Parasocial relationships and social media interactions: building brand credibility and loyalty

Purpose – This paper aims to explore how parasocial relationships with Korean celebrity endorsers on social media result in brandcredibility andloyalty. Design/methodology/approach – The participants were identi ﬁ ed through a purposive sampling approach, and they were composed of consumers who purchased Korean-celebrity-endorsed products and services of a telecommunications company. The hypothesized relationships were gauged using a predictive approach asa researchdesignvia partial leastsquares (PLS)path modeling. Findings – The ﬁ ndings show that all hypothesized relationships are supported. In particular, social media interaction was found to have a substantial, positive and signi ﬁ cant effect on self-disclosure. Moreover, self-disclosure has a considerably signi ﬁ cant and direct effect on parasocial relationships and was found to indirectly affect the link between social media interactions and parasocial relationships. The results further reveal that social media interactions and parasocial relationships predict source trustworthiness, leading to brandcredibilityandloyalty. Originality/value – To the best of the authors ’ knowledge, the present undertaking is the only study that examined how parasocial relationships on social media are built when foreign celebrities, in this case, the well-known Korean popular group BTS, endorse telecommunications products and services.

and marketing professionals in maximizing the use of social media and celebrity endorsers, which shall benefit the brand and the business organization's bottom line.

Literature review
The present study used source models as theoretical frameworks which determine the factor influencing attitudes and behaviors toward a brand and loyalty. Under source models, there are two underlying frameworks: source attractiveness which deals with physical attributes, and source credibility, which deals with perceived expertise and trustworthiness. Notably, the source attractiveness model proposed by McGuire (1985) explains the benefit of engaging celebrity endorsers as familiarity and likeability are linked to the effectiveness of a persuasive message. Endorsers' popularity and appeal lead to a higher brand attitude, including loyalty (Nguyen, 2021). The source credibility model, on the other hand, elucidates that a message's efficacy will depend on the perceived level of trustworthiness and expertise of a celebrity endorser (Hovland and Weiss, 1951).
Using the source modelssource attractiveness and source credibilitythe undertaking explores how endorsers' qualities and level of acceptance from the followers may trigger an auspicious outlook of the brand (Ananda and Wandebori, 2016), leading to favorable brand credibility and loyalty. Furthermore, in evaluating trustworthiness and credibility, the source models help us connect the trustworthiness of celebrity endorsers in making interactions related to building credibility and loyalty.

Social media interactions and self-disclosure
Social media interaction, as defined, is a measure of how people interact with their social media accounts and content, covering a broad range of actions across all platforms (Eckstein, 2022). These interactions become essential in establishing strong customer relations through likes, comments, shares and mentions (Casal o et al., 2017;Wellington, 2022). As a two-way communication, social media interactions happen reactively and proactivelyreactive when individuals message the business about its brand, and proactive when businesses, especially their ambassadors or endorsers, reach out to individuals. Considering these interactions, businesses see them as a strategy to strengthen brand awareness, gain leads and develop relationships with other potential customers (Caramela, 2021).
Concerning interactions, providing sensitive and personal information is always critical as it may lead to possible negative consequences (Casal o et al., 2021;. As Chung and Cho (2017) described, self-disclosure pertains to revealing hidden and personal ideas and feelings to others. In the online context, celebrities self-disclose as they continuously share unique content and provide information on their social media profiles (Ampong et al., 2018). Ampong et al. (2018) hypothesized that interactions significantly and positively affect selfdisclosure. Accordingly, the frequency of interaction and maintaining social relationships with others increase an individual's trust, which motivates them to disclose more personal information (Debatin et al., 2009). Similarly, Special and Li-Barber (2012) highlighted that those individuals who use social media for enjoyment tend to interact and engage more. That is, as they obtain the best entertainment through their participation in an ongoing conversation, they are more likely to disclose and be open about their information (Burke et al., 2010).
Moreover, Chung and Cho (2017) pointed out that social media interactions have a selfdisclosing nature in which targeting in social media creates interaction which is relevant in pushing trigger to disclosure (Bazarova and Choi, 2014). As per the study of Fazli-Salehi et al. (2022), when celebrities or vloggers constantly interact with their social media followers, self-disclosure typically occurs. Leite and Baptista (2022) even argued that social SJME media influencers tend to self-disclose when they interact more with their followers, which translates to consumers' behavioral intentions in the long run. With the prevalence of social media, users can freely express their thoughts and emotions to the public. The degree of social media interaction, familiarity and intimacy that may be built between and among users may result in self-disclosure (Lin and Utz, 2017). The more self-disclosure exists or is present in the engagement or relationship; the more emotionally invested those parties involved in the relationship will be (Ferchaud et al., 2018). Thus, we postulated that: H1. Social media interactions have a significant effect on self-disclosure.

