The effect of country-specific factors on the
adoption rate of infrequently purchased
consumer durables in the saturation phase
Master’s Thesis
June, 2013
University of Groningen
Faculty of Economics and Business MSc BA Strategy & Innovation First Supervisor: Dr. Florian Noseleit Second Supervisor: Dr. Charlie Carroll Bas Raats
Abstract
This study expands the literature in the area of cross-country innovation diffusion by providing insight into how economic, cultural and demographic factors affect innovation adoption. Research has shown that country-specific factors explain difference in adoption rates of consumer durables. The aim of this research is to determine the effect of these factors in saturated markets. The adoption rates of households of microwave ovens, personal computers, refrigerators, televisions and washing machines for 25 countries are collected in order to perform statistical analyses. My study provides evidence that the country-specific factors prove to be useful in explaining difference in innovation adoption rates between countries in the saturation phase. GDP, tourist arrivals, population size and urbanization have a positive effect on the adoption rate, while individualism has a negative effect. Managers can incorporate this new knowledge in segmenting strategies for saturated markets.
Acknowledgement
This contribution is my final effort to graduate as a Business Administration student at the University of Groningen. Hereby, I would like to take this opportunity to express my appreciation to the many people who contributed to the realization of this study.
My appreciation is for Florian Noseleit, one of the most flexible and motivating supervisors. Thank you for your great support and guidance, which enabled me to learn and develop myself as much as possible.
Furthermore, I would like to express my gratefulness for the unconditional love of my parents, Harry and Esther, and my sisters, Floor and Merel. Thank you for your support and encouragement to always get out the best of me.
Also, a special thanks goes out to you, who has been supportive and understanding to me throughout the entire process last year.
Looking back at a great journey I can tell time has flown, since I moved from Enschede to Groningen. Here, I had the chance to meet new friends, experience extracurricular activities and, above all, enjoy Groningen.
Lastly, I want to make a statement. Recently, I lost someone important to me. Prior to this, I equalized successfulness to happiness. Not anymore. I want to encourage everyone to achieve goals you are intrinsically motivated for and to be successful according to your own norms and values, not those set by others around us.
Table of Content
Page
1.
Introduction
1
1.1 Initial Motive 1
1.2 Problem Statement and Research Objective 1
1.3 Structure of the Paper 2
2.
Literature Review
3
2.1 Innovation Diffusion 3
2.2 Cross-Country Innovation Diffusion 6
2.3 Hypotheses Building 8 2.3.1 Economic Factors 8 2.3.2 Cultural Factors 9 2.3.3 Demographic Factors 10
3. Research Design
13
3.1 Data 13 3.1.1 Collection 13 3.1.2 Description 13 3.2 Variables 15 3.2.1 Dependent Variable 15 3.2.2 Independent Variables 15 3.2.3 Control Variables 163.3 Procedure of Data Analysis 16
3.4 Models 17
3.4.1 Multivariate log-log Regression 17
3.4.2 Fixed Effects Regression 18
4.
Results
19
4.1 Descriptive Statistics 19
4.2 Correlation Analyses 19
4.3 Univariate log-log Regression 20
4.4 Multivariate log-log Regression (Model 1) 21
4.5 Variance Inflation Factor 22
4.6 Multivariate log-log Regression (Model 2) 22
4.7 Standardized Coefficient Ranking 23
4.8 Robust Multivariate log-log Regression 23
4.9 Fixed Effects Regression 25
5. Discussion
29
5.1 Findings 29
5.2 Conclusion 30
5.3 Theoretical and Practical Relevance 30
5.4 Limitation and Suggestions for Further Research 31
References
33
1. Introduction
1.1 Initial Motive
Globalization urges organizations to introduce products across countries. Globalizing markets increase the speed at which innovations diffuse, mature, and decline (Chandrasekaran & Tellis, 2008). For organizations to compete successfully, they need to understand the diffusion patterns cross countries and how they can adjust their strategy to increase the adoption rates of their innovation.
Several studies examined diffusion patterns cross countries and tried to explain the difference in adoption rates (Gatignon et al., 1989; Helsen et al., 1993; Takada & Jain, 1991; Kumar et al. 1998). For many innovations launched in multiple countries one can observe high adoption rates in some countries, but low in others (Waarts & van Everdingen, 2005).
Previous research in the area of cross-country innovation diffusion has extensively described country-specific factors that explain difference in adoption rates (Talukdar et al., 2002; Tellis, Stremersch & Yin, 2003). While the take off phase has been widely studied, only few studies have looked into the saturation phase of innovation diffusion. Research on saturated markets is highly relevant for organizations, since this is the natural habitat for many product categories.
One of the most influential contributions in marketing is the Bass diffusion Model (1969), which describes the process of how innovations get adopted as an interaction between users and potential users. Prior to this, Rogers (1962) published a highly influential paper that illustrates the different stages of innovation adoption. Bass (1969) mathematically formalized the innovation concept. Innovation diffusion is an innovation that is communicated through channels over time among the members of a social system (Rogers, 1983). A communication channel is the means by which messages get from one individual to another. Furthermore, a social system is a set of interrelated units that are engaged in joint problem solving to accomplish a common goal.
Country-specific factors that affect innovation diffusion can be economic, like GDP (Sundqvist et al., 2005; Sood et al., 2009; Steenkamp and Burgess, 2002; Arts, Frambach & Bijmolt, 2011; Chinn & Fairliey, 2006; Tellis et al., 2003; Talukdar et al., 2002), cultural, like uncertainty avoidance (Yaveroglu & Donthu, 2002; Waarts & van Everdingen, 2005; Yeniyurt & Townsend, 2003; Van Everdingen, Fok & Stremersch, 2008) and demographic, like urbanization (Sood et al., 2009; Dekimpe et al., 1998; Talukdar et al., 2002; Alesina & Wacziarg, 1998; Alesina, Spolaore & Wacziarg, 1997).
1.2 Problem Statement and Research Objective
the saturation phase. Country-specific factors are linked to the adoption rates of infrequently purchased consumer durables in the saturation phase in order to explain difference between countries. Furthermore, a more comprehensive model is needed in the area of cross-national innovation diffusion to include the effect of previously discussed determinants like economic, cultural, and demographic factors (Yaveroglu & Donthu, 2002).
The findings of this contribution clarify whether country-specific factors explain difference in adoption rates between countries in the saturation phase and which country-specific factors have an effect on the adoption rate. Finally, managers can incorporate this new knowledge in segmenting strategies for saturated markets.
