The Impact of Proximity on Consumer Fair Trade Engagement and Purchasing Behavior: The Moderating Role of Empathic Concern and Hypocrisy

Proximity is defined as the “feeling that the moral agent has for victims (beneficiaries) of the evil (beneficial) act in question” (Jones 1991, p. 376) and comprises physical, social, and psychological dimensions. Contextualizing proximity in FT purchase decision-making implies that consumers (moral agents) who are physically, socially, and psychologically closer to marginalized producers/workers could be affected more by the issues around their living conditions.

Physical proximity refers to the geographical nearness between someone making a decision or performing an action and those affected by this decision or act (Jones 1991). When examined in an ethical or moral context, it was found that physical proximity enhances helping behavior (Grau and Folse 2007; Adams and Raisborough 2008; Chang 2012). For instance, Grau and Folse (2007), in examining the impact of nearness of donation activity and the nature of messages on consumers’ attitudes and intentions, revealed that a donation is more likely when local people are involved and hence, identify with the cause. Although FT is a movement initially developed to connect consumers in the global north to the marginalized producers/workers of the global south, recent developments in consumer markets indicate that FT is no longer limited to the north but it also extents to emerging economies with serious implications for physical proximity between consumers and marginalized producers/workers (Alexander and Nicholls 2006; Doherty et al. 2013). Although the role of physical proximity has not yet been examined within a FT context, Adams and Raisborough’s (2010) study among British consumers indirectly found that ‘closer to home’ describes a consumer type whose consumption preferences are predicated upon products being local. They also revealed that most participants expressed a common concern, namely promoting and looking out for the welfare of others. However, there was a vast difference in whom and why they perceived some parties as a deserving beneficiary, which then had an impact on whether they backed FT products from marginalized producers/workers or supported local British goods. The above suggest that:

Social proximity exists when people share a social familiarity based on collective experiences, deriving from closeness in terms of values, norms, customs, family, background, and beliefs (Boschma 2005). Ghorbani et al. (2013) suggest that social proximity is an empowering factor in ethical behavior, grouping the social ties into strong and weak. It is proposed that social identification increases as a function of commitment or a feeling of familiarity to a group and this influences behavioral outcomes (Mencl and May 2009). It is the founding intention of FT that the norms and values of society be changed to reveal people, places, and relations responsible for the commodities, which can facilitate the change where consumers could be more actively engaged with FT instead of passive involvement (Golding and Peattie 2005). As Raynolds (2000) puts it, greater understanding of the socio-cultural traits of marginalized producers/workers by consumers helps in understanding the invisible forces that govern the marketplace. In other words, consumers’ social proximity, that is, familiarity with the values, norms, customs, and beliefs of marginalized producers/workers, is more likely to influence their engagement with FT issues (Narasimhan et al. 2015). Hence, the following hypothesis can be made:

Psychological proximity is described as an affective propinquity or a ‘high level of commitment’ (Mencl and May 2009, p. 206). This commitment is directed towards those who are close to the decision-maker, such as family members and close friends (Ghorbani et al. 2013). In a FT context, it is the subjective experience of perceived closeness to marginalized producers/workers from developing countries. Psychological proximity can be generated, for example, through media coverage, awareness programs, and personal visits to these countries. FT organizations, striving to increase awareness of their raison d’être, generate national marketing awareness campaigns, demonstrating the quintessential differences between the FT process and the traditional model, thus enhancing the psychological proximity between consumers and impoverished producers/workers (Nicholls and Lee 2006). Thus, to some extent, the distancing, out-group aspects of the far-away producer/worker (Turner and Tajfel 1986) can be ameliorated as the consumer begins to develop a genuine rapport, overcoming the out-group aspects of ‘disadvantaged producers in far-flung corners of the globe’ (Doran 2010, p. 538), which s/he demonstrates by engaging in FT concerns (Doran 2009). Further, De Ferran and Grunert (2007) found that consumers supporting FT products were motivated by a desire for greater equality among mankind, resulting in an enhanced psychological proximity with the marginalized producers/workers. In other words, increased psychological proximity between consumers and marginalized producers/workers results in heightened involvement and investment in the FT ethos. Therefore, we may posit the following:

Customer engagement is defined as a process by which a customer becomes actively involved with the brand, including and beyond purchase (Hollebeek et al. 2014; Vivek et al. 2014; Pansari and Kumar 2017; Gupta et al. 2018). It is a multifaceted concept, which includes such dimensions as cognitive, behavioral, social, and emotional aspects (Vivek et al. 2012), as well as brand interaction and identification (So et al. 2014). Customer engagement creates value for businesses (Pansari and Kumar 2017) and as such it can be demonstrated as having links with customer satisfaction and loyalty (Bowden 2009; Van Doorn et al. 2010; Vivek et al. 2012). From an ethical consumption perspective, customer engagement influences FT purchase behavior. This is because ethical engagement permits consumers to demonstrate a feeling of responsibility toward society and their admiration of businesses employing socially responsible approaches, as expressed by purchasing products which have positive, moral, and ethical qualities. This contrasts to exploitative and unethical firms (such as those classified as ‘modern slavery’), which are scrupulously avoided (de Pelsmacker et al. 2005; Andorfer and Liebe 2015). In fact, there is empirical evidence showing that consumers, for whom ethical issues are of concern, are likely to change their behavior and buy FT products to support marginalized producers/workers (e.g., Nicholls and Lee 2006; Ozcaglar-Toulouse et al. 2006; Bray et al. 2011; Hassan et al. 2016). Although the addition of an ethical premium to the price of FT merchandise as a means of providing a remittance to marginalized producers/workers may be considered negatively by some consumers, others may regard it as a validation of the product which represent the fight against poverty, exploitation, and slavery (Thompson and Coskuner-Balli, 2007). Hence, we can argue that:

Empathy is defined as a moral emotion deriving from sentiments of wishing to benefit a person in need, thus demonstrating concern for the wellbeing of others (Batson et al. 1988). Empathy comprises two distinct dimensions, perspective taking and empathic concern. While perspective taking (or cognitive empathy) suggests that the empathiser is not only aware of, but also adopts the perspectives of the other person, in the case of empathic concern (or affective empathy) the empathiser has moral feelings and emotions of warmth and compassion toward others (Davis 1980, 1983; Kamdar et al. 2006; Chowdhury and Fernando 2014). In this study, we focus on empathic concern, because it allows the consumers to express the emotions of sympathy and compassion toward the plight of marginalized producers/workers, rather than perspective taking, which involves stepping in others’ shoes which, for proximity reasons (e.g., physical, social, etc.), may be unachievable for many consumers. Support for the argument that moral emotions of empathic concern, including compassion and care for others, enhance engagement with ethical decision-making is found in a number of studies (Batson et al. 1988; Basil et al. 2008; Mencl and May 2009; Kim and Kou 2014). For example, Mencl and May (2009), in a study involving HR professionals, found that empathic concern moderates the relationship between physical proximity and ethical decision-making process. In addition, in examining children’s attitudes toward FT, Nicholas and Lee (2006) found that empathic concern led to involvement with FT issues as they could psychologically relate to the issues of child labor.

Notably, high empathic concern will strengthen the positive impact of proximity on customer FT engagement, because it allows individuals to feel the plight of others and generate an affective state of unhappiness and sadness (Hwang and Kim 2018). To alleviate these negative feelings, consumers will seek to collaborate with and actively participate in FT networks (as activists, buyers, etc.), expressing in this way their affiliation and sympathy toward these marginalized producers/workers (Ballet and Carimentrand 2010; Peattie and Samuel 2018). On the other hand, lower levels of empathic concern dampen the impact of physical, social, and psychological proximity on consumer engagement with FT issues due to an indifference to the working conditions and living of marginalized producers/workers. In fact, people with low-empathic concern tend to care more about their self-interests rather than the interests of the society, weakening in this way the favorable impact of proximity on consumers engaging with the FT ethos (Doran 2010; Peattie and Samuel 2018). Hence, we can hypothesize that:

Hypocrisy is defined as a “motivation to appear moral yet, if possible, avoid the cost of actually being moral…. it allows one to engender trust, and still relentlessly pursue personal gain” (Batson et al. 2006, p. 321). It takes place when a person’s deeds do not align with their words (Wagner et al. 2009). In this vein, research pertaining to ethical consumption suggests that often consumers’ words and actions are misaligned, also known as the ‘attitude–behavior gap’, and that the desire to appear ethical is such that the consumer fabricates an untrue attitude—a false persona (Boulstridge and Carrigan 2000; Carrington et al. 2010, 2014; Hassan et al. 2016; Park and Lin 2018: Gamma et al. 2018). For example, in the context of FT, while on one hand some consumers are willing to pay more for local products (and therefore, price appeared less of an issue), they are not willing to pay extra for FT products since the consumer finds it easier to relate to buying local products (Adams and Raisborough 2010). Hypocritical consumers may also use the uncertainties associated with geographic and psychic distance from marginalized producers/workers as a pretext for failing to engage with FT.

This moderating role of hypocrisy can be justified by the fact that consumers who show inconsistent behaviors when compared to their attitudes toward FT are less likely to engage with the FT ethos, even though they perceive themselves to be close to marginalized producers/workers. Indeed, consumers who demonstrate higher levels of physical, social, and psychological proximity, when coupled with hypocrisy, tend not to tie in their behaviors the moral principles and social norms they advocate. As a result, this perceived proximity with the marginalized producers/workers may not necessarily translate into active involvement with the FT ethos, even if the consumers harbor positive attitudes toward FT (Carrington et al. 2014; Shaw et al. 2016). In contrast, in the case of consumers characterized by lower levels of hypocrisy, their proximity to marginalized producers/workers will show stronger effects on engagement with the FT ethos. This is because when consumers are less hypocritical, their moral beliefs are well aligned with their behaviors, thus displaying more commitment toward marginalized producers/workers (Hassan et al. 2016; Park and Lin 2018). Based on the above, we can hypothesize that:

Data were collected through a Qualtrics panel, a research agency having branches in multiple countries, using various sources to offer respondents for completing online surveys. This method of recruiting panelists for data collection has been used widely in business research recently and thus was deemed appropriate for the cross-country focus of our study (Ward and Dahl 2014; Kumar and Pansari 2016; Merz et al. 2018). Respondents were qualified to take part in the study if they were aware of the concept of FT and FT products. To avoid leading the consumers, we gave them options of ethical issues, such as animal rights, recycling, climate change, and FT, and only when the FT option was selected could consumers proceed to participate in the study.

While we acknowledge the limitations related to the use of such data collection methods, such as identity checks, response integrity, sampling issues, and data quality issues (Lowry et al. 2016), these were addressed by folllowing Smith et al.’s (2016) recommendations: first, we utilized attention check items to filter panelists that did not devote adequate time to answer all questions; second, we included reverse coded items for various constructs to avoid random responses; third, we eliminated responses that were completed very quickly as compared to the average response time, thus avoiding speeding issues; fourth, we checked for response duplications by reviewing respondents’ IP addresses; finally, we included several screening questions to ensure response quality.

With regard to the operationalization of constructs, we used only published, validated measures, with the exception for proximity where there was no readily available scale (see Table 5 in Appendix). Consumer FT engagement comprised 19 items adopted from So et al. (2014), measuring the four dimensions of engagement, namely, enthusiasm (4 items), attention (5 items), absorption (5 items), and interaction (5 items). FT purchasing behavior includes five items taken from Hollebeek et al. (2014); Sudbury-Riley and Kohlbacher (2016) and Zanjani et al. (2016) and adapted it to our context. Empathic concern was measured using a 7-item scale adopted by Kamdar et al. (2006), which in fact is a short version of the interpersonal reactivity index (IRI) developed by Davis (1980), which comprises 28 items. For the hypocrisy construct, we reviewed the measure of corporate hypocrisy developed by Wagner et al. (2009) and adjusted it from the perspective of consumers.

For the construct of proximity, in the absence of established scales due to limited empirical research on the relationship between proximity and ethical consumer decision-making, we had to develop a new scale that captures its three dimensions, namely physical, social, and psychological. In doing so, we followed the scale development and validation procedure suggested by Churchill (1979). First, we based the item generation phase on thoroughly reviewing the existing literature and carrying in-depth interviews with eight British and six Indian consumers. This has contributed to generating an exhaustive list of items that could be of relevance to each of the three dimensions of proximity.¹ This list of items was subsequently presented to a panel of five academic experts, specializing in ethical business issues, to obtain their views about their suitability to operationalize proximity. Some items had to be deleted, others had to be adjusted, while some other new items had to be added. The resulting list of items was allocated by the panel experts to each of the three dimensions of proximity. These were in turn thoroughly reviewed with the assistance of 15 British and 12 Indian consumers to confirm their nomological validity. This process resulted in obtaining four items for physical proximity, five items for social proximity, and five items for psychological proximity.