Social media interactions and source trustworthiness
Social media platforms provide people with options for obtaining and sharing news and information and lead them to interact online (Sterrett et al., 2019). They interact through sharing images, videos and news articles on social media platforms. Interaction in social media, and the convenience of accessing information, allow individuals to build relationships arising from a strong sense of belongingness and well-being (Belanche et al., 2020;Thamik and Wu, 2020). This social presence or the need to feel that an actual human being, such as influencers and celebrity endorsers, intensively interacting with an individual is a factor of trust (Balaban et al., 2020) where the level of cooperation between individuals in social communities is defined (Ridings et al., 2002). For instance, on Instagram, consumers react positively or negatively to product placements depending on whether the influencers appear with the products they endorse (Jin et al., 2019). Likewise, Instagram and other social media platforms allow consumers to interact with their favorite influencers (Weismueller et al., 2020). Tran et al. (2019) contended that an individual's attitude toward a celebrity or endorser is mainly due to their connection. Moreover, Parmar and Mann (2021) argued that the expectation of consumers of sincere and positive messages from celebrity endorsers increases their likeability toward them. Thus, trustworthiness shows consumers' trust and confidence in celebrity endorsers and influencers (Chen et al., 2022).
According to Rashedi and Siahi (2020), source trustworthiness is the extent of the authenticity of someone's statements. In an online environment where everyone can freely express their opinions and feelings by staying anonymous, individuals evaluate the source's trustworthiness to determine whether they accept or reject the information provided. The conveyor as an information source is considered trustworthy if recipients of the message find it valid and honest, which is formed through constant communication and interaction between them (Chung and Cho, 2017). Trust is generally formed when authentic social media exchanges happen between users (Fu et al., 2019), such as interactions between a personality and his/her followers. Social media personalities' online presence and constant interaction with their followers can build trust, which may translate to better credibility and good behavioral intentions (Sokolova and Kefi, 2020). When the interaction that happens in social media is perceived to be authentic, social media influencers can easily attain trust from their followers . Therefore, we hypothesized that: H2. Social media interactions have a significant effect on source trustworthiness.
1.4 Self-disclosure and parasocial relationships Self-disclosure, which can be defined as making others know an individual, is a central component for individuals in relationship-building (Kaloydis et al., 2017). It is a vital factor in developing and maintaining relationships (Collins and Miller, 1994). It is accomplished by sharing information about oneself, opinions and attitudes with another person (Chen et al., 2016).
Self-disclosure plays an essential role in developing parasocial relationships. Horton and Wohl (1956) stated that media personalities personally and privately converse with an audience, making them more likely to anticipate a response, thus, creating parasocial relationships quickly. Moraes et al. (2019) further noted that successful celebrity endorsers could easily connect with consumers by being role models.
Parasocial relationships are one-sided relationships that people establish with media personae, such as show characters, news anchors, talk show hosts and celebrities (Horton and Wohl, 1956). It focuses on how audiences interact, relate to and develop relationships with a celebrity. Liu et al. (2019) identified that parasocial interaction might lead to favorable perceptions toward brand quality, brand affect and brand preference of the consumers. Collins and Miller (1994) disputed that individuals who intimately disclose themselves tend to be liked more. When social media personalities reveal themselves and shares their sentiments, emotions and feelings on an online platform, their followers tend to feel that they are building a form of relationship, such as a parasocial relationship (Chung and Cho, 2017). According to Kim and Song (2016), the parasocial interaction experiences with celebrities are increased by self-disclosure, enhancing the user's feelings. Hence, we posited that: H3. Self-disclosure has a significant effect on parasocial relationships.
1.5 Social media interactions, self-disclosure and parasocial relationships Self-disclosure is divulging personal information such as beliefs, thoughts, values and feelings about ourselves to others (Mulayousef, 2018). It also plays a significant role in building and upholding intimate relationships because individuals like people who disclose personal information to them (Mulayousef, 2018). The things that people disclose evolve as a relationship develops. These things include the observable and hidden personal beliefs that people can only discuss in a deep, intimate relationship. Self-disclosure is embedded in social media as celebrities often engage in high levels of self-disclosure (Chung and Cho, 2017).
Social media provides a safe and convenient way for celebrities to interact with many fans (Chung and Cho, 2017) through various platforms such as Facebook, Twitter and Instagram. Through social media, celebrities constantly converse with their fans to promote or create awareness. Many celebrities also use social media for personal communication (Stever and Lawson, 2013). This interactive conversation with celebrities excites fans and encourages them to remain loyal to the personalities they follow. The self-disclosing nature of social media interactions enhances perceived intimacy and bonding, thus, fostering parasocial relationships.
A parasocial relationship is best promoted through social media for its frequent, intimate and confessional messaging characteristics (Chung and Cho, 2017). When individuals are frequently exposed to celebrities, they develop a sense of intimacy, perceived friendship and identification with them (Horton and Wohl, 1956). Through social media interactions, consumers feel up close and personal with celebrities as they can hear directly from them. Hence, we postulated that: H4. Self-disclosure has a mediating effect on the relationship between social media interactions and parasocial relationships. SJME 1.6 Parasocial relationships and source trustworthiness Celebrities can attract clients, which has been linked to their credibility regarding the company they are supporting. When celebrities promote a brand, it indicates that they believe in the brand. Similar to the concept of endorser credibility, brand credibility is related to a company's dependability in delivering its promises to its customers (Erdem and Swait, 2004). It is one of the key reasons why potential consumers' buying intentions are impacted. Even if they do not interact or meet in real life, the spectator feels a strong connection to the figure.
On the other hand, thinking that a celebrity is trustworthy suggests respect for the character, life choices and even principles. As a result, when brand trustworthiness (Goldsmith et al., 2000) and brand beliefs (Yoon and Weatherwax, 1998) are created, the endorser's credibility is expected to be evaluated. Others may distrust the items being advocated since celebrities are paid, yet trustworthiness is attained at a greater level due to the significant quantity of interpersonal interactions. In this regard, celebrity endorsement is a good aspect of trustworthiness. Fostering parasocial interactions helps to enhance celebrity trustworthiness and brand credibility, which act as cushions for brand trust (Chung and Cho, 2017). Thus, we posited that: H5. Parasocial relationships have a significant effect on source trustworthiness.

