The Influence of Culture on Cosmetic Consumption

(1)

The Influence of Culture on Cosmetic Consumption

Master Thesis

June 2007

Student:

A.J.M. van der Lande Student no: 1177907 Supervisors: Dr. K.S. Muehlfeld Dr. G. Peli

Rijksuniversiteit Groningen

Faculty of Economics

(2)

3. METHODOLOGY AND DATA COLLECTION 26 3.1 Sampling ... 26 3.2 Data collection... 27 3.3 Variables... 28 3.3.1 Dependent variables ... 29 3.3.2 Independent variables... 30 3.3.3 Control variable ... 33 3.4 The Model ... 33 3.5 Data analysis ... 35 4. RESULTS 42 4.1. Results from the regression of the first model... 42

4.2 Results from the regression of the second model ... 46

4.3 Summary of findings ... 53

5. CONCLUDING REMARKS 55 5.1 Limitations of the research and suggestions for further research ... 55

5.2 Implications for research ... 56

5.3 Interpretation of results, implications for cosmetic companies ... 56

References... 60

(3)

Abstract

(4)

1. Introduction

East Asia is one of the biggest markets for luxury and prestige brands from the West. East Asians are particularly avaricious luxury- and brand-name consumers (Wong and Ahuvia 1998). For example Chinese consumers tend to latch on to well known brands far quicker than those in other countries (Hare, 2004). Several cultural factors lie behind the desire for luxury and prestige brands in Asia (Wong and Ahuvia, 1998). Those cultural factors have their implications for growth possibilities of luxury and prestige brands on Asian markets.

An exemplary case of the penchant for brand-name luxury goods by East Asians is the consumption of cosmetics in China. Cosmetic consumption in China has grown in the last 20 years from 25 million USD to more than 6 billion USD. Many expect this to increase to 14 billion USD in 2010 (Hare, 2004). Despite its fast growth, the Chinese cosmetic market is still at a very low level of development and is far from saturation. With potentially the largest customer base in the world and currently a low level of per capita spending of only $4 on cosmetics, China provides tremendous business opportunities for foreign cosmetic producers (Strategis, 2003). Due to the enormous development in Asian markets and the penchant for brand-name luxuries by Asian consumers, Western cosmetic companies are currently focusing on Asian markets in order to take advantage of the increase in sales of prestige brands and cosmetics. For example, L’Oreal’s acquisition of the Japanese company Shu Uemura Cosmetics and Chinese skin care brands Mininurse and Yue Sai provided an opportunity for the company to gain greater access to Asia (L’Oreal, Annual report 2004). But also Western brands are extended to Asian Markets.

The cosmetic industry is a very international industry. More and more cosmetic companies are enlarging their possibilities by entering new markets. Apart from Asia, Western cosmetic companies are moving to many other markets such as Latin America, The Middle East and Eastern Europe.

The combination of the penchant for branded luxury goods by Asian consumers, the growth possibilities of the cosmetic industry in upcoming markets, and the internationalisation of the cosmetic industry drove me to focus this research on the international cosmetic industry.

1.1 Problem statement

(5)

behaviour (De Mooij, 2003). Many authors (e.g. Levitt (1983), Jain (1989); Assael (1998)) expect this convergence to lead to homogeneous consumer needs, tastes and lifestyles. For example, Harvard Professor Ted Levitt (1983) wrote in his article “The Globalization of markets”, that new technology (such as TV and email) would lead to homogenization of consumer wants and needs, in other words, convergence of consumer behaviour in the world. But, as De Mooij (2004, p4) wrote, “In the foreseeable future economic development and technology are not leading to homogenization of consumption. Increased global mobility for business and vacations will not cause people to homogenize”.

Ideas on homogenization of consumption are generally based upon macro economic factors. Until recently, the impact of culture was not well understood (De Mooij, 2003). The Mooij’s ideas against convergence of consumer behaviour are based upon the idea that micro factors such as culture are not converging while those cultural aspects, such as value systems, strongly influence consumer tastes and needs. “There are wide differences among the value systems of consumers in different countries – value systems that are rooted strongly in history and appear to be resistant to change. These differences were predicted to disappear, but to date they have not. Although there is evidence of converging economic and demographic systems, there is no evidence of converging value systems” (De Mooij, 2003).

National culture is a fundamental factor that distinguishes consumers of one country from those of another country (Dwyer, Mesak and Hsu, 2005). Indeed, a country’s culture has long been identified as a factor that influences consumer behaviour (Roth 1995; Sommers and Kernan 1967). Research has shown that national culture has considerable influences on consumer behaviour (Dwyer, Mesak and Hsu (2005). And thus it is reasonable to suppose that cultural differences pose barriers to internationalisation. See Shaw and Clarke 1998, p163: “Cultural differences across countries remain a significant barrier to internationalisation…. the importance of culture as a potential barrier to market development is clear. To overcome the cultural barrier, retailers and manufacturers must develop a deeper understanding of local cultures and their influence upon purchasing decisions.”

(6)

An example that shows how cultural differences imply a barrier to internationalisation is the case of the Dutch cosmetic company Rituals. In Western Europe Rituals has been quite successful because of its products in skin care, body care and bath & shower. Indeed, in several European countries, it has been relatively easy to enter the market. However, while attempting to enter the Middle East and Russia, management discovered that an adapted approach was necessary due to different consumer habits and preferences. It turned out that, contrary to the Western European market, make-up (colour cosmetics) is much more important in Russia and the Middle East and that colour cosmetics is the category that brings in the highest turnover in the cosmetic shops in Russia and the Middle East. Apparently women in Russia and the Middle East highly value make-up, while to them, Bath & Shower products are of less importance.

As Rituals did not have a colour cosmetics product line, potential local investors in Russia and the Middle East hesitated to invest in Rituals. Raymon Cloosterman, CEO of Rituals says: “We found that it is important to account for cultural differences. These differences impact on cosmetic consumption and affect our ability to attract potential local investors.” As a result Rituals will now shortly launch a full colour cosmetics product line, which it hopes to offer on the Russian market as well.

As this example shows, despite forces of globalisation, there are still significant cultural differences that constitute barriers to international extension for cosmetic companies. Therefore, the focus of this research will be the implications of cultural differences between societies to cosmetic consumption.

1.2 Research objective

Literature describes cultural differences from different points of view (Hofstede 1980, 2001; Schwartz 1994; Hall 1976), but “little has been done to compare consumer behaviour across cultures” (De Mooij 2004, p18). Several researchers suggest that further research should be done on the influence of culture on consumer behaviour and product choice (Shaw and Clarke 1998). Therefore, in this research I want to address the topic of the influence of culture on consumer behaviour.

