1
University of Groningen
Faculty of Economics and Business
Cultural influence on eco-innovation
Supervisor:
Bart Los
Student name: Liana Stefan (S207726)
Student email: l.b.stefan@student.rug.nl
2
Cultural influence on eco-innovation
Abstract
The environmental problems that have surfaced in the last years have encouraged enterprises as well as governments and different global organization to focus more on the environmental results from the creation of new technology and business processes. Eco-innovation is seen as the way to achieve both economic and environmental advantages. A typical influence on innovation that has been long studied is the cultural characteristic of a country. The aim of the thesis is to determine the implications of Hofstede’s cultural dimensions on countries’ interest in engaging in eco-innovation production and development. The analysis concentrates on the results of the Community Innovation Survey 2008 for 19 European countries, using cross- section data. My overall results show that national cultural characteristics can have some influence on the number of enterprises that choose to introduce innovations with environmental benefits.
Key words: eco-innovation, Hofstede’s cultural dimensions, CIS 2008
3
Table of Contents
1. Introduction ... 4
2. Theoretical Background ... 7
2.1. Eco-innovation ... 7
2.2. The Hofstede Cultural Dimensions ... 10
2.3. Theoretical hypotheses ... 12
3. Proposed measures of eco-innovation ... 13
4. Data and Methodology ... 14
4.1. Sample ... 14
4.2. Dependent Variable ... 17
4.3. Independent Variables ... 18
4.4. Control Variables ... 18
4.5. The econometric model ... 21
5. Results ... 24
5.1. Descriptive statistics ... 24
5.2. Regression 1 ... 24
5.3. Regression 2 ... 26
5.4. Regression 3 ... 27
6. Limitations ... 29
7. Conclusions ... 30
8. References ... 32
Appendix ... 37
4 1. Introduction
As economic growth has continued to rise, so has the attention given to the health of the natural environment and sustainable development (OECD, 2009). Either locally or globally, environmental issues represent challenges that the world is currently facing. Some of these challenges, exposed by the environment, vary from limited resources availability, pollution of water or air, emission of green house gases, climate changes, to the on-going search for a fossil –fuel replacement and so on (OECD, 2009, Jansson et al, 2011).
Nonetheless, the ability to acknowledge and manage this phenomenon is increasing. As countries evolve and become more technologically advanced so is their desire and capacity to deal with environmental pressures and finding more innovative technological solutions to solve such challenges (OECD, 2010).
Fig. 1: Number of entreprises with innovations with environmental benefits/millions of people
Observing the data gathered by Eurostat on innovations performed by enterprises in European
Member States, we can