Source trustworthiness and brand credibility
One of the determinants of source credibility is when the consumer accepts the message's accuracy and honesty because of the endorser's positive characteristics (Chung and Cho, 2017). The consumers perceive the information provided by this endorser as practical, valid and persuasive because it is from a highly trustworthy source (Wang et al., 2007), hence, emphatically affecting their "internalization" process that deals with their beliefs, attitudes, opinions and behavior (El Hedhli et al., 2021).
Similar to endorsers' credibility, brand credibility is how the company proves its trustworthiness by delivering its claims and expertise through executing its brand promise (Erdem and Swait, 2004). It is formed through the influence of effective marketing strategies that includes celebrity endorsement (Baek and King, 2011). Gong and Li (2017) claimed that parasocial interactions and the credibility of an endorser lead to endorsement effectiveness. The transfer model of McCracken (1989) shows that some personality traits of celebrity endorsers may be transferred to their endorsed products. It is aligned with Chung and Cho's (2017) argument that endorser and brand credibility are positively associated. Thus, the credibility of endorsers will likely be assessed when brand credibility perceptions are formed.
Overall, it is crucial to identify the trustworthiness of the source of information (the endorsers; Wang and Scheinbaum, 2017) to increase brand credibility and recognition. Thus, a trusted celebrity that becomes the source of the brand information should lead to higher brand credibility of the endorsed brand. There we hypothesized that: H6. Source trustworthiness has a significant effect on brand credibility.

