, can see the potential and actual damage with long-term negative effects caused by the operations and occasional negligence of this industry [42]. Hence, different stakeholder groups, including consumers, have been exerting pressure on this industry to be accountable and do better [43]. Third, most of the companies in this industry possess substantial financial resources. For example, Exxon Mobil’s earnings are far greater than the GDP of a number of developing countries [44,45]. There is an implied social obligation and responsibility, which comes with this kind of financial power, to do good for society and the environment, the mining industry is known to spend a significant amount of their resources to maintain relationships with their various stakeholders and have considerably well-planned CSR programs.

A large pool of CSR initiatives of the six chosen firms was drawn from various secondary sources, including CSR reports, sustainability reports, annual reports and corporate websites. This list was further sorted by critically analysing the perceived scope of each initiative. Items appearing multiple times were noted for emphasis and repetition was removed.

Table 1. Broad themes for corporate social responsibility (CSR) initiatives from three Australian industries.

Industry	CSR Themes
Retailing	▪ Sustainability and Environment ▪ Health and Well-being ▪ Education and Employment ▪ Rural and Regional Communities
Mining	▪ People ▪ Environment ▪ Society ▪ Governance ▪ Economic
Automotive	▪ Customers First ▪ Respect for Employees ▪ Safety and Health ▪ Fair and Transparent Business ▪ Timely and Proper Disclosure ▪ Management of Confidential Information ▪ Compliance ▪ Social ▪ Environment ▪ Quality and Safety

presents the broad CSR themes derived from six large firms across three Australian industries.

On closer examination of the CSR initiatives under each theme some repetition was found with at times, different terms used for the same theme.

Table 2. Item pool for CSR initiatives.

Type	CSR Initiatives
Social	▪ Social welfare ▪ Support for youth sports activities ▪ Supporting employee participation in local sports activities ▪ Charity drive for cancer research ▪ Children’s health ▪ Providing presents for disadvantaged and disabled children ▪ Associate volunteers travel to disaster areas ▪ Developing our employees ▪ Investing in graduate development ▪ Committed to broad-based community support ▪ Increase direct indigenous employment ▪ Investing in youth studentship program ▪ Inclusive workplace for indigenous and disabled ▪ Donation for flood relief efforts ▪ Employment to aboriginal groups ▪ Support the communities in which we operate
Environmental	▪ Forest conservation ▪ Environmental protection/nature conservation ▪ Response to large scale earthquake ▪ Tree plantation ▪ Watershed preservations ▪ Beach cleaning activities ▪ Managing water in our operations ▪ Initiate development of a water management strategy ▪ Water efficiency

presents the items belonging to social and environmental CSR initiatives originating from the broad CSR themes from the three industries.

An iterative process produced a final pool of 20 items, with 12 items belonging to social initiatives (representing actions taken by these firms for the community) and six items belonging to environmental initiatives. The third main theme of “economic initiatives” is embedded in the sharing-in (an expectation of direct economic benefit) and sharing-out (an expectation of no direct economic benefit) aspect of both social and environmental CSR initiatives and that is the main point of difference with previous CSR initiatives that do not take the scope of benefit of these initiatives into account. These items were repeated with emphasis on either a limited scope of sharing (sharing-in) or a wider scope of sharing (sharing-out). The categorisation with each item for the sharing-in was “only in the areas where the company operates its business” and the sharing-out was “anywhere in Australia”. Three researchers took part in the iterative categorisation of CSR initiatives. We used a Likert-type scale anchored by “strongly disagree” (1) and “strongly agree” (9). The measurement scale is presented in the results section.

4.1. Data Collection

In order to test the hypothesised model, we conducted a survey using an online consumer panel. The use of an online panel eliminated the dependency on obtaining accurate email addresses and other time and administration issues that come with incorrect emails lists, as well as ensuring a higher response rate [46]. As most of these panels are recruited through telephone or postal mail, the sample can be seen as representative of the broader community [47]. A research panel company was appointed to recruit consumers to take part in the online survey. To obtain a representative sample from across Australia, the sample distribution was based on the most recent census state-wide split. The minimum age requirement was put at 18 years. Recruiting potential respondents through a panel company also ensured the privacy and anonymity of all respondents.