(7)

The Asian market is exemplary of the expansion of the cosmetics industry. Within this market, I have identified a particularly important cultural element driving consumer behaviour and contributing significantly to the opportunities for cosmetic brands. This element, i.e. a strong penchant for luxury products, explains why East Asia currently is the biggest market for prestige brands (Wong and Ahuvia, 1998). The first part of my research will therefore focus on this element, which I will further indicate as “status consumption”. I will explore the relationship between culture and status consumption.

Next I will address whether consumers of different cultures respond differently to the introduction of new products. If such cultural differences exist, this may pose additional barriers to the diffusion of innovations, or, conversely, smooth the progress of introducing new products. The cosmetic industry is particularly known for its high frequency of new introductions and for that reason the correlation between culture and product diffusion seems highly relevant for its internationalisation.

Summarizing, this thesis studies the influence of culture on consumer behaviour, specifically in the cosmetic industry, with a particular focus on two aspects:

1. Does a correlation exist between culture and status consumption? 2. Does culture have an influence on product diffusion?

(8)

2. Literature review and hypotheses development

In this chapter, I will first introduce the concept of culture and the approach of Hofstede to culture that will further be used in this research. Subsequently, I will elaborate on the concept of status consumption after which a literature review concerning the influence of culture on status consumption is given. Two hypothesises are developed on this topic. Next follow a discussion of the concept of product diffusion as well as literature review on product diffusion and the influence of culture on product diffusion. Two final hypothesises will conclude this part of the research.

2.1 Culture

As argued before, culture can be an important barrier to international market development. “To overcome this barrier, retailers and manufacturers must develop a deeper understanding of local cultures and their influence upon purchasing decisions” (Deidre, Shaw and Clarke, 1998, p163). Before going into differences in national culture, one should have a good understanding of the definition of national culture.

2.1.1 Culture, many definitions approaching a complex construct

Culture is a construct that is difficult to define. There is no universally accepted definition of culture (McAuley, 2001. p47). Definitions of culture speak of the complex whole of aspects such as knowledge, beliefs, arts, laws, morals, customs, and habits acquired by human members of society. Many different definitions exist. Kroeber and Kluckhohn (1952) identified over 160 definitions of culture. To give some examples, culture was defined as “a set of control mechanisms for the governing of behaviour” (Geertz, 1973), or as “the set of psychological characteristics manifested by a giva s system of ideas that provide a design for living. Hall and Hall (1990) view culture as a system for creating, sending, storing and processing information. Despite the diversity of approaches, there are many common threads, which the various definitions share. And so it is generalized as a shared pattern of being, thinking and behaving; something learned from childhood through socialization; something deeply rooted in tradition that permeates all aspects of any society (McAuley, 2001. p48).

(9)

models have been developed. One of these models is the model developed by Geert Hofstede (1980, 2001).

Hofstede’s model is one of the few dimensional models that provide country scores. These can be used as independent variables for the analysis of phenomena that vary across nations. As a result data on consumption and other aspects of consumer behaviour can be correlated with cultural variables. Since the dimensions are measured relative to other cultures, they are well positioned for comparative, cross-cultural research purposes (Dwyer, Mesak, and Hsu 2005). Hofstede’s dimensional framework of culture is widely accepted in marketing and other international disciplines (Nakata and Sivakumar 2001) and his five dimensions were empirically found and validated (Hofstede, 2001, Hoppe 1990, Nakata and Sivakumar 2001). Therefore, in this research, Hofstede’s model will be used to measure culture.

2.1.2 Hofstede’s dimensions of national culture

Geert Hofstede defines culture as “the collective programming of the mind which distinguishes the members of one group or category of people from another” (Hofstede 2001, p9). Hofstede developed a model of five dimensions of national culture that helps to understand basic value differences between cultures. His model distinguishes cultures according to five dimensions:

1. Power distance stands for “the extent to which the less powerful members of institutions and organisations within a country expect and accept that power is distributed unequally” (Hofstede, 2001, p98).

2. “Masculinity stands for a society in which social gender roles are clearly distinct: Men are supposed to be assertive, tough, and focused on material success; women are supposed to be more modest, tender and concerned with the quality of life. Femininity stands for a society in which social gender roles overlap: Both men and women are supposed to be modest, tender, and concerned with the quality of life” (Hofstede, 2001, p297).

3. “Individualism stands for a society in which the ties between individuals are loose: Everyone is expected to look after him/herself and his/her immediate family only. Collectivism stands for a society in which people from birth onwards are integrated into strong, cohesive in-groups, which throughout people’s lifetime continue to protect them in exchange for unquestioned loyalty” (Hofstede, 2001, p225).

4. Uncertainty avoidance stands for “the extent to which the members of a culture feel threatened by uncertain or unknown situations” (Hofstede, 2001, p161).

(10)

The model is based on an empirically based, landmark study, which was completed in the 1970s and involved IBM employees in more than 50 countries. The survey among IBM employees was conducted twice, around 1968 and 1972, producing a total of more than 116.000 questionnaires. Later on, additional data were collected from other populations unrelated to IBM but matched across countries (Hofstede 2001). Through theoretical reasoning and statistical analysis, dimensions were revealed on which country cultures differ. The study resulted in a categorization of national culture into the first four dimensions of culture, a framework that was later expanded to five dimensions (Hofstede 2001).

Hofstede’s model has been criticized however. For example, McSweeney (2002) criticised Hofstede’s model by evaluating his research methodology. According to McSweeney (2002) “Hofstede’s claims are excessive and unbalanced; excessive because they claim far more in terms of identifiable characteristics and consequences than is justified; unbalanced, because there is too great a desire to ‘prove’ his a priori convictions rather than evaluate the adequacy of his ‘findings’” (McSweeney, 2002, p112). McSweeney argues that Hofstede used a limited characterization of culture in his work; he questions its determination of culture within the territory of states. And he suggests that there are methodological flaws (for example, the number of respondents to the questionnaire in some countries would be too small) in his work.

Although Hofstede’s model has been criticized, as explained before, Hofstede’s dimension of national culture is widely accepted in marketing and other international business disciplines (Nakata and Sivakumar, 2001). The model can be used to position each country on the scale represented by each dimension. The dimensions are measured on a scale from 0 to 100.

The fifth dimension of Hofstede, i.e. long-term versus short-term orientation, was introduced some time after the first four dimensions. Consequently, only a limited number of country scores of this dimension are available. Therefore, in this research, I will focus on the first four dimensions of the Hofstede’s cultural framework.