The questionnaire included a list of pre-coded questions which were reflecting the scales derived from the prior literature or were self-developed. These were measured on a 5-point Likert type scale, ranging from 1 (strongly disagree) to 5 (strongly agree). We also included a section in the questionnaire requiring information about the demographic characteristics of respondents, namely age, gender, education level, and residential area. Prior to starting the complete study, we pilot-tested the questionnaire with five British and five Indian consumers (using face-to-face interviews) to confirm its workability in terms of flow, structure, and length, revealing no problems.

Our fieldwork took place in two countries, namely the United Kingdom (a developed country) and India (an emerging economy), which overall are characterized by different levels of proximity from marginalized producers/workers based in developing countries. Altogether, we received 323 responses, 211 from the United Kingdom and 112 from India.² The demographic profile of the UK sample versus that of the Indian sample exhibited differences, attributable mainly to differences in the overall population structure of the two countries. Specifically, British respondents had the following demographic profile: gender (male: 46.2%, female: 53.8%), age group (18–24 years: 21.5%, 25–34 years: 25.7%, 35–44 years: 15.0%, 45–54 years: 20.0%, 55–64 years: 17.8%), and level of education (primary: 0.5%, secondary: 56.0%, undergraduate: 29.6%, postgraduate: 13.9%). On the other hand, the demographic profile of the Indian respondents was the following: gender (male: 58.5%, female: 41.5%), age group (18–24 years: 14.5%, 25–34 years: 59.1%, 35–44 years: 18.2%, 45–54 years: 4.6%, 55–64 years: 3.6%), and level of education (primary: 13.8%, secondary: 65.1%, undergraduate: 14.1%, postgraduate: 7.0%).

We adopted structural equation modeling (SEM) to test the proposed conceptual model and the hypothesized relationships between the constructs, owing to three major reasons: (a) it allows simultaneous analysis of relationships hypothesized between the constructs of the model in a comprehensive manner; (b) it is based on confirmatory instead of an exploratory approach to the data analysis; and (c) it considers explicit estimates of measurement error, whereby the model fit is regarded (Hair et al. 2014). The analysis was performed using the EQS statistical program, which, compared to other available programs, places less stringent assumptions on the multivariate normality of the data (Bentler 2008).

With regard to the measurement model, using confirmatory factor analysis (CFA), each item was restricted to load on its a priori specified factor, while the underlying factors were allowed to correlate (Anderson and Gerbing 1988) (see Table 1). To estimate the model, we used the elliptical reweighted least-squares (ERLS) procedure, which is superior to other estimation techniques (Stump and Heide 1996). The Chi square statistic was found to be significant (χ² = 1173.62, p = 0.000), which is not surprising given the sensitivity of this index of fit to sample size (Hair et al. 2014). However, the results of the alternative fit indices pointed to a good model fit, as all values were well within the commonly accepted critical levels (Byrne 2006). Specifically, the non-normed fit index (NNFI) was 0.95, the comparative fit index (CFI) was 0.97, and the root mean square error of approximation (RMSEA) was 0.07.

Construct reliability was satisfactory because all constructs had Cronbach’s alphas greater than the critical value of 0.70. The same was also true with regard to composite reliability, because all coefficients being greater than 0.65. In addition, we have examined convergent validity, which was verified because the t value for each item was high and significant, all standard errors of the estimated coefficients were very low, and the average variance extracted for each construct was equal to or above 0.45 (Hair et al. 2014). With regard to discriminant validity, this was verified because the confidence interval around the correlation estimate for each pair of constructs examined never included 1.00, while the squared correlation for each pair of constructs never exceeded their average variance extracted (Fornell and Larcker 1981; Anderson and Gerbing 1988).