Cross-country panel data on household’s adoption rates of microwave ovens, personal computers, refrigerators, televisions and washing machines for 25 countries between 1991 and 2011 is collected from Datamonitor (2012).The data is commonly used to expand knowledge on cross-country innovation diffusion (Gatignon et al., 1989, Takada & Jain, 1991 and Helsen et al., 1993). In order to achieve the research objective a multivariate log-log regression analysis and a fixed effects regression analysis are performed to determine the relation between country-specific factors and innovation adoption in the saturation phase.
According to Chandrasekaran & Tellis (2008) many cross-country diffusion contributions exclude some of the largest economies like Japan, China and India, but also fast growing economies as Korea and Brazil. These countries are included in this research. Such an approach is needed since a limited focus on developed countries is seen as symptomatic of much of the prior research on diffusion (Sarvary et al., 2000, Hauser et al., 2006).
1.3 Structure of the Paper
2. Literature Review
This chapter consists of the literature background. The study explains how country-specific factors affect innovation adoption. The characteristics of the population of one country differ from another. As a result, adoption rates and diffusion speed differ per country. Section 2.1 explains innovation diffusion. Section 2.2 describes two approaches for explaining difference in cross-country innovation adoption. The section elaborates on country-specific factors that explain difference in innovation adoption rates. This section is complemented by segmenting strategies. Section 2.3 discusses the hypotheses.
2.1 Innovation Diffusion
Innovation diffusion and innovation adoption are closely related to each other and are both central to this study. Innovation adoption is actually driven by innovation diffusion. The definition of innovation diffusion, according to Rogers (1962), is:
an innovation that is communicated through channels over time among the members of a social system.
In more depth, an innovation is an idea or object perceived as new by a potential adopter. A communication channel is the means by which messages get from one individual to another potential adopter. The time is the process through which an individual or other decision-making unit passes from first knowledge of an innovation to forming an attitude toward the innovation, to a decision to adopt or reject, to implementation of the new idea, and to confirmation of this decision. A social system is a set of interrelated units that are engaged in joint problem solving to accomplish a common goal. Every social system has a certain structure, that is defined as the patterned arrangements of the individuals or potential adopters in a system. This gives some form of stability to the behavior and expectations of individuals or potential adopters in the system (Rogers, 1962).
The social and communication structure of a system facilitates the diffusion of innovations in a particular system. One important aspect of social structure is the norms, which are the established behavior patterns for the individuals or potential adopters within a social system. As mentioned previously, members of a social system communicate with each other. There are two types of communication: homophily and heterophily. Firstly, Homophily is the degree to which individuals who communicate are similar. Secondly, Heterophily is the degree to which individuals who interact are different from each other. The diffusion networks are most of the times homophilous (Rogers, 1962).
mainly applied to forecast the sales of products and technologies. Bass (1969) assumes that there are two groups in the diffusion process, namely the innovators and the imitators. This is shown in Figure 1. Additionally, there are two transfer mechanisms by which an innovation diffuses. Firstly, there is mass media, like advertising, which is considered an external influence that only affects the innovators. Secondly, there is word of mouth, which is an internal influence of the social system that only affects imitators. This means that the imitators are influenced by other members of the system through social interaction. In general, the number of innovators is assumed to be a small compared to the group of imitators. According to Bass (1969), the people who adopt an innovation in the saturation phase are imitators. The innovation diffusion among imitators arises mostly by word of mouth, but also by advertising. This means that social interaction within a social system plays an important role in innovation diffusion and stimulates innovation adoption in the saturation phase
The spread of an innovation is not instantaneously. Instead, the rule seems to be that the diffusion starts with an exponential increase, followed by a slowdown. This is a typical S-shaped pattern. (Grübler, 1997). Figure 2 shows the S-shaped pattern, which is a cumulative translation of the Bass Diffusion Model. The slowdown phase of the S-shaped diffusion curve, as indicated in the figure, is the saturation phase. Every stage in the diffusion consists of adopters with different characteristics.
Figure 1: Bass Diffusion Model Figure 2: S-shaped diffusion pattern
Rogers (1983) complements the Bass Diffusion model by proposing a classification for the adopter categories. Adopter categories are the classifications of the members of a social system on the basis of innovativeness; the degree to which an individual or adopter is relatively early in adopting innovations as compared to other members of a system.
(1962), adopters within the saturation phase can be characterized as the late majority and laggards. The people who adopt the innovation relatively earlier in a country do not differ with the later adopters in age, however the earlier adopters are expected to be literature, have a higher social status and have more years of proper education.
A study of Levitt (1965) explores the diffusion process from a product’s perspective: the product life cycle. Figure 4 represents the product life cycle, which denotes similarities with the Bass Diffusion model. Golder & Tellis (2004) describe the different stages in the product life cycle. First, the introduction stage, which is the period from commercialization until the takeoff of an innovation. Second, the growth stage, which is the takeoff phase until it slowdowns in sales. Third, the maturity stage, which is the phase from the innovation’s slowdown until sales begin a steady decline. Fourth, the decline stage, which is the period of decreasing sales until the innovation demises.
Three of these events are defined again by Golder & Tellis (2004), who mark the beginning and end of the first two stages: firstly, Commercialization is the moment at which an innovation is sold to consumers for the first time. Secondly, Takeoff is transition from the introduction to the growth stage of the product life cycle. It is the first significant increase in the sales volume of an innovation. In this phase 46% of the new products fail (Catellion & Markham, 2013). Thirdly, slowdown is the beginning of a period in which the sales volume of an innovation slowly increases or even temporarily decreases. Golder & Tellis (2004) define the slowdown phase as the saturation phase in innovation diffusion.
Chandrasekaran & Tellis (2007) argue that based on several empirical studies, on average, the duration of the introduction stage is six to ten years; the growth stage is eight to ten years; and the early maturity stage is five years for consumer durables. Consumer durables, for example refrigerators and televisions, are commercialized respectively in 1918 and 1962 and reached the slowdown phase in 1938 and 1969, are established in saturated markets for years (Golder & Tellis, 2004). Thus, for many consumer durables the saturation phase is their natural habitat.
Despite of the fact, that extensively research has been done on the adopters and different diffusion stages, little knowledge is gained about the saturation phase. Saturated markets can be characterized as highly competitive and low priced (Barrena & Sánchez, 2009). Aumann (1966) calls this the continuum-of-traders. The concept of competitive equilibrium is generally agreed to be significant only in a market with ‘perfect competition’. The saturated market can be measured in terms of number of organizations active on a market with a particular product (Stassen & Grünhagen, 2011). A different approach to measure a saturated market from an innovation diffusion perspective is the extent to which the innovation reaches a hundred percent adoption rate (Golder & Tellis, 2004).
