Outcomes for self-expressive brands followed on social media: Identifying different paths for inner self-expressive and social self-expressive brands

Brands provide a means for consumers to present the self, expressing their inner selves (the ‘real me’), or their social selves (how others see me). With the rise in social media use, academics and practitioners seek to understand how consumers’ self-expression through following brands online leads to brand outcomes. This study investigates the inner and social self-expressiveness of brands as separate antecedents of two outcomes: intention to co-create value and willingness to pay a premium price. Insights from 332 followers of brands on social media reveal that the inner and social self-expressiveness of brands drive these outcomes differently, as online brand engagement and brand trust mediate the relationship between self-expressive brands and brand outcomes, but only when the brand has an ability to express the inner self. The findings have implications for the study of selfexpressive brands and their influence on brand outcomes.


Introduction
Consumer behavior research has long recognized that consumers seek to express themselves through brands (Belk, 1988). Self-expressive brands are defined by the extent to which consumers perceive that a brand reflects their inner self ('the kind of person I truly am inside') or enhances their social self ('the way society views me') (Carroll & Ahuvia, 2006, p. 84, 85). As social media platforms have fundamentally altered the way that people interact with each other and how they present themselves, research has sought to explore how people associate with brands online and how they use brands for self-expression (Hollenbeck & Kaikati, 2012). The need for self-expression has actively motivated consumers to co-create brand value, as people driven by selfexpression motives are likely to be highly engaged with brands online (de Vries, Peluso, Romani, Leeflang, & Marcati, 2017). Studies have also demonstrated the effect of self-expressive brands on WOM, brand advocacy, brand love, and knowledge sharing with the firm (Carroll & Ahuvia, 2006;Wallace, Buil, & de Chernatony, 2014;Leckie, Nyadzayo, & Johnson, 2016;Choi & Burnham, 2020). Recent studies of selfexpressive brands in an offline context have distinguished between inner and social self-expressive brands (Leckie et al., 2016;Choi & Burnham, 2020), focusing on inner self-expressive brands and arguing for further consideration of social self-expressive brands (Leckie et al., 2016).
Research also acknowledges that consumers engage in different forms of self-expression through brands followed on social media (Hollenbeck & Kaikati, 2012). Recent studies have focused on inner selfexpression motives (de Vries et al., 2017;Algharabat, Rana, Alalwan, Baabdullah, & Gupta, 2020), and other research suggests that selfexpressive brands may have different effects on brand outcomes, depending on whether the brand is inner self-expressive or socially selfexpressive (Wallace et al., 2014). The distinction between inner selfexpressive and social self-expressive brands is arguably more important in the context of the social network, which is a very conspicuous form of self-presentation where consumers are freer in their presentation of the self and where they can associate with brands that are outside of their material realities (Schau & Gilly, 2003;Hollenbeck & Kaikati, 2012).
In line with recent studies of self-expressive brands and drawing on self-congruity theory (Sirgy, 1982;Malär, Krohmer, Hoyer, & Nyffenegger, 2011;Gaustad, Samuelsen, Warlop, & Fitzsimons, 2019), we assert the inner and social self-expressiveness of brands should be considered separately to determine the effects of brands that have an ability to express the inner or social self, on key brand outcomes. The nascent literature on the role of the brand's self-concept suggests that there is a need to further investigate these two forms of self-expressive brands, especially the relationship between the social selfexpressiveness of brands, and key brand outcomes. This study, therefore, tests hypotheses that investigate inner self-expressive brands and social self-expressive brands as separate antecedents of brand outcomes, specifically with the intention to co-create value with the brand (Tajvidi, Wang, Hajli, & Love, 2017) and a willingness to pay a price premium for the brand (Netemeyer et al., 2004). Willingness to pay a premium is a strong indicator of brand loyalty and is considered a summary measure of brand equity (Netemeyer et al., 2004). However, brands followed on social media may simply serve as identity cues (Hollenbeck & Kaikati, 2012) without any consumption in the consumer's material reality (Schau & Gilly, 2003). Therefore, the relationship between selfexpressive brands followed and consumers' willingness to pay is investigated. Furthermore, for brands on social media, consumer co-creation benefits corporate reputation, store visits, sales, and profitability (de Vries et al., 2017). We investigate the intention to co-create value as an outcome for self-expressive brands followed on social media.
The relationships we reveal in this study offer valuable and timely contributions to the literature on self-expressive brands followed on social media. We hypothesize that self-expressive brands that have the ability to reflect the inner self or to reflect the social self are separate antecedents of key brand outcome measures. Our results support our hypotheses. First, we show that online brand engagement and brand trust influence both the intention to co-create value and a willingness to pay when the brand followed has the ability to express the inner self. In contrast, we find that when brands followed express the social self, there is no significant association with online brand engagement or brand trust.
Second, we show that those who follow brands that have the ability to express a social self will co-create value and are willing to pay a premium price. We suggest that when the customer's focal brand is socially self-expressive, co-creating value facilitates further social interaction, for example through word of mouth, which may support greater social self-expression. Moreover, the willingness to pay a premium for brands with an ability for social self-expression, as revealed in our study, may indicate a desire to pay more for items that help them express a social image.
Third, the findings indicate that consumers whose focal brand has an ability for inner self-expression will only co-create value or pay a premium price if they are engaged with the brand or trust the brand. These findings are important, as they show a different path to value co-creation and a willingness to pay for brands with an ability for inner or social selfexpressiveness. As inner self-expressive brands reflect the 'real' self, they may indicate longer-term relationships, suggested by our finding that trust and engagement are required before the individual will co-create value for the brand or pay more for it. In contrast, social selfexpressive brands that are followed may create a more short-term relationship formed by a need to connect with others. The findings also open new avenues for future research to consider the inner and social self-expressiveness of brands, and to investigate different antecedents and outcomes.
Managerially, we suggest ways that strategies can be informed by the distinction between brands with the ability to express the inner or social self. We posit that the inner self-expressiveness of brands may help to indicate longer-term relationships, as trust and engagement are required for value co-creation and willingness to pay for these brands. Therefore, it would be desirable for managers to proactively attract more followers who believe the brand is inner self-expressive, and we offer suggestions for brand managers to attract these followers on social media. In addition, this research advocates that online brand engagement and brand trust would be incorporated as performance indicators, in addition to word of mouth or willingness to pay metrics. Finally, we show the value of followers of brands with an ability for social self-expression, as advocates and influencers on social media. We suggest ways that managers can harness the enthusiasm of these followers to help to share the brand's message across the social network.
The paper opens with a literature review, and we present our research hypotheses and our conceptual framework. Then, we describe the method used to elicit the attitudes of 332 followers of brands on social media, and we present the results of our measurement and structural models. Finally, our findings are discussed, and conclusions are drawn.