In the next part I will further elaborate on Geert Hofstede’s dimensions of culture and their supposed relation to consumer behaviour. However, following De Mooij (2004) I will link the elements masculinity and power distance to status consumption while the elements individualism and uncertainty avoidance are linked to product diffusion.

2.2. National culture and status consumption

2.2.1 Status consumption

(11)

used to describe the lavish spending on goods and services that are acquired mainly for the purpose of displaying income or wealth rather than to satisfy a real need of the consumer (Van Kempen, 2005). Status consumption by an individual is purposed to obtain or demonstrate a certain social position relative to other individuals (Hyman, 1942). Veblen (1953) argued that many or most consumers purchase certain items to gain status benefits rather than or not only, to enjoy qualities intrinsic to the good.

Chao and Schor (1996) studied consumer behaviour differentiating between status motivated behaviour and consumer purchasing in which no status motives are present. They hypothesised that more “status” is purchased with the more socially visible products. Patterns of brand buying are one of the indicators of status consumption, and a pattern of brand buying favouring higher-priced brands is a characteristic of status consumption.

(12)

When stating this relationship, Chao and Schor imply that higher priced brand buying always involves status consumption. However, brands have several other functions apart from the “symbolic devise”, allowing consumers to project a specific self-image (Keller, 2003, p9). Other functions of brands include identification of source of the product, reducing risks, reducing search costs (the effort to search the right product), a signal of quality. Also a brand functions as a promise, or it functions as a pact with the maker of the product (Keller, 2004). Thus apart from the functions of a brand that can serve as status signals (the symbolic function and the signal of quality) there are several other reasons to buy a brand (Keller 2004).

A high price is not necessary to fulfil these other functions of a brand. Indeed, the cosmetic market also offers several well-known mass brands which are not high priced. Those well-known mass brands are able to fulfil each of the above mentioned brand functions, except the status function. As a consequence the other functions of brands can be eliminated as a possible explanation for consumers buying expensive cosmetic brands. Only, the search for status will probably better be fulfilled by expensive (prestige) brands than by mass brands.

2.2.2 Power distance

As mentioned before, power distance is defined as “the extent to which the less powerful members of institutions and organisations within a country expect and accept that power is distributed unequally” (Hofstede, 2001, p98). This cultural dimension measures how far social inequalities are accepted by a society (Hofstede, 2001).

In low power distance cultures, inequality is considered an undesirable condition (Hofstede, 2001; Dwyer, Mesak, and Hsu, 2005). Thus in these cultures, powerful people should attempt to appear less powerful (Hofstede 2001, De Mooij, 2004) and people attempt to minimize social inequality (Hofstede 2001). Indeed, status symbols are frowned upon in those cultures (Dwyer, Mesak, and Hsu, 2005).

(13)

2.2.3 Power distance and cosmetic consumption

While low power distance cultures attempt to minimize social inequality, powerful people in high power distance cultures tend to preserve, if not increase, their power (Hofstede 2001, p84). Thus, the powerful in high power distance cultures should be predisposed to acquiring new products to display their power and position. In those cultures, people are more concerned with other people’s perceptions of them and with the maintenance of their status. Consequently, status is a motive for consumption in high power distance cultures.

De Mooij (2004) supports the idea that status consumption is related to culture. According to De Mooij, “there are substantial differences in status needs across countries, and income differences have no explanatory power” (p.139). As de Mooij (2004) explains, “in high power distance cultures, positions and social status are not fluid, thus people want to demonstrate their position in society” (p141). Luxury products fulfil the desire for this type of status (De Mooij, 2004). De Mooij tested this assumed relationship between luxury product consumption and high power distance for Scotch Whiskey. Scotch Whiskey is a luxury product that fulfils the desire for status by demonstrating ones position in society (De Mooij, 2004). De Mooij found a strong correlation (r=0.74) between power distance and consumption of Scotch Whiskey.

Cosmetics are also a luxury product category, especially premium brand cosmetics. This suggests that for cosmetics a relationship may exist between cultural differences in terms of the degree of power distance, and status consumption. This is confirmed by Chao and Schor (1996). They found that patterns of brand buying of visible cosmetic products are related to the degree of power distance that exists in a culture.

De Mooij, cited before above, has extensively studied the influence of culture on consumer behaviour. In her book “consumer behaviour and culture” she puts forward different aspects of culture and consumer behaviour, supported with empirical findings in which she links cultural data with consumption data. In various paragraphs, she concentrates on cosmetic consumption. De Mooij (2004, p302.) confirms the importance of status brands for colour cosmetics in high power distance societies. She also explains that conformity of brands is important in high power distance cultures, i,e. users look for brands that are in keeping with their social status and position. This explains why status brands play an important role in high power distance societies. By contrast, as stated before, status consumption is frowned upon in low-power distance cultures. In those societies symbolic goods that mark economic differences should not be consumed excessivily. Thus expensive brand buying of visible products is not likely to be widely accepted in those societies.

(14)

H1: The cultural dimension of power distance is positively related to premium brand purchases of visible cosmetic products.

2.2.4 Masculinity

Masculinity focuses on the degree to which the society reinforces, or does not reinforce, the traditional masculine work role model of male achievement, control, and power. A high masculinity ranking indicates a country experiences a high degree of gender differentiation (Hofstede, 2001). In these cultures, males dominate a significant portion of the society and power structure, with females being controlled by male domination. A low masculinity ranking indicates a country has a low level of differentiation and discrimination between genders. In these cultures, females are treated equally to males in all aspects of the society (Hofstede, 2001).

Masculine societies have greater emphasis on competition, ambition, and career advancement (Van Everdingen and Waarts, 2003). Dominant values in masculine societies are achievement and success (De Mooij, 2004). Considerable importance is placed on individual accomplishment and the acquisition of material goods that signal such success (Dwyer, Mesak, and Hsu, 2005). Hofstede notes that in masculine countries, “money and things are important” (Hofstede, 2001, p299). Status is important to show success (De Mooij, 2004). Achievement is measured in terms of wealth and the recognition that it brings with it. The materialistic, possession orientated nature of masculine cultures suggests that the acquisition of goods and, in particular, new products are valued in these societies. In masculine cultures, people consume for show (De Mooij, 2004). In contrast, feminine societies stress modesty, equality, and concern for the living environment. They place a primary interest in people and environment, not in “things” (Dwyer et al 2005). The dominant values in a feminine society are caring for others and quality of life. Status is not so important to show success (De Mooij, 2004).