Brand credibility and brand loyalty
Building solid and positive brands leads customers to prefer a particular brand, which might lead to loyalty over time. Loyalty is made up of a high degree of bonding between a customer and a brand (Shin et al., 2019). It is a powerful marketing term, reflecting the degree to which a brand is committed to a customer. Executives make projections based on brand loyalty levels to ensure that the buyer stays on the mark, even when costs or accomplishments change (Kwon et al., 2021). Moreover, brand loyalty is a two-dimensional structure that contains attitudes and behavior reflected in repetitive purchase actions (Arifine et al., 2019).
Brand credibility is the legitimacy of consumer data in a market to meet expectations or promises.  stated that, to develop customer loyalty, companies need brand credibility. Similarly, Mathew and Thomas (2018) argued that there is a positive relationship between brand credibility and true loyalty. That is, sellers need to know when customer expectations for comfort in buying a product increase for innovation and trust to be developed (Duarte et al., 2018). Moreover, Jun (2020) highlighted that brand credibility and loyalty have a direct relationship and should be considered in consumer decision-making. According to Kashif et al. (2018), brand credibility is a state where customer needs significantly affect loyalty. Therefore, we posited that: H7. Brand credibility has a significant effect on brand loyalty.
Provided with the identified research hypotheses, Figure 1 illustrates the proposed conceptual framework depicting the relationship of social media interactions on self-disclosure and source trustworthiness; the mediating effect of self-disclosure on the relationship of social media interactions and parasocial relationships; the influence of parasocial relationships on source trustworthiness; the relationship of source trustworthiness and brand credibility; and the relationship of brand credibility and brand loyalty.

Participants and sampling technique
This study used a purposive sampling technique as specific characteristics were set to determine the study's respondents. In particular, the target respondents were Filipinos who have recently bought products and services from Smart Communications, Inc.
Filipinos were identified as participants of the study because of two primary reasons: (1) international celebrities, including Korean pop groups and idols, have been evident in the Philippine local advertising landscape in the past years (Quieta and Godinez, 2021); and (2) the Philippines, as of 2022, has 76.01 million social media users (Amurthalingam, 2022).
The study's conceptual model SJME Hence, carrying out a study on parasocial relationships with Korean celebrity endorsers and the Philippines as the setting, is timely and relevant. The gathering of data was done through an online survey. The respondents were asked about their full intention to participate in the study. They were then asked if they were Smart subscribers and had bought its products and services in the last three months. Once the respondents' preliminary requirements were met, they were asked about their awareness of the Smart Communications, Inc. endorsement of BTS, where a sample print ad was attached to confirm alignment. After all the evaluations, only participants who met the set criteria were directed to assess the items and statements in the online survey. Out of 653, 542 respondents answered the online survey thoroughly, resulting to 83.38% response rate.
To attest to the sufficiency of the sample size, inverse square root and gamma exponential methods were used through WarpPLS 8.0 (Kock, 2022). The illustration below shows the minimum absolute significant path coefficient of 0.162 at a 0.05 significance level and a power level of 0.800; thus, the inverse square root method's minimum required sample size is 236, while the gamma-exponential method is 223. Thus, with 542 valid responses, sufficiency of the sample size was attained ( Figure 2).

