Global/Local brands and performance
The role of Conspicuous consumption, Consumer ethnocentrism
and Product category cultural embeddedness
Tijmen Oudshoorn - 11948663 Supervisor: Vittoria Scalera
Program: MSc Business Administration – International Management Version: Final version
Statement of originality
This document is written by Tijmen Oudshoorn who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Abstract
Around the world consumers constantly choose between global and local brands. What drives consumers to choose for either local or global products and does this differ among different product categories?
Two highly relevant concepts in this context are Consumer Ethnocentrism (CET), the tendency to purchase local brands due to economic nationalism and cultural conservatism, and Conspicuous Consumption (CC), the consumption of goods to display the ability of purchasing those goods. The literature remains unclear about how those concepts relate to each other, and to global and local brand performance. Both concepts, however, cannot be analyzed without involving product category. A new construct is introduced to capture the effect of the degree to which a product category is embedded in an ethnic or national culture; Product Category Cultural Embeddedness (PCE).
In this study, the effect of CET, CC and PCE on global and local brand performance is researched. Next to that, it is tested whether these relationships differ for developing and advanced economies and high or low levels of PCE. Data from 766 brands in 11 Asian countries was analyzed.
Evidence was found that CC influences global brand performance positively in developing economies and negatively in advanced economies. CET influences local brand performance positively, but no separate effects have been found for developing and advanced economies. PCE also influences global and local brand performance and moderates the relationship between CET, CC and local brand performance although further research is required to understand these effects better.
Keywords: Global brands; local brands; consumer ethnocentrism; conspicuous consumption product category; product category cultural embeddedness
Contents
Introduction ... 5
Literature review ... 7
Global Brand Architecture ... 7
Global brands ... 8
Local brands ... 9
Factors influencing the G/L – Performance relationship... 10
Advanced/developing economies ... 12
Literature gap ... 13
Theoretical background and hypothesis development ... 15
Research Design ... 19 Dependent variable ... 20 Independent variable ... 20 Moderators ... 21 Control Variables ... 21 Preparation data ... 24 Empirical Strategy ... 24 Results ... 26
Results global brand performance ... 30
Results local brand performance ... 34
Additional analyses ... 37
Discussion ... 38
Conclusions ... 44
References... 45
Introduction
An essential element leading to high performance of a Multinational Enterprise (MNE) is the possession of one or more strong brands. They can help the MNE to build a strong identity in the marketplace and develop a strong customer base (Keller, 2008). Because brands are essential for MNE’s, an extensive amount of research has been done into what determines consumers to choose for a certain brand. A key element in this is whether this brand is domestic or global (Chabowski, Samiee, & Hult, 2013). Globalization has led to the emergence of more and more global brands. Although local brands still outnumber global brands, MNEs are reducing the number of local brands in favor of global ones (Kapferer, 2002).
Global brands, in general, have been found to lead to a higher performance (Holton, 2000; Steenkamp, Batra, & Alden, 2003; Talay, Townsend, & Yeniyurt, 2015). Already in 1983, Levitt mentioned that consumers worldwide prefer the greater reliability and superior quality that global brands would offer (Levitt, 1983). Because consumers constantly choose between global or local brand, various consumer characteristics impact the performance of brands, such as animosity against brands due to political conflicts, consumer ethnocentrism, attitude towards globalization and conspicuous consumption.
One of the concepts that has been mentioned the most in this area is Consumer Ethnocentrism (CET) (Steenkamp et al., 2003; Strizhakova & Coulter, 2015). CET is the tendency to perceive domestic brands as having a higher quality. Highly ethnocentric consumers will therefore often choose local products over their global counterparts (Steenkamp et al., 2003). Another country characteristic related to CET that has been found to influence the local/global choice is conspicuous consumption (CC). This entails the tendency to provide visible evidence of the ability to afford luxury goods (Piron, 2000).
Because in advanced economies, domestic products are often perceived as better, more premium goods, CC and CET are usually correlated here. But, in developing economies, domestic products are usually perceived as having a lower quality than imported products, especially when they come from a country of a higher origin (Wang and Chen, 2004). As these global brands are thus perceived as of a higher quality, it would mean that global brands perform better in a developing economy with a high level of conspicuous consumption. Both concepts are clearly related to each other and to brand performance, but the interplay between them remains unclear. As most MNEs aim to use their global brands in as much countries as possible to maximize brand awareness and economies of scale, it is key to realize what determines the performance of global brands at the country level.
Next to CET and CC, consumer preferences for local/global brands are largely formed at the product level (Davvetas & Diamantopoulos, 2016). Still, the literature on how product category influences brand preference is limited. Some product categories appear to be dominated by global brands and in others the majority of brands is local. Certain product categories, especially food and drinks, are very much related to culture and historical context for instance, which results in more local brands (Özsomer, 2012; Winit, Gregory, Cleveland, & Verlegh, 2014). A logical next question is what it would mean when a product category is highly embedded in a culture.
Take for instance Starbucks, one of the most international brands with a presence in 50 countries through 15.000 outlets (Starbucks, n.d.). Yet, they have entered Italy only in 2018, years after their entry into other European countries. Because coffee and the routine of drinking it is highly embedded in the Italian culture, it has been extremely challenging to enter this market for Starbucks (The Economist, 2016).
A new construct is introduced in this study: product category cultural embeddedness (PCE). This construct captures the extent to which a product category is culturally grounded in a country, such as coffee in Italy. The construct is introduced to be able to measure whether and how this phenomenon influences brand performance.
Because the choice for global or local brands is very much related to the product category, the dynamics of CET and CC in relation to global and local brand performance cannot be studied without taking product category into account. This study will synthesize the concepts of CET and CC into a framework including economy advancedness and PCE.
Contributions will be made to the existing literature by researching the moderating effects of the role of PCE and economy advancedness on the relationship between CET, CC, global and local brand performance. It is highly relevant because it provides MNE’s with determinants to decide on whether to enter a foreign market through its own global brand or to acquire an existing local brand for instance. It can provide tools for firms to design their global brand architecture.
This study researches to what extent global and local brand performance is influenced by PCE, CET and CC and whether the relationship between CET, CC and brand performance is moderated by PCE and economy advancedness. Two separate panel regression analyses have been performed on a sample (N = 8.718) of global brands and a sample of local brands active in the beer, wine and spirits markets in 11 Asian countries over the years 2006 - 2015. In short, the results indicate that global brand performance is positively influenced by a high degree of
CC in developing economies, and by low degree of CC in advanced economies. For local brand performance, this effect is the other way around, a high degree of CC in developing economies influences it negatively and a high degree of CC in advanced economies influences it positively. For CET, a general positive effect for local brand performance has been found, but this does not differ for developing and advanced economies. PCE also influences brand performance, although more research is necessary to grasp the full meaning of those results.
First, an extensive review of the existing literature will be provided, followed by the theoretical framework with the hypotheses. After this, the methods employed in this study will be discussed. Then a thorough analysis will be performed, followed by some additional analyses. Finally, the results will be discussed to assess the contributions and managerial relevance.