Brands and the virtual self
Cognizant that consumers may seek to express who they would like to be rather than who they actually are (Markus & Kunda, 1986) and that individuals use brands for self-expression (Aaker, 1999), the extant literature considers self-expressive brands in the cultivation of positive brand outcomes. Self-expressive brands that have an ability to reflect the social self are those that the consumer perceives to contribute to their image, reflect their social 'role', and positively impact how they are perceived by others or society (Carroll & Ahuvia, 2006). In contrast, selfexpressive brands that are able to reflect the inner self mirror the 'real' self, form an extension of their personality, and symbolize the kind of person they are inside (Carroll & Ahuvia, 2006, p.84/85).
Online social media also allows individuals to express the self through their posts and through the connections they make on the network (Schau & Gilly, 2003). When an individual follows a brand on a social network such as Facebook, the brand appears in their news feed, forming part of their online 'virtual self', as Belk (2013, p. 484) noted, 'Facebook is now a key part of self-presentation for one-sixth of humanity'.
Individuals who follow brands on social media may carefully select brands to curate a virtual identity that makes an impression on others because their behavior online is highly visible to their reference group, yet idealized versions of the self can be presented without disapproval from offline social circles (Hollenbeck & Kaikati, 2012). Therefore, it is possible that the brands that are followed may represent an aggregated self, and in such instances, the brand may serve a self-expressive function, allowing for online impression management without any offline consumption (Schau & Gilly, 2003). Indeed, Park and Kim (2014, p.460) caution that many individuals are not 'brand enthusiasts' or 'active loyalists' when connecting with brands online. In such instances, the potential 'superficial' nature of consumers' interactions with brands online raises unique and interesting questions about consumer brand relationships, whether consumers engage with or trust the brands they follow, and the implications of such engagement and trust for brand outcomes.
The conceptual model explored in this study investigates whether self-expressive brands positively influence brand outcomes, specifically the intention to co-create brand value and a willingness to pay a premium price. Moreover, it distinguishes between the inner and social selfexpressiveness of brands by investigating brands with these abilities as separate antecedents of brand outcomes. As previously noted, earlier studies of self-expressive brands investigated both inner and social selfexpressive brands as a global construct (Carroll & Ahuvia, 2006). More recent studies have sought to further understand how the presentation of the inner or social self with brands on social media may inform brand outcomes by separating inner and social self-expressive brands.
For example, research suggests that consumers' participation in fan pages is enhanced when they portray their own self-concept, leading to customer engagement (Jahn & Kunz, 2012). In other research, Wallace et al. (2014, p.39) found that followers of brands who believed they expressed their inner self on Facebook would not accept wrongdoing by the brand. In contrast, in the same study, consumers whose focal brand was socially self-expressive offered WOM while accepting wrongdoing by the brand, because the brand was not internalized (Wallace et al., 2014). Therefore, Wallace et al. (2014) advocated that further research investigating self-expressive brands and brand outcomes would distinguish between brands that have the ability to express the inner and social self in their hypotheses. Choi and Burnham (2020) considered inner and social self-expressive brands as different mediators in a model of brand reputation and consumer sharing behavior, revealing a stronger relationship between social self-expression and sharing behaviors when customers had higher levels of status seeking. Leckie et al. (2016) focused on the inner selfexpressive brand, revealing a positive relationship with the affective and cognitive processing dimensions of consumer brand engagement. While their study did not investigate socially self-expressive brands as a separate antecedent, Leckie et al. (2016) recognized a distinction between brands with the ability for inner or social self-expression. Clearly, extant research supports the consideration of the brand's ability for inner or social self-expression as separate antecedents of brand outcomes.
Furthermore, Leckie et al. (2016) argued for the consideration of selfcongruity theory as a way to inform research investigating separate components of self-expression and their effect on brand outcomes. Selfbrand congruence is the fit between the consumer's self-image and the brand image (Sirgy, 1982). The connection that a consumer feels with a brand affects the importance of the brand as a vehicle for their selfexpression (Gaustad et al., 2019). Strong consumer identification with a brand, characterized by a strong affiliation with the brand, results in positive brand outcomes (He et al., 2012). This insight is salient for our understanding of brands with the ability to express the inner self in particular. When brands can reflect the consumer's true selves, the consumer focuses on the fit between the actual self and the brand, and they are motivated to sustain the brand as it supports verification of the self (Malär et al., 2011). In contrast, when a consumer has strong selfenhancement motives, they may perceive a match between their ideal self and a brand (Malär et al., 2011). These consumers may be less concerned with changes to the brand (Gaustad et al., 2019). Therefore, self-congruity theory also suggests that brands reflecting the inner self and brands reflecting a social self may invoke different consumer behaviors towards the brand (Gaustad et al., 2019).
In line with recent approaches to self-expressive brands in the literature (for example Wallace et al., 2014;Leckie et al., 2016;Choi & Burnham, 2020) and drawing on self-congruence literature (Sirgy, 1982), we distinguish between the inner and social self-expressiveness of brands, investigating both as separate antecedents of both value cocreation and willingness to pay.
We also investigate the role of online brand engagement and brand trust as partial mediators in these relationships. We find that the mediating role of brand engagement and brand trust is different when brands have an ability for inner or social self-expression, and we suggest that followers of brands with an ability to express an inner self may be more focused on long-term relationships with the brand, necessitating trust and engagement, while followers of brands with an ability to express a social self may be more focused on building short-term relationships. The conceptual model and the hypotheses to be tested are presented in Fig. 1.

