The previous research related to brand extension has demonstrated the relevance of product category fit but has also highlighted the need to evaluate the brand concept consistency (Lanseng & Olsen, 2012). The brand concept is the main aspect that helps position a product in the consumer's mind and differentiates it from other brands in the same product category (Park, Jaworski, & Maclnnis, 1986). It represents the core character of the brand composed of the symbols and the set of brand associations (Aaker, 1996; Black & Veloutsou, 2017; Coleman, De Chernatony, & Christodoulides, 2011). Brand concept has a unique abstract meaning, which can be maintained with time through consolidated marketing actions that lend consistency to the brand. Brand concept consistency, therefore, encompasses the extended product line; therefore, if it evokes the same associations in the consumer as the parent brand, it will probably similarly influence consumer reaction to brand line extensions (Park et al., 1991).

Luxury items are purchased because they are social indicators, and some other attributes attach distinction and differentiation to the consumers of luxury brands and predispose them to pay more, as extreme expensiveness enhances these qualities (Zhang & Kim, 2013). Then, the role of brand concept consistency directly affects the consumer's evaluation of brand extensions (Kressmann et al., 2006; Liu et al., 2012; Park et al., 1991). The extension of brand names has resulted not only in the reconsideration of the concept of luxury but also as a strategy of mass marketing associated with the weakness of the image of exclusivity through the introduction of more economical versions of luxury branded goods.

According to the text exposed, the brand contains associations and attributes, which are well-known and perceived by the consumer through its concept consistency. Therefore, in the context of downward price brand extension, the brand concept consistency will be affected. The concept change also modifies the product attributes, provoking a new evaluation of the brand extension and a possible new opinion of the extended brand, if the concept of a luxury brand is not similar to the luxury brand extension, due to a different concept perception, which could decrease the purchase intention and consequently affect the loyalty toward the brand. Hence, we define the first hypotheses:

H1a-. The lesser the brand concept consistency, the lesser the purchase intention.

H1b-. The lesser the brand concept consistency, the lesser the loyalty towards the brand.

Brand personality furnishes the brand with human-like attributes (Aaker, 1997; Wee, 2004). For some brands, marketing communication manages to acquire for the brand a personality recognised by the consumer. This works better for those brands that invariably appear similar to consumers’ perceptions of themselves. This personality helps to link the brand concept and brand image to the consumer's self-concept. Brand personality communicates associations through traits, just like a human being, and when consumers perceive these personality traits, they compare them with themselves. If the comparison is positive, then the brand becomes a vector of expression of the person, and brand value appears (Dubois & Duquense, 1995). Then, the value of a luxury brand is based on the personality traits that the luxury products are bought for (Haasler, 2006; Heine, Atwal, Crener-Ricard, & Phan, 2018; Phau & Pendergast, 2000).

Value can be considered the “capacity that a brand has to transmit to its clients’ personality the brand associations, and to be converted into the vector of their personal expression” (Dubois & Duquense, 1995). The latter is possible through the use/consumption of certain brands, which possess brand personality (e.g. Carolina Herrera portrays a distinguished person). In a context of downward price brand extension, the brand concept and image may be affected. Due to these changes, some traits of the brand personality may also be affected and consequently may shift more away from the consumer's personality (Royo-Vela & Voss, 2012, 2015). Therefore, we establish the next hypotheses:

H2a-. The lesser the coincidence between the brand's personality and the consumer's personality, the lesser the purchase intention.

H2b-. The lesser the coincidence between the brand's personality and the consumer's personality, the lesser the loyalty towards the brand.

Brand image is the subjective concept of a brand maintained by the consumer (Dobni& Zinkhan, 1990), which is affected by marketing activities, due to the variable contexts and characteristics of the perceiver. The brand image, according to the consumer's self-image, generates a subjective experience referred to as brand-image congruence or self-image congruence (Sirgy, Grewal, Mangelburg, & Park, 1997). According to the level of congruence perceived between the attributes and certain factors related to the consumers, they will tend to consume one brand or another, preferring the one whose image is similar to theirs. Consumers evaluate the fit between the brand's image and their own concept to decide if that brand image enhances their self-image expression (Kang, Hong, & Lee, 2009). Consumers prefer brands with images congruent with self-image (Douglas & George, 1970; Kressmann et al., 2006; Liu et al., 2012).