Many organizations are aware of the problems of saturated markets. Saturated markets affect the effectiveness of marketing efforts (Liu & Yang, 2009). Liu & Yang (2009) found that the impact of an individual loyalty program decreases as the marketplace becomes more saturated. The study of Liu & Yang (2009) discuss a market environment that is changing. The markets are characterized by highly competitiveness, more demanding consumers and developments towards relation marketing. As a results organizations adopt three growth strategies. First, products are intentionally designed so that they need a replacement at some point.
Second, a limitation that derives from saturated markets is limited growth opportunities in the home market. As a result organizations strive for globalization for additional growth (Williams, 1992). The other motives for organization to globalize are an internationally appealing and innovative concept, but also a surplus of resources. Williams (1992) studied the motives for internationalization from a retailer’s perspective. Retailers expand globally because the saturated markets are characterized by increased competition and unsuitable diversification possibilities. Thus, saturation and globalization are closely related.
2.2 Cross-Country Innovation Diffusion
Innovation diffusion in an international context has been studied before (Yaverogly & Donthu, 2002). The contributions that elaborate on cross-country innovation diffusion discuss two different approaches for explaining difference in adoption rates of consumer durables.
The first approach explains the diffusion patterns across countries by a time lag. Takada and Jain (1991) observe that when an innovation is introduced first in one country, the lead country, and with a time lag in subsequent countries, the lag countries, the diffusion rate in the lag countries is faster than the rate in the lead country and a higher adoption rates is accomplished.
with mass media from other countries and, moreover, they can benefit from economies of scale. The findings of Helsen et al. (1993) are contradictory to previous arguments. They illustrate a negative relation between innovation diffusion and lag time. They found that there are certain constructs that may relate to the patterns of innovation diffusion. However, it is not always clear when such constructs have an impact on the innovation diffusion and in which direction, especially among different product categories.
The second approach explains the different diffusion patterns across countries by certain country-specific factors. This approach is adopted in this paper because it is the foundation for developing segmenting strategies, which is an important tool for organizations in saturated markets. This approach illustrates that the adoption rate within a country is affected by a set of country-specific factors that determine the characteristics of a country. For example, Gatignon et al. (1989) argue that an essential concept of diffusion theory, relevant to explaining the difference of cross-country innovation diffusion, is the source of information about the innovation. The diffusion pattern tends to be exponential shaped, when the most influential source of information is external to the social system. Examples of external influences are change agents and previous adopters in different social systems.
Communication between groups of adopters and potential adopters, in different geographical locations within the social system, is a necessary condition for innovation to spread. Therefore, Gatignon et al. (1989) examined the influence of cosmopolitan on the different diffusion patterns across countries. They found that this variable positively influences innovation adoption among innovators and imitators.
Furthermore, Takada and Jain (1991) discussed the effect of a communication system within a country on the diffusion pattern. They found that the adoption rate of consumer durables in countries that are characterized by high context cultures and homophilous communication is higher than in countries that are characterized by low context culture and heterophilous. Van Everdingen, Fok & Stremersch (2008) proposed a conceptual model where they include economic, cultural and demographic variables as important determinants to characterize a country. In addition, Stremersch & Tellis (2004) supported this approach by adopting a model where they include economic and cultural variables. Therefore, this approach is adopted in this paper.
communicate and influence potential adopters in the lag countries. This provides an additional external source of information to the lag countries, which ultimately has a positive effect on the adoption rate and diffusion rate among the potential adopters of an innovation..
The current global competition requires an organization to follow the sprinkler strategy, when introducing an innovation to its global markets. In the past it was possible that, based on game theory frameworks, market condition are derived under which the waterfall strategy is optimal in a competitive game between two competitors. However, these market conditions, where two organizations compete against each other, may not exist anymore for consumer durable markets (Kalish, Mahajan & Muller, 1995).
2.3 Hypotheses Building
This section elaborates on the country-specific factors that are expected or proven to affect adoption rates. The factors are categorized in economic, cultural and demographic variables, as proposed by Van Everdingen, Fok & Stremersch (2008). The variables are: GDP, import, export, tourist arrivals, uncertainty avoidance, individualism, power distance, population size and urbanization.
2.3.1. Economic Factors
The wealth level of a country positively influences the adoption rate of an innovation and the late, uncertainty avoiding adopters have a greater coefficient of imitiation (Sundqvist et al., 2005). So, wealth has an effect in all stages of the diffusion process. Consumer innovativeness has a positive effect on innovation adoption, where country difference determine the consumer innovativeness (Sood et al., 2009). Income is identified as one of the determinants which differ per country and a high income increases consumer innovativeness (Sood et al., 2009). This theory is supported by Steenkamp and Burgess (2002), who denote that income has a positive effect on the consumer innovativenesss, which ultimately has a positive effect on innovation adoption.
Arts, Frambach & Bijmolt (2011) show that the level of income of adopters positively influences innovation adoption. Chinn & Fairliey (2006) observed a positive effect of GDP on computer and internet adoption rates. This argument is supported by Tellis et al. (2003) who prove that a country’s economy is directly related to the affordability of an innovation and time to takeoff. Thus, hypothesize is that:
H1. The higher the GDP in a country, the higher the adoption rate in the saturation phase
determinant for innovation and innovation diffusion rates. Populations in open economies will be more enabled to share information with foreigners, because they have developed more relationship heuristics (Wuyts et al., 2004).
According to Van Everdingen, Fok & Stremersch (2008) an economy can be open in terms of international trade, like imports or exports of goods and services, or in terms of international traffic of people, like tourist arrivals. A positive relation between tourist arrivals and innovation diffusion has been found, but no statistical evidence has been found for import or export (Van Everdingen, Fok & Stremersch, 2008). From this the following hypotheses are derived:
H2. The higher the percentage of import in a country, the higher the adoption rate in the saturation phase H3. The higher the percentage of export in a country, the higher the adoption rate in the saturation phase H4. The higher the percentage of tourist arrivals in a country, the higher the adoption rate in the saturation
phase
2.3.2. Cultural Factors
A country’s culture is related to the degree to which consumers are innovative and on the other hand socially connected (Van den Bulte & Stremersch, 2004). Takada & Jain (1991) discovered in early cross-country diffusion research that the adoption rate in countries characterized by high context culture and homophilous communication is higher than in low context culture.