2.2.5 Masculinity and cosmetic consumption

(15)

De Mooij (2004) found empirical evidence for the relation between masculinity and ownership of luxury products. For 44 countries, she correlated the ownership of luxury products such as a camera and expensive suit or dress, and sales of jewellery with GNP per capita and with the score on the masculinity index. She used secondary data of Euromonitor and of EMS (European Media and Marketing Survey), which she correlated with Masculinity scores. For all products, cameras, expensive suits, dresses and jewellery, the correlation of masculinity and sales or ownership of the products exceeds the correlation of GNP/capita with sales or ownership of the products.

Many luxury articles serve as status needs (De Mooij, 2004). Also the wish to wear the latest fashion is a matter of status (De Mooij, 2004). Colour cosmetics and fashion are closely related. Status consumption occurs with socially visible products. A pattern of high-priced brand buying signifies status consumption. This leads to the assumption that, in masculine societies, where manifestations of one’s material success through materialism occur, it can be expected that people are more likely to buy expensive brands when buying visible cosmetic products. This position is reflected in the following hypothesis.

H2: The cultural dimension of masculinity is positively related to premium brand purchases of visible cosmetic products.

2.3. National culture and product diffusion

2.3.1 Diffusion of innovation processes

Product diffusion is the process by which a product innovation is communicated and accepted through certain channels among the members of a social system over time (Rogers, 1962). There are four essential elements in product diffusion. Firstly, the innovation, “an idea perceived as new by an individual” (Rogers, 1962, p19). Secondly, its “communication (of the new idea) from one individual to another” (p19). Thirdly, the social system in which the communication takes place. A social system is “a population of individuals who are functionally differentiated and engaged in collective problem solving behaviour” (p20). And fourthly, the time during which the diffusion process takes place (Rogers, 1962).

(16)

of the individual), internal influence affects individuals through the personal network (Delre et al, 2005), in other words the social system.

A social system is defined as a population of individuals who are functionally differentiated and engaged in collective problem-solving behaviour (Rogers, 1962). The members of a social system are individuals, although these individuals may represent informal groups, industrial firms, or schools.

Several diffusion studies (Gatignon, Eliashberg, and Robertson 1989; Helsen, Jedidi, and DeSarbo 1993; Takada and Jain 1991) indicate that consumers in different countries respond in different ways to the introduction of a new product (Tellefsen and Takada 1999).

A number of studies have empirically explored diffusion on a cross-national basis (Dwyer, Mesak and Hsu, 2004). Takada and Jain’s (1991) study was among the first to examine culture’s impact on the cross-national diffusion of products, exploring the influence of high-context versus low-context aspects of culture. In general their results support the hypothesis that the rate of diffusion is greater in countries that are characterized by a high-context culture (cultures in which very little is made explicit as part of a message, most information is part of the context (Hall, 1976)). Van Everdingen and Waarts (2003) also investigated the role of national culture to explain differences in adoption rates across countries. They used the national culture classifications of Hofstede (2001) and Hall (1976), and demonstrate to what extent these classifications help to explain variations in cross national innovation adoption rates. Their results indicate that variables describing national culture have a significant influence on the country adoption rates (Van Everdingen and Waarts, 2003).

Furthermore, Dwyer Mesak and Hsu (2005) examined the direct influence of national culture on the cross-national diffusion of innovations. They find support linking cultural dimensions to cross national product diffusion. Their study, by explicitly measuring culture, presents empirical evidence that suggests that dimensions of national culture directly influence product diffusion patterns on a cross national basis.

Thus national culture is suggested to influence product diffusion especially the dimensions collectivism and uncertainty avoidance. In the next paragraph, the two cultural dimensions and their possible relationship with product diffusion and the cosmetic marked will be elaborated.

2.3.2 Individualism and collectivism

(17)

In individualistic cultures, values are in the person, and people want to differentiate themselves from others. In individualistic cultures, people express their private opinion, and self-actualisation is important. Individual decisions are higher valued than group decisions. Individualists attach priority to variety and adventure, whereas collectivists prefer harmony (Hofstede, 2001, De Mooij, 2004).

In collectivistic cultures, identity is based in the social network to which one belongs. People are “we” conscious – their identity is based on the social system. Harmony with in-group members and avoiding loss of face is important (Hofstede, 2001, Wong and Ahuvia, 1998).

In the West, individualism is a dominant attitude among consumers (Melewar, Meadows, Zheng and Rickards 2004). In Asian and Latin American countries collectivism dominates. South European countries are less individualistic than those on north of Europe (De Mooij, 2004).

2.3.3 Individualism, collectivism and the cosmetic market

Also this Hoftsede dimension influences consumer behaviour. As Melewar, Meadows, Zheng and Rickards (2004) say, collectiveness is reflected in shopping patterns. To give an example, collective purchasing power of companies and institutions should not be underestimated. In China, a strong collective orientated country, most state-owned companies buy large quantities of certain products (e.g. shampoo and soap bars) and distribute them to employees as benefits (Melewar, Meadows, Zheng and Rickards 2004). In collective countries, this occurs since preferences among individuals are more or less the same. In individualistic countries this is not likely to occur, since preferences differ among individuals.

Presumably, shopping patterns will be influenced by collectivism on a much broader scope than the above-mentioned example (Melewar et al 2004). For example, while in the Netherlands, an individualistic country, individual members of one family may have their own shampoo according to their personal product preferences, in a collective country such as China; all members of the family will use the same shampoo according to their collective product preferences.

In this paragraph, literature is consulted to discuss the influence of collectivism versus individualism on consumer behaviour of cosmetic products in different cultures, specifically the influence on product diffusion.

(18)

the consumers and each link between the nodes representing the relation between consumers. The networks and the links between the nodes can differ according to the type of network that is simulated. The model studies how the penetration of a product in a population of consumers is affected by the structure of the network.

Social networks connect the consumers through links that are relations that consumers have among themselves. Networks can have different structures; sometimes the links between members of a network are very tight, thus influence is strong or sometimes there are many links among members of a network, thus information travels fast through the group, etc. According to Delre et al (2005) the structure of the network determines the speed and degree of diffusion. The social network structure is the structure of links between agents. The structure of the network depends on the number of links, the distribution of the links and clustering coefficients (Delre, Jager, and Janssen, 2005).

For example, having a network with large clustering coefficients, thus a high clustered network, means that social influence within the network is strong. Thus, if a fashion enters in a cluster, the social influence towards the non-adopters is very strong and it is very likely that the fashion involves the entire cluster (Delre et al, 2005). But contrarily, if all friends of a consumer belong to completely different groups (like in a random network, thus not a clustered network), the consumer will not feel a strong social pressure to adopt (Delre et al, 2005).