Instrumentation
A self-administered survey was used in the present undertaking. More specifically, the instrument consisted of three parts: the first part contained the letter of request to the participants and the gatekeeper questions to filter the qualified respondents of the study, while the second part composed of the statements to assess the six constructssocial media interactions, self-disclosure, parasocial relationships, source trustworthiness, brand credibility and brand loyalty. The third part contained questions about the participant's profilesex, age, location and employment type.
Relative to the research instrument, we considered the work of Chung and Cho (2017) to evaluate social media interactions, self-disclosure, parasocial relationships, source trustworthiness and brand credibility. For social media interactions, two statements using a five-point Likert scale were operated; three statements for self-disclosure using a seven-point Likert scale; five statements for parasocial relationships using a seven-point Likert scale; five statements for source trustworthiness using a five-level scale; and five statements for brand credibility using a five-point Likert scale. Moreover, to evaluate brand loyalty, we used the work of Kim and Song (2016) with four statements using a five-point Likert scale.

Research design and statistical approach
The study used a predictive research design to determine the interrelationships of the six constructs: social media interactions, self-disclosure, parasocial relationships, source trustworthiness, brand credibility and brand loyalty. Moreover, this was used as the study aimed to determine how one variable affects the other, providing a forecast of future outcomes about the variables (Iacus et al., 2018). The partial least squares-structural equation modeling (PLS-SEM) using WarpPLS 8.0 software was applied to provide estimates relative to the proposed model with latent variables. The present study also applied the mediation analysis to measure the mediating effect of self-disclosure on the hypothesized relationships.

Assessment of the measurement model
The assessment of the measurement model involves examining internal consistency through composite reliability, convergent validity through average variance extracted and factor loading and discriminant validity through the heterotrait-monotrait (HTMT) ratio of correlations. Table 1 presents the internal consistency and validity (convergent and discriminant) of the lower-order reflective constructs of the study. Based on the results, the constructs obtained composite reliability (CR) values ranging from 0.914 to 0.941. This indicates that the items of each construct showcased a more relaxed measurement model (Kock and Lynn, 2012) and demonstrated internal consistency and reliability; that is, the high CR values are reasonable indications that all the scale items persistently measure the same variable. Relative to convergent validity, the variables were assessed by factor loading, average variance extracted and p-value. Considering the factor loadings, all items were significant at p < 0.001. Also, the factor loadings were beyond the threshold of 0.50 (Amora, 2021;Kock, 2014), of which the resulting factors range from 0.727 to 0.935. Likewise, the average variance extracted for the six constructs obtained values ranging from 0.707 to 0.875, Notes: SM = social media interactions; SD = self-disclosure; ST = source trustworthiness; BC = brand credibility; BL = brand loyalty; FR = friendship; UI = understanding/identification. All factor or indicator loadings are significant, p < 0.001. Friendship and understanding/identification are the two dimensions of parasocial relationships. All loadings are significant (p < 0.001) Source: Authors' compilation SJME exceeding the threshold of 0.50 (Amora, 2021;Kock and Lynn, 2012). In conclusion, the measurement model demonstrated convergent validity. Using the HTMT ratio of correlations, Table 1 displays the ratio values to attest to the discriminant validity of the construct. As per Henseler et al. (2015), HTMT ratios provide estimates of the relationship between two latent variables. Based on the result, the constructs obtained ratio values ranging from 0.460 to 0.863, all below the conservative threshold of 0.85 (Voorhees et al., 2016). A hierarchical component model assessment was also performed to measure the higherorder construct used in the present study. To evaluate parasocial relationships as a higherorder formative construct, a disjoint two-stage method (Agarwal and Karahanna, 2000;Lacap and Sicat, 2022) was used.
In evaluating a hierarchical component model, factor weights, p-values, variance inflation factor (VIF) and full-collinearity VIF (FCVIF) were gauged. According to Ramayah et al. (2018), the factor weights of each indicator must be significant (p < 0.05). Based on the results in Table 3, the two indicators (friendship and understanding/identification) of the higher-order formative construct (parasocial relationships) passed the requirement for the factor weights.
VIFs were measured for the collinearity test. To say that the indicators are free from collinearity problems, each VIF must be at most 3.30 (Diamantopoulos and Siguaw, 2006). Moreover, the full collinearity test was also performed by assessing the higher-order formative construct's FCVIF (Isip and Lacap, 2021;Rasoolimanesh, 2022). According to Kock (2015), FCVIF must be at most 3.30. Based on the results in Table 2, parasocial relationships as a higher-order formative construct passed the hierarchical component model assessment requirements. Table 3 presents the direct and indirect effects of the PLS-SEM model. The evaluation shows the relationship and the magnitude of the path coefficient where social media interactions positively and significantly affect self-disclosure (b ¼ 0.609, p < 0.001) with an effect size of large (Cohen's f 2 ¼ 0.370), which, therefore, supports H1.