Self-expressive brands and consumer brand engagement
Customer engagement is a 'psychological state that occurs by virtue of interactive, co-creative customer experiences with a foci agent/object (e.g., a brand)' (Brodie, Hollebeek, Jurić, & Ilić, 2011, p. 260). Central to the concept is that the relationship is interactive (Hollebeek, Glynn, & Brodie, 2014) and social (Vivek, Beatty, Dalela, & Morgan, 2014). In the context of branding, Hollebeek (2011, p. 790) defines brand Conceptual model(SEI = Selfexpressiveness (Inner Self); SES = Selfexpressiveness (Social Self); OCBE = Online consumer brand engagement; OCBE Cog = Online consumer brand engagement cognitive; OCBE Aff = Online consumer brand engagement affective; OCBE Act = Online consumer brand engagement active; BT = Brand trust; BT Cred = Brand trust credibility; BT Inte = Brand trust integrity; BT Bene = Brand trust benevolence; WPP = Willingness to pay a premium price; ICC = Intention to cocreate brand value.) engagement as 'the level of a customer's motivational, brand-related and context-dependent state of mind that is characterized by specific levels of cognitive, emotional, and behavioral activity in brand interactions.' In this study, we posit a positive relationship between the selfexpressiveness of brands and consumer brand engagement. In two recent studies of mobile phone users, self-expressive brands were associated with customer brand engagement (Leckie et al., 2016;Nyadzayo et al., 2020). In the Leckie et al. (2016) study, the research focused on inner self-expressive brands as they argued that this form of selfexpressive brands is an extension of the self, and as self-endorsement motives increase engagement, inner self-expressive brands would increase engagement. Their findings supported this hypothesis, and they advocated that future research would also separately consider social self-expressive brands as an antecedent.
Our study investigates customer engagement with brands followed on social media. In the context of online brand engagement, Mollen and Wilson (2010) describe sustained cognitive processing as central to engagement. In the current study, we conceptualize consumer brand engagement in line with Brodie et al. (2011). That is, we investigate individuals' engagement with the brand that they follow on social media rather than with the social media itself. Central to this view of engagement is the contention that it is active, motivational, and multidimensional, comprising cognitive, affective, and behavioral dimensions (Hollebeek et al., 2014). As engagement relates to brands followed online, we use the term online consumer brand engagement in our study.
To date, much of the online brand engagement literature focuses on the brand community (e.g., Cova, 1997;Muniz & O'Guinn, 2001;Dessart et al., 2015), which includes groups identified on the basis of the identification or commonality of members (McAlexander, Schouten, & Koenig, 2002). Dessart et al. (2015) identified the relationship with the brand and the congruence between oneself and the brand's values as an impetus for engagement within the community. In studies of fan pages, Jahn and Kunz (2012) identified a positive association between the expression of the self-concept on fan pages and consumer engagement with fan pages. However, these studies investigate engagement with the community or the fan page rather than with the brand that is followed.
Research on Facebook brand followers by Wallace et al. (2014) shows that following self-expressive brands on social media was positively associated with affective and behavioral components of engagement. Further studies have investigated the relationship between selfexpressive brands and customer brand engagement (Algharabat et al., 2020;France, Merrilees, & Miller, 2016). Notably, Algharabat et al. (2020) focused on inner self-expressive brands. They explained that when social media pages allow consumers to express their inner selves, this enhances their engagement, as they can express their values and self-concepts (Algharabat et al., 2020).
Therefore, recent literature suggests that there is merit in the consideration of self-expressive brands as an antecedent of online consumer brand engagement. In particular, the findings of Algharabat et al. (2020) and Leckie et al. (2016), would support the separate consideration of the inner and social self-expressiveness of brands as antecedents of online consumer brand engagement. To our knowledge, while the inner self-expressiveness of brands has received attention, less is known about the role of brands with the ability for social self-expression as an antecedent of online consumer brand engagement. The current study contributes to existing knowledge by exploring the relationship between self-expressive brands followed on social media and online consumer brand engagement. Building on the extant literature (for example Wallace et al., 2014;Leckie et al., 2016;de Vries et al., 2017;Algharabat et al., 2020), we also consider the inner and social self-expressiveness of brands as separate antecedents of online consumer brand engagement. We hypothesize the following: H1a: There is a positive relationship between the inner selfexpressiveness of brands followed on social media and online consumer brand engagement. H1b: There is a positive relationship between the social selfexpressiveness of brands followed on social media and online consumer brand engagement.

Self-expressive brand and brand trust
Brand trust is a core construct in our study because trust is central to forming strong brand relationships (Morgan & Hunt, 1994). Brand trust has been defined as a psychological state or 'confident expectations of the brand's reliability and intentions in situations entailing risk to the consumer' (Delgado Ballester, Yagüe, & Munuera-Alemán, 2003, p. 37). Implicit in this definition is a belief that the brand is reliable, and consumers are motivated to retain a relationship with the brand, even in the event of unexpected problems associated with the brand (Delgado Ballester et al., 2003). Trust is a vital component of consumer-brand relationships (Fournier et al., 2012). Brand trust is also a recognized antecedent of positive outcomes such as brand loyalty, market share, and relative price (Chaudhuri & Holbrook, 2001); therefore, it is a useful construct in the investigation of brand outcomes, such as willingness to pay and value co-creation.
Although the role of brand trust in interactions such as technology acceptance, online brand experience (Morgan-Thomas & Veloutsou, 2013), and e-commerce (Ha, 2004;Becerra & Korgaonkar, 2011) has been explored, an area where brand trust has received less attention is in the following areas: brands on social media. This is surprising given the ubiquity of social media, the ease of individuals' ability to connect with brands on sites such as Facebook and Instagram, and the importance of social media for the self (Hollenbeck & Kaikati, 2012).
In this study, we are focused on trust in the brand followed by social media rather than trust in the social medium itself. In particular, we focus on self-expressive brands. In the extant literature, self-expressive brands on social media have been considered 'superficial' or 'virtual', where consumers follow the brand to curate a self that may be unrelated to the real self (Schau & Gilly, 2003). Consumers have an opportunity to display either an authentic self or a strategically chosen self when representing themselves through brands they associate with on social media (Hollenbeck & Kaikati, 2012). Brand trust is associated with honesty (Chaudhuri & Holbrook, 2001); therefore, we considered it interesting to investigate self-expressive brands and, especially, to consider whether those who would express their 'real self' or a social self through the brands that they follow, would exhibit brand trust. The extant literature suggests that self-expressive brands influence brand trust. Social identity theory advocates that consumers who identify with a company are motivated by self-definitional ('who am I?') needs (Bhattacharya & Sen, 2003). He et al. (2012) found a strong relationship between identity and brand trust, with brand trust mediating the identification effect on brand outcomes. When consumers identify with a brand, this engenders their trust in the brand (He et al., 2012). Drawing on this finding, Leckie et al. (2016) argued that consumers who had strong brand identification are more likely to develop trust, ultimately leading to greater loyalty. They hypothesized that selfexpressive brands would therefore have a positive effect on loyalty, and for the same reason, they advocated that further research would investigate brand trust (Leckie et al., 2016). Additionally, Baek et al. (2010) found that consumers who purchased more self-expressive products tended to focus on brand credibility, a component of brand trust (Gurviez & Korchia, 2002). Belk (1988) posited that the things people attach to become part of an extended self, with items having varying degrees of centrality to that self-concept. Extending these ideas to the self on social media, Belk (2013) highlighted that different facets of the self can be expressed online, and the disinhibition provided by social media allows consumers to reveal a 'true self' better than in an offline context. Additionally, Belk (2013, p. 483) explains that our control of our digital self leads us to feel that 'these things (possessions) are part of us'. If we express our true selves through the items we associate with, would we have greater trust in these items? Both the self-congruity literature and brand community literature are helpful in considering the role of brand trust for brands expressing the self on social media. Rabbanee, Roy, and Spence (2020) found that, on social networks, self-congruence with the actual or social self leads to brand attachment, and they asserted that this attachment may lead to the creation of brand trust. Akrout and Nagy (2018) found that consumers who connect with communities on social media gain hedonic benefits through affective gratification, leading to greater trust in the community. Furthermore, Laroche, Habibi, Richard, and Sankaranarayanan (2012) found that brand use in online brand communities positively influences brand trust.
We propose that the self-expressiveness of brands (distinguishing between inner or social self) is positively associated with brand trust when brands are followed on social media. We suggest that the ability for inner or social self-expressive brands would be considered separate antecedents of brand trust. Wallace et al. (2014) found that followers of brands with an ability for inner self-expression would not accept wrongdoing by the brand and posited that this may suggest a 'real' brand relationship between those brands and brand outcomes. This is in contrast to their finding regarding brands with an ability for social selfexpression, where followers accept wrongdoing and engage in WOM because the brand is not internalized (Wallace et al., 2014). Earlier, we noted that brand trust implies that consumers are motivated to retain a relationship with the brand, even in the event of unexpected problems associated with the brand (Delgado Ballester et al., 2003). Therefore, the relationships between the inner or social self-expressiveness of brands and brand trust warrant further investigation.
Given the importance of brand trust for consumer brand relationships, the relationship between self-expressive brands and brand trust (Baek et al., 2010;Bhattacharya & Sen, 2003;He et al., 2012;Leckie et al., 2016), and the nascent studies that suggest different effects on brand outcomes between inner and social self-expressive brands (Wallace et al., 2014;Leckie et al., 2016;Choi & Burnham, 2020), we hypothesize a relationship between self-expressive brands and brand trust, distinguishing between brands' ability for inner or social selfexpression. Our hypotheses are as follows: H2a: There is a positive relationship between the inner selfexpressiveness of brands followed on social media and brand trust. H2b: There is a positive relationship between the social selfexpressiveness of brands followed on social media and brand trust.