The culture dimensions of Hofstede (1980) are applied in studies regarding consumer innovativeness (Steenkamp et al., 1999), business to business adoption and diffusion (Everdingen & Waarts, 2005), diffusion patterns (Yeniyurt & Townsend, 2003) and innovations (Kaasa & Vadi, 2010). Hofstede’s cultural dimension can be found in Table A1 in the appendix.
The five dimensions are frequently adopted as indicators when testing for cultural difference between countries. The five dimensions are uncertainty avoidance, individualism, power distance, masculinity and long term orientation. Three of them, uncertainty avoidance, individualism and power distance are intensively emphasized and proven in empirical studies to influence innovation adoption.
al., 1999). Yaverogly & Donthu (2002) also observed a negative link between uncertainty avoidance and innovation diffusion. Hence, hypothesized is that:
H5. The lower the uncertainty avoidance in a country, the higher the adoption rate in the saturation phase
The second dimension of Hofstede is individualism. The high side of this dimension, called individualism, that can be defined as a country or society in which individuals are expected to take care of themselves and their immediate families only. Its opposite, collectivism, represents a preference for a country or society in which individuals can expect their relatives or members of a particular social environment to look after them in exchange for loyalty.
This dimension reflects whether people in a society believe and act in terms of “I” or “we” (Hofstede, 1980).
Yaverogly & Donthu (2002) observed that imitation is high in countries that are low on individualism. In addition, a positive impact of individualism on the innovativeness of consumers has been observed (Steenkamp et al., 1999). These findings are supported by Yeniyurt & Townsend (2003), who observed a positive effect of individualism on the adoption rate of innovations. Hence, hypothesized is that:
H6. The higher the individualism in a country, the higher the adoption rate in the saturation phase
Thirdly, the power distance is discussed. This dimension expresses the degree to which the less powerful members of a society accept and expect that power is distributed unequally. The fundamental issue is what the exact opinion of a country or society is about inequalities among people. People in countries or societies that respect a large degree of power distance also accept a hierarchical order in which everybody has a place. Furthermore, for this they need no further justification for the unequal distribution of power. In countries where there is low power distance, people try to equalize the distribution of power and, additionally, demand justification for inequalities of power. (Hofstede, 1980).The coefficient of innovation is high for countries that are low on power distance (Yaverogly & Donthu 2002).
H7. The lower the power distance in a country, the higher the adoption rate in the saturation phase
2.3.3. Demographic Factors
Assumed is that small countries are typically less self-centered than large countries (Alesina and Wacziarg 1998), which may make them more receptive for foreign influence.
In contrast, the larger countries are expected to have a more diverse population than the smaller countries (Alesina, Wacziarg & Spolaore, 1997), which may lead to contacts with foreigners, this can have a positive effect on the adoption rates. Van Everdingen, Fok & Stremersch (2008) already described that an open economy is important for innovation diffusion within a country and is expected to have more foreign contacts, which positively influence innovation diffusion. Therefore, the following hypothesis is defined:
H8. The higher the population size in a country, the higher the adoption rate in the saturation phase
Rogers & Shoemaker (1971) argue that innovativeness correlates with urbanization. Thus, rural areas innovate less than urban areas. Moreover, Talukdar et al. (2002) determined that urbanization positively influences the adoption rate. They argued that access to a product is essential for innovation adoption speed.
Research in economics has shown that urban areas can benefit more from the production and distribution efficiencies, this because of better infrastructure and economies of scale (Calem & Carlino, 1991). Rural areas are difficult to reach due to poor infrastructure, self-centered communities, low interconnectedness with urban areas, less exposure to mass media and often are characterized by lower spending power than the population living in urban areas. Therefore, hypothesized is that:
In Figure 5 the conceptual model is presented based on the hypotheses developed in the section above.
Figure 5: Conceptual Model
1. GDP 2. Import 3. Export 4. Tourist arrivals 5. Uncertainty avoidance 6. Individualism 8. Population size 9. Urbanization
3. Research Design
The hypotheses that are derived from the literature study are being analyzed with statistical tests. The research design describes the data and methodology of this study. In section 3.1 the data collection process and data characteristics are presented. Section 3.2 emphasizes on the independent and dependent variables in this study. In section 3.3 the procedure of statistical analyses is mentioned. Section 3.4 displays the statistical models.
3.1 Data
3.1.1. Collection
A literature search is executed towards empirical studies published since 1960, that addressed innovation diffusion and related topics. As part of this work a series of search strategies are carried out. First, papers published in acknowledged scientific journals are accessed via business source premier and purple search by focusing on keywords such as ‘adoption’, ‘innovation’ and ‘diffusion’. Second, references in the publications that were obtained are examined to find additional empirical research.
The decision to adopt a research in this paper is based on two criteria. First, studies that report new empirical findings are included and, secondly, studies that focus on innovation adoption in a business to business context are not included.
The data for the statistical analyses is collected from different sources. First, data for the dependent variable, adoption rate, is gathered from Datamonitor (2012). The data consists of the household’s adoption rates of consumer durables. The availability of data limited the time span of the diffusion curve. Therefore, the data is mainly a reflection of the saturation phase.
Furthermore, the data for the independent variables is collected from World Databank (2012). For the variable tourist arrivals, data is collected from the year 2010 to replace the missing data of 2011. The data for the independent variables that represent cultural factors are derived from Hofstede’s (1980) prior research on cultural dimension. Recently, these scores were updated by Hofstede et al. (2011) (see Table A1 in the appendix).
3.1.2. Description
As mentioned, the panel data includes adoption rates of infrequently purchased consumer durables by households. Consumer durables can be defined as luxury goods. Consumer durables might be continuous innovations, like televisions, or discontinuous innovation, like VCR’s, which eventually is replaced by disrupting technologies. This study focuses on continuous innovations.
knowledge on cross-country innovation diffusion (Gatignon et al., 1989; Takada & Jain, 1991; Helsen et al., 1993). The variables are measured at three points in time: 1997, 2004 and 2011. The particular points in time are chosen to obtain 291 observations (N) of the dependent variable: adoption rate.
Golder & Tellis (2004) investigated the commercialization timing of consumer durables (see Table 1), where slowdown means that the innovation reached maturity and, therefore, the diffusion is close to the saturation point. Personal computers and washing machines are not included in the study by Golder & Tellis (2004), but known is that the personal computer and the washing machine are commercialized in 1975 and 1930. Thus, it is expected that the consumer durables are well beyond the takeoff phase and reached the saturation phase.