Evaluation of the structural model and its direct and indirect effects
H2 is also supported as social media interactions positively and significantly affect source trustworthiness (b ¼ 0.079, p ¼ 0.032) with a small effect size (Cohen's f 2 ¼ 0.039). Moreover, self-disclosure positively and significantly affects parasocial relationships (b ¼ 0.781, p < 0.01) with a substantial effect size (Cohen's f 2 ¼ 0.610), thus, supporting H3.
In addition, the results show that parasocial relationships significantly and positively affect source trustworthiness (b ¼ 0.603, p < 0.001), and its effect size is large (Cohen's f 2 ¼ 0.397), which supports H5. Likewise, source trustworthiness positively and significantly affects brand credibility (b ¼ 0.539, p < 0.001) with a medium effect size (Cohen's f 2 ¼ 0.291) supporting H6. And brand credibility showed a significant and positive effect on Meanwhile, H4 was supported since self-disclosure as a mediating variable has an indirect effect between social media interaction and parasocial relationships (b ¼ 0.475, p < 0.001) with a medium size of effect (Cohen's f 2 ¼ 0.328).

Discussion
The study revealed that social media interactions directly and significantly affect selfdisclosure. This finding suggests that constant interactions on social media platforms play a vital role in influencing how celebrity endorsers disclose information. Celebrity endorsers tend to share more professional life updates and personal experiences when interactions become more personal, intimate and inviting. Thus, social media interactions trigger celebrity endorsers to disclose. Previous studies strengthen the idea that participation in an ongoing conversation bolsters celebrity endorsers to disclose more and become more open about their information (Ferchaud et al., 2018;Ampong et al., 2018;Burke et al., 2010;Fazli-Salehi et al., 2022). Likewise, the result is aligned with the study of Debatin et al. (2009) andEllison et al. (2007), which highlighted that maintaining social relationships with others increases the level of trust of an individual, which further serves as a motivation in order for the celebrity endorsers to unveil more personal information. As consumers build higher levels of interactions with celebrities, higher levels of self-disclosure will also exist. Hence, the more self-disclosure is present in an engagement; the more emotionally invested the parties involved will be.
The findings also showed that social media interaction significantly and positively affects source trustworthiness. The positive relationship between social media interaction and source trustworthiness indicates that celebrity endorsers gain more trust and confidence among their followers when they constantly interact with them through social media, which is similarly discussed in the previous studies of Ridings et al. (2002), Jin et al. (2019) and Balaban et al. (2020). The findings validate the unique bond between celebrity endorsers and their followers (Yuan and Lou, 2020) and how Instagram celebrities who have grown closer to their followers can build emotional attachments, increasing their trustworthiness (Jin et al., 2019). It allows consumers to consider information from their trusted endorsers when presented with information (Gong and Li, 2017) from different sources. Thus, the level of trust accorded to celebrity endorsers depends on the Notes: SM = social media interactions; SD = self-disclosure; ST = source trustworthiness; BC = brand credibility; BL = brand loyalty; PS = parasocial relationships. *p < 0.05; **p <0.01; ***p < 0.001 Source: Authors' compilation SJME sustainability of their relationship with their followers (AlFarraj et al., 2021) through social media interaction. Furthermore, self-disclosure has a significant and positive impact on parasocial relationships. It implies that self-disclosure is embedded in social media. Celebrities on social media often engage in high levels of self-disclosure through seemingly honest expressions of emotion, the sharing of highly opinionated statements on various topics, and uploads of selfies of themselves and their families (Marwick and Boyd, 2011). It has been found that public personalities' self-disclosures on social media may improve relationships. Likewise, self-disclosure plays an essential role in developing parasocial relationships. As per Horton and Wohl (1956), media celebrities who personally speak with an audience increase the likelihood that the public would anticipate a response, which makes it easier for media personalities to establish parasocial connections. Self-disclosure increases parasocial interactions with celebrities, which boosts user sentiments (Kim and Song, 2016;Jackson and Lilleker, 2011).