Literature review
In this chapter, the existing literature about the main topics within this research will be reviewed. First, the concept of global brand architecture will be covered, followed by global and local brands and their relationship with performance. Then, the concepts of CC and CET will be explained. Furthermore, the literature on product category and product role in a society will be covered. Finally, the literature gap will be justified.
Global Brand Architecture
MNE’s often possess various different brands dispersed over the countries where they perform activities. This is the result of an evolutionary process following decisions to enter a country or to acquire a brand following the incremental internationalization model (Johanson & Vahlne, 1977). Brand strategy used to be decided at the country or brand level, without considering the overall structure over the various international markets where the MNE is active. Nowadays, a global brand architecture (GBA) is often installed to manage this portfolio. GBA refers to “a formal process and outcome by which management rationalizes the firm's brands and makes explicit how brand names at each level in the organization will be applied” (Douglas, Craig, & Nijssen, 2001). Having a GBA enables the MNE to differentiate the brands and to ensure consistency across countries (Douglas et al., 2001). This is said to be a response to the fact that product segments now transcend boundaries (F. Ter Hofstede, Wedel, & Steenkamp, 2002).
Brands within the MNE’s GBA can have four different strategies: global, multi-regional, regional and domestic (Townsend, Yeniyurt, & Talay, 2009). A global brand is
characterized by a presence in all major market regions of the world and by a standardized approach across markets. Multi-regional brands are not present in all of the world regions, such as Asia, North America or Europe, but just in one or two (Rugman & Verbeke, 2004). A brand is also multi-regional when its marketing program is not standardized or centralized across markets. Regional brands subsequently are present in only one region, but in multiple countries. Local brands are present only in individual national markets (Townsend et al., 2009). Somewhat similar categorizations have been developed by several other scholars (Douglas, Craig and Nijssen, 2001; Alden, Steenkamp and Batra, 2006), each of which can be classified along a continuum from local to global.
The reason why a GBA is a key component of the MNE’s international marketing strategy is because it enables them to leverage their brand’s equity across markets, integrate acquired brands and rationalize global strategies (Douglas et al., 2001). MNEs can employ global, regional or local strategies for their brands and the literature is somewhat divided on which strategies lead to the highest performance. Below, the key studies on these different ideas will be reviewed.
Global brands
Fundamental changes over the past two decades have changed the international business environment, causing a continually expanding number of firms to internationalize, with the formation of more global marketing strategies as a consequence (Chabowski et al., 2013; Douglas et al., 2001). This makes sense as MNEs can achieve significant economies of scale not only in marketing, but also in manufacturing and R&D (Kapferer, 2002). Due to their large size global brands can create barriers to entry and have a unique image worldwide, leading to a standardized and recognizable brand image throughout the world (Schuiling & Kapferer, 2004). It is found that global brands are key in building a sustainable competitive advantage and that they lead to a higher esteem by consumers (Johansson & Ronkainen, 2005). In the branding literature, there is a difference between how consumers perceive a brand’s global- or localness and how global or local a brand is based on its geographic scope. Perceived globalness is not necessarily the same as a global availability because a brand’s domestic origin or cultural adaptability can also be a source of perceived localness (Dimofte, Johansson, & Ronkainen, 2008). Consumers use cues to evaluate a brand when they are unfamiliar with it. These cues can be extrinsic to the product or intrinsic (Richardson, Dick, & Jain, 1994). Globalness can then be seen as an extrinsic cue, which brings a certain value.
that consumers around the world prefer the so-called greater reliability and superior quality that global products offer (Levitt, 1983). Twenty years later this topic was revisited by the HBR, posing that consumers worldwide associate global brands with three main characteristics: brand quality, global myth and social responsibility (Holt, 2004). The country of origin effect as an indicator for brand quality is only one third as strong as a brand’s globalness as an indicator. Global brands are global myths because consumers look to them as cultural ideals. One respondent best expressed this dimension by saying: “local brands show what we are; global brands show what we want to be.” Finally, consumers believe that global brands have a greater social responsibility due to the extraordinary influence they wield on society’s wellbeing (Holt, 2004).
Another study also found that a high perception of globalness leads to a higher perception of quality and prestige and through those to a higher purchase likelihood (Steenkamp et al., 2003). Furthermore, evidence exists that a large portion of people around the world have a preference for globally accepted consumer cultural images and symbols, which are characteristics of global brands, over traditional and local ones (Alden et al., 2006). It is even argued that the consumption of global goods is associated with ‘good life’ perceptions such as modernity, progress and the promise of abundance (Holton, 2000). This is further acknowledged by a study that argues that global brands in developing economies can serve as a global passport (Strizhakova, Coulter, & Price, 2008). Young consumers used these global brands to create imagined global identities. In a study on developing economy consumers, it has been found that those consumers attitudinally prefer nonlocal brands, especially coming from the West (Batra, Ramaswamy, Alden, Steenkamp, & Ramachander, 2000). Finally, in a worldwide, longitudinal study in the automotive industry, it has been found that global brands perform better than their local counterparts (Talay et al., 2015). Because of these benefits MNEs are reducing their number of local brands significantly and focus more on their brands that can be globalized (Kapferer, 2002).
Local brands
Douglas & Craig (2011) argue that although in general markets are converging, some markets actually are becoming more diverse, resulting in marketers having to deal with economic and cultural heterogeneity more and more. Therefore, they suggest MNEs should focus on each of the spheres they operate in separately and manage these strategies locally (Douglas & Craig, 2011).
intimate relationships with local markets resulting in a longer time to react when problems occur. They also argue that local brands can offer certain strategic advantages such as a greater strategic flexibility and the possibility to hedge against the risks of a portfolio with mostly global brands (Schuiling & Kapferer, 2004). Local brands are perceived as unique and original and traditionally benefit from high awareness and close relationships with their consumers (Özsomer, 2012). Besides reflecting the character of the national market, local brands can even help define it (Dimofte et al., 2008). Therefore, some local brands can be perceived as local icons because they are associated with the local culture, heritage and country (Özsomer, 2012). ‘Local iconness’ has been mentioned more often in the literature. Local Icon brands have been defined as “consumer brands that carry consensus expressions of particular values held dear by some members of a society” (Holt, 2004, p. 4). Similar to Özsomer, Ger has posed that a closer connection to local culture, heritage and national identity is a major strength of local brands (Ger, 1999).
Factors influencing the Global/Local – Performance relationship
There are many factors that influence the global/local – performance relationship. Some are specific for the host country, some for the firm owning the brands and others are specific consumer characteristics. For instance, it is proven that self-concept and identity significantly influence consumer behavior. Two key constructs that have been identified in relation to global and local brand performance are animosity and consumer ethnocentrism (Gurhan-Canli, Sarial-Abi, & Hayran, 2017).
Animosity concerns the remains of dislike related to previous or ongoing military, political or economic events (Klein, Ettenson, & Morris, 1998). In that particular study, the animosity of Chinese consumers against Japanese brands was tested. It was found that those consumers held negative values against buying Japanese brands, irrespective of the quality of the products.