Self-expressive brands and brand outcomes: Intention to co-create value and willingness to pay a premium price
The literature highlights the dissonance between brands followed on social media and consumers' offline behavioral intentions towards those brands. In particular, outcomes such as advertisement by word of mouth and brand love (Carroll & Ahuvia, 2006) have been well explored. However, two brand outcomes that have received less attention are the intention to co-create brand value and the willingness to pay a premium price for the brand. This is surprising given the importance of consumers' willingness to pay for the financial success of the firm and the emphasis on the importance of value co-creation in the extant literature. For example, Grönroos (2011) argued that the unique contribution of value cocreation was not just that customers create value but that firms obtain opportunities to co-create value with their customers. In the same way, consumers who intend to co-create allow firms to co-create value with them, and this may be especially important when those consumers are connected to the company through its brands on social media. Moreover, de Vries et al. (2017) highlighted that self-expression motives were drivers of consumers' content creation on social media.
Brand value co-creation has its origins in the value co-creation literature (e.g., Vargo & Lusch, 2004), which considers the customer as a source of value through collaboration, dialogue, and interaction. On social media, brand value co-creation is value that is co-created 'though consumers' engagement in specific interactive experiences' and specific brand-related experiences, 'triggered by the new design features of social commerce' (Tajvidi et al., 2017, p. 2). Co-creation involves a deep sense of participation and immersion for the consumer, where the audience coproduces content (Thorpe & Roper, 2019). However, while value co-creation is often investigated offline, there is a paucity of information about consumers' value co-creation on social media (Tajvidi et al., 2017). Moreover, value co-creation has been considered in its role of self-concept expression, but only in the context of social support for the self, rather than expression of the self (Tajvidi et al., 2017). This is surprising, as the online space presents a useful medium for value cocreation, and the concept of co-creation has received increasing research attention from the perspective of consumers' motivated and active participation in response to marketing activity (Thorpe & Roper, 2019). Furthermore, value co-creation has been found to enhance consumer happiness (Hsieh, Chiu, Tang, & Lin, 2018); therefore, positive interactions with brands and with other consumers of those brands may be self-reinforcing. The current study therefore investigates value cocreation as an outcome of self-expressive brands followed on social media in the presence of brand relationships.
To expand these insights, we considered it interesting to investigate whether following self-expressive brands would influence value cocreation, considering brands' ability for inner or social self-expression as separate antecedents. In the related literature, both inner and social self-expressive brands have influenced customer voluntary sharing behaviors. Moreover, under the condition of status seeking, social selfexpressive brand perceptions translate into social influence behaviors (Choi & Burnham, 2020). Although the focus of this study is on the intention to co-create value and we are not investigating status seeking, we nevertheless considered that an investigation of the relationships between the inner or social self-expressiveness of brands and consumers' intention to co-create value would add to this literature. Co-creation of value may be inspired by inner self-expressive motives, as these activities may help to express one's personal identity (Belk, 1988). As cocreating value with others also allows the consumer to connect with others, it can be hypothesized that this co-creation also helps to communicate a social self to others. We therefore hypothesize the following: H3a: There is a positive relationship between the inner selfexpressiveness of brands followed on social media and the intention to co-create brand value. H3b: There is a positive relationship between the social selfexpressiveness of brands followed on social media and the intention to co-create brand value.
The study also investigates willingness to pay a premium price as an offline outcome of following brands with an ability for inner or social self-expression. Willingness to pay a premium price is 'the amount a customer is willing to pay for his/her preferred brand over comparable/ lesser brands of the same package size/quantity' (Netemeyer et al., 2004, p. 211). Willingness to pay a premium price is also a direct antecedent of brand purchase behavior. Given the asserted discrepancy between what one follows on social media and what one buys (Schau & Gilly, 2003), we ask: would individuals pay more for a self-expressive brand that they follow on social media? We draw on extant literature related to brand loyalty to support our hypothesis. Kressmann et al. (2006) concluded that when consumers expressed themselves through a brand, they had more brand loyalty. When consumers have greater brand loyalty and perceive that a brand offers some unique value, they are willing to pay more for it (Chaudhuri & Holbrook, 2001). Indeed, willingness to pay a premium price is considered an indicator of brand loyalty (Netemeyer et al., 2004). Swimberghe, Astakhova, and Woolbridge (2014) found that selfexpressive brands were positively associated with a greater willingness to pay a premium under conditions where the consumer experienced passion for the brand. However, their study focused on offline brands, and they did not consider the direct effects of self-expressive brands on consumers' willingness to pay a premium. To further advance this theory, we posit that those who follow brands on social media that have an ability for inner or social self-expression will also be willing to pay more for those brands in an offline context. We hypothesize that: H4a: There is a positive relationship between the inner selfexpressiveness of brands followed on social media and the willingness to pay a premium price for that brand. H4b: There is a positive relationship between the social selfexpressiveness of brands followed on social media and the willingness to pay a premium price for that brand.