4.2. Sample

The sample was composed of 691 usable responses from across Australia. Distribution was based on the latest census state-wide split to ensure a representative sample of Australia. The criterion for selection was purposefully broad and inclusive. Cochran’s [48] formula N_o = Z²pq/e² for calculation of sample size presented the desired sample size of 385. However, as the study involved testing a number of variables using structured equation modelling, a decision was made to get more sample. Given the reasonably large sample size obtained in the present study (n = 691), it was possible to use the most widely adopted estimation method of maximum likelihood (ML) [49]. ML estimation is used to test the one-factor measurement models for each of the latent constructs to confirm the factor structure. ML freely estimates all factor loadings and error variances. The conventional approach was adopted, and the data set was split in half with one set used to perform exploratory factor analysis (EFA) and the other confirmatory factor analysis (CFA). Details of respondents’ demographics are presented in

Table 3. Demographics of respondents.

Particulars	Respondents
Age	Range 18–71
Gender	59 per cent female and 41 per cent male
Income	From less than $10k to above $100k per annum Half earning between $25–75k per annum
Education	From completed Year 9–11 to postgraduate degree 27 per cent had TAFE (vocational) qualification
Place of birth	82 per cent were born in Australia
Relationship status	40 per cent married and 29 per cent single

.

It is important to check for outliers as they can skew the results. These are the scores which fall at the “outer ranges of the distribution” [50]. When a score is noticeably distinct at a single variable level, it is termed a univariate outlier, whereas multivariate outliers exist across a set of variables [51]. As the present study explores factor level scores, it was deemed appropriate to examine only univariate outliers. Data were checked for univariate outliers through visual analysis of histograms and Q-Q plots. After examining these, it was decided that univariate outliers would not pose any problem in the analyses as the higher/lower values are genuine responses of certain individuals and therefore, none were deleted.

Normality of the data was checked through tests of skewness and kurtosis. Histograms were also produced and examined as they show a clear indication of the distribution of any data. According to Hair et al. [52], skewness is a measure of the symmetry of a distribution. A positively skewed distribution of data shows fewer large values, whereas, a negatively skewed distribution of data shows fewer small values. Upon examination of the initial descriptive, most of the items showed some level of skewness, although these generally fell within the acceptable limit of −1 to +1.

Detailed descriptive statistics for each construct, including the mean, median, mode, skewness and kurtosis of each item in each measurement scale, are produced below. The following

Table 4. Mean, median, mode, skewness and kurtosis for social CSR initiatives.

Coded Items	Mean	Median	Mode	Skewness	Std. Error of Skewness	Kurtosis	Std. Error of Kurtosis
S1	7.58	8	9	(1.681)	0.094	2.562	0.187
S2	4.53	5	1	0.164	0.095	(1.218)	0.189
S3	5.78	6	9	(0.370)	0.094	(0.934)	0.187
S4	5.75	6	9	(0.0484)	0.094	(0.958)	0.187
S5	6.36	7	9	(0.650)	0.094	(0.520)	0.187
S6	5.30	5	5	(0.263)	0.094	(1.075)	0.188
S7	6.76	7	9	(0.928)	0.093	0.056	0.186
S8	5.21	5	1	(0.250)	0.094	(1.200)	0.188
S9	6.79	7	9	(0.952)	0.093	(0.019)	0.187
S10	5.54	6	9	(0.375)	0.094	(1.154)	0.188
S11	6.13	7	9	(0.556)	0.093	(0.821)	0.187
S12	4.86	5	5	(0.054)	0.094	(1.197)	0.187
S15	6.17	7	9	(0.584)	0.093	(0.837)	0.187
S16	4.94	5	5	(0.045)	0.094	(1.172)	0.188
S27	6.35	7	9	(0.658)	0.093	(0.671)	0.187
S28	5.05	5	1	(0.125)	0.094	(1.323)	0.188
S29	6.90	8	9	(1.027)	0.093	0.192	0.186
S30	5.18	5	1	(0.203)	0.094	(1.289)	0.188
S31	6.73	7	9	(0.877)	0.094	(0.250)	0.187
S32	5.26	5	9	(1.197)	0.094	(1.197)	0.188

and

Table 5. Mean, median, mode, skewness and kurtosis for environmental CSR initiatives.