Table 1 : Commercialization timing of consumer durables
Product category Commercialization Takeoff Slowdown
Microwave ovens 1966 1972 1988
Color Televisions 1954 1962 1969
Refrigerators 1918 1926 1938
It is a prerequisite that the panel data represents countries with difference in culture and include industrialized countries as well as emerging countries. According to Chandrasekaran & Tellis (2008) many cross-country diffusion contributions exclude some of the largest economies like Japan, China, India but also fast growing economies as Korea and Brazil. The countries selected are listed in Table 2.
Table 2: Overview of selected countries
Nr. Country Nr. Country Nr. Country Nr. Country Nr. Country
1 Australia 6 China 11 India 16 New Zealand 21 Spain
2 Austria 7 Denmark 12 Ireland 17 Norway 22 Sweden
3 Belgium 8 Finland 13 Italy 18 Mexico 23 Switzerland
4 Brazil 9 France 14 Japan 19 Republic of Korea 24 United Kingdom
3.2 Variables
The dependent and independent variables are summed up in this section. In addition, the indicators that measure the variables are described. The independent variables are categorized by economic-, cultural- and demographic factors. The last part of this section is dedicated to the control variables.
3.2.1. Dependent Variable
The indicator that measures the effect of adoption rate is the total percentage of households in a country that adopted a consumer durable at a particular year. The consumer durables are: microwave ovens, personal computers, refrigerators, televisions and washing machines.
3.2.2. Independent Variables
The independent variables are summed in Table 3. In addition, the indicator that measures the effect of the independent variables is described.
Table 3: Overview of independent variables
Category Independent variable Indicator
Economic GDP
This variable represents the income level of the individual within a country. The indicator that measures the effect is the country’s gross domestic product per capita in American dollars in a particular year.
Import
This variable represents the openness of the economy in terms of international trade. The indicator that measures the effect is import as a percentage of the country’s gross national product in a
particular year, as proposed by Talukdar et al. (2002).
Export
This variable represents the openness of the economy in terms of international trade. The indicator that measures the effect is export as a percentage of the country’s gross national product in a particular year, as proposed by Talukdar et al. (2002).
Tourist arrivals
Cultural Uncertainty avoidance
This variable represents the country’s cultural characteristic. The indicator that measures the effect is the uncertainty avoidance index on a scale from 0-100, where 0 is low uncertainty avoidance and 100 is high uncertainty avoidance, as described by Hofstede (1980).
Individualism
This variable represents the country’s cultural characteristic. The indicator that measures the effect is the individualism index on a scale from 0-100, where 0 is high collectivism and 100 is high individualism, as described by Hofstede (1980).
Power distance
This variable represents the country’s cultural characteristic. The indicator that measures the effect is the power distance index on a scale from 0-100, where 0 is low power distance and 100 is high power distance, as described by Hofstede (1980).
Demographic Population size
This variable represents the country’s demographic factors. The indicator that measures the effect is the country’s total population in a particular year.
Urbanization
This variable represents the country’s demographic factors. The indicator that measures the effect is the number of people that live in an urban area as a percentage of the country’s total population in a particular year
3.2.3 Control Variables
The study includes two dummy variables to control for unobserved country effects. Firstly, product dummies are introduced to control for the difference between the five consumer durables. Secondly, dummies for regions are included to check if there is regional difference between: North America, South America, Europe and Asia-Pacific.
3.3 Procedure of Data Analysis
simplified model, followed by a robust regression analysis. Finally, a fixed effects regression analysis with dummies is performed, to control for unobserved country effects.
3.4 Models
This section is dedicated to explaining the multivariate log-log regression analysis and the fixed effects regression analysis. As indicated, the diffusion process is a nonlinear function, meaning that the saturation phase of the diffusion curve is also a nonlinear function. Therefore, all variables are log transformed.
3.4.1 Multivariate log-log Regression
This paper assumes that household’s adoption of consumer durables is a function of numerous country-specific explanatory variables. The standard multivariate log-log regression model is:
(1)
The interpretation of beta in a log-log regression analysis is based on the percentage change of the explanatory variables:
If the explanatory variable increases by one percent, then this causes the dependent variable to increase by percent.
This paper includes country-specific factors that differ between countries over time. The specific multivariate log-log regression equation underlying this paper is:
(2)
Where
= household’s adoption rate of a specific consumer durable in a country at a particular
year= intercept or constant
3.4.2 Fixed Effects Model
The second model denoted in this paper is the fixed effects regression analysis. Fixed effects methods completely ignore the between-country variation and focus only on the within-country variation. A motive to include the fixed effects model is that the between-country variation is very likely to be contaminated by unobserved regional effects that are correlated with the explanatory variables.
Time invariant variables are omitted from the model, because fixed effects regressions analyses are designed to study the causes of changes within a country. A time invariant characteristic is captured by the country-specific fixed effect. Compared to the random effects model the fixed effects model is generally much less restrictive than the random effects model and, thus, these models are more likely to represent the data in a realistic way (Allison, 2005). The basic equation of the fixed effect regression is:
(1)
Consider the equation where the subscript refers to different countries and
refers to different
measurements within countries. Moreover, is the unobserved country effect. In order to avoid the problem of heterogeneity bias, all higher-level variance, and with it any between effects, are controlled out using the higher-level entities themselves (Allison, 2009). The method to achieve this is the inclusion of dummy variables for each higher-level entity, such that:
Where is a series of dummy variables, one for each higher-level entity , and is the average effect for each of these. This study includes dummies for product and regions. These are expected to influence the adoption rates. As such, the equation can be reduced to:
Where
= household’s adoption rate of a specific consumer durable in a country at a particular
year
= intercept or constant
= coefficient of dummy variable product = coefficient of dummy variable regions
= coefficient of a country-specific factor for a specific consumer durable at a particular year
= error term
(2)
4. Results
This chapter presents the results of the statistical analyses that are performed. The focus lies on testing the hypotheses by conducting the following analyses: descriptive statistics, the univariate log-log regression and a multivariate log-log regression analysis. The remainder of this chapter describes the fixed effects regression analysis including dummies for product and regions to control for unobserved country effects.
4.1 Descriptive Statistics
This study includes 25 countries, of which 15 are European, 6 are Asia-Pacific, 3 are North American and 1 is South American. From the 25 countries, there are five countries that can be categorized as developing countries, but which are also categorized as emerging markets, according to the IMF (2012). These are India, Russia, China, Brazil and Mexico.
The average adoption rate of the five consumer durables in Europe is 88% against 61% in the emerging markets in 2012. However, the average adoption rate for India is only 20%. By taking a closer look at the product categories an average adoption rate of 95% of televisions can be identified for all the twenty-five countries combined.