Consumer ethnocentrism (CET) is defined as “the beliefs held by consumers about the appropriateness, indeed morality, of purchasing foreign-made products” (Shimp & Sharma, 1987). Ethnocentric consumers prefer domestic goods because they think products from their country are the best (Klein et al., 1998). Ethnocentric consumers may even choose for a lower quality or higher price, just to avoid contact with the global brand. Non-ethnocentric consumers have a more cosmopolitan outlook and are said to have a higher degree of cultural openness (Baughn & Yaprak, 1996). CET negatively moderates the aforementioned relationship between perceived brand globalness and perceived quality (Steenkamp et al., 2003). In another study it
has been found that indeed ethnocentric consumers remain biased towards purchasing a local brand over a global brand, even when the price is higher (Winit et al., 2014). Also, the attitude towards local products of consumers is positively influenced by CET and the attitude towards global products in turn is negatively influenced by it (Steenkamp & de Jong, 2010).
Opposed to the preference for local brands of ethnocentric consumers, consumers positive towards globalization exhibit a preference for global brands (Riefler, Diamantopoulos, & Siguaw, 2012). The same has been found for consumers appreciating the emerging global consumer culture (Alden, Steenkamp, & Batra, 1999).
Another cultural concept that is related to the previously discussed concepts is conspicuous consumption (CC), the consumption of goods to show the ability of purchasing (luxury) goods, using prominent visible evidence (Piron, 2000). People from all social classes, richest to poorest can be engaging in CC. By showing their wealth, they achieve a greater social status (Bagwell & Bernheim, 1996). It is especially interesting because the effect it has on the choice between global and local brands is mixed. In advanced economies countries, CC values and CET values are usually correlated, as the purchase of the ‘superior’ domestic goods can also be used to show off (Wang and Chen, 2004). However, in general, in developing economies global or imported products are perceived as higher quality products. Even ethnocentric consumers may then evaluate the quality as better, especially if the country of origin has a better image (Yagci, 2001). Therefore, in developing economies CC is found to have a mitigating effect on CET in (Wang and Chen, 2004). In another study, it has even been found that in developing countries, nonlocal brands are often preferred because of social status-enhancing reasons, which was amplified by the degree to which those consumers were sensitive to normative influences (Batra et al., 2000). The social signaling power of the product category of the brand also amplifies this preference.
Opposed to the abundance of literature on the brand- and consumer-related determinants of global/local brand preference, few studies have investigated the effect that product category has on these choices (Davvetas & Diamantopoulos, 2016). It is relevant because product category seems to influence both the concept of CC and CET. It has been found that consumers often make choices that diverge from those of other consumers to communicate their desired identities effectively. This effect is stronger for product categories that are seen as highly symbolic for the identity, such as music or apparel (Berger & Heath, 2007). Moreover, it has been found that CET is also contingent on product category symbolism (Strizhakova & Coulter, 2015). In another study, global brands were found to be superior in categories where
consumption is visible to others or products that have the capacity to provide identity signals, even in advanced economies (Davvetas & Diamantopoulos, 2016).
When discussing product category in general, it has been found that food products are more often promoted using local cultural aspects opposed to durables and technology products, which are promoted as symbols of the universal consumption culture, such as Apple or Samsung (Alden et al., 1999). Also, in their study, Johansson & Ronkainen explicitly use the label of ‘global brand’ for product categories like automobiles and computers and the label ‘multi-domestic brand’ for food and drinks (Johansson & Ronkainen, 2005). Moreover, they find that the effect of global brands leading to a higher brand esteem is stronger in global product categories. Similarly, another study finds that consumers place a high importance into purchasing global brands in categories such as electronics but a low importance when it comes to categories such as fast food or chocolate (Winit et al., 2014). It is also argued that these differences between product categories are due to the country of origin effect or perceived brand equity (Winit et al., 2014). Özsomer finds that local brands are often preferred in food products as those are often culturally grounded product categories (Özsomer, 2012).
In a recent study on how product category shapes preferences for local/global brand preferences, it has also been found that product category is an important determinant in consumer choices between local and global brands. Moreover, global brands were perceived as better than local ones, when the purchase entails product risk and provides functionality instead of experiential enjoyment (Davvetas & Diamantopoulos, 2016).
Advanced/developing economies
Often, a distinction between countries is made in the literature by the degree to which their economies are advanced. Most of the times, this is a dichotomous distinction, either a country is advanced or it is developing. Developing countries are defined by the international monetary fund (IMF) as having a low GDP per capita, a low export diversification and a low integration into the global financial system. Advanced economies are in turn defined by the IMF as having a high GDP per capita and a significant degree of industrialization (IMF, 2016). Among developing economies, there are emerging economies, transition economies and least developed countries. The second group, transition economies, are countries that previously were a planned economy, a communist country, and now are committed to strengthen their market mechanisms through liberalization, stabilization and the encouragement of the private enterprise (Hoskisson, Eden, Lau, & Wright, 2000). Later, a more fine-grained understanding of this country context has been developed. Four types of developing economies can be
identified along two dimensions: institutional development and the development of infrastructure or factor markets (Hoskisson, Wright, Filatotchev, & Peng, 2013).
These differences between economies are relevant, as it determines what the business climate of a country looks like. Because developing economies are so different from advanced economies, much of the literature on MNE strategy for advanced economies is not valid for developing economies. Because of the different characteristics of both types, theories that hold in advanced economies, might not hold in developing economies. Ramamurti (2012) acknowledges this and states that some theories might still be valid but other might not be anymore (Ramamurti, 2012). Consequently, those theories need to be tested for developing economies as well.
Literature gap
The literature discussing the effects of local or global brands on performance is abundant. Although some argue that local brands could lead to a higher performance (Kapferer, 2002; Schuiling & Kapferer, 2004), most of the literature argues that global brands are usually perceived as of a higher quality and in general perform better (Alden et al., 2006; Steenkamp et al., 2003; Strizhakova et al., 2008). The relationship between local and global brands and their performance is influenced by various factors at the country-, firm- and consumer-level, such as consumer animosity or CET on the one hand and CC or a positive attitude towards globalization on the other hand. It has been proven that CET usually has a negative effect on the performance of global brands as these consumers favor local brands (Steenkamp et al., 2003). At the same time, CC mitigates this effect on the performance of global brands in the case of developing economies (Batra et al., 2000; Wang and Chen, 2004). These concepts have been proven to influence the performance of both global and local brands. However, CET has only been tested on an individual level, using questionnaires to obtain the data. Therefore, it would be interesting to look at this phenomenon on a country level. Next to that, the combination of CET and CC in one model is expected to yield an interesting dynamic that is different for advanced and developing economies. Which countries score high on CET or high on CC and what does that mean for the brand performance in those countries? With that information, it would be possible for MNE’s to adjust their global brand architecture accordingly. Also, CC has only been tested as a moderator of CET and it could very well be that CC influences the global and local brand – performance relationship directly. Moreover, as the effects appear to be different for developing and advanced economies, testing both CET and CC moderated by the advancedness of the economy would provide a complete synthesis of
previous works and some interesting research opportunities. However, this framework would not be complete if product category is not involved, as the effects of CC and CET on brand performance are contingent on product category (Berger & Heath, 2007; Davvetas & Diamantopoulos, 2016; Strizhakova & Coulter, 2015).