Moreover, the results showed that self-disclosure positively mediates the relationship between social media interactions and parasocial relationships. This means that celebrities strengthen their parasocial relationship with their fans by disclosing themselves by sharing their professional and personal information through social media platforms such as Facebook, Twitter and Instagram, which serve as their primary tools to communicate and interact with their fans. This connection between the three constructs also suggests that the self-disclosure of an endorser can positively affect the parasocial relationship with the consumers when using social media platforms. It is also mentioned in the prior studies that the self-disclosure of an endorser signifies a direct influence on the parasocial relations with their fans and consumers (Kim and Song, 2016;Liew, 2018).
Likewise, the findings showed that parasocial interactions improved source trustworthiness toward consumer behavior. This result suggests that trustworthiness is the basis for parasocial interaction. The positive correlation between these two factors implies that consumers see celebrities as trustworthy to some extent when a poster or advertisement offers information with honesty and goodwill (Ismagilova et al., 2019). Furthermore, celebrity endorsement may increase the commercial's credibility and the effectiveness of the advertising message (Biswas et al., 2006). It implies that if the endorser is trustworthy, customers will think the product is trustworthy (Gilal et al., 2020), and, therefore, trustworthiness is attained at a higher level of human connections. Consequently, developing parasocial contacts promotes celebrity and brand credibility, both serving as brand trust buffers (Chung and Cho, 2017).
Regarding source trustworthiness and brand credibility, the result showed that source trustworthiness significantly and directly influences brand credibility. This finding signifies that consumers will have more trust in the brand if it contains accurate information about the products from a reliable source (Erdem and Swait, 2004), which also creates a good brand reputation, increasing consumer trust (Ngo et al., 2020). Moreover, the positive correlation between the two constructs means that the reliability and integrity of the source of information and presentation of a brand substantially affect its brand credibility. In short, Korean celebrities who have gained global popularity create a more trustworthy, valuable image for the company by creating brand credibility through endorsement. It is also the case in previous studies regarding the significance of source trustworthiness and how it directly influences and increases brand credibility and recognition (Wang and Scheinbaum, 2017;Chung and Cho, 2017).
Lastly, brand credibility has a positive and significant effect on brand loyalty. Therefore, a higher company's brand credibility increases brand loyalty. Companies and products endorsed by Korean celebrities increase brand appeal loyalty. People are more likely to recommend a brand when they believe in it. It is in line with Rizwan (2014) and Alam et al. (2012), which established that brand credibility makes consumers loyal, even in the presence of competitors. Sweeney and Swait (2008) stated that quality brands build sustainable and robust trust between manufacturers and customers, increasing brand loyalty. Hence, it proved the argument of  that to develop customer loyalty, companies need brand credibility. Brand credibility is the legitimacy of consumer information in a market to meet expectations. Furthermore, there is a positive relationship between brand credibility and true loyalty (Mathew and Thomas, 2018).