The mediating role of online consumer brand engagement and brand trust in influencing the intention to co-create value and the willingness to pay a premium price
The study considers the effect of online consumer brand engagement and brand trust on brand outcomes. While one might reasonably expect that brand engagement would lead to positive brand outcomes, given this outcome in the extant literature (see, for example, Dessart, Veloutsou, & Morgan-Thomas, 2016;Brodie et al., 2011;Hollebeek et al., 2014), this study also focuses on the mediating role of brand trust. We investigate brand trust as an antecedent of the intention to co-create value and the willingness to pay a premium price in the context of followers of self-expressive brands because the literature asserts that those followers may be presenting a curated self (Hollenbeck & Kaikati, 2012). It is recognized that the version of the self that is presented on social media may or may not be an accurate reflection of those individuals' material realities (Schau & Gilly, 2003). Little is known about this relationship. Will these consumers pay more for the self-expressive brands that they follow? Will they proactively co-create value for the brand? To investigate these questions, we present further hypotheses below.
Value co-creation studies have taken an engagement perspective, where the consumer generates value-in-use through their engagement (Grönroos & Voima, 2013). More recently, Tajvidi et al. (2017) noted that engagement arising from consumer-to-consumer interaction can facilitate value co-creation.
Conceptualizations of consumer brand engagement consider that it requires a level of energy, effort, and time through sharing and endorsing the brand (Hollebeek et al., 2014;Dessart et al., 2016). Indeed, Brodie et al. (2011) definition of engagement suggests that engagement involves co-creation with a focal object such as a brand. In addition, given the importance of engagement for value co-creation in the broader marketing literature (Vargo & Lusch, 2004;Christodoulides & de Chernatony, 2010;Tajvidi et al., 2017), we posit the following: H5: For brands followed on social media that have an ability for selfexpression, the level of online consumer brand engagement has a positive impact on the intention to co-create brand value.
As the intention to co-create value is a relatively new idea, we also drew on the research on word-of-mouth advertising to consider the relationship between brand trust and the intention to co-create value on social media. Offline, Albert, Merunka, and Valette-Florence (2013) identified a positive association between brand trust and word-ofmouth, which suggests that those who trust the brand feel motivated to advocate for it to others. In their recent conceptualization of the intention to co-create value for the brand on social networks, Tajvidi et al. (2017) theorize a likely relationship between trust and the intention to co-create brand value. To further investigate the role of brand trust in this relationship, we hypothesize the following: H6: For brands followed on social media that have an ability for selfexpression, the level of brand trust has a positive impact on the intention to co-create brand value.
The extant literature has identified that consumer brand engagement has a significant relationship with positive behavioral outcomes for the brand, including consumers' willingness to pay a premium price for the brand (Risitano, Romano, Sorrentino, & Quintano, 2017). Less is known about the relationship between online consumer brand engagement and willingness to pay a premium price. Park and Kim (2014) found that interactions with brands on social media were not a strong driver of the willingness to pay a premium price, and they advocated further research to investigate the association between relationships with brands on social media and an offline willingness to pay. As the relationship between brand engagement and loyalty is well established (Hollebeek et al., 2014) and loyalty is associated with the willingness to pay a premium price (Chaudhuri & Holbrook, 2001), we suggest that online consumer brand engagement would be positively associated with a willingness to pay a premium price, and we hypothesize the following: H7: For brands followed on social media that have an ability for selfexpression, the level of online consumer brand engagement has a positive impact on consumers' willingness to pay a premium price for that brand.
The study also considered whether brand trust would be associated with a willingness to pay more for the brand. Offline, brand trust has long been associated with a willingness to pay a premium, due in part to its role in supporting brand loyalty (Chaudhuri & Holbrook, 2001). In the context of the virtual world, where we investigate the relationship between following self-expressive brands and the willingness to pay a premium price, we question whether trust in those brands would foster a willingness to pay.
H8: For brands followed on social media that have an ability for selfexpression, the level of brand trust has a positive impact on consumers' willingness to pay a premium price for that brand.
Finally, we explore the direct relationship between online brand engagement and brand trust. Although the extant literature questions whether customer engagement has stronger ties with brand trust than brand engagement (Dessart et al., 2016), trust has been identified as a relational consequence of consumer engagement, even when the engagement with the focal object is with a brand (Brodie et al., 2011). Interactive communication between those connected to a brand on social media can enhance brand trust (Park & Kim, 2014). Therefore, to test the relationship between engagement with the brand and brand trust, and in the context of self-expressive brands, we hypothesize the following: H9: For brands followed on social media that have an ability for selfexpression, online consumer brand engagement positively influences brand trust.

Research method
The study was conducted among social media users who followed brands in Portugal. Participants were invited to complete a survey issued through their social network and were asked to answer questions while thinking about the brands they follow on social media. Where more than one brand was followed, participants were asked to think about the brand they followed that was most salient to them. The answers were collected from the 12th of February to the 21st of May 2018. From the 391 answers obtained, 332 were considered valid for the purpose of the study. Those cases omitted did not mention a brand or had a high number of missing values. All the participants included in the sample followed at least one brand on social media. Female respondents accounted for 55% of the respondents. The majority of respondents' ages, approximately 67%, were in the range of 23-37 years old. Workers accounted for 39% of all respondents, students accounted for 59% of respondents, and 4% of those who replied were unemployed. The sample was composed of mostly heavy social media users, with 56% of respondents spending at least three hours online each day. In terms of social media platforms, participants followed brands on Instagram (50%) and Facebook (40%), with the remainder of brands followed on other social media sites, including Twitter and Pinterest. The demographic profile of the respondents is presented in Table 1.

Measures
We used existing scales to measure all of the constructs. For selfexpressive brands, we used the established measure by Carroll and Ahuvia (2006). The scale incorporates four items to measure brands that have an ability for inner self-expression, including statements like 'this brand mirrors the real me,' and four items to measure brands that have an ability for social self-expression, including statements like 'this brand adds to the social 'role' I play'.
For brand trust, we used the scale by Gurviez and Korchia (2002). This scale captures the complexity of brand trust by considering three second-order dimensions: credibility, integrity, and benevolence (Gurviez & Korchia, 2002); it is in line with other existing literature (Albert et al., 2013). Items in this scale include statements like 'I trust the quality of this brand's products' (credibility), 'this brand is sincere with its customers' (integrity), and 'I think this brand improves continuously its response to customers' needs' (benevolence).
Online consumer brand engagement was measured based on the items presented in the scale by Hollebeek et al. (2014). The scale captures cognitive ('when I am interacting with the brand on social media, I think a lot about this brand'), affective ('I feel very positive when I am interacting with this brand'), and active ('whenever I am online on social media, I usually look for that brand') dimensions of online consumer brand engagement. It should be noted that Hollebeek et al. (2014) original scale was developed specifically featuring LinkedIn as the brand; however, in this study, participants answered the questions thinking about the most prevalent brand they followed on social media. Consumers' willingness to pay a premium price was measured using the scale by Chaudhuri and Ligas (2009) and adapted by Park and Kim (2014). Items include statements like 'I would be willing to pay a higher price for this brand over other brands' and 'I prefer to purchase this brand even if another brand advertises at a lower price'.
Finally, consumers' intention to co-create brand value was measured using the consumer-consumer interaction scale by Hajli (2013) as presented in Tajvidi et al. (2017) and adapted for social media interaction with a friendship social group. Items on the original scale include statements like 'I would like to use people's online recommendations to buy a product from a brand', which was adapted to 'I am willing to buy the products of a brand recommended by my friends on my favorite social networking site'.
All items were measured on 7-point Likert scales with anchors of 'strongly disagree' and 'strongly agree'.