Several studies have focused on the concept of product category and its influence on CC, CET and local/global brand preference but they only partially explain what the determinants of these differences are. Some determinants are the country of origin, product risk, identity signal capacity or the degree of symbolism of the product category (Davvetas & Diamantopoulos, 2016; Strizhakova & Coulter, 2015; Winit et al., 2014). It has also been found that food and drinks product categories are usually more culturally grounded due to more heterogeneous customer demands (Özsomer, 2012). A relatively unexplored phenomenon within this literature, is the degree to which a product category is or has been consumed in society. In Germany, for instance, Beer is a well-known alcoholic beverage, rooted in the German culture. This is confirmed by the average per capita consumption of beer over the past 50 years, which is well above the worldwide average (WHO, 2014). To capture this effect, the concept of cultural embeddedness of products (CEP) has been introduced by Jakubanecs & Supphellen (2016). It is defined as ‘the degree to which a product is connected with ethnic/national culture’ (Jakubanecs & Supphellen, 2016). Although they do theorize about the effect it has on quality perceptions of global and local brands, this concept has not been researched as directly influencing global and local brand performance and CET and CC. Davvetas and Diamantopoulos mention that an interesting future research opportunity could be the influence of the importance of a product category for a local culture on global brand preference (Davvetas & Diamantopoulos, 2016). Moreover, several scholars in the cross-cultural research domain, have argued for the development of new theoretical constructs with the power to explain consumer behavior within and across cultures (Maheswaran & Shavitt, 2000; Singelis, 2000) Therefore, a new construct, product category cultural embeddedness (PCE), will be introduced. This construct is similar to CEP in that it concerns the degree to which a product (category) is connected to a culture, but it is broader than this. Whereas CEP is defined as a product being a true symbol for the ethnic/national culture, PCE is defined as a product being consumed higher than average in the ethnic/national culture. For instance, hamburgers are seen as a symbol for the U.S. culture, but they are also consumed above average in most other western world regions. Australia turns out to be even higher on per capita
hamburger consumption and the top ten consists only of the U.S. and European countries with the exception of Russia and Japan (Ferdman, 2016).
Through this study, the knowledge about whether and how PCE influences local/global brand performance can be expanded. MNE’s can use this to design or evaluate the GBA as it could inform them whether either local or global brands would be performing better in a particular country. Also, MNE’s will be able to assess how the concepts of CET and CC will influence their GBA.
Hence, this study will focus on the effect of PCE, CET and CC on global and local brand performance, moderated by PCE and economy advancedness. The following research question has been drafted and the corresponding conceptual model can been found in figure 1, after the hypotheses development.
To what extent is global and local brand performance influenced by product category cultural embeddedness (PCE), consumer ethnocentrism (CET) and conspicuous consumption (CC) and to what extent is the relationship between CET, CC and global and local brand performance moderated by product category cultural embeddedness (PCE) and economy advancedness (EAD)?
Theoretical background and hypothesis development
Conspicuous consumptionAs discussed before, CC is the consumption of products with the goal of showing off to others, possibly to climb up the social status ladder. Consumers use products as a means to construct, strengthen and communicate their desired identities (Arnould & Thompson, 2005). Therefore, consumers prefer brands that match their identity and with which they can communicate their identity most effectively (Reed, Forehand, Puntoni, & Warlop, 2012).
The most extreme form of communicating a desired identity and consuming with the purpose of enhancing one’s status, would then be conspicuous consumption. Although it seems that this is only a phenomenon for the richest group of people, it occurs in all social and income groups from richest to poorest (Wang and Chen, 2004). Consequently, if a country holds high CC values in general, one would then say that global brands are more popular in that country, due to the fact that conspicuous consumers prefer these global brands over local brands.
In developing economies, consumers often prefer nonlocal brands to local brands that fit the identity they want to communicate (Batra et al., 2000). The reason behind this is that
global products are in general more expensive and they promote a foreign lifestyle (Ger, 1999). In turn, that would then lead to a higher global brand performance in developing economies.
Conversely, in advanced economies, local products or brands are often even perceived as of a higher quality. Consumers perceive brands from developing countries as more economical, but also poorer in quality, performance and originality compared to brands from advanced economies (Ahmed & d’Astous, 2001). Therefore, in advanced economies, conspicuous consumers will probably choose for local brands over global ones, as these are perceived as of the highest quality and as these are the most congruent with their desired identities. In advanced economies with high CC values, the expectation would then be that local brands are more popular and thus have a higher performance. Therefore:
H1a: In developing economies, conspicuous consumption will influence global brand performance positively
H1b: In developing economies, conspicuous consumption will influence local brand performance negatively
H2a: In advanced economies, conspicuous consumption, will influence global brand performance negatively
H2b: In advanced economies, conspicuous consumption, will influence local brand performance positively
Consumer Ethnocentrism
As discussed previously, CET is defined as ‘the beliefs held by consumers about the appropriateness, indeed morality, of purchasing foreign-made products’ (Shimp & Sharma, 1987). It can be explained as consumers taking pride in their country’s brands, symbols and culture. Economic nationalism lies at the basis of CET (Baughn & Yaprak, 1996). Economic nationalist consumers tend to think that purchasing local brands leads to more individual wellbeing, through a higher economic prosperity. Global brands even can be seen as a threat to this prosperity. With CET, however, this is taken one step further and global brands are even seen as a threat to the culture of the country (Steenkamp et al., 2003). This can result in the irrational tendency to purchase local brands even if the quality is lower or the price is higher (Baughn & Yaprak, 1996). As ethnocentric consumers prefer to buy domestic brands, global brands are expected to perform worse in countries with high CET values.
In developing economies, domestic products are usually perceived as of a lower quality than foreign products. Therefore, one would expect the relationship between CET and global
brand performance to be less negative. However, as CET is partly irrational, consumers might still be choosing for local brands. This implies that the lower quality of local brands will mitigate the effects of CET on local brand performance in developing economy, but there will always be the effect of the cultural threat.
Because of this, the effect of CET on global and local brand performance still is expected to be in the same direction as for advanced economies. However, the effect is expected to be less strong than in advanced economies due to the mitigating effect of the general preference for global products. Therefore:
H3a: In advanced economies, consumer ethnocentrism will influence global brand performance negatively
H3b: In advanced economies, consumer ethnocentrism will influence local brand performance positively.
H4a: In developing economies, consumer ethnocentrism will influence global brand performance less negative than in advanced economies.
H4b: In developing economies, consumer ethnocentrism will influence local brand performance less positive than in advanced economies.