Cowin and Jonard (2003) examined the effect of network structures on the knowledge diffusion process. They consider the space of structures “that fall between, at one extreme, a network in which every agent is connected to n nearest neighbours, and at the other extreme a network with each agent being connected to, on average, randomly chosen agents” (Cowin and Jonard, 2003, p1558). The cliquishness (which in a social network can be defined as “the share of friends of one individual who are also friends of each other “(p1560)) and the average path length (the distance, in terms of number of links, between agents) of the network influence the diffusion of knowledge within a network (Cowin and Jonard, 2003).

A short path increases the diffusion power of the system; knowledge will move to different parts of the graph more quickly. Thus in small world networks (networks “where the proportion of links between an agent and other agents not in his neighbourhood is between 1 and 10 percent of all direct links between agent pairs” (Cowin and Jonard, 2003, p1572)), paths between all agents are shorter, since agents are not only linked to close neighbours but also to agents elsewhere in the network. Thus, there is more diffusion of knowledge in small world networks than in a regular network (in which only close neighbours are linked)

(19)

network, an agent will receive more knowledge from his neighbours. However, he will not receive knowledge from agents that are not close neighbours. Accordingly, results show that very highly cliquish networks have poor diffusion properties (Cowin and Jonard, 2003).

Culture plays an important role in the diffusion process through its influence on network structures. The social network structure is influenced by the culture in a society. Especially, the degree of collectivism and individualism determines the network structure of a social system; as Van Everdingen and Waards (2003) explain, in collective countries, individuals act conform the group. The importance of “social roles and social expectations” (Hofstede, 2001) in collective societies involve high social influence with strong links between members and highly clustered networks. According to Delre et al (2005), consequences of strong social influence in collective societies for adoption of innovations are twofold. In networks with strong social influence, the adoption of innovations by individuals depends on the number of adopters in his of her network. On one hand, each individual has his own personal threshold, only if the number of adopters in his/her network exceeds the personal threshold, the individual will decide to adopt (Delre et al, 2005). In collective cultures social influence is important, therefore personal thresholds are high. Thus, in collective societies with strong social influence, the adoption of innovation is impeded due to high personal thresholds. Consequently, it can be difficult and time consuming to reach the critical mass within highly clustered networks.

On the other hand, once the fraction of adopters in a network exceeds personal thresholds of agents to adopt; agents will feel strong social pressure to adopt (Delre et al, 2005). Thus, once a certain threshold is reached in collective societies; the innovation will involve the entire cluster due to the strong social pressure.

Van Everdingen and Waarts (2003) hypothesized a similar relation between individualism or collectivism and the diffusion of innovations. They found empirical evidence for this hypothesis. They studied adoption of innovation of 2647 medium sized firms from ten European countries, and linked this information to score’s in cultural dimensions of Hofstede (2001). Their data survey showed that individualism plays a positive role in early adoption of innovations. But results showed that individualism does not positively influence late adopters. Individualism apparently works in a positive way in getting the diffusion process started; at later stages of the diffusion curve this positive influence decreases which may be due to the fact that later adopters are more influenced by personal networks, make decisions in consultation with peers or friends (Van Everdingen and Waarts 2003) or have less need to differentiate themselves.

(20)

has reached the certain threshold in a collective society, strong social influence will stimulate the adoption of the innovation by late adopters of the network.

Consequently, it is likely that, once thresholds are exceeded, product adaptation involves the entire cluster in collective societies. The effects of network structure decrease as markets are more individualistic. In such markets, social influence towards non-adopters is of less importance, and therefore, fashion does not necessarily involve the entire cluster in individualistic societies. Therefore, one could believe that in different countries, due to cultural differences, the dynamic process of diffusion could create a continuous different market situation in terms of market shares (as for example in some networks (influences by culture) fashion always involves the entire network, thus increases market shares of popular brands, while in other networks this is not the case).

Furthermore, the population of individualistic countries is more heterogeneous as regards product preferences. In collective societies, product choices often reflect social norms, while in individualistic societies; product choices often reflect individual attitudes and tastes (Wong and Ahuvia 1998). Wong and Ahuvia (1998) suggest that (Western) individualistic consumers place more emphasis on the private meaning of their possessions and that (Asian) collective consumers place more emphasis on the public meaning of their possessions.

Private meanings are meanings that are idiosyncratic to an individual. Public meanings are meanings that are widely shared within a culture or group (Wong and Ahuvia 1998). Private meanings may comprise a wider variety of product characteristics than public meanings since private meanings can differ for each individual within a group. Public meanings are the same among a group of individuals. As Dwyer, Mesak and Hsu (2005) state: “opinions are largely established by the group in collectivist societies. When a product is accepted by the group, it is likely to gain rapid acceptance by its conformity-minded members. Conversely, in individualistic cultures, varying opinions are respected, if not expected” (p9).

While in individualistic societies, consumers may pursue many different personal values when buying their products, collective societies allow for only a few collective values that are pursued by consumers while buying products. This may imply that the range of motivations for product choices is wider in individualistic societies than in collective societies. Individualistic societies admit a more diverse range of product preferences while preferences are individually orientated and not limited to socially orientated and socially accepted product preferences.

(21)

texture, while individual B attaches more to a specific point of sale of a lipstick. Public product preferences tend to be more of the same among individuals.

Thus, consumers in collective society’s value public meanings of their possessions and individualistic consumers value private meanings. Since public meanings are shared within a group and private meanings may be idiosyncratic for each individual, product preferences are more uniform in collective societies and individualistic societies admit a wider variety of product preferences.

This finding may have implications for the cosmetic industry. On the market level, this could be reflected by the number of successful brands in a certain market. Introduction of a new brand will most likely be more difficult in collective societies than in individualistic societies. It is reasonable to assume that this has a bearing on the number of brands that are successfully introduced. However, once introduced, the brand quickly reaches the entire social network in collective societies, as opposed to the effect in individualistic societies. This implies stronger market domination by successful brands in collective societies.

Market domination is reinforced by uniform product preferences in collective societies, while the wider variety in product preferences in individualistic markets leaves more room for a larger number of successful brands. Individualistic societies generally leave more room for different brands. The more collective a society, the more the market will be dominated by a few brands.

H3: The cultural dimension of individualism is negatively related to the market share of the largest brands on a market.