The average GDP of the developed countries is $50.000 in 2011 against $8.500 in the developing countries. Moreover, the urbanization rate in developed countries is 81% in 2011, against 63% in developing countries. These numbers clearly show the difference between developed and developing countries.
It is expected that adoption rates of consumer durables in the saturation phase are closely concentrated to the right of the diffusion curve with relatively high values. This is supported by some descriptive statistics, which can be found in Table A2 in the appendix. The kurtosis of adoption rate (20.3) shows that a large part of the observations are closely concentrated. Moreover, the data is negatively skewed, indicating that the mass distribution is concentrated on the right of the figure, which means that it has relatively few low values.
4.2 Correlation Analyses
in a country, measured in this study as population size, has a negative correlation with adoption rate with a value of -.43
4.3 Univariate log-log Regression
The output of the univariate log-log regression analyses can be found in Table 4. The assumption for this test is that an independent variable individually explains the variation in adoption rates without controlling for other explanatory variables:
Table 4 indicates significant (p < 0.01) findings for eight of the nine country-specific factors. The most important findings are summed up in this section. The findings for GDP (Adj. R2: .4336) implicate that countries where the population has a higher income also have higher adoption rates. GDP explains 43.36% of the variation in adoption rates without controlling for other explanatory variables. Since, the variables are log transformed it means that 1% increase in GDP causes an increase of 0.38% in adoption rates.
Tourist arrivals explain 30.13% of the variation in adoption rates and assume that countries with a higher number of tourist arrivals as a percentage of the total population, have higher adoption rates. If tourist arrivals increase with 1%, then the adoption rate increases with 0.23%.
Urbanization (Adj. R2: .4134) denotes that countries with a higher urbanization rate also have a higher adoption rate. Thus, in countries with rural areas, adoption rates are lower.
Individualism (Adj. R2: .0953) shows that 1% increase in the level of individualism in a country results in 0.53% increase in the adoption rate. So individualism is positively associated with the adoption rate. Population size (Adj. R2: .1813) has negative coefficient of -.16, which indicates that population size has negative influence on the adoption rates in a country.
Table 4: Univariate regression of adoption rate with the nine independent variables
Variable name N Coef.(β) Std. Err Constant Adjusted R2
GDP 291 .39*** .026 .388 .4336 Import 291 .22*** .072 3.521 .0276 Export 291 .20*** .063 3.571 .0305 Tourist arrivals 279 .23*** .021 3.458 .3013 Uncertainty avoidance 286 .09 .090 3.897 .0003 Individualism 286 .53*** .095 2.078 .0953 Power distance 286 -.38*** .067 5.668 .0977 Population size 291 -.16*** .0196 7.021 .1813 Urbanization 291 1.65*** .115 -2.815 .4134
Note: *** significance at the 0.01 level (2-tailed)
4.4 Multivariate log-log Regression (Model 1)
In the previous section the univariate log-log regression analyses were performed to test the relation between adoption rate and the nine independent variables separately. Now, the multivariate log-log regression analysis is performed to examine the effect of an independent variable, when controlling for the other explanatory variables. The output is summed up in Model 1 in Table 5. The assumption is that there are numerous explanatory variables that affect the dependent variable:
Model 1 shows significant findings for six of the nine independent variables, when controlling for other explanatory variables. GDP, tourist arrivals and urbanization have a positive coefficient and, thus, an increase in GDP, tourist arrivals or urbanization leads to an increase of adoption rates. Importantly, the effect of GDP is smaller (β = .27) compared to the univariate log-log regression analysis (β = .39). Thus,
indicate, when controlling for other country-specific factors, that the coefficient sign of population size changes to positive, meaning that in combination with other explanatory variables population size has a positive effect on the adoption rates, thus the greater the size of the population the higher the adoption rate. The same change in sign, but then in the opposite direction, counts for individualism. Therefore, high individualism negatively affects the adoption rate of consumer in the saturation phase. A 1% increase in individualism means a decrease in the adoption rate of 0.35%. Concluding, Model 1 including nine country-specific factors explains 53.28% (Adj. R2: .5328) of the variation in adoption rates.
4.5 Variance Inflation Factor
This section elaborates on the variance inflation factor, which is a post regression analysis to check for multicollinearity. Based on Model 1 in Table 5 multicollinearity is expected, due to insignificant p-values for import, export and power distance. The rule of thumb is that variables that score a five or higher on the test might cause problems in the model. Therefore, variables with extreme scores are dropped from the model. The outcome of the test is summarized in Table A4 in the appendix. Table A4 clearly indicates that import and export have extreme scores (17.02 and 20.41) compared to the other variables. The findings are supported by the correlation matrix in Table A3 in the appendix, which indicates that there is a strong correlation (.99) between import and export. Thus, export is dropped from the model, because economic inflow is considered to have a larger effect on adoption rate than outflow.
4.6 Multivariate log-log Regression (Model 2)
This section reshapes Model 1 into Model 2, which is strongly simplified. The art of model building recognizes the impossibility of representing all of the many influences on a dependent variable and tries to pick out the most influential variables (Newbold et al., 2006). Therefore, a simple model is build that is easy to interpret but not so oversimplified that important influences are ignored.
Based on the results of univariate log-log regression analyses, multivariate log-log regression analysis and the variance inflation factor, independent variables are dropped from the model. Hence, uncertainty avoidance is dropped due to economic and statistical insignificance in the univariate log-log regression analysis and power distance due to statistical insignificance in the multivariate log-log regression analysis. Import is eliminated from the model due to economic insignificance in the univariate log-log regression analysis and export is excluded due to an extreme variance inflation factor. A new multivariate log-log regression analysis, Model 2, is performed and the findings are presented in Table 5.
only decreases by 0.52% when excluding uncertainty avoidance, power distance, import and export. Furthermore, the statistical significance of the independent variables in Model 2 improves. Dropping more variables would lead to a significant decrease in the goodness of fit (i.e. Adj. R2).
4.7 Standardized Coefficient Ranking
The previous sections discussed Model 2 which includes five independent variables. This section elaborates on the importance of each independent variable in Model 2. The coefficients (β) of the independent variables indicate the strength with the dependent variable. However, the independent variables differ in variance. Therefore, the coefficients are standardized, so that the variances are 1. The standardized coefficient refers to how many standard deviations a dependent variable will change, per standard deviation increase in the predictor variable.
Table A5, in the appendix, ranked the beta coefficients and the standardized coefficients for comparison. Hence, observed is that an increase in GDP and urbanization has the largest effect on adoption rates. Moreover, based on Model 2, individualism is expected to be a more important determinant, than tourist arrivals. However, the standardized coefficient ranking shows the opposite result. Thus, tourist arrivals have a larger effect on adoption rates, than the level of individualism in a country.