Coded Items	Mean	Median	Mode	Skewness	Std. Error of Skewness	Kurtosis	Std. Error of Kurtosis
S13	7.04	8	9	(1.082)	0.094	0.397	0.187
S14	5.26	6	1	(0.229)	0.094	(1.302)	0.188
S17	6.80	7	9	(0.885)	0.093	(0.177)	0.186
S18	5.03	5	5	(0.078)	0.094	(1.236)	0.188
S19	7.22	8	9	(1.278)	0.093	0.759	0.186
S20	4.98	5	1	(0.056)	0.094	(1.350)	0.188
S21	7.31	8	9	(1.311)	0.093	1.046	0.186
S22	5.28	5	9	(0.194)	0.094	(1.334)	0.188
S23	7.12	8	9	(1.185)	0.093	0.679	0.186
S24	5.11	5	1	(0.120)	0.094	(1.320)	0.188
S25	7.46	8	9	(1.540)	0.093	1.737	0.186
S26	5.26	5	9	(0.188)	0.094	(1.395)	0.188

show the descriptive statistics including mean, mode, median, skewness and kurtosis of each item in the measurement scale of consumers’ preferences for social and environmental CSR initiatives, respectively.

To test the scale dimensionality, exploratory factor analysis (EFA) and confirmatory factor analysis (CFA) were conducted on the data. An EFA was applied to the new measurement scale for capturing the dependent variables as part of scale development. An EFA is used to allow the data to define the nature of the relationships and as a data reduction technique to reduce a large number of items into a more meaningful smaller set [52,53]. Several criteria determine whether the data are appropriate for factor analyses. These include the observation of correlation matrix, Kaiser Meyer–Olkin (KMO) and Bartlett’s test of sphericity, the anti-image matrix, observation of communalities, total variance explained and the scree plot.

Confirmatory factor analysis (CFA) is typically used to confirm the structure of a developed scale, ensuring it “holds” its robustness with each cohort or context in which a scale is used. CFA is thus used to statistically confirm a pre-specified relationship [52] and determine whether the data fits the theoretically proposed model. According to Cunningham [54], CFA is based on theory, whereas an EFA is statistically driven. After loading all items on latent variables in EFA, CFA calculates factor loadings for those items that are directly related to the factor. CFA was performed using the statistical package, Analysis of Moment Structures (AMOS—version 21).

5. Results

Data were subjected to principal component analysis with Promax (Oblique) rotation using SPSS (version 20) as the factors were expected to correlate [52]. The correlations in the correlation matrix were above 0.30, and so the data were deemed suitable for factor analysis. The Kaiser Meyer–Olkin (KMO) measure of sampling adequacy (MSA) value was 0.873 and Bartlett’s test value was 0.000, which further suggests that the data were suitable for factor analysis. A two-factor result was expected with a split showing the aspects of sharing-in and sharing-out confirming the theoretical framing of the items. However, the analysis resulted in a four-factor solution with the expected split of sharing-in and sharing-out, thus confirming the theoretical underpinning but also providing a deeper understanding of the consumers’ expectation of social CSR initiatives. The results of the exploratory factor analysis (EFA) for social CSR initiatives are presented in

Table 6. Factor loadings, eigenvalues and average variance extracted of social CSR initiatives.

No.	Items	RSSO	RSSI	DSSI	DSSO
S3	Supporting junior sports clubs anywhere in Australia	0.801
S9	Providing employment to disadvantaged groups anywhere in Australia	0.801
S7	Providing scholarships for further education anywhere in Australia	0.797
S1	Supporting health anywhere in Australia	0.686
S15	Providing shelter facilities for lost or injured domestic pets anywhere in Australia	0.685
S5	Supporting the conservation of heritage buildings anywhere in Australia	0.629
S11	Providing support for victims of drug and alcohol abuse anywhere in Australia	0.606
S4	Supporting local junior sports clubs only in the areas where the company operates its business		0.878
S8	Providing scholarships for further education only in the areas where the company operates its business		0.799
S10	Providing employment to disadvantaged groups only in the areas where the company operates its business		0.754
S6	Supporting the conservation of heritage buildings only in the areas where the company operates its business		0.742
S12	Providing support for victims of drug and alcohol abuse only in the areas where the company operates its business		0.597	0.323
S2	Supporting health only in the areas where the company operates its business		0.594
S16	Providing shelter facilities for lost or injured domestic pets only in the areas where the company operates its business		0.530
S30	Providing support for those in poverty and/or homelessness only in the areas where the company operates its business			0.893
S32	Providing housing, medical aid, food supplies and/or counselling to those in need only in the areas where the company operates its business			0.883
S28	Supporting indigenous people only in the areas where the company operates its business			0.752
S29	Providing support for those in poverty and/or homelessness anywhere in Australia				0.858
S27	Supporting indigenous people anywhere in Australia				0.831
S31	Providing housing, medical aid, food supplies and/or counselling to those in need anywhere in Australia				0.821
Eigenvalue		5.7	4.6	1.2	1.1
Percentage of Variance Explained		28.3	22.8	6.1	5.6
Cronbach Alpha		0.86	0.83	0.84	0.86

RSSO = relatable social sharing-out; RSSI = relatable social sharing-in; DSSO = distant social sharing-out; DSSO = distant social sharing-in.

with their loadings, eigenvalues, total variance explained and Cronbach’s alphas.