According to the above, when a price-based downward extension of a luxury brand is launched, the brand image may be affected and luxury consumers may perceive and feel that the brand fits their own image in a lesser way than before launching and consequently may become less loyal to the luxury brand (Royo-Vela & Voss, 2012, 2015). In other words, luxury consumers will purchase a premium luxury brand if its image coincides with their self-image, which they wish to show to the public. Hence, we establish the next hypotheses:

H3a- The lesser the perception of congruence with self-image, the lesser the purchase intention.

H3b- The lesser the congruence of the consumer's self-image, the lesser the loyalty towards the brand.

The context within which the research emerges is the luxury market; this sector traditionally is characterised by the exclusivity, scarcity and uniqueness of its products, characteristics usually related to their expected high price. The perception of luxury brands by luxury consumers may be affected when the brand concept, personality and image do not match their perception of what a luxury brand is supposed to be.

When the parent brand has a brand extension with extensive price, which means that consumers can buy luxury goods at a reduced price or that they can buy in a lower price retail format, the consequence could be that the consumers perceive that the brand expresses different brand associations and attributes, and as a consequence, the concept of the brand does not anymore fit the previous consumer perception, not with the brand personality or with their self-image.

This research focuses on accurately understanding consumers’ perception changes from the moment they learn about the existence of the downward price-based brand extensions in the market and its effects on brand concept consistency, brand personality changes and self-image congruency. It proposes, based on the literature (see Table 1), an integrative five-factor model of antecedents (brand concept consistency, brand personality and congruence of self-image) and consequences (purchase intention and brand loyalty) and their structural relationships in the form of hypotheses (see Fig. 1).

Fig. 1.

The proposed model and its structural relationships.

(0.28MB).

Hence, these dimensions and their relations were not chosen at random but instead based on previously proposed conceptual models. The work by Liu et al. (2012)) was observed, who considered image congruence to be made of three constructs not related to the product. This research considers only two of those attributes: brand personality congruence and user congruence together with brand personality. All three affect purchasing intention and loyalty. The symbolic congruence of the use of the brand was not considered because the concept itself already contains the symbolic considerations that luxury brands imply. Another dimension not considered is the fifth dimension established by Aaker (1997) in the scale of measurement of the brand personality construct. The work by Liu et al. (2012)) demonstrated that this last dimension is not significant in prestigious brands.

Additionally, the qualitative investigation carried out by Royo-Vela and Voss (2012, see Method section) is particularly relevant as the precedent for the current model, which aims to develop a quantitative model that allows the comparison of the results with their findings. It could also sustain and enhance their previously proposed theoretical model. Actually, this study adopted a confirmed approach in its aim to validate and enhance some of the conclusions obtained by the previous qualitative research in the form of propositions (Royo-Vela & Voss, 2012, 2015). A synthesis of the analyses in both qualitative propositions and quantitative hypotheses research is presented in Table 4 in the Results and Discussion Section.

As a first research step, qualitative techniques, i.e. focus group and in-depth interview, were applied. The objectives of the focus group were the following: a) understand the consumers’ perception of luxury goods, revealing important brand personality traits and brand associations that define luxury for the consumers; b) identify the underlying motivations for luxury goods purchase and the value proposition they offer to the consumer; c) comprehend and evaluate how consumers attach their own personality to the luxury goods they buy and try to reveal the consumers’ emotions and attitudes towards price-based brand line extensions in connection to their self-image and the overall brand concept.