4.8 Robust Multivariate log-log Regression
The robust regression analysis indicates whether Model 2 is affected by residuals that create leverage. Figure A1 and Table A6 in the appendix indicate that the distribution of the residuals is negatively skewed (-.19) with a kurtosis of 18.68. This evidence supports the assumption that the diffusion curve in the saturation phase is a non-linear function. Moreover, the Cook’s D (see Figure A2 in the appendix), the basis for the robust regression analysis, indicates that India creates leverages.
Note: * significance at the 0.10 level (2-tailed) Note: ** significance at the 0.05 level (2-tailed) Note: *** significance at the 0.01 level (2-tailed)
Table 5: Multivariate regression of adoption rate with the nine independent variables
Model 1 Model 2
Variable name Coef.(β) Std. Err Variable name Coef.(β) Std. Err
GDP .27*** .061 GDP .29*** .057
Import .02 .222 Tourist arrivals .10*** .033
Export -.07 .215 Individualism -.35*** .099
Tourist arrivals .13*** .038 Population size .06** .026
Uncertainty avoidance -.18** .081 Urbanization .86*** .169
4.9 Fixed Effects Regression
In order to control for the unobserved country effects a fixed effects regression analysis is performed. As discussed in Chapter 3 the equation for the fixed effects regression in this paper is:
The results of the fixed effects regression analysis are summed up in Table 6. The independent variable export is excluded due to multicollinearity as discussed in Section 4.5.
Compared to Model 1 (R2 = .5328 | DF 35.60) in Table 5, the fixed effects regression analysis increases the goodness of fit (R2 = .7104 | DF 13.23). As a result the R2 is higher, when controlling for the unobserved regional and product category effects.
In order to simplify the fixed effects regression analysis the same approach is adopted as discussed in Section 4.6. Therefore, import, export, uncertainty avoidance and power distance are dropped based on the same motivation as noted in the sections above. Table 6 shows, that by simplifying the fixed effect regression analysis to model 4, a slight decrease of .0234 (=R2) goodness of fit is observed. This result supports the decision to drop the variables.
The dummies for product difference indicate that personal computers do not explain variation in adoption rates between countries. The dummies for product indicate a higher intercept in ranked order for television, refrigerator, washing machine and personal computer compared with the micro wave oven. These findings suggest that difference in products explain the variation in adoption rates between countries. This means that, for example, the adoption rates of televisions are higher, than the adoption rates of washing machines.
Note: * significance at the 0.10 level (2-tailed) Note: ** significance at the 0.05 level (2-tailed) Note: *** significance at the 0.01 level (2-tailed
Note: Microwave oven is omitted from the model, hence functions as the reference dummy for product Note: Asia Pacific is omitted from the model, hence functions as the reference dummy for regions
Table 6: Robust fixed effects regression of adoption rate with the nine independent
variables with dummies for product and regions
Model 3 Model 4
Variable name Coef.(β) Std. Err Variable name Coef.(β) Std. Err
GDP .40*** .080 GDP .30*** .054
Import .18** .079 Tourist arrivals .18*** .041 Tourist arrivals .18*** .043 Individualism -.30*** .098 Uncertainty avoidance -.18*** .061 Population size .09*** .021 Individualism -.40*** .115 Urbanization .75*** .228 Power distance .03 .054
Population size .15*** .031 Product
Urbanization .83*** .248 Personal computers .09 .085 Refrigerators .39*** .083
Product Televisions .53*** .085
Personal computers -.06 .084 Washing machines .30*** .081 Refrigerators .39*** .081 Microwave ovens Reference
Televisions .53*** .083
Washing machines .30*** .0763 Regions
Microwave ovens Reference Europe -.28*** .069
North America -.19*** .069
Regions South America -.01 .148
Europe -.38*** .092 Asia Reference
North America -.301*** 081
South America .13 .164 Constant -2.90*** 1.070
4.10 Results and Hypotheses
In this section it is argued whether or not the hypotheses are supported by the analyses in the previous section.
Economic Variables
H1. The higher the GDP in a country, the higher the adoption rate in the saturation phase
The univariate log-log regression analysis indicates an Adjusted R2 of .4336 and a positive coefficient. This means that, there is a positive relation between GDP and adoption rates in the saturation phase and the variable explains 43,36% of the variation in adoption rates between countries. This shows that the level of income of adopters positively influences innovation adoption. The positive relationship is supported by the fixed effects regression analysis. Hence, the findings support the hypothesis.
H2. The higher the percentage of import in a country, the higher the adoption rate in the saturation phase
The univariate log-log regression analysis shows a positive coefficient and an Adjusted R2 of .0276, which indicates an economically irrelevant relationship with adoption rates. This means that an open economy, measured as import as a percentage of the country’s gross national product, does not influence the adoption rate in the saturation phase. Hence, the findings do not support the hypothesis.
H3. The higher the percentage of export in a country, the higher the adoption rate in the saturation phase
The univariate log-log regression analysis shows a positive coefficient and an Adjusted R2 of .0305, which indicates an economically irrelevant relationship with adoption rates. If we control for other variables export becomes statistical insignificant and the variance inflation factor indicates that the variable correlates with import. Therefore, it is dropped from the model, which showed no decrease in Adjusted R2. Hence, the findings do not support the hypothesis.
H4. The higher the percentage of tourist arrival in a country, the higher the adoption rate in the saturation
phase
Cultural Variables
H5. The lower the uncertainty avoidance in a country, the higher the adoption rate in the saturation phase
The univariate log-log regression analysis shows a positive coefficient and an Adjusted R2 of .0003, which indicates an economically irrelevant relationship with adoption rates. If we control for other variables uncertainty avoidance becomes statistical insignificant. The extent to which people feel uncomfortable in the presence of vagueness and ambiguity does not influence the adoption rate in the saturation phase. Hence, the findings do not support the hypothesis
H6. The higher the individualism in a country, the higher the adoption rate in the saturation phase
The univariate log-log regression analysis indicates an Adjusted R2 of .0953 and a positive coefficient. If we control for the other variables individualism has a negative coefficient significant, so the coefficient changes by inclusion of other variables. The variable explains variation in adoption rates and adds value to the model, although the negative coefficient conflicts with the hypothesis. Hence, the findings do not
support the hypothesis. However, the variable is relevant in explaining the variation in adoption rates
between countries, when controlling for the other variables.