Another EFA was conducted on the scale items measuring environmental CSR initiatives. As the correlations in the correlation matrix were above 0.30, the data were deemed suitable for factor analysis. The Kaiser Meyer–Olkin (KMO) measure of sampling adequacy (MSA) value was 0.85 and Bartlett’s test value was 0.000, which again suggests that the data are suitable for factor analysis. A two-factor split confirmed the capture of sharing-in and sharing-out was achieved. The results of the EFA are presented in

Table 7. Factor loadings, eigenvalues and average variance extracted of environmental CSR initiatives.

No.	Items	ESI	ESO
S18	Protecting endangered plants only in the areas where the company operates its business	0.875
S20	Protecting endangered animals only in the areas where the company operates its business	0.865
S24	Supporting programs for the promotion of cleaner air only in the areas where the company operates its business	0.819
S26	Supporting programs for the promotion of cleaner water only in the areas where the company operates its business	0.817
S22	Support research in new warning systems for bush fires, cyclones, etc. only in the areas where the company operates its business	0.810
S14	Supporting the planting of more trees only in the areas where the company operates its business	0.799
S23	Support research in new warning systems for bush fires, cyclones, etc. anywhere in Australia		0.870
S19	Protecting endangered animals anywhere in Australia		0.829
S17	Protecting endangered plants anywhere in Australia		0.818
S25	Supporting programs for the promotion of cleaner water anywhere in Australia		0.811
S21	Protecting endangered animals anywhere in Australia		0.765
S13	Supporting the planting of more trees anywhere in Australia		0.757
Eigenvalue		4.4	3.7
Percentage of Variance Explained		36.9	30.7
Cronbach Alpha		0.91	0.89

ESI = environmental sharing-in; ESO = environmental sharing-out.

with their loadings, eigenvalues, total variance explained and Cronbach’s alphas.

A confirmatory factor analysis (CFA) was conducted on the factors emerging from the EFA to statistically confirm the pre-specified relationship [52] and determine whether the data fit the theoretically proposed model. Several other tests such as composite reliability (CR) and average variance extracted (AVE) were conducted to achieve internal consistency [50]. The recommended thresholds are values greater than 0.6 for CR and values greater than 0.5 for AVE. Validity was demonstrated by showing that the AVE by the latent construct was greater than the squared correlation between two latent measures.

Further testing was conducted to ensure sound validity. Convergent validity looks at how much the measure correlates positively with other measures of the same construct [52]. Discriminant validity is the other side of convergent validity as it looks at how the measure is different from other similar measures of the same construct [52]. Validity was demonstrated by showing that the AVE by the latent construct was greater than the squared correlation between two latent measures. The formula for measurement is produced below.

$$\text{Average Variance Extracted}=\frac{\sum (\text{standardised loadings}{)}^2}{\sum (\text{standardised loadings}{)}^2+\sum (\text{indicator measurement errors}).}$$

All constructs revealed good reliability and validity, both convergent and discriminant.

Table 8. Correlations among all latent variables.

	CR	AVE	ESI	RSSO	RSSI	DSSO	DSSI	ESO
ESI	0.877	0.641	0.801
RSSO	0.790	0.556	−0.217	0.746
RSSI	0.795	0.564	0.744	−0.186	0.751
DSSO	0.854	0.662	−0.164	0.703	−0.082	0.814
DSSI	0.851	0.658	0.796	−0.093	0.648	−0.006	0.811
ESO	0.884	0.656	−0.107	0.714	−0.013	0.755	−0.064	0.810

presents the correlations among latent variables.