PCE
The last hypotheses have to do with product category cultural embeddedness (PCE). The consumption of a certain product category in a certain society is often determined by the history of that product category and by the role that it plays in the societies’ culture. It is likely that when a category plays a significant role or when it is highly embedded, that there is a high consumption or demand of that product category, compared to other countries. This high demand, causes that manufacturers establish themselves in those countries with a high demand or a large domestic market. This argument is based on the home market effect, a term from international economics (Krugman, 1980). An industry will base itself in a country where most of its products are consumed to minimize transportation costs. Consequently, strong domestic brands emerge in that industry. According to the well-known five forces framework by Porter, a high degree of rivalry in a country, will make it more difficult for foreign firms to enter that market (Porter, 2008). It is expected that a high degree of PCE in a country will mean that the domestic product market is well developed. This in turn means that foreign global brands will have more difficulties with entering such a product market on the one hand, and more
difficulties performing well once they have entered the market on the other hand. Therefore, this will mean that global brands have a lower performance in countries with a high PCE.
Also, the quality of locally produced products can be perceived as higher due to the country of origin effect (Elliott & Cameron, 1994). Consumers evaluate the quality of products using stereotypes associated with certain countries (Schooler, 1965). A high PCE level and a strong connection between a product and a culture are usually the source of such stereotypes. Therefore, it can be expected that foreign- or global brands are at a disadvantage against local brands because the quality is perceived as higher than their foreign counterparts.
Moreover, as described above, a high level of CET is related to the purchase of local brands. However, for local brands with a high PCE, this effect could be even stronger. If the product category is very much culturally embedded, the threat to the culture of foreign or global brands is even larger. Therefore, it is expected that the negative effect of CET on global brand performance becomes stronger for a high degree of PCE.
For CC, a similar argument can be made. Generally, a conspicuous consumer tends to choose global over local brands, due to the higher perception of quality, prestige and power of global brands (Steenkamp et al., 2003). However, a high PCE would mean that the quality perceptions of local brands are also high and together with the low presence of global brands that could mean that conspicuous consumers will stick to the local brands more often, even in developing economies. PCE would then have a mitigating effect on the relationship between CC and global or local brand performance. Therefore,
H5a: The embeddedness of a product category in a culture (PCE) will negatively influence global brand performance.
H5b: The embeddedness of a product category in a culture (PCE) will positively influence local brand performance.
H6a: The embeddedness of a product category in a culture (PCE) will positively moderate the relationship between consumer ethnocentrism and global brand performance.
H6b: The embeddedness of a product category in a culture (PCE) will negatively moderate the relationship between consumer ethnocentrism and local brand performance.
H7a: The embeddedness of a product category in a culture (PCE) will negatively moderate the relationship between conspicuous consumption and global brand performance.
H7b: The embeddedness of a product category in a culture (PCE) will mitigate the negative effect of conspicuous consumption on local brand performance.
Figure 1 – Conceptual model
Research Design
To be able to study the research question above, a quantitative analysis will be done using secondary data from databases. The research setting is the beer, wine and liquor market in Asia. This geographical setting has been selected due to the presence of countries with both advanced and developing economies. The product segment of alcoholic beverages has been selected because consumption differs highly per country, especially in Asia, partly due to religious attitudes. Also, alcohol consumption has historically been relatively low, but is on a rise over the past year. In China for instance, the per capita alcohol consumption was 0,5 liters of pure alcohol in 1960 and 4.2 liters in 2015 (WHO, 2014).
The brands of alcoholic beverages operating in Asia and their market shares have been retrieved from the Passport database. Passport is a global market research database providing insights on industries, economies and consumers worldwide. The sample consists of 991 unique brands operating in 16 different countries (China, Hong Kong, India, Indonesia, Japan, Kazakhstan, Korea, Malaysia, Pakistan, Philippines, Singapore, Taiwan, Thailand, Uzbekistan, Vietnam). It is a panel dataset with observations of each brand in each country for the years 2006 – 2015. The dataset in total contains 17.351 observations and contains the following variables: brand category (Beer, Wine, Spirits), brand, host country, brand type (local, regional, global), company name, country of origin (COO) of the company, year of observation, market share (%). In table 1 below, an example of two observations is given.
Product category cultural embededness (PCE) Conspicuous Consumption (CC) Consumer Ethnocentrism (CET) Advancedness economy
Global and local brand performance
Table 1 – Two examples of observations
Category Brand Country Type Company COO Year Mshare Beer Heineken China Global Heineken NV Netherlands 2006 0,001 Beer Anchor China Regional Heineken NV Netherlands 2012 0,002
Dependent variable
The performance of a brand will be measured using the market share of brand I in country J. Market share provides a robust measurement as this is based on actual objective data instead of subjective data based on surveys. Market share has been used as a proxy for brand performance extensively, which allows this study to be compared to other studies (Guo, 2013; Iversen & Hem, 2011; Swoboda, Pennemann, & Taube, 2012; Talay et al., 2015). In the first part of this study, all the global brands will be the dependent variable. The rest of the sample consists of both local and regional brands. In this study, as most regional brands are only active in 1.5 countries on average, those brands will be treated as local brands as well. Therefore, in the second part the local brands (including regional brands) will be the dependent variable.
Independent variable
The independent variables in this study are CC and CET. First of all, CC will be measured using Hofstede’s scores for the Power distance (PD) and the Masculinity – Femininity dimension (Masc) as a proxy. Power distance because of the fact that in countries scoring high on this, prestige, wealth and authority are seen as crucial factors in forming social classes as well as shaping the relationships between them. People are more inclined to consume conspicuously to follow and imitate their aspirational social classes (Talay et al., 2015).
In countries with a high Masculinity, very distinct gender roles, the emphasis is on success, achievement and the acquisition of wealth (G. Hofstede, Hofstede, & Minkov, 2010). Purchasing highly qualitative or high-reputational products is a way of demonstrating success and achievement in highly masculine cultures (de Mooij & Hofstede, 2011). Therefore, the cultural scores for both of these dimensions are used to assess the level to which CC takes place in a country. For both variables (PD, MAS), every country can have a value between 0 and 100. Both variables will be combined into the variable CC by computing their standardized average. Next to that, CET will be measured using the collectivism – individualism dimension (COLIND) by Hofstede and data from the World Values Survey (WVS), as a proxy. Collectivist cultures experience significantly higher levels of patriotism and CET than individualist cultures and collectivist people often subordinate their personal interests for the country’s welfare (G.
Hofstede et al., 2010). Therefore, this dimension can be used to measure the degree of CET in a country.
The WVS features a question about the proudness of people of their nationality, which is also related to CET. The question is as follows: “How proud are you to be [Nationality]?” The respondents can choose between the answers “Very proud”, “Quite proud”, “Not very proud”, “Not at all proud” or “I am not [Nationality] (WVS, 2014). Consequently, this variable yields a number of respondents per answer, which have been averaged. This variable will be named NatWVS. For the first variable, COLIND, countries can have a value between 0 and 100 and for the second variable, NAT, the lowest value is 271, and the highest value is 382,7. Both variables will be combined into the variable CET by computing the standardized average.