In total, six people attended the Focus Group, four women ranging from 32 to 60 years old and two men ranging from 41 to 57 years old. All of them were current consumers of luxury brands such as Gucci, Prada, Valentino, Chanel, Porsche, Mont Blanc, Dior, Louis Vuitton, Lancome or Strenesse among others. The choice of the participants depended on the specific luxury brand and the respective product category of the brand of which they were consumers. The brands were chosen considering both their symbolic personality/image associations and their prestige value-creating attributes, as opposed to the functional ones (Kressmann et al., 2006). The Focus Group was conducted in Hamburg, Germany. The session was taped with a voice recorder, as two participants did not feel comfortable being filmed and did not want to be recognised. The participants were only told though that the research was conducted to expose consumers’ attitudes towards luxury goods and their underlying motivations for buying them. The exact details about the concepts and constructs involved were not directly mentioned, though, to prevent biased answers. To evaluate the brand concept and the self-image congruity (actual and ideal self-congruity), the brand personality framework developed by Aaker (1997), as suggested by Kressmann et al. (2006), was applied.

Additionally, two in-depth interviews were conducted to uncover deeper insight into the research topic and the findings of the focus group discussions. A 48-year-old man and a 54-year-old woman were interviewed. As in the focus group discussion, the actual known and purchased luxury brands of the participants were taken as the main stimulus to establish the necessary familiarity with the brand and the image of the particular brand. The interviews were also recorded. The focus group and in-depth interviews proceeded through the following steps: introduction, warm-up (writing down three to five words participants associated with luxury and discussion), projective (arrange those words according to their importance in concentric circles the core one being the most important and discussion), open-ended questions and discussion, and brand extension explanation and discussion.

According to the Bain and Company study (2018), Spain was placed fifth in luxury market consumption within the European Union. In the year 2017, the annual spending on the luxury market in Spain reached €9200 million, with an increase of 9% since the previous year (Bain & Company, 2018).

The target population of this study comprises luxury brand consumers, with a focus on the “Prosper Medium Class” segment consisting of families with an annual income exceeding €100.000, which includes about 400,000 families in Spain (INE, 2017).

The sample was selected by a non-probabilistic approach, applying a purposive sampling technique. The survey and its interviews followed a structured questionnaire, applied face-to-face to consumers in premium luxury brand stores such as Chanel, Gucci, Loewe, Lladró, Louis Vuitton, Prada, Porsche, Rolex, Tiffany & Co, Versace and so on. Some years ago, it was relatively unproblematic to identify the wealthy population; however, nowadays, due to the emergence of the nouveau-rich and brand extensions, neither exact numbers nor a classification is available. When selecting the survey respondent, nationality, country, gender, balanced gender proportion and an additional criterion based on whether luxury store visitors already wore luxury brand products, i.e. car, portfolio, handbag, clothes, were considered.

Finally, a valid sample size of 158 premium luxury (mother) brand consumers in the three main Spanish cities was obtained, among them 46% men and 54% women. The more representative age group range at 44% was between 47 and 57 years old, followed closely by the 36- to 46-year-old band with 43%. Most of the respondents, 44%, have at least a university degree, and 43% have completed a master's degree. Most of the respondents, 73%, have a full-time job, while 15% are self-employed, and 48% of the respondents reported an average monthly income of over €6000. In comparison, both the €4000–5000 and €5000–6000 ranges represented 20% of the respondents.

The research measurement scales have been chosen from the related constructs proposed in the literature and previous research work, predominantly using a seven-point Likert-type response format. Thus, to measure brand concept consistency, the scale developed by Park, McCarthy, and Milberg (1993) is used. In the case of brand personality, a scale proposed by Liu et al. (2012) was considered. Additionally, for the study of self-image congruence, the scale conceived by Sirgy et al. (1997) is applied. Purchase intention was measured on a scale designed by Hung et al. (2011). Finally, brand loyalty was measured on a scale devised by Kim, Han, and Park (2001). These scales were chosen for this study because of the positive results and validity as well as their precision.

Respondents were told about the meaning of a brand extension and more precisely a downward price-based brand line extension. To reinforce the understanding of the concept, two graphic examples were displayed, a parent brand with a downward price-based brand extension. The respondent was then asked to recall their favorite luxury brand, to keep it in their mind during the interview, and then to imagine the company deciding to sell a brand extension based on price reduction and asked to mention both the mother brand and the extension. Thus, the aim was always to provoke the participant to focus on the luxury brand and the vertical extension when answering the questionnaire.