A number of widely-used goodness-of-fit indices were used. The absolute fit indices measure the whole model including the measurement and path models and include CMIN, CMIN/df [51,52]. The chi-square statistic provides a “goodness-of-fit” comparison between the covariance matrix for the observed data and the covariance matrix derived from a theoretically-derived structure model [54]. Relative fit indices compare the incremental model with the null model and vary in value from 0 to 1 and include the Tucker–Lewis index (TLI), comparative fit index (CFI) and normed fit index (NFI) [55]. However, there is little consensus as to the cut-off value for an acceptable fit. Reisinger and Mavondo [56] state that 0.95 can be taken as an acceptable fit. Another fit measure that is widely used is the root mean squared error of approximation (RMSEA). This index is based on the non-central chi-square distribution [52]. Different researchers have given a range of acceptable limits; for example, Reisinger and Mavondo [56] suggest that values <0.08 are acceptable and values <0.05 represent a good fit, whereas Hu and Bentler [57] prefer values <0.06. A CFA presented reasonable acceptable indices of goodness-of-fit with χ² 511.64, df 155, sig 0.000, χ²/df 3.3, CFI 0.95, NFI 0.94, TLI 0.94 and RMSEA 0.05.

The results provided confirmation of the soundness of the theoretical framing of including the scope of benefit into the CSR initiatives. Findings reveal a clear split of sharing-in and sharing-out in consumers’ preferences for social and environmental CSR initiatives. Further addition of two more factors in social initiatives make the results robust. The clear delineation of everyday relatable CSR actions from somewhat distant, though still important CSR actions, presents a deeper insight into the scope of CSR initiatives firms might consider if achieving a favourable response among specific stakeholders, such as consumers, is a CSR-related goal.

6. Discussion

Consumer preferences for CSR initiatives are varied and there is no one-size-fits-all approach. The results confirm the theoretical conception of embedding the scope of “sharing” in both social and environmental CSR initiatives. We developed a measurement scale to test consumers’ preferences for different types of social and environmental CSR initiatives, based on their scope and spread of benefit. Results support our hypothesis that consumers will distinguish between the sharing-in and sharing-out CSR initiatives for both social and environmental domains. These CSR domains are termed social sharing-in and out and environmental sharing-in and out.

The group preferring sharing-in CSR initiatives may focus on a more pragmatic view of firm motivation to adopt CSR initiatives, i.e., longevity and sustainability of the firm by getting its return-on-investment through only catering for its direct stakeholders. The saying “charity begins at home” might be at play here.

Firms do not engage in CSR initiatives for purely economic reasons but may also have an altruistic motivation. These firms are more likely to embed long-term and wide-reaching strategies in their CSR programs and providing the CSR benefits to the disparate cohort (sharing-out) may be one of these strategies. Such strategic CSR aligns well with the notion of firms using their resources for the community betterment (environment included). There might also be an expectation from the cohort preferring sharing-out CSR that firms should not limit their reach to a few local groups or nearby geographic regions. These findings align with Ackah-Baidoo [58] who argued for wider-reaching CSR initiatives programs from oil companies operating in the sub-Saharan region because of some companies’ tendency to keep their CSR programs proximate to their operations.

Interestingly, our findings differ from those of a somewhat similar study [59] in which consumers were found to prefer local to non-local CSR initiatives, indicating a sharing-in response. Respondents were provided fictitious scenarios based on geographic proximity differentiation of CSR benefits: “Domestic versus foreign”, “home state versus distant state” and “home city versus distant city”. Consumers’ egocentrism was considered to be a major contributor to those results, restricting the benefits of CSR initiatives to their proximate in-group.

Consumers have preferred both sharing-in and sharing-out CSR initiatives. The distinction between “relatable” and “distant” was found only in the social CSR initiatives and not in environmental CSR initiatives. This is an important result in itself, as it might be due to the fact that there is an implicit admission/understanding that the environmental issues affect us all, irrespective of their geographical location. It might also reflect a growing global concern about the environmental well-being of the planet and consumers preference for firms to do all they can to mitigate negative environmental externalities. Another possible reason for such results might be the recent environment-related events in Australia. For example, the on-going occurrence of bush fires in the Australian states of Victoria and Tasmania and floods in Queensland and severe drought conditions in New South Wales around the time of data collection may have heightened respondents’ sensitivity to firms’ environmental CSR initiatives. There is also vast media coverage of the immediate and long-term devastation these natural events bring to the areas.

A clear distinction was found in responses to social CSR initiatives. The CSR initiatives found to resonate most with respondents were on themes around employment, children’s sports, education, health and programs to address drug and alcohol abuse. According to one estimate, around thirteen per cent of children live in an Australian household with at least one adult regularly affected by alcohol [60].