Moderators
The first moderator, product category cultural embeddedness (PCE) has been measured using the per capita consumption of pure alcohol in liters of a certain product per year in a country. The goal is to measure the embeddedness of a product category in a culture. Therefore, the per capita consumption of each product category (Beer, Wine & Spirits) has been used to assess the role that product plays in that country. Data of the per capita consumption from 1965 to 2015 has been used to calculate the average per capita consumption of a product category over 50 years. This is used as an indicator of how ‘popular’ or embedded that product category is in a country. The data is retrieved from the databank of the WHO, which is accessible through the website. For this variable, countries have three scores, one for each product category. For instance, in China, the average per capita consumption of beer from 1960 to 2015 is 0.57 (Measured in liters of pure alcohol). Even though the current per capita consumption of beer is 1.88, the per capita consumption of beer in China in 1965 was 0.01. The variables have absolute values which represent the alcohol consumption in liters of pure alcohol.
Consequently, the level of advancedness of a country’s economy will moderate the moderation effects of CC and CET. For this, the country classification in the world economic outlook by the international monetary funds from 2016 will be used. There are two classifications: advanced economy and developing economy (IMF, 2016). Therefore, this will be a dummy variable with 1 for advanced and 0 for developing economies.
Control Variables
Several control variables will be used that could potentially influence the dependent variable. A brand’s home country, for instance, could influence the negative relationship between CET
and global brand performance, as in that case an ethnocentric consumer will have no problem with the consumption of that global brand. This variable will be operationalized as a dummy variable. The variable takes on a value of 1 in case a global brand originates from the same country as the host country of that observation.
On the country level, population size, GDP per capita, the median age of a country, the education level, the urbanization rate and the market commitment of a company to a country will be added. Also, more common variables such as firm size and age of the firm owning the brand will be included.
First of all, firm size and age have to be controlled for because these could influence the way the brands of that firm are marketed or the brand image of those brands. Firm size has been retrieved from the Orbis database. Orbis uses a classification of firms in the database ranging from small company, medium sized company, large company to very large company. This distinction is based on several measures such as the number of employees, operating revenue and total assets (Orbis, 2018). Firm age has been calculated using the date of incorporation found in the Orbis database. Next to that, the population size, GDP per capita, median age and education level could influence the purchasing behavior of the consumers. A large population means a larger market size making that market more attractive for global brands to enter and a higher per capita income could cause consumers to be able to buy more global brands. A high median age could imply that consumers are more accustomed to local brands and a high education level could mean that consumers are more conscious about the choices that they make regarding their buying behavior. The population size and GDP per capita, measured in US dollars, have been retrieved from the world databank for each of the years in the sample (The World Bank, 2016b, 2016a). The median age and education level have been retrieved from the United Nations human development index (HDI). The education level is based on the HDI education index and can yield a value between 0 and 1 (United Nations, 2015). The data was only available for the years 2005 and 2010 – 2015. Therefore, for the years 2006 – 2009 the value of 2005 has been used. For the median age, only data from 2005 and from 2010 was available. Urbanization has been measured by taking the percentage of the population that lives in urban areas. The final control variable, market commitment, is the degree to which a firm employs all of the brands in its portfolio in a certain host country, as found in Talay et al. (2015). This is the total number of brands in the portfolio of a company and the number of brands of a company in a specific host country, resulting in a percentage of the brands of a firm that are
available in the country (Talay et al., 2015). In table 2, an overview of all variables, their abbreviations, definitions and sources is presented.
Table 2: Overview of variables
Variable Abbreviation Definition Source
Dependent variable
Brand performance Mshare The market share of a brand Passport Database
Independent variables
Consumer Ethnocentrism CET & NatWVS)
Average of standardized variables COLIND & NatWVS
Collectivism – Individualism COLIND Collectivism dimension score of host country
Hofstede
Nationalism question WVS NatWVS Weighted average of the answers to nationalism question of host country
World Values Survey Conspicuous Consumption CC
(PD & Masc)
Power distance and Masculinity dimension
Hofstede
Power distance PD Power distance dimension score of host country
Hofstede
Masculinity – Femininity Masc Masculinity dimension score of host country
Hofstede
Moderators
Product category embeddedness in host country
PCE Average per capita product category consumption 1960 – 2015
WHO Database
Economy Advancedness EAD Advanced economy (1) & Developing economy (0)
IMF
Control variables
Home country advantage HCA Whether brand is active in its own home country in an observation
Passport Database Firm size F_size Firm size classification Orbis Orbis Firm Age F_age Firm Age from date of
incorporation
Orbis Population Size Pop Population of the host country The World Bank GDP per capita GDP GDP per capita of the host
country
The World Bank Median Age Age Median age of the host country United Nations Education level Edu Education index score of the host
country
United nations Urbanization Urb Percentage of population that
lives in urban areas
The World Bank Market Commitment Mcom Percentage of brands active in a
host country of total brands owned by firm
Passport Database
Preparation data
As said, the total number of observations is 17.351. However, for some of the independent variables there is no data available. The data on the cultural dimensions by Hofstede is not available for Kazakhstan and Uzbekistan. Therefore, these countries have been deleted from the sample. Similarly, no data was available on the consumption of alcohol in Taiwan and Hong Kong, possibly due to differing sovereignty perceptions. However, as both Taiwan and Hong Kong are substantially different from mainland China on a number of aspects, the analysis would be flawed if the consumption data of China had also been used for Taiwan and Hong Kong. Therefore, these countries have also been excluded from the sample.
Next to that, some firms in the sample have started or seized to exist during the sample time period. Also, firms have merged or firms have been acquired by other firms in the sample. However, this has not been processed in the dataset. For instance, in 2008 two of the largest beer producers, Anheuser-Busch and Inbev merged into AB Inbev. Therefore, in the dataset the activities of three different companies has been documented for the period 2006 – 2015. The market shares of the brands that were not actually in possession of that company at that time were left blank. Therefore, by removing all observations with a blank market share, this problem has been solved.
Next to that, some of the brands in the dataset were named as “others”, which most likely is some kind of rest category representing the remaining number of brands in a country. However, as it is impossible to find the corresponding values for the other variables, these observations cannot be used and have been deleted as well. Furthermore, for several very small companies, it was not possible to find the data for the control variables due to incomplete names or language barriers. Therefore, these observations have also been deleted from the dataset. In the end, this leaves us with a total of 8.718 observations of 766 brands in 11 countries. Of these observations, 1598 are of global brands, 3256 are of regional brands and 3863 are of local brands, equaling 7119 brands that are treated as local.
Empirical Strategy
First, the data has been explored to find and remedy irregularities and prepare the data for analysis. Then, the analysis has been performed using the four models displayed in table 3. In model 1, all control variables and the two moderators predict brand performance and hypothesis 5a and 5b are tested. In model 2, both CET and CC are added. Both CET and CC are tested as a grouped variable in all models. In Model 3, the first interaction effect of PCE is added and in Model 4 the interaction effect of EAD has been added.