Nonresponse bias and common method bias
We followed the approach recommended by Armstrong and Overton (1977) to test nonresponse bias. This approach is widely used (e.g., Alayo, Maseda, Iturralde, & Arzubiaga, 2019; Augusto, Godinho, & Torres, 2019). The means obtained in two subsamples (the total sample was split in half) were compared using the t-test for equality of means. The results of the t-test show, with three exceptions, no significant differences at the conventional significance level (5%) or between the means of the two groups of the 26 items used to measure the model constructs. Thus, nonresponse bias was not a major problem in this study.
Common method variance bias can occur when the variance of the responses is systematically attributable to the single measurement method used (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003;Podsakoff, MacKenzie, & Podsakoff, 2012). The sample was collected using selfreported data from the same respondents using the same instrument; therefore, common method variance was assessed. To address this potential bias, ex ante and ex post procedures were employed. Ex ante, following Podsakoff et al. (2003), used control procedures which included the following steps: (1) the survey was pretested to define ambiguous terms and avoid vague concepts, complex syntax, and double-barreled questions, and keep each question simple, specific, and concise; and (2) on the first page of the questionnaire, respondents were assured that the answers were anonymous and advised that there were no right or wrong answers to each question. According to Podsakoff et al. (2003), ex post, the CMV can be tested using different techniques, such as Harman's single factor test, the correlational marker technique, a single unmeasured latent method factor, and multiple method factors. Harman's single-factor test, which is the most frequently used approach (Fuller, Simmering, Atinc, Atinc, & Babin, 2016), was employed in this study. The exploratory factor analysis without rotation shows four factors with eigenvalues greater than 1.0 that accounted for 74.11% of the variance. The first factor accounts for 45.68% of the variance, which is below the 50% threshold suggested by Podsakoff and Organ (1986). Therefore, based on both the ex-ante and ex post procedures used, we have determined that common method variance is not a major issue in this study.

Results
Our study considers the extent to which a consumers' focal brand on social media is inner self-expressive and the extent to which it is socially self-expressive, and the impact of this on outcome variables, including the intention to co-create value and the willingness to pay a premium price. The extant literature considers that self-expressive brands encompass both inner self-expressive and social self-expressive Notes: Stand. loads = Standardised loadings. CR = Composite reliability; AVE = Average variance extracted. SEI = Self-expressiveness (Inner Self); SES = Self-expressiveness (Social Self); OCBE = Online consumer brand engagement; OCBE Cog = Online consumer brand engagement cognitive; OCBE Aff = Online consumer brand engagement affective; OCBE Act = Online consumer brand engagement active; BT = Brand trust; BT Cred = Brand trust credibility; BT Inte = Brand trust integrity; BT Bene = Brand trust benevolence; WPP = Willingness to pay a premium price; ICC = Intention to co-create brand value. Model global fit: Chi-square (χ2) = 578.73; df = 278; goodness of fit index (GFI) = 0.878; incremental fit index (IFI) = 0.962; Tucker-Lewis index (TLI) = 0.955; comparative fit index (CFI) = 0.961; root mean square error approximation (RMSEA) = 0.057. components, and both components are presented in the measure presented by Carroll and Ahuvia (2006). That is, brands that are selfexpressive can reflect the 'true' self (the 'real me') or the social self (the social role I play) (Carroll & Ahuvia, 2006).
As noted earlier, to fully explore the potential for self-expressive brands to influence brand relationships and brand outcomes, and building on nascent literature (Wallace et al., 2014, Leckie et al., 2016, Choi & Burnham, 2020, we consider the inner and social selfexpressiveness of brands as separate antecedents in our research hypotheses. In line with the literature, online consumer brand engagement was considered a second-order construct (e.g., Hollebeek et al., 2014); it includes the following dimensions: cognitive (OCBECog), affective (OCBEAff), and active (OCBEAct). Brand trust was also treated as a second-order construct, taking into account the related literature (e.g., Gurviez & Korchia, 2002). Brand trust therefore included the following first-order constructs: credibility (BTCred), integrity (BTInte), and benevolence (BTBene).
The proposed global model comprises two components: the measurement component and the structural component. Following Anderson and Gerbing (1988), the two components of the global model were estimated separately. These estimations were performed using AMOS 25.0 software and the maximum likelihood estimation method. This method relies on the hypothesis of multi-normality distribution of the observed variables. However, the literature (e.g., Kline, 2017) advocates that this estimation method provides robust estimates for both parameters and standard errors when the departure of the multi-normality assumption is not severe (skewness < 3.0 and kurtosis < 20.0). The skewness ranges from − 0.99 to 0.34, and the kurtosis ranges from − 1.17 to 0.38. Thus, the departure from the multi-normality distribution of the observed variables is not a major problem in the use of the maximum likelihood estimation method.

Measurement model results
A preliminary data analysis was then performed to detect items that were poorly correlated with the other items of the same scale. This analysis led to the elimination of some items from the original scales. From the original scale used to measure self-expressiveness, one item was dropped in both the brand's ability for inner self-expression and the brand's ability for social self-expression: 'this brand symbolizes the kind of person I truly am inside' and 'this brand contributes to my image', respectively. Regarding online consumer brand engagement, each of the three dimensions includes three items; only one item was dropped in OCBEAff, which has 4 items in the original scale. The dropped item was 'following this brand's Facebook account makes me happy'. The dimensions of brand trust include two items. For BTCred, the item 'This brand's products make me feel safe.' was deleted. Regarding BTInte, one item, 'this brand expresses an interest in its customers', was dropped. The scales for the intention to co-create value and willingness to pay a premium price are consistent with the extant literature.
To assess the global fit of each measurement model, multi-fit measures were used, including, chi-squared (χ 2 ), the goodness of fit index (GFI), the incremental fit index (IFI), the Tucker-Lewis index (TLI), the comparative fit index (CFI), and root mean square error approximation (RMSEA). Table 2 shows the theoretical constructs (1st and 2nd order), the final items used in the analysis, the estimated results of the measurement model (standardized loadings, the t-statistics, and the R 2 ), Cronbach's alpha coefficients, the composite reliabilities (CR), and the average variance extracted (AVE). Although the chi-square was statistically significant (χ2 = 578.73; df = 278, p < 0.01), the model performed well to the data, considering the other most popular goodness-offit statistics (IFI = 0.962, GFI = 0.878, TLI = 0.955, CFI = 0.961, and RMSEA = 0.057). Regarding the particular aspects of the model fit, the standardized loadings were larger (all exceeded the 0.50 threshold) and were all highly significant (p < 0.01), and R 2 estimates were all above the 0.20 threshold. These results provided support for the convergent validity of the measured variables. Cronbach's alpha values and CR and AVE estimates are also presented in Table 2. The value of Cronbach's alpha ranged from 0.813 to 0.968, and the CR varied between 0.813 and 0.968; both exceeded the 0.70 threshold. These results support the internal consistency of the scales (Fornell & Larcker, 1981). The AVE estimates range from 0.611 to 0.847; therefore, they are larger than the maximum shared variance (see Table 3). Thus, discriminant validity is also supported (Fornell & Larcker, 1981). Table 4 shows the standardized structural coefficient estimates, the tstatistics, and the summary of the hypotheses tests. The different goodness-of-fit statistics used provide support for the structural model. Although the chi-square is statistically significant (χ2 = 584.11; df = 279, p < 0.01), the remaining overall model statistics suggest an acceptable model fit to data collected in the sample (IFI = 0.961, GFI = 0.877, TLI = 0.954, CFI = 0.961, and RMSEA = 0.057). From the 13 hypotheses, 9 are statistically significant at the conventional significance level. An inspection of the modification indices reveals that no other path is statistically significant at the conventional significance level. Therefore, the results support the proposed model.