As far as the research instrument is concerned, a structured survey was applied. A pilot study was conducted involving 20 people. On obtaining the data, we decided to modify the questionnaire by reducing the number of questions without altering the scales. Following the data collection phase, we ended up with a small but sufficient sample of 158 answered questionnaires. Various minimum sample sizes have been suggested for a valid study, e.g Hair, Sarstedt, Hopkins and Kuppelwieser (2014). suggests 130 instruments, Westland (2010) indicates 128, while Sivo, Fan, Witta, and Willse (2006) recommend 150 elements.

Regarding convergent validity, a bootstrap of over 5000 resamples was conducted with no sign changes in the resampling, which is the most conservative method. After that, we compared the results with sign changes at a construct level and individual changes. We used a one-tailed test at a significance level of 0.05. The results were consistent among the three methods, and all indicators obtained outer loadings higher than 0.70 as suggested by Carmines and Zeller (1979), so the reflective constructs were accepted. The reliability and validity of the scales were confirmed by the information obtained from values, which were over the established coefficients criteria.

Below are the details of standardised loadings, average variance extracted (AVE), composite reliability (CR) and Cronbach's alpha (CA) for each concept. As shown in Table 2, all item loads on their predicted factor as well as the loads for brand personality dimension on the formative second-order factor are significant.

As regards reliability and convergent validity evaluation, Cronbach's alpha scores are greater than 0.70, the suggested standard criteria by Nunally and Bernstein (1994). The composite reliability (CR) values exceed 0.89, far above 0.60, the proposed threshold by Bagozzi and Yi (1998). Finally, the average variance extracted (AVE) values are greater than 0.64, more than 0.50, the benchmark recommended by Fornell and Lacker (1981).

Evaluation of the discriminant validity (see Table 3) was based on chi-square differences and confidence interval (CI) (Anderson & Gerbing, 1988). The extracted variance for each construct was higher than the square root of the correlation between that construct and any other, as is suggested by Fornell and Larcker (1981).

Moreover, we ran the heterotrait-monotrait ratio (HTMT; Henseler, Ringle & Sarstedt, 2015) to test the discriminant validity, and the ratios for Brand Concept Consistency (BCC) and Self-Image Congruence (SIC), 0.697 and 0.898, respectively, were below the threshold of 0.90 (Fornell & Larcker, 1981). This means that all the pairs of reflective constructs share discriminant validity.

For assessing the structural model, we must analyse the dependent latent variable variance explained by the predictive constructs. Therefore, the criterion to apply is R2 and must be higher than 0.1 (Falk & Miller, 1992). In addition to R2, Hair, Sarstedt, Ringle, and Mena (2012) suggest the use of the effect size (f2), as well as path coefficients with their respective t-values for models with reflective indicators. We also assessed the cross-validated redundancy index (Q2) by blindfolding (Geisser, 1975; Stone, 1974) that will inform together to R2, about the predictive relevance of endogenous constructs, with values above zero indicating the predictive relevance of the model (Chin, 1998; see Table 5).

To assess validity for the formative brand personality construct, collinearity and weights’ statistical significance criteria were analysed (Chin, 1998). Most of the variance inflation factor (VIF) indicators were less than 5 and higher than 0.20, which is the recommended range to leave the preselected variables (Hair, Ringle & Sarstedt, 2011). The only exception shows a light collinearity tension through the second-order factors, with a 5.299 value. However, this does not affect the content validity criteria established by Hair et al. (2011). Regarding weights, all of them are statistically significant (see Table 2).

Regarding qualitative results, as summarised in Table 2 in which the relevant variables identified in the qualitative research are in the middle column, Propositions 1 and 2 were observed for luxury goods that are solely bought for emotional or symbolic motives. Here the downward price-based line extension most likely boosts the availability of the brand, which in turn diminishes the exclusivity character and thus alters the brand concept, rendering it inconsistent from the viewpoint of the consumer. Luxury goods that fall out of the pure image associations, i.e. brands bought for their superior quality, do not suffer from an inconsistent brand concept after a price-based line extension as this does not automatically diminish the quality proposition or promise of the brand towards the consumer. For this reason and only for functional brands, Propositions 1 and 2 were partially observed.