To check for common method bias, we first employed the Harman’s single-factor test (Podsakoff and Organ 1986), where all questionnaire items were included in a principal component analysis with varimax rotation. Five separate factors with eigenvalues greater than 1.0 emerged from the unrotated factor solution, with these factors explaining 65.9% of the total variance (of which 19.8% was explained by the first factor). We also used a confirmatory factor approach, in which all items included in the measurement model were restricted to load on a single factor (Venkatraman and Prescott 1990). The model fit indices revealed very poor values, well below the commonly acceptable cut-off points (i.e., χ² = 3396.23, p = 0.000; df = 527; NFI = 0.78; NNFI = 0.79; CFI = 0.80; RMSEA = 0.14). Finally, we used the partial correlation technique (Lindell and Whitney 2001), where a theoretically unrelated construct served as a marker variable. This neither exhibited a significant correlation with any other constructs used in the model nor changed the significance of the correlation coefficients after having the partial correlation adjustments. Based on the results of the above three tests, we can safely argue that there is no problem with common method bias.

We tested the hypothesized associations between the constructs by estimating the structural model using the elliptical re-weighted least squares (ERLS) technique. While the Chi square for the model was found to be significant (χ² = 1230.74, p = 0.000), this was anticipated owing to the shortcomings of this statistic and thus is not solely relied upon when accepting or rejecting the model (Kline 2005). Therefore, we used the alternative fit indices, which showed an acceptable structural model fit (NFI = 0.94; NNFI = 0.96; CFI = 0.96; RMSEA = 0.08). The standardized path coefficients, along with corresponding t values of the structural model, are reported in Table 2. All the standardized path coefficients are statistically significant, thus providing support for the hypothesized paths.

The first three hypotheses examined the relationship between the three dimensions of proximity (i.e., physical, social, and psychological) with consumer FT engagement. With regard to H₁, physical proximity was found to be conducive to the formation of consumer FT engagement (β = 0.17, t = 2.74, p = 0.01). This suggests that physical proximity enhances the consumer engagement directed toward FT issues, thus providing support for our first hypothesis. This is in line with Adam and Raisborough’s (2010) study, where it was found that British consumers framed ethical consumption in terms of supporting physically proximate local British farmers because it benefits the local communities. Social proximity was also revealed to be positively linked with consumer FT engagement (β = 0.56, t = 6.76, p = 0.00), thus confirming H₂. This is concurrent with previous FT research which contends that, consumers may benefit individuals within their social group (in group), and hence may distance themselves from engagement with FT as they consider marginalized producers/workers out group (Doran 2010). Thus, increased levels of social proximity can also lead to consumer FT engagement. H₃ was also supported, since a positive relationship was established between psychological proximity and consumer FT engagement (β = 0.42, t = 5.36, p = 0.00). Our finding is consistent with the previous studies that the closer psychologically an individual feels toward the beneficiary and understand their situation, the more likely they are to engage with their cause (Mencl and May 2009; Ghorbani et al. 2013). Among the three dimensions of proximity, social proximity was the most significant predictor of consumer FT engagement, followed by psychological proximity and physical proximity.

Our study also confirmed that consumer FT engagement is conducive to the purchase of FT products (β = 0.84, t = 7.74, p = 0.00), thus confirming H₄. Indeed, this is in line with previous empirical research that argues that consumers who care about FT issues are more motivated to purchase FT products (Nicholls and Lee 2006; Bondy and Talwar 2011; Chatzidakis et al. 2016; Yamoah et al. 2016). Clearly, this indicates that consumers who actively engage with and support FT issues are willing to purchase FT products, even if this means paying a slightly extra price.

To examine moderating effects, we adopted the split-group approach, where the initial sample was divided into two subgroups using the median (Sharma et al. 2009). Two models were subsequently estimated for each hypothesized moderating effect: (a) a constrained model, where the path affected by the moderating variable was fixed to 1; and (b) a free model, where all paths of the structural model were allowed to be freely estimated. A significant Chi square difference between the two models implies that the moderator variable has a significant effect on the hypothesized association between constructs. The outcomes of this analysis are presented in Table 3.³