2.3.4 Uncertainty avoidance

(22)

2.3.5 Uncertainty avoidance and cosmetic consumption

Literature offers several assumptions about the influence of uncertainty avoidance on consumer behaviour. De Mooij (2004) specifies uncertainty avoidance in relation to cosmetic consumption, but the relationship does not seem to be unambiguous. The fact that cosmetic products are artificial and not pure, leads to limited acceptance of cosmetic products in high uncertainty avoidance cultures (De Mooij, 2004). And on the other hand, in high uncertainty avoidance cultures there is a stronger need for purity, which stimulates the usage of certain cosmetic products (De Mooij, 2004).

Thus, as regards cosmetic consumption, uncertainty avoidance influences purity needs among consumers. High uncertainty avoidance increases the need for purity, which in the end increases the need for cosmetic products such as soap and bathing products. For example, in Japan, a society high on uncertainty avoidance, cleanliness has historically been an important need. “The Japanese have a horror of filth. Most people have at least one bath a day” (De Mooij, 2004, p142). Also, varying purity needs are reflected in the differences in mineral water consumption and in the varying volume of soap powder used. In Europe, in high uncertainty avoidance cultures people drink more mineral water and use more soap powder and household cleaning products. Cross-cultural variations in purity needs are explained by uncertainty avoidance (De Mooij, 2004). This relationship between uncertainty avoidance and needs for purity suggests that high

uncertainty avoidance positively influences cosmetic consumption.

However the need for purity is not the only characteristic of cultures that are high in uncertainty avoidance. Uncertainty avoidance influences the trust or distrust in cosmetic products. Cultures high in uncertainty avoidance distrust the artificial. This limits the acceptance of cosmetics (De Mooij, 2004). This would contradict the above-suggested relationship. Companies such as L’Oreal try to reach consumers in high uncertainty avoidance cultures with scientific arguments, thus reducing the distrust for artificial products. Scientific arguments seem to be effective in high uncertainty avoidance cultures (De Mooij, 2004). Vichy is an example of a brand that uses scientific arguments to take away the distrust of high uncertainty avoiding consumers for artificial products. So the distrust against artificial products may partly be taken away by scientific

arguments that reassure consumers in high uncertainty avoidance cultures. The relationship between cosmetic consumption and uncertainty avoidance is probably not unambiguous.

(23)

In other words, low uncertainty avoidance cultures more readily tolerate uncertainty and have a willingness to take unknown risks (Hofstede 2001), high uncertainty avoidance cultures are less likely to adopt innovations. This is reinforced by the fact that cultures with low uncertainty avoidance want variety in their lives (Dwyer, Mesak and Hsu, 2005) and consequently will be searching for more variety in their product and brand choices.

Thus, in low uncertainty avoidance cultures, it is easier to introduce a new brand or product since people are less averse of the risk of buying a new product. Furthermore, those consumers tend to look for variety, and this offers market possibilities for more brands. As for the cosmetic market, this may imply that it is easier to introduce a new brand in low uncertainty markets. Furthermore, since low uncertainty consumers search for variety, a larger variety of brands can be offered on the market. This suggests that a high degree of uncertainty avoidance is negatively associated with the number of successful cosmetic brands on a market.

H4: The cultural dimension of uncertainty avoidance is negatively related to the number of successful cosmetic brands on a market.

2.4 Control variable

As Cooper and Schindler (2003) say, “an almost infinite number of extraneous variables exist that might conceivably affect a given relationship….Most can be safely ignored… Others may be important”. Thus there may be extraneous variables to consider as confounding variables to the hypothesized relationships between independent and dependent variables.

Therefore, a control variable is introduced in this research. Since De Mooij states that the assumed strongest influence on consumer behaviour is income, a control variable is introduced to account for this influence.

National income can be measured in many different ways. A commonly used measure to compare national income is the Gross Domestic Product (GDP) per capita. “GDP includes all incomes earned within a country’s borders – by residents and non-residents” (Burda and Wyploz, 2001, p21).

Annual data that measure GDP per capita have been used as control variable in several other studies that are relevant to this research. Dwyer, Mesak, and Hsu (2005), and Kumar and Krishnan (2002) used this variable as a control variable in diffusion studies.

(24)

(25)

2.5 Conclusion

To summarize this chapter, the Hofstede framework (2001) is used to analyse how cultural differences are linked to consumer behaviour in terms of status consumption and product diffusion. Cultural differences cause different motivations for cosmetic consumption. This results in differences in consumption patterns, in terms of premium brand buying. Cultural differences also influence product diffusion processes, which may affect the ability of the largest brands to gain market share. These assumed relationships are presented in figure 2.1.

(26)

3. Methodology and data collection

3.1 Sampling

This research focuses on cosmetic markets in different cultures. The target population of this research consists of all cosmetic markets in different cultures in the world.

Although culture and country are not the same, I used country as a proxy for culture for practical purposes as largely has been done in prior research (Hofstede, 2001, Dwyer, Mesak and Hsu, 2005). Hofstede (2001) argues that cultural data should indeed be collected at the national level. “For example, most countries have a dominant language, identifiable education systems, and other integrating mechanisms that inculcate members with common values and outlooks” (Dwyer, Hsu and Mesak, p13, and Dawar and Parker 1994). Furthermore, Hofstede (1980) and Clark (1990) note that the country is a political entity and that politically created boundaries are usually stable over time and can be defined and precisely identified in space and time. This makes countries appropriate for defining culture.

Two main data sources were consulted for this research; market studies of Euromonitor and country scores on Hofstede dimensions published on the official website of Geert Hofstede (www.geert-hofstede.com). Since both sources do not include the entire world, the total population is not available for study. Data are available only from a limited number of countries. Hofstede provides country scores of 69 countries and Euromonitor provides market reports of 58 countries. Therefore a non-probability sample, namely a convenience sample, was taken from the population (Cooper and Schindler, 2003).

Data of Hofstede were matched with data of Euromonitor to develop a complete database with data of countries on culture and data on the cosmetic markets. When matching data of those two different sources, several things have to be taken into account. First of all, the Hofstede data are based on findings in 2001, while the findings of Euromonitor are based on figures of 2005. As there are no updated figure available on culture in 2005 and as cultures is assumed to be relatively stable over time, it was assumed that this would not cause considerable problems. Also, it has to be said that Hofstede’s research was based on employees of large international companies such as IBM, while Euromonitors figures are purely concentrated on the cosmetic industry. It is not certain whether findings on culture that were found in large international companies can be applied to the cosmetic industry without problems. However, it was assumed that this would not cause important differences.

(27)

it is a relatively small sample since the sample covers only about a quarter of all countries in the world. Also, a larger sample would have been better for the statistical analyses.