In order to test the hypotheses, a random effects panel regression analysis has been performed. The random effects method has been used because some of the variables of interest in this study are time invariant. That is, they do not change over time. The fixed effects model would in case of a panel regression omit time invariant variables. Because the random effects model assumes that α" and x"( are uncorrelated, this assumption has to be tested. This is done
by comparing the estimates from both a fixed effects and a random effects regression, the Hausman test (Hausman, 1978). It assesses the null hypothesis that the coefficient estimates of the random effects estimator are similar to the ones of the fixed effects estimator. In this case the result is that the null hypothesis can be strongly rejected ()* = 169.42, p = .000), meaning
that according to this test it cannot be assumed that α+ and x"( are uncorrelated. Although methodologically this indicates that a random effects model cannot be used, theory does indicate that a random effects panel regression should be used. As part of the data for the independent variables and moderators is time invariant, this would mean that these variables would be omitted from the analysis. Therefore, the random effects method is used and a fixed effects regression will be done in the additional analysis, where the results will be compared. To test for the hypothesized moderation effects, interaction effects have been computed. The interaction effect is the product of the standardized independent variable and the standardized moderator variable.
As can be seen in table 4, the variables firm age, population size and GDP per capita have rather large numbers, resulting in extremely low coefficients. Therefore, these variables have been transformed into their standardized versions. Next to that, firm size is a categorical variable. Therefore, 3 dummy variables have been created to be able to analyze them in these models. Also, to control for the unobserved year fixed effects that could occur, N-1 dummies for each year have been added to the regression.
As suggested by Wooldridge (2002), the errors of the panel regression can also be correlated within the clusters, serial correlation, leading to biased standard deviations (Wooldridge, 2002). In this research, it is likely that this is the case, as the clusters are the different brands. The market shares of observations in different countries or years of the same brand are probably not independent from each other, as they are owned by the same firm. Therefore, the Wooldridge test is done, in order to test for serial correlation. The results indicate that the hypothesis of no serial correlation in the composite error is strongly rejected (F (1, 59) = 1072.900, p = 0.000). This indicates that there is serial correlation in the composite error implying that generalized least squares will be a better estimation technique and an
autoregressive disturbance term should be added in the model. This is reinforced by the modified Durbin-Watson test statistic (.24) and the Baltagi-Wu locally best invariant test (.34). Also, the results of the random effects estimation indicate a first order serial correlation of , = .913. To allow for first order autoregressive correlation among the disturbance terms, the xtregar procedure in stata 15.0 is used to analyze the data.
Table 3: Panel regression models
1. /01230 & 51630 27389 :;7<17=386;>? = A +
C1 DEF>? + C2 F6181=H I9J386;98;KK>?+ C3 E18M710 J37N320;K>?+ O>? 2. /01230 & 51630 27389 :;7<17=386;>? = A + C1 DEF>?+
C2 F6181=H I9J386;98;KK>?+ C3EFP >?+ C4 EE >?+ C5 E18M710 J37N320;K>?+ O>? 3. /01230 & 51630 27389 :;7<17=386;>? = A +
C1 DEF>? + C2 F6181=H I9J386;98;KK>?+ C2EFP >?+ C3 EE >?+ C4 EFP ∗ DEF >?+ C5 EE ∗ DEF >? + C6 E18M710 J37N320;K>?+ O>?
4. /01230 & 51630 27389 :;7<17=386;>? = A +
C1 DEF>? + C2 F6181=H I9J386;98;KK>?+ C3EFP >?+ C4 EE >?+ C5 EFP ∗ F6181=H I96386;98;KK >?+ C6 EE ∗ F6181=H I96386;98;KK >?+
C7 E18M710 J37N320;K>?+ O>?
Where b = brand (x), t = firm (a) in country (b) in year (c) and O = A + V. µ = the disturbance term, A = individual specific components, V = remainder components
Results
In the following section the results are presented. First, a preliminary analysis has been conducted. Below the descriptive statistics are presented in table 4. From the descriptive statistics, it can be concluded that there is no missing data anymore. However, looking at the extremes, observations that have a value more than 3 standard deviations from the mean could be outliers, which could bias the results. For the global brands sample, 96,43% of the data falls within the range of 3 standard deviations from the mean. For the local brands sample, 97,75% accounts for all the data within 3 standard deviations of the mean. When looking at some of the outliers, a reason why the observations are outliers becomes clear. Some brands appear to be formerly state owned, resulting in the fact that the population of that country is very much accustomed to that brand, as there were no alternatives available. Other reasons include a brand having a leading role within a country’s history, possibly due to a colonial past. One of the assumptions of this study is a fair competition and therefore, these outliers could bias the
estimates in the regressions. However, the threshold of 3 standard deviations seems too conservative as that would exclude several seemingly valid observations. Therefore, the threshold will be increased to 6 standard deviations in order to only exclude true extremes. In the global brands sample no observations have a market share larger than 6 standard deviations of the mean but in the local brands sample 57 observations have been excluded.
Table 4: Descriptive statistics
Global Brands sample Local Brands sample
Var N Mean STD Min Max N Mean STD Min Max
Mshare 1.598 0,03 0,06 0,00 0,37 7.119 0,025 0,071 0,000 0,817 F_Age 1.598 68 54 0 198 7.119 48 45 0,000 320 EAD 1.598 0,60 0,49 0,00 1,00 7.119 0,608 0,488 0,000 1,000 HCA 1.598 0,05 0,21 0,00 1,00 7.119 0,578 0,494 0,000 1,000 PCE 1.598 0,94 0,90 0,00 5,11 7.119 1,01 0,873 0,000 5,11 Pop 1.598 210 M 387 M 4,4 M 1.3 B 7.119 451M 550 M 4,4M 1,3B GDP $ 1.598 19.320 19.466 779 56.336 7.119 17.739 18.300 779 56.336 Age 1.598 32 6,8 20 46 7.119 33 7,6 20 46 Urb 1.598 0,67 0,12 0,33 0,87 7.119 0,661 0,136 0,331 0,867 Mcom 1.598 0,67 0,26 0,28 1,00 7.119 0,634 0,248 0,279 1,000 F_Size 1.598 0,67 0,29 0,09 1,00 7.119 0,870 0,252 0,000 1,000 CET 1.598 0,00 0,47 -1,12 1,36 7.119 0,00 0,947 -0,90 1,14 CC 1.598 0,00 0,56 -0,93 0,92 7.119 0,00 0,587 -0,96 0,81
Next, a correlation matrix of both the sample of global brands and local brands is presented in table 5a and 5b. Some of the control variables correlated highly with the moderator variable EAD. As an extra check, the variables have been tested for multicollinearity, using the common cutoff value of 10 as a threshold for the variance inflation factors (VIF) (Hair, Black, Babin, & Anderson, 2010). The VIF values have been presented in the bottom row of table 5a and 5b. Two control variables, GDP and Urbanization have been excluded from this analysis due to multicollinearity. As can be seen in table 5, no multicollinearity exists anymore and all VIF’s are below 5.