Practical implications
The current study offers practical implications for businesses and the industry in general. The massive influence of social media and the admiration of consumers for international celebrity endorsers can translate to brand credibility and loyalty if suitable marketing campaigns are implemented. Getting a well-known celebrity as an endorser or ambassador only sometimes translates to good behavioral intentions among consumers. Companies need to consider that it is vital to identify the right personalities to represent the brand; the prospective endorsers must be individuals with a positive social media reputation; they must have a social media presence as evident with their number of followers; they must be imbued with a good attitude based on their social media interactions with their followers; and they must be authentic in sharing their feelings and sentiments online. Popularity alone does not guarantee that social media users or consumers will patronize the brand the personalities endorse. Brand credibility and loyalty only happen when endorsers build trust and parasocial relationships through their online favorable social media presence and selfdisclosure. Furthermore, brand managers should take a strategic approach to manage the parasocial relationships between Korean celebrity endorsers and their followers as a model for building brand loyalty. Likewise, the telecommunications industry should maximize not just their Korean celebrity endorsers but other celebrity endorsers by allowing them to fully exploit the available technologies as platforms to engage with their followers, which in turn can help the company expand its reach and market share.

Theoretical contributions
Grounded on the source models (source attractiveness and source credibility) and drawing from the conceptual framework of Chung and Cho (2017), an essential theoretical contribution of this study establishes the importance of parasocial relationships of the consumers with Korean celebrity endorsers on social media. These relationships affect how consumers see the brand's credibility and become loyal. The findings support the source attractiveness model where Korean celebrity endorsers influence the consumers to patronize their endorsed product with the help of their physical attributes (Nguyen, 2021). The findings also support the source credibility model on which the consumers can see the reliability of the Korean celebrity endorsers, which directly leads to high levels of commitment (Ananda and Wandebori, 2016). These parasocial relationships can imply that consumers feel connected and gain intimacy with Korean celebrities through the identification process and think they are related or friends with these Korean celebrities (Chung and Cho, 2017;Lee and Um, 2014). Moreover, acknowledging the contribution of celebrity endorsers, the present study agrees with the source credibility model, which signifies that as consumers identify themselves with the celebrity, the more they will give credit to the personalities with favorable attitudes. Thus, the current study contributes to the development of the source models (source attractiveness and source credibility models) SJME where attractiveness must go hand-in-hand with credibility, and this can only be attained when endorsers favorably and genuinely interact with their followers and build some form of relationships such as friendship and understanding with their online supporters (Table 4).

Limitations and future research direction
There are limitations in the present study. First, the study has few respondents from Visayas and Mindanao, given that the study focuses on the Philippines. Hence, future researchers may conduct in-depth studies in these two locations mentioned since most brands that use K-pop celebrities as their endorsers operate nationwide. Second, the study concentrated on the most recently endorsed brand of a popular K-pop male group. Future researchers may conduct related studies on other K-pop groups endorsing brands from different sectors or industries, such as the food and fashion industries. It is also an opportunity to investigate local celebrity endorsers and compare their degree of relevance in building parasocial relationships with foreign endorsers. Moreover, third, as the study mainly focused on customer loyalty in the context of parasocial relationships, future research may be expanded by exploring the relationship between brand love and parasocial relationships and their impact on customer lifetime value as a measure of customer profitability and the business's long-term growth. Table 4.

Conclusions
Theoretical and managerial implications Social media interactions among endorsers lead to selfdisclosure and higher trust among consumers Brand credibility and loyalty only happen when the endorser can build trust and parasocial relationships through good online social media presence and self-disclosure Self-disclosure of endorsers results in parasocial relationships, which leads to consumers' trust. Self-disclosure serves as a mediator on the link between social media interactions and parasocial relationships Constant engagement between consumers and celebrities, regardless of the information shared, develops into an intimate connection between the two parties, thus, boosting celebrities as a trustworthy source When endorsers' trustworthiness is established, brand credibility rises, leading to brand loyalty The massive influence of social media and the admiration of consumers for international celebrity endorsers can translate to brand credibility and loyalty if suitable marketing campaigns are implemented Source: Authors' own work