Health-related initiatives also resonated with respondents. According to Diabetes Australia [61], an estimated 280 people develop this ailment each day. Therefore, it is logical to assume that a considerable number of consumers are not only well aware of health issues but also live with a patient in a supporting role. Again, it is easy to understand why a number of social CSR initiatives were categorised as relatable. Firms may be able to tap this sensitivity towards an awareness of some of these issues and plan their CSR strategies accordingly to capture the preferences of their consumer base.

Results show that some social CSR initiatives were grouped together, and we labelled them as “distant”. These items were found to be issues around poverty and homelessness, housing, food supplies and indigenous people. This grouping presents an interesting finding, which may be used to generalise countries with similar demographics and social issues. Australia is a developed country and has a relatively low percentage of its population, 13.2 per cent living below the poverty line [62]. However, this shows that 86.8 per cent of the population may not be confronted by this issue and thus find it more difficult to relate to. This cocooning might be compounded by the fact that people tend to associate with similar socio-economic background, people in poverty feel shame in disclosing their state of economic affairs and this issue is kept insular.

Another important social CSR initiative was around indigenous people. Why results have shown it to be part of a smaller group might be due to their significantly small numbers in the total population. Indigenous and Torres Strait Islander communities only make up of approximately 3 per cent of the total population [63]. Further compounding is the fact that this population is dispersed throughout the country with a large number living in either remote or very remote areas [63], thus making this group of people less visible to average Australians.

The findings of this study suggest that Australian consumers, at least, prefer wide-ranging CSR programs, that include sharing-in and sharing-out initiatives. Conversely, failure to develop CSR programs that resonate with key stakeholder groups, based on the concept of sharing, may alienate stakeholder groups with which the firm seeks to develop evidence- and trust-based relationships.

7. Conclusions

Based on the conceptualisation of sharing [12], the purpose of this paper was to investigate consumer responses to firms’ CSR activities based on whether the benefits of those activities were perceived to be shared-in or shared-out. We distinguished between two concepts of sharing-in (benefit limited to direct stakeholder groups) and sharing-out (indiscriminate spread of benefit). Data were collected and analysed to investigate whether consumers differentiate between sharing-in and sharing-out CSR initiatives in both social and environmental domains and whether their responses differed, based on the distinction.

Results highlighted an important categorization in the social domain with consumers distinguishing between “relatable” (everyday issues) and “distant” (remote) CSR initiatives. Interestingly, this classification was present in social CSR domain alone.

The information gleaned from this study has practical managerial implications. Bespoke CSR programs are more likely to engage with firm’s target market and can be bestowed with favourable response manifested as purchase intention, willingness to pay more, brand recognition and loyalty and resilience in bleak times. Firms should apportion sharing-in and sharing-out CSR initiatives in order to develop more effective CSR programs that align more closely to target stakeholders CSR expectations of firms in particular industries. Such an alignment is not only important for maintaining its license-to-operate but may also serve to inform marketing strategy focused on particular targeted market segments.

This research presents another way to understand what motivates consumers, arguably the most important stakeholder group of a firm when they are evaluating different types of CSR programs, particularly when they may be comparing CSR records of competing firms in the same industry. As such, the study’s findings have a practical value for CSR managers. The managerial impact of the research is to better understand their target market and their CSR preferences, which may/may not be the same as the course of action that the firm has taken. The misalignment of the firm–consumer CSR expectations can have less favourable results for the firm. The research findings also show a heightened sensitivity to social CSR initiatives, not only based on the scope of perceived benefit, but also on the relatability of the CSR actions to the majority of the study’s participants.

These findings indicate that consumers are willing to think in altruistic and aggregate terms and some may expect firms to undertake CSR initiatives that benefit a wider spectrum of people and areas. This provides a clear indication for firms to strategically develop their CSR initiatives. Future research can explore how consumers perceive firms investing in communities by applying the concepts of sharing-in and sharing-out through a range of alternative CSR initiatives to those included here. Potential issues around different marketplaces, specific geographically-bound communities, varying demographics, for example, age, and specific industries might also be investigated in order to further our knowledge of the CSR–stakeholder relationship, particularly the CSR–consumer relationship.

This research was conducted in Australia, and thus exploring the impact of sharing-in and sharing-out of firms’ CSR program benefits warrants investigation. The sharing lens adds to our knowledge of the CSR–consumer behaviour literature, but further work is needed to determine how generalized the findings reported here may prove to be, across markets and cultures.