Also regarding Propositions 3 and 4, if both an inconsistent brand concept (P3) occurs due to a downward price-based line extension and an incongruent self-concept or mismatch results between the consumers’ image and the image of the brand (P4), loyal attitudes towards the luxury brand will diminish and so will consumer repurchase behaviour. These results are confirmed in the quantitative research (see H1b and H3b below).

Regarding quantitative results, as we can see in Table 5, brand concept consistency has a negative and direct effect on purchase intention; hence, the hypothesis H1a is supported. Thus, the lower the brand concept consistency, the lesser the purchase intention. Similar results although related to reactions towards brand extensions were observed by Park et al. (1991). On extending the parent brand into reduced-price luxury goods due to brand extension, certain characteristics could change, affecting the consumer's concept of the parent brand. This means that from the downward-price brand extension, the concept will be affected, and then, consequently, the purchase intention will be affected negatively.

The hypothesis H1b, unlike the P3 observed in the qualitative research, is not supported. The brand concept consistency weakly affects brand loyalty. The standardised beta and the t value are insignificant. In the qualitative research, P3 was observed for luxury goods solely bought because of the image, emotional and prestige value proposition of the brand. If consumers perceive the concept of the entire brand to be inconsistent and not clear anymore, their emotions towards the brand may change and attitudinal loyalty may diminish resulting in weakened behavioural loyalty. However, as Royo-Vela and Voss (2015) stated:

“Goods that are bought for functional reasons or rational motives, i.e. because a superior quality, have to be taken into account differently. It was argued that if the quality of the extended and original product stays the same, a question about the repurchase decision would be answered with yes if the brand is bought for quality attributes. In this case, the concept of the entire brand evolves mostly around the superior quality offering as value proposition and not around any symbolic, emotional or hedonic attributes. Therefore, if the quality of the good stays the same, the brand concept is not altered. If the quality instead would diminish due to an extension, the quality concept of the overall brand would also suffer and a repurchase would not be as likely as before”. (p.154)

This could explain why this relationship was not significant.

The hypothesis H2a is rejected. We observed a weak effect of brand personality on purchase intention. Therefore, rather than a direct effect on purchase intention, a moderating role between self-image congruency and purchase intention seems more accurate (see Aaker, 1997). Brand personality, which is studied as a formative concept, has a similar correlation with its four dimensions; all the results are significant at the 5% level. Overall, the four factors have a direct and positive effect on brand loyalty. Then, hypothesis H2b is supported. As stated by Liu et al. (2012), as long as the brand personality dimensions of sincerity, excitation, competence and sophistication reflect the personality of consumers, they will remain loyal to the brand. Thus, the lower the resemblance between the brand's personality and the consumer's personality, the lesser the loyalty towards the brand. The typical luxury brand buyer's loyalty will be significantly affected depending on the relative perceived difference between them and the brand extension buyer.

The highest result we observed was concerning self-image congruence (henceforth SIC) and its effect on purchase intention and brand loyalty. It significantly affects both concepts. Therefore, hypotheses H3a and H3b are accepted. Thus, the lower the perception of SIC, the lower the purchase intention. Self-image is strengthened by the consumption of brands that reflect user traits, and therefore its weakness will directly affect the purchase intention positively Davis and Lang (2013). also found a positive effect of SIC on product usage and purchase. Regarding H3b, the lesser the consumer's SIC, the lesser the loyalty towards the brand. Similar direct effects of SIC on brand loyalty were observed both in the qualitative research (see P4 above) and by Kressmann et al. (2006), Sirgy et al. (2008) and Liu et al. (2012). The loyalty towards a brand is negatively affected by the inconsistency between self-image and brand image following the introduction of the brand extension.