In the local brands sample, some of these variables correlated even higher. Therefore, also based on the threshold of 10 for the variation inflation factors (VIF), the variables GDP, urbanization and education have been excluded from the analysis for local brand
performance. The correlation coefficients and VIF’s of the local brand sample have been presented in table 5b
Table 5a: Pearson correlation matrix & VIF values for global brands 1. M sha re 2. F _S iz eV L C 3. F _S iz eL C 4. F _S iz eM S C 5. F _A ge 6. P op 7. H C A 8. A ge 9. E du 10. M com 11. P CE 12. E A D 13. Z CE T 14. Z C C 1 1,000 2 0,057* 1,000 3 -0,028 -0,118* 1,000 4 -0,021 -0,089* -0,003 1,000 5 0,197* 0,430* -0,059* -0,034 1,000 6 -0,117* 0,017 0,183* 0,144* 0,051* 1,000 7 0,473* 0,056* -0,015 -0,011 0,079* -0,057* 1,000 8 -0,163* 0,103* -0,019 0,003 -0,119* -0,177* -0,044 1,000 9 -0,085* 0,053* -0,076* -0,053* -0,155* -0,446* -0,047 0,778* 1,000 10 0,147* -0,046 0,075* 0,057* 0,054* -0,198* 0,240* 0,077* 0,064* 1,000 11 0,044 0,226* -0,047 -0,030 0,066* -0,155* 0,343* 0,346* 0,314* 0,049* 1,000 12 0,115* -0,084* 0,055* 0,041 0,151* 0,345* 0,044 -0,820* -0,793* -0,131* -0,438* 1,000 13 0,165* -0,064* -0,015 -0,080* 0,033 0,092* 0,160* -0,448* -0,392* -0,119* -0,167* 0,428* 1,000 14 0,091* 0,041 0,039 0,043 0,048 0,179* 0,104* -0,129* 0,056* -0,060* -0,016 0,145* 0,274* 1,000 VIF - 1,34 1,07 1,06 1,31 2,00 1,35 4,33 4,81 1,21 1,58 5,00 1,47 1,00 *p < 0,05
Table 5b: Pearson correlation matrix & VIF values for local brands 1. M sha re 2. F _S iz e1 3. F _S iz e2 4. F _S iz e4 5. HC A 6. A ge 7. Mc om 8. F _A ge 9. P op 10 . P C E 11. EA D 12. ZC C 13. Z CE T 1 1,000 2 0,051* 1,000 3 -0,049* -0,460* 1,000 4 -0,025* -0,388* -0,080* 1,000 5 0,135* 0,006 0,156* 0,116* 1,000 6 -0,141* 0,037* 0,171* -0,108* 0,171* 1,000 7 0,029* -0,117* 0,159* 0,092* 0,493* 0,085* 1,000 8 0,088* 0,242* -0,01 -0,085* -0,035* -0,048* -0,074* 1,000 9 -0,092* 0,056* -0,01 -0,076* 0,167* -0,283* 0,152* -0,059* 1,000 10 -0,014 0,189* -0,090* -0,213* 0,169* 0,350* -0,069* -0,033* -0,141* 1,000 11 0,076* -0,046* -0,127* 0,102* -0,092* -0,829* -0,040* -0,057* 0,527* -0,282* 1,000 12 0,053* 0,020 -0,031* 0,066* 0,109* -0,419* 0,037* 0,131* -0,049* -0,067* 0,197* 1,000 13 -0,163* -0,095* 0,065* 0,168* 0,087* -0,016 0,01 -0,026* 0,297* 0,035* 0,153* 0,046* 1,000 VIF - 1.62 1.53 1.87 1.62 5.09 1.41 1,13 1.87 1.34 5.03 1,22 1.47 *p < 0,05
Results global brand performance
In table 6a, the results of the panel regressions for global brand performance have been presented. The coefficients (!) and p value of each variable in each model are displayed here. Also, the constant, the Wald "#, the overall R2 and the sample size (N) of each model is
displayed here. The Wald "# indicates whether all estimates are significantly different than
zero.
The first model includes the control variables, PCE and economy advancedness. Hypothesis 5a is tested in this model. The model was significant ("# = 1925,88, p = 0,000) with
an R2 of 0,201. Hypothesis 5a predicted PCE to have a negative influence on global brand
performance. The results do not provide enough evidence to support this hypothesis as there is a negative effect, but it is far from significant (! = -0,001, p = 0,655). Economy advancedness does have a significant, negative effect on global brand performance (! = -0,016, p = 0,002). Some of the control variables also have a significant effect. The years 2010 to 2015 have an increasing positive effect on global brand performance, as can be seen in table 6a. Firm age seems to have a negative significant effect (! = -0,009, p = 0,066), just like population size (! = -0,005, p = 0,003) Market commitment (! = 0,037, p = 0,000) and median country age (! = -0,002, p = 0,000). A large positive effect can be seen between home country advantage and global brand performance (! = 0,223, p = 0,000).
The second model tests the general relationship of CET and CC with global brand performance without taking the moderating effect of economy advancedness into account yet. The model is significant ("# = 1939,39, p = 0,000) with an overall R2 of 0,204. From the results
it can be concluded that CC generally has a significant, positive effect on global band performance (! = 0,007, p = 0,011). CET does not have a significant effect (! = 0,001, p = 0,681). The effects of the control variables on global brand performance are almost identical to the effects in model 1.
Model 3 tests the interaction effects of the first moderator, PCE. Here hypothesis 6a and 7a are tested. The model is significant ("# = 1938,48, p = 0,000) with an overall R2 of 0,203.
Both the interaction effects CET*PCE and CC*PCE are not significant (! = 0,001, p = 0,708; ! = -0,002, p = 0,411). Therefore, hypothesis 6a and 7a have to be rejected and no moderation is occurring between PCE and the CET- and CC - global brand performance relationships.
In model 4, the interaction effects of the second moderator have been added, economy advancedness. Here hypothesis 1a, 2a, 3a and 4a are tested. The model is significant ("# =
CC*EAD are significant (! = -0,197, p = 0,000; ! = 0,140, p = 0,000). Hypothesis 3a and 4a posed that CET would have a negative effect on global brand performance in advanced economies and a less negative effect in developing economies. The interaction effect has been plotted in figure 2a and the graph shows that for advanced economies, the influence of CET on global brand performance is actually strongly positive and for developing economies this relationship is slightly positive as well (! = -0,197, p = 0,000). Therefore, hypothesis 3a and 4a are rejected, as moderation is occurring but not in the hypothesized direction.
Hypothesis 1a and 2a posed that CC would have a positive effect on global brand performance in developing economies and a negative effect in advanced economies. In figure 2b, the interaction effect has been plotted. This graph shows that the relationship between CC and global brand performance is strongly negative for advanced economies and slightly positive for developing economies (! = 0,140, p = 0,000). Therefore, both hypothesis 1a and 2a can be accepted.
The effect of the control variables on global brand performance is different in model 4. None of the year dummy effects are significant anymore just like firm age. Population size, Market commitment and host country advantage still have similar effects as in the previous models.