Structural model
The results show that there are different paths for a brand's inner or social self-expressiveness to effect brand outcomes. Brands with inner self-expressiveness have a positive effect on both online consumer brand engagement and brand trust, but the effect on the intention to co-create value and the willingness to pay a premium price is mediated by online consumer brand engagement and brand trust. In contrast, brands with social self-expressiveness have no direct effect on online consumer band engagement and brand trust, yet they positively impact both the intention to co-create value and the willingness to pay a price premium. Therefore, hypotheses 1a, 2a, 3b, and 4b are supported, while hypotheses 1b, 2b, 3a, and 4a are not supported. All the other hypotheses are supported. Online consumer brand engagement has a positive effect on the intention to co-create value (H5), willingness to pay a premium price (H7), and brand trust (H9). Brand trust positively influences the intention to co-create value (H6) and willingness to pay a premium price (H8).

Discussion
Our study reveals a number of findings to advance the extant literature. For self-expressive brands followed on social media, the findings reveal different outcomes in terms of a brand's ability for inner or social self-expression. We hypothesized that the inner and social selfexpressiveness of brands were separate antecedents of online brand engagement and brand trust. willingness to pay a premium price and the intention to co-create value. To the best of our knowledge, this study is the first to separately investigate the inner and social self-expressiveness of brands as antecedents of these brand outcomes.
The main contribution of our study is showing that brands followed on social media that have an ability for inner or self-expression follow different paths to brand outcomes. The findings show that a consumer whose focal brand is inner self-expressive will engage in value cocreation, as the brand reflects the 'real me' (Carroll & Ahuvia, 2006). When the consumer perceives that the brand is intrinsic to themselves, they intend to co-create value, but only when they are engaged with the brand and trust the brand. Likewise, the results show that followers of brands that reflect an inner self are willing to pay a premium price, but only when they are engaged with, and trust the brand. Therefore, we identify a vital mediating role for online brand engagement and brand trust in the relationship between inner self-expressive brands and both the intention to co-create value and the willingness to pay a premium price.
Our findings also support a positive relationship between brands that consumers perceive are socially self-expressive and both intention to cocreate value and willingness to pay a premium price. However, in contrast to brands with an ability for inner self-expression, where this relationship was positive only when the consumer trusted the brand or engaged with it, we find an insignificant relationship between brands with an ability for social self-expression and both brand trust and online consumer brand engagement. However, followers of brands that reflect a social self will still pay more for the brand and will co-create value for the brand. We draw on the extant literature to explain this finding. Consumers who wish to express a social self may have a deep need to connect to others and to create a positive impression on others (Carroll & Ahuvia, 2006). Individuals with a need for close interpersonal relationships may be more willing to adjust their spending to be in line with those they wish to interact with, or they may engage in conspicuous consumption if they feel that their social status is threatened (Ahuvia, 2015). We suggest that these motives also explain the relationship between those who follow brands with an ability for social self-expression and their willingness to pay more for those brands. Furthermore, cocreating value with others allows the consumer to connect with others. This co-creation may express a social self to others, just as following the brand on social media expresses their social selves. Therefore, the social self-expressiveness of brands allows the individual to connect with others, fulfilling their own needs, resulting in the intention to co-create value and a willingness to pay a premium as positive brand outcomes.
Our findings also reveal a positive relationship between both online brand engagement and brand trust on the outcomes intention to cocreate value and willingness to pay a premium. The findings also support the assertion that brand engagement is positively associated with brand trust, supporting Dessart et al. (2016), who also proposed brand engagement as a predictor of online brand trust. From an ethical perspective, Thorpe and Roper (2019) caution against firms encouraging 'hyper-engagement', for instance through gamification, and advocate that firms should focus on transparency instead of trying to meet advertising tenets such as honesty and truthfulness, and to avoid any forms of coercion. In our study, we find that fostering online consumer brand engagement actually fosters trust, but only where the brand has an intrinsic meaning for the self (inner self-expressiveness). This finding may suggest that the consumer is best placed to judge the trustworthiness of the brand and when they follow a brand that has intrinsic meaning, they engage with it and trust it.
A key question emerges from our findings. Given that following brands with an ability for inner or social self-expression leads to positive brand outcomes (intention to co-create value and willingness to pay a premium), does it matter that those who follow social self-expressive brands may not engage with the brand or trust the brand? We suggest that the relationship between followers of brands with inner selfexpressiveness and brand outcomes, such as the intention to co-create value and the willingness to pay a premium price, may indicate a longer-term relationship, as those brands reflect the 'real' self, and we find that trust and engagement in the brand are required before the individual will co-create value for the brand or pay more for it.
In contrast, when the focal brand has an ability for social selfexpression, this may indicate a more immediate and potentially shorter-term relationship, that contributes to the follower's image, improves how they are seen by others, and illustrates their social role (Carroll & Ahuvia, 2006). Therefore, the individual's relationship with those brands may be borne out of a need to connect with others, and the social self-expressiveness of a brand may serve a valuable purpose in this regard, without the consumer's need to trust it or engage with it. Returning to Belk (1988) concept of the 'person-thing-person' relationship, we suggest that the 'person-thing' aspect of this relationship is more important for brands with an ability for inner self-expression ('I need to fully connect with it (the brand) before I connect with you about it'), whereas the 'person-person' aspect of the relationship has a greater import for brands with an ability for social self-expression on social media ('I want to connect with you, and the brand helps me to do this').
In short, viewing the intention to co-create value or a willingness to pay a premium price as positive outcomes of self-expressive brands may be an incomplete interpretation of the consumer's relationship with these brands, without also considering the distinction between the inner or social self-expressiveness of brands and the mediating roles of online brand engagement and brand trust in those relationships.

Implications for managers
Managers seeking to harness social media to build brand relationships and encourage positive brand outcomes should be aware of the distinctions between the inner or social self-expressiveness of brands. Specifically, they should be aware that those consumers who follow a brand to express their social self may do so because they are motivated Note: Diagonal entries are average variance extracted (AVE) and the body of table are the estimated correlation square. MSV = Maximum shared variance; SEI = Selfexpressiveness (Inner Self); SES = Self-expressiveness (Social Self); OCBE = Online consumer brand engagement; BT = Brand trust; WPP = Willingness to pay a premium price; ICC = Intention to co-create brand value.