With regard to consumer empathic concern, our results show a strong moderating effect on the link between the three proximity dimensions and consumer FT engagement. In specific, the effect of physical proximity and consumer FT engagement is stronger when the consumers demonstrate higher levels of empathic concern (β = 0.36, t = 3.10, p < 0.01) than when they express lower levels of empathic concern (β = 0.23, t = 2.07, p < 0.05) (Δχ²= 2.94, p < 0.10). Besides, the interaction effect between social proximity and consumer FT engagement was marginally higher, when consumers display higher levels of empathic concern (β = 0.47, t = 4.33, p < 0.01) than they show lower levels of empathic concern (β = 0.45, t = 3.24, p < 0.01) (Δχ² = 0.52, p < 0.10). Further, empathic concern also exhibited significant associations between psychological proximity and consumer FT engagement. Results reveal that psychological proximity was more strongly related to consumer FT engagement when consumers demonstrated higher levels of empathic concern (β = 0.47, t = 3.51, p < 0.01) than when they feel lower levels of empathic concern (β = 0.18, t = 1.74, p < 0.10) (Δχ² = 5.07, p < 0.05). These findings concur with previous research, which suggests that consumers with empathic concern are more likely to recognize the distress of marginalized producers/workers, which enhances their involvement with FT issues (Nicholls and Lee 2006; Doran and Natale 2011; Lee 2016; Hwang and Kim 2018).

The moderating role of consumer hypocrisy on the relationship between each of the three proximity dimensions and consumer FT engagement was also supported. Our results confirmed that the effect of physical proximity on consumer FT engagement weakens in the case of consumers in the high hypocrisy group (β = 0.22, t = 1.98, p < 0.01), as opposed to consumers who are characterized by low hypocrisy (β = 0.36, t = 3.61, p < 0.01) (Δχ² = 3.13, p < 0.10). Similarly, the relationship between social proximity and consumer FT engagement wanes when consumers display higher levels of hypocrisy (β = 0.39, t = 3.38, p < 0.1) as opposed to lower levels of hypocrisy (β = 0.49, t = 4.42, p < 0.01) (Δχ² = 0.70, p < 0.10). Finally, the effect of psychological proximity and consumer FT engagement is weaker in consumers demonstrating higher levels of hypocrisy (β = 0.27, t = 2.24, p < 0.05) than those showing lesser levels of hypocrisy (β = 0.52, t = 4.28, p < 0.01) (Δχ² = 5.23, p < 0.10). Although there are hints in the literature that hypocrisy could influence ethical consumption (Carrington et al. 2016; Gamma et al. 2018), our findings offer a new explanation for the relationship between proximity dimensions and consumer FT engagement. This is in line with the arguments of Reczek et al. (2017), who found that people wilfully ignore information about ethical attributes to avoid emotionally difficult ethical information (e.g., child labor, exploited marginalized farmers) in purchase decisions, suggesting that wilfully ignorant memory is a more morally acceptable form of coping with ‘want/should’ conflict. In doing so, people stray from embedded beliefs and behavior to satisfy the want self’s desire to feel good, even though they may be physically, socially or psychologically proximate to the marginalized producers/workers (Hassan et al. 2016). Hence, the effect of proximity on consumer FT engagement is weakened by hypocrisy as the consumer is able to reconcile self-perception as hypocrite, by justifying and rationalizing his or her actions.

The role of consumer nationality, whether British or Indian, as a control variable on proximity, was also examined, revealing a significant effect on all three proximity dimensions. Specifically, compared to their British counterparts, Indian consumers perceived higher levels of physical (β = 0.86, t = 7.08, p = 0.00), social (β = 0.83, t = 8.37, p = 0.00), and psychological (β = 0.88, t = 7.46, p = 0.00) proximity to the marginalized producers/workers. This reinforces that India, being itself an FT producer and a consumer market for FT goods, is geographically nearer to marginalized producers/workers in the ‘global south’ and thus its consumers are more likely to be engaged with FT issues than British consumers (Doherty et al. 2015). The results relating to social proximity can also be explained by the fact that Indian consumers tend to share similar social experiences with marginalized producers/workers and show higher commitment to their wellbeing, as they are more likely to perceive the poor producers/workers as an in-group than British consumers (Doran 2009). Finally, the greater psychological proximity of Indian consumers to FT issues can be attributed to the fact that they may be able to understand plight of the marginalized producers/workers and show higher commitment to their wellbeing, thus, they would want to support them by engaging more with FT products (Nicholls and Lee 2006; Doherty et al. 2015).⁴