To assess the structural model, we have analysed the dependent latent variables variance explained by the predictive constructs (see Table 3). Therefore, the criterion to apply is R2, and it must be higher than 0.1 (Falk & Miller, 1992). In addition to R2, we also assessed the cross-validated redundancy index (Q2) by blindfolding (Geisser, 1975; Stone, 1974), which will inform, together with R2, about the predictive capacity of endogenous constructs, with values above zero indicating the model's predictive relevance (Chin, 1998) Hair et al. (2011). suggest that the use of the effect size (f2) is somewhat redundant for the size of the path coefficients with their respective t-values for models with reflective indicators. Regarding the proposed model, the results aim to explain the general model fit. The indicators support the fact that buying intention and brand loyalty are moderately explained by the antecedents with an R2 equal to 0.41 and 0.32, respectively (Hair et al., 2011). A medium predictive accuracy is confirmed by Q2 scores higher than 0.25 and lower than 0.5 (Hair, Sarstedt, & Ringle, 2019).

The luxury sector has historically proven to be a social and personal phenomenon, which despite its fragility in the face of the economic crises (Zhang & Kim, 2013), has demonstrated steady growth, becoming one of the most significant commercial activities. This development considerably impacts the demand in the luxury market, as the introduction of extensions such as mass-market strategy dilutes and weakens the brand image through various inconsistencies (Matthiesen & Phau, 2010). Indeed, the current literature related to brand extensions reveals the benefits and the inconveniences of the strategy, the consequences of the lack of fitting between the brand concept or the brand image and the consumer's self-image, especially with respect to the luxury market. In this sense, we advise to qualify the use of this strategy as a very sensitive issue.

As all the messages will be observed by the consumer, who creates a general perception made up of a set of beliefs, associations and impressions about the brand, the brand concept, personality and image should be observed carefully, analysing the synthesis of all the signals emitted by the brand very well. When price comes into play, its reduction could create a different impression of these luxury brands, affecting the perception of exclusivity and scarcity that drives the consumer to achieve higher social status through differentiation with respect to other consumers. Then, for luxury brands operating across multiple product categories, the challenge is to differentiate between the parent product line and the extension but always alluding to the core brand perceptions (Liu et al., 2012) and to create more affordable product categories (Loken & Deborah, 1998), while generating marketing strategies and communications to preserve the overall parent brand image (see below).

First, the reasons for purchasing the line extension could be either hedonistic/emotional reasons or functional/rational ones. The current study has not separated these two issues, which could be considered in the future. The congruence of the brand image analysis refers to associations through the typical user perceptions about the brand and its perceived appropriate use (Liu et al., 2012; Sotiropoulos, 2003). This study has analysed only the perception of typical brand users and the way they evaluate the correspondence between the brand image and personality and the self-image and personality.

Second, the consumer's brand involvement is another aspect that could be considered in future studies. According to Royo-Vela and Voss (2015), only known brands can create links that may be the key to the acceptance of brand extensions. Additionally, the symbolic nature of the luxury brands suggests that other constructions of self could be more relevant with real consumers.in studies related to luxury brands and their extensions

Third, profound study is required of the different parent brand segments, including the luxury brand extension users. Although various product categories have been researched, this methodology of segmentation needs application in various differentiated groups so that their reasons for purchase can be established more clearly between expensive and new cheaper buyers (see Dall'Olmo et al., 2013; Kapferer & Laurent, 2016).

Fourth, in this study, we focused on brand concept consistency and the attributes or associations of the brand concept. Naturally, one could argue that the brand can keep its identity even after the appearance of the brand extension, depending on the company's communication strategy and its efforts to preserve the parent's brand concept. However, these communication strategies necessary to maintain the brand concept after the brand extension are not part of this study; rather, they could be considered for future research.

Finally, the small sample size, although enough for modelling testing, is limited to only one country and should be increased to become more proportional and representative of the luxury consumer population.

This paper was partially funded by the Corvinus Institute for Advanced Studies (CIAS) and its Fellowship program. Corvinus University of Budapest.