Notes:
Stand. coeff. = standardised coefficient; two-tailed significant testing: * p ≤ 0.10; **. p ≤ 0.05; p ≤ 0.01; S = Supported; NS = Not supported; SEI = Selfexpressiveness (Inner Self); SES = Self-expressiveness (Social Self); OCBE = Online consumer brand engagement; BT = Brand trust; WPP = Willingness to pay a premium price; ICC = Intention to co-create brand value. by a need to interact with others and express a social self to others. While followers of brands with the ability for inner or social self-expressiveness will pay more for a brand and will co-create value, our findings suggest that those consumers who follow a brand to express the inner self are more likely to need to engage with and trust the brand to generate these positive outcomes. A challenge for brand managers, therefore, is to identify whether a brand is followed to express the inner self or the social self. It would be easy to consider that a hedonic brand may be more socially selfexpressive, while a utilitarian brand or a nonprofit brand may be more inner self-expressive, but this cannot be assumed. For example, it is difficult for managers to identify whether the brand 'mirrors the real me' (Carroll & Ahuvia, 2006, p. 84), indicating an inner self-expressive brand, or whether the brand 'adds to the social role I play' (Carroll & Ahuvia, 2006, p. 85), indicating a social self-expressive brand (see Table 2 for the measures of inner and social self-expressive brands).
In our findings, we suggest that following a brand with an ability for inner self-expression is a longer-term relationship, as followers need to trust in and engage with the brand before they are willing to pay a premium price or intend to co-create value with the brand. We suggest ways that brand managers proactively seek to attract more consumers who follow their brand because they believe it expresses the inner self. For example, messages of brand authenticity, such as social media posts about the brand by its employees, may reinforce a belief that the brand has an ability for inner self-expression (Wallace et al., 2014). Alternatively, engaging in authentic brand storytelling may be helpful. Storytelling is where the brand's origin, product designs, or other personalized narratives related to the brand are chronicled (Parahia, Keinan, Avery, & Schor, 2011). Research indicates that a storytelling approach to branding may have rhetorical power, encourage consumers' feeling of connection with the brand through the narrative, and make the brand story more believable (Parahia et al., 2011). We posit that storytelling would be a helpful approach to enhance consumers' perception that the brand is expressive of their inner selves. It may be worthwhile for managers seeking to enhance the inner selfexpressiveness of their brands would also consider measuring online brand engagement and brand trust as part of their performance indicators, rather than relying on intention to co-create value or willingness to pay measures when investigating the relationship between online brand followers and offline brand outcomes.
In contrast, if the consumer's focal brand is socially self-expressive, this may be invaluable to the brand manager because of these consumers' desire to connect with others on social networks. As they are also willing to engage in co-creation and believe that the brand adds to their own image (Carroll & Ahuvia, 2006), they are potentially powerful advocates and influencers on social media. For example, managers could harness the enthusiasm of these brand followers by encouraging them to post selfies or other socially self-expressive content involving the brand, which would help to share the brand across the user's social network.

Limitations and directions for further research
Like most studies, our research is not without limitations. First, the study is conducted among consumers who follow brands on social media in Portugal. We sought to ensure that a diverse sample was included, and therefore, we included workers, students, and unemployed persons in our sample, with an almost 50:50 split of males to females. While the sample profile is consistent with the profile of social media users globally (for example, see Statista, 2020), we nevertheless advocate for further research in other countries to explore the generalizability of the results.
Further, the study did not measure brand loyalty, as we considered the willingness to pay a premium price a strong indicator of loyalty, in line with the literature (Netemeyer et al., 2004). We suggest that research investigating the intention to co-create value and the willingness to pay a premium price could consider the possibility that followers of brands with inner or social self-expressiveness may have different motives for their willingness to pay a premium price, as we suggested in our Discussion. Therefore, the relationship between a willingness to pay and brand loyalty (Netemeyer et al., 2004) may require further investigation in the context of brands followed on social media.
As we find a distinction between the inner and social selfexpressiveness of brands for brand outcomes, we suggest that further research should also investigate brand loyalty to measure the relationship between a willingness to pay a premium price and brand loyalty and compare the results when brands are perceived to express inner or social selves.
Moreover, the study did not consider the valence of posts on social media. Recent research (Sundar & Cao, 2020) suggests that brand trust underlies negative consumer attacks on the brand, when consumers tend to be more critical, and when brands communicate in more polite ways. We advocate that further research would investigate our model when engagement and trust with the brand is high but the brand engages in wrongdoing. It would also be interesting to consider how different forms of social distance, evidenced through the use of polite language by marketers on social media (Sundar & Cao, 2020), might influence the relationship between inner self-expressive brands and brand trust in our model.
In line with Wallace et al. (2014), our research shows that the inner self-expressiveness of brands has a greater effect on positive brand outcomes, and we show a positive association between the inner selfexpressiveness of brands and brand engagement and brand trust. In contrast, our results indicate that the social self-expressiveness of brands does not influence brand engagement or brand trust, although it is an antecedent of the intention to co-create value and a willingness to pay a premium price. To our knowledge, the extant literature considering selfexpressive brands has focused on the global construct of 'self-expressive brands' or on inner self-expressive brands (see Leckie et al., 2016;Algharabat et al., 2020). We advocate that research investigate the inner or social self-expressiveness of brands as two separate constructs.
Furthermore, research has identified a distinction between individualism and collectivism in relation to customer participation in brand outcomes such as value co-creation (Hsieh et al., 2018). As ours is the first study to distinguish between the inner or social self-expressiveness of brands and their impact on brand outcomes (intention to co-create value and a willingness to pay a premium price), we suggest that further research is needed to investigate the follower's motives for value co-creation and willingness to pay and investigate this link for consumers whose focal brand is socially self-expressive.
Finally, we posited that those consumers whose focal brand is inner self-expressive may have a longer-term relationship with the brand, as brand trust and brand engagement influence them to pay more or to engage in brand co-creation. While this is a suggestion to explain our findings, we propose that further research might investigate whether there are differences in the relationship between brands with an ability for inner or social self-expression and online consumer brand engagement and brand trust, by taking a longitudinal approach.

Conclusion
This study shows that a brand's ability for inner or social selfexpression has a different impact on outcome variables. The findings indicate that the inner or social self-expressiveness of brands followed on social media influence intention to co-create value and a willingness to pay a premium. However, we reveal differences in the role of online brand engagement and brand trust as mediators in these relationships. Both online brand engagement and brand trust influence the intention to co-create value and willingness to pay a premium when a brand has an ability inner self-expression. In contrast, brands with an ability for social self-expression are not associated with brand trust or online brand engagement, yet they have a direct effect on the intention to co-create value and the willingness to pay a premium price. Our findings provide greater insights into the impact of following a self-expressive brand on social media for brand outcomes and we present new distinctions between the inner and social self-expressiveness of brands. We argue for consideration of these forms of self-expressiveness of brands as separate constructs in future studies. When brands followed on social media have an ability for inner self-expression, we posit that they reflect the 'real' self. These brands could be more focused on long-term relationships; therefore, trust and engagement in the brand might be required before the individual will co-create value for the brand or pay more for it. In contrast, focal brands with an ability for social selfexpression may be reflective of a short-term relationship formed by a need to connect with others. The brand's ability for social selfexpression may serve a valuable purpose in this regard, as followers may be powerful advocates or influencers on the social network.
Declaration of funding: This work has been funded by national funds through FCT -Fundação para a Ciência e a Tecnologia, I.P., Project UIDB/05037/2020