Secondly, the subjective selection of countries in the sample may create a bias in the sample. The sample includes 16 Western European countries, 7 Eastern European countries, 12 Asian Pacific countries, 5 Latin American countries, 2 countries the Middle East, and Australia, New Zeeland, Canada and the US (See appendix A for an overview of the countries included in this research).

The relatively large number of Western European countries may create a bias. For example, those countries all have relatively high scores on individualism. Consequently, there may be European consumption patterns, which might be caused by other factors than individualism such as European legislation or European taxes. Thus this research might find a correlation between a consumption pattern and individualism while individualism is not the cause of the consumption but European legislation being the cause of a specific consumption pattern in Europe.

Also, the lack of a representation of countries from Africa and the limited number of countries from the Middle East countries may be an important limitation since data on those countries would give insight to cultures that are under represented in the current sample. Since information on many countries of the world is missing, it is not possible to generalize the results of this research for the entire world.

3.2 Data collection

(28)

collected from Euromonitor’s markets reports. Most of the country data concern 2005, but a two country reports, those of South Korea and Portugal, date from 2004. For those two countries I checked whether data of preceding years are indicating that large differences can be expected between 2004 and 2005. For both countries, an estimation on premium brands sales in 2005 was made, based on data on previous years (1999-2004). The average annual change in premium brand sales over the preceding 5 years in both countries was taken to estimate premium brand sales in 2005. Concerning market shares of cosmetic brands in both countries, I assumed that market shares of 2004 give a good estimation for market shares in 2005, since, based on previous years, it can be expected that no important shifts of market shares would take place. Also for measuring culture, I consulted secondary data. Country index scores on Hofstede’s dimensions were used to measure and compare culture. Hofstede based the scores on quantitative research and gives scores for different countries. The model can be used to position each country on the scale represented by each dimension. The dimensions are measured on a scale from 0 to 100. The Hofstede index score’s are not updated for 2005, but given that “culture remains relatively stable over time” (Hofstede 2001), it seems that the country scores will be appropriate to measure culture of the countries that are being analysed in this research. The index scores that have been published by Hofstede in 1995, 2001 and on the official website of Hofstede are not showing many differences, although in the mean time additional research has been done. Therefore, I assume that data that have been published by Hofstede in 2001 give an adequate representation of the index scores in 2004.

Data on the control variable were gathered by consulting the database of the IMF. The world economic outlook database (September 2005) provides an overview of Gross Domestic Product (GDP) figures. As explained before, the control variable is the Gross Domestic Product based on purchasing power parity per capita. Since the data of Euromonitor were based on 2005, we will also take GDP data of 2005 of the countries of the sample of this research.

3.3 Variables

(29)

3.3.1 Dependent variables

Percentage premium brand sales of total sales in colour cosmetics

As explained in chapter 2, premium brand sales of visible cosmetic products are associated with status consumption. We concentrate on colour cosmetics being a category of visible cosmetic products. First of all, since Chao and Schor (1996) defined lipstick, a colour cosmetic product, as one of the most visible cosmetic products as it is often used in public. Secondly, since De Mooij also focused on colour cosmetics. She confirmed the importance of status brands for colour cosmetics in high power distance cultures. And finally, since accurate data are available for colour cosmetics.

Thus for the purpose of this research, I want to measure premium brand consumption of colour cosmetics. Cosmetic brands can be divided into two brand types: premium brands and mass brands. Premium brands are those brands that are distributed in selective distribution channels, and brands that apply a premium price and positioning strategy (Euromonitor (answer to email send to Euromonitor in this topic)). Mass brands are those bands that are distributed in mass distribution channels, and brands that apply a competitive mass market price and positioning strategy (Euromonitor). Since distribution channel, prices and label positioning are all taken into account when measuring premium and mass brand sales, the definition of premium versus mass is not likely to create problems or country biases, such as biases created by different price and positioning strategies of one single brand in different countries or biases created by the fact that a perfumery in one country exclusively sells premium brands, while in other countries perfumeries are distribution channels for both mass and premium products. By taking into account the distribution channel, as well as the price, as well as the positioning strategy of the brands, Euromonitor has taken into account possible disturbing differences among countries.

So for measuring premium brand consumption I take the percentage of total colour cosmetic sales that are premium brand sales. For each country I gathered the percentage of premium brand sales of colour cosmetics. Figures on these percentages are provided by Euromonitor. Euromonitor has developed market reports on the cosmetic and toiletry industry. I consulted 52 reports of the different countries of the research sample, and collected data on premium brand sales of colour cosmetics in the different countries.

(30)

Market share of the largest brands

In order to measure the degree to which the market is dominated by its largest brands, I will compare the market shares of the largest brands of different countries. I will do this by counting up the market shares of the largest cosmetic brands in the market. Than, I will test whether those market shares are significantly different between countries with different cultures.

This measure is very sensitive to the number of brands that are taken into account. For example, when calculating the market share of the three largest brands, outcomes may be completely different compared to taking into account the 10 largest brands. Ideally, this problem could be resolved by using a Lorenz curve and calculating the Gini coefficient (a measure of inequality of a distribution) instead of counting up market shares. A Gini coefficient of 0 indicating a completely equal distribution of market shares among all competitors in the market. The closer the Gini coefficient would be to 1, the larger the market share of a small number of brands with a relatively small market share of the other brands. Additionally, a comparison of the number of brands on the market would also be necessary, since a Gini coefficient of 0, which suggests equal distribution of market shares, would also be possible for a market with two brands, both having 50% of market share.

However, unfortunately, market shares of the smallest brands are not available. Therefore necessary data are missing for calculating Gini coefficients. Thus I will be restricted to the data that are available; data on the brands with the larger market shares. As explained above, this measure is very sensitive to the decision how many brands take part in the comparison of largest market shares. Therefore I will use different measures.

I will calculate the market share of the three largest brands in each country, and take this as a dependent variable. I will do the same for the market share of the largest 5, 10 and 30 brands to account for the notion that this way of measuring is very sensitive to the decision on the number of brands included.

Euromonitor provides information of market shares of cosmetic brands in different countries. The number of brands that are included in overview of brand shares varies between about 100 and 200 brands, which account mostly for about 75% or 85% of the market share. From each country I will take the largest 3, respectively 5, 10 and 30 brands and calculate their market shares. This way, it is possible to see whether in some countries, the power of the largest brands is larger than in other countries.

3.3.2 Independent variables

Hofstede scales are used to measure the independent variables, which are the four dimensions of Hofstede. Thus this research has four independent variables, namely

(31)

3. Individualism score

4. Uncertainty avoidance score

As explained in chapter 2, Hofstede (1980, 2001) conducted a research among employees of companies (such as IBM) in different countries. On the basis of this research he developed scales for culture, on which countries can be positioned. This way, cultures of different countries can be compared and measured. The scores are measured on a scale from 0 to 100.

As Hofstede (2001) indicates, his findings “can be used for quantitative applications such as using the culture dimension index scores of countries as potential explanatory variables for various measured phenomena” (p461).

Hofstede indicates several important things that have to be taken into account while using his dimensions. First of all, he says that his model does not represent a finished theory; all conclusions are falsifiable, and his model presented in his book “Cultures consequences” (Hofstede 2001) is meant as a step in an ongoing exploration. This is a limitation of using the model for this research, however the model has recently been used in other researches interesting for this study (Dwyer, Hsu and Mesak, 2005).

Secondly, the time dimension should be expanded. The cultural dimensions have been studied in 1980 and have been republished after an extra survey in 2001. There are no historical data on

Fig 4.1 Power distance and Masculinity scores

SLV UK SWZ FNL SWD AUT ISR DMK NZL IRL NRW GER AUL NTL CAN USA HUN ARG ITL JAP CZE ESPTWN GRE S.KO POR CHIL THA BEL COL POL TUR FRA HKG BRZ BUL MRC SIN IND INDO CHIN ARAB MEX VNZ ROM RUS PHI MAL 0 50 100 0 50 100 PDI M A S

(32)

culture that give information on culture over time (Hofstede 2001), and data are not updated in 2005. Furthermore, culture is measured for countries, thus information on subcultures in not reflected in the index scores.

Thus while the index score of Hofstede (1980, 2001) make culture measurable and while Hofstede suggest that the index score could be used as explanatory variables, as in this research, there are however some limitations to using Hofstedes model.

Figure 3.1 provides an overview of the countries of the sample and their scores on power distance and masculinity. Figure 3.2 provides an overview of the countries of the sample and their scores in individualism and uncertainty avoidance (See appendix A for an explanation of the abbreviations of the countries). In paragraph 3.5, I will shortly describe the sample of this research on the basis of the country scores.

Figure 3.2 Individualism and Uncertainty avoidance scores

(33)

3.3.3 Control variable GDP per capita

A control variable was introduced to account for the notion that “the assumed strongest influence on consumer behaviour is income. The amount of disposable income is expected to determine buying behaviour once the basic needs are fulfilled” (De Mooij, 2004, p65).

Because disposable income was not available for all of the countries in this sample, another measure was chosen to quantify income. As explained before, GDP per capita is a commonly used measure to compare national income. A country’s GDP is a measure of its factor incomes earned from economic activities within a geographic location during a period of time (Burda and Wyplosz, 2001). GDP measures the total wage and profit income earned by different groups in a country over a period of time producing GDP (Hall and Papel, 2005).

“The definition of GDP contains a fair amount of arbitrariness, and it is open to debate whether every positive movement in DGP constitutes in improvement in national wealth” (Burda and Wyplosz, 2001, p21). GDP concerns only recorded market transactions. This leaves out many activities that are not carried out through legal channels. It is not an exact measure of disposable income (the amount of national income available for households to spend (Hall and Papel, 2005)). But GDP is seen as an indication of disposable income.

As explained before, in this research GDP per capita at PPP is taken as the control variable. Purchasing power parity of GDP takes into account differences in the relative prices of goods and services—particularly non-tradable—and therefore provides a better overall measure of the real value of output produced by an economy compared to other economies (World bank). PPP provides a better measure of the standard of living of residents of an economy (World bank). PPP GDP is measured in current international dollars which, in principal, have the same purchasing power as a dollar spent on GDP in the U.S. economy (World Bank). This makes GDP at PPP per capita an appropriate tool to compare countries.

3.4 The Model

(34)

Hypothesis Independent variable

(IV) Dependent variable (DV) Expected relationship (IV on DV) variable Control H1 Power distance score

PDI % Premium brand sales of total sales in colour cosmetics

Pr

Positive GDP/Capita PPP

H2 Masculinity score

MAI % Premium brand sales of total sales in colour cosmetics

Pr

Table 3.4 a Assumptions of the first regression model

H3 Individualism score _IDV Market share of largest _{cosmetic brands*}

Ms

Negative GDP/Capita PPP

H4 Uncertainty avoidance _score

UAI

Market share of largest cosmetic brands*

Ms

Table 3.4 b Assumptions of the second regression model * As “largest brands” I will take respectively the three, five, ten and thirty largest brands.

In the first two hypothesises, it was hypothesised that premium brand sales in colour cosmetics are related to power distance and masculinity. Additionally GDP/capita at PPP would be included as a control variable.

Thus the first model is:

ε

β

+

=

C

₁

PDI

₂

MAI

₃

GDP

Pr

Where:

Pr = Premium brand sales PDI= Power distance score MAI= Masculinity score GDP= GDP/Capita at PPP

In the second two hypothesises it is hypothesised that the market share of the largest brands is related to individualism and uncertainty avoidance. Additionally, DGP/Capita at GDP will be taken as a control variable.

(35)

ε

β

+

=

C

IDI

UAI

GDP

Ms

₄ ₅ ₆ Where:

Ms* = Market share of largest brands IDI = Individualism score

UAI = Uncertainty avoidance score GDP = GDP/Capita at PPP

* For this model four separate tests will be done in which Ms represents respectively the three, five, ten and thirty largest brands in a country. This will be indicated by Ms(3), Ms(5), Ms(10), and Ms(30).

In both models it is assumed that the model’s error term is both normally distributed and has a zero mean. Some errors will be positive and some negative and will therefore average out to zero. This assumption implies that the average of all omitted variables and other errors made when estimating the model is zero. If this assumption holds, then the model is on average correct. Furthermore, the model must be tested for homoskedasticity. Each random error has a probability distribution with a variance measuring the uncertainty in the statistical model. The variance must be the same for each observation so that for no observations will the model uncertainty be mode, or less, nor is it directly related to any of the variables. Errors with this property are said to be homoskedastic (Carter Hill, Griffiths, and Judge, 2003). Therefore, a test for the presence of heteroskedasticity will be performed, which can be done with a scatter plot of the least squares residuals against the dependent variable. When no patterns of any kind can be detected, then the model exhibits homoskedasticity satisfying the assumption.

Additionally, the explanatory variables should not be a linear function of each other. If two or more independent variables are highly correlated, this can have damaging effects on multiple regressions. In case correlations at a 0.80 or a greater level one of the two variables should be deleted (Cooper and Shindler, 2003). Thus the variables will be tested for collinearity.

The Influence of Culture on Cosmetic Consumption