H7. The lower the power distance in a country, the higher the adoption rate in the saturation phase
The univariate log-log regression shows an Adjusted R2 of .0977 with significance of 1% and a positive coefficient. If we control for other variables power distance becomes statistically insignificant. How a society handles inequalities among people does not influence the adoption rate in the saturation phase. Hence, the findings do not support the hypothesis
Demographic Variables
H8. The higher the population size in a country, the higher the adoption rate in the saturation phase
The univariate log-log regression analysis indicates an Adjusted R2 of .1813 with a negative coefficient. So, there is a negative relation between population size and adoption rates in the saturation phase and the variable explain 18,13% of the variation between countries. The sign of the coefficient changes after controlling for the other variables. The positive relationship is supported by the fixed effects regression analysis. Thus, the findings support the hypothesis.
H9. The higher the urbanization in a country, the higher the adoption rate in the saturation phase
5. Discussion
This section highlights the findings and addresses the conclusions that can be drawn from this study. Thereafter, the theoretical and practical relevance, limitations and suggestions for further research are discussed.
5.1 Findings
There is direct evidence for the theory of Steenkamp & Burgess (2002) that income, in this paper measured as GDP, explains variation in adoption rates. In countries with a higher average income the adoption rates in the saturation phase are also higher, because people can afford to buy an innovative product. The data shows that international traffic of people, in my study measured as the number of tourist arrivals, influences adoption rates, as illustrated by Tellis et al. (2003). International traffic results in higher adoption rates in saturation phase among households. Therefore, international contacts stimulate innovation diffusion. In other words, the population within a country is stimulated to adopt an innovation by external influences through word of mouth from people outside their country.
This study shows contradictory findings for individualism, since it finds a negative relation with adoption rate. Results indicate that a positive relation exists between collectivism and adoption rates in the saturation phase. This contradicts findings of various studies: Steenkamp et al. (1999); Yeniyurt & Townsend (2003). They assume that individualism has a positive effect on the adoption rate, because individuals are driven to take an adoption decision by themselves. However, the outcome of this contribution is in line with studies that focus on the coefficient of imitation. Yaverogly & Donthu (2002) showed that countries that are low on individualism are high on imitation. As indicated by the Bass Diffusion Model (1969) the adopters in the saturation phase are typified as imitators. Therefore, this study confirms that collectivism is positively related with the adoption rate in the saturation phase.
Results for population size as an explanatory variable support the theory of Van Everdingen, Fok & Stremersch (2008), that larger populations show higher adoption rates in the saturation phase, because larger populations increase the chance of innovation diffusion within a country. Findings indicate that urbanization has a positive influence on the adoption rate of the innovation. This is in line with theory of Talukdar et al. (2002), who found that urbanization is likely to increase the market penetration potential. This means that for a country, that is characterized by high urbanization, the innovation diffusion evolves more efficient and, thus, results in higher adoption rates in the saturation phase.
The results for power distance and uncertainty avoidance showed no significant findings for explaining adoption rates. This means that the degree to which people accept a hierarchical order in which everybody has a place and which needs no further justification does not influence the adoption rates in the saturation phase. The same counts for the degree to which people are comfortable with unfamiliar situation. This can be explained by the fact that cultural factors are mainly studied as determinants for innovativeness (Yaverogly & Donthu, 2002). However, the saturation phase is characterized as a phase where innovations are mainly adopted by imitators, who are being characterized as not innovative (Bass, 1969). Lastly, this study found evidence for regional difference in explaining variation in adoption rates in the saturation phase. Especially, emerging markets are significantly different from developing markets, when comparing innovation adoption rates. This is mainly caused by a low GDP and low urbanization rates for the emerging markets.
5.2 Conclusion
The study provides compelling evidence that the characteristics of a country prove to be useful in explaining the difference in innovation adoption rates between countries in the saturation phase. The goal was to expand knowledge on cross-country innovation diffusion by exploring which country-specific factors explain variation in adoption rates in the saturation phase. Data on household’s adoption rates of microwave ovens, personal computers, refrigerators, televisions and washing machines for 25 countries were analyzed with multivariate log-log regression analysis and a fixed effects regression analysis. Positive relations are found between the adoption rate and GDP, tourist arrivals, population size, urbanization and a negative relationship with individualism. Thus, economic, cultural and demographic factors that characterize a country explain difference in adoption rates between countries in the saturation phase.
5.3 Theoretical and Practical Relevance
Previous research in the area of cross-country innovation diffusion has largely focused on determinants, like economic, cultural and demographic factors, to explain difference in adoption rates between countries (Sood et al., 2009; Yaveroglu & Donthu, 2002; Dekimpe et al., 1998). Although the take off phase has been widely mentioned (Talukdar et al., 2002; Tellis, Stremersch & Yin, 2003), only few studies have looked into the saturation phase of the diffusion curve (Golder & Tellis, 2004).
The findings of this research are beneficial to managers who formulate segmenting strategies for international saturated markets. The knowledge gained in this study supports management to segment countries. It supports, especially, segmenting strategies for consumer durables in the saturation phase. A critical question for managers is how to allocate their marketing budget. By understanding how difference in adoption rates between countries in the saturation phase is caused, a manager can identify opportunities for marketing efforts.
Prior research of Kalish, Mahajan & Muller (1995) discusses that international organizations can choose between two strategies: the waterfall strategy, where the markets are entered sequentially, and the sprinkler strategy, where markets are entered simultaneously. They argue that current nature of global competition requires a multinational firm to follow the sprinkler strategy when introducing an innovation to its global markets. These strategies are entering strategies. However, the saturated markets need a different strategy, since a sprinkler strategy simply results in high costs.
Based on the findings, a suggestion for strategic efforts for international saturated markets is carefully mentioned. Since, the findings support segmenting strategies based on country-specific factors the best fit is a segmented diversification strategy. Research showed that country-specific factors can be the basis for segmenting countries. Furthermore, in the saturation phase, the competitive focus of a company shifts to economies of scale and differentiating initiatives. Prior research has indicated a strong positive link between a diversification strategy and organizational performance, which is essential in high competitive markets (Antoncic, 2006). The diversification strategy is a common strategy from an organization’s perspective to help overcome the challenges of a saturated market.
5.4 Limitations and Suggestions for Further Research
The scope of this study incorporates countries with different characteristics, but the study is not fully representative of the variety of levels of development and cultures existing in the world. The analyses showed that countries like India, Russia, China, Brazil and Mexico show a great variety in characteristics, which differ significantly from developed countries. Therefore, other developing markets like Africa need to be represented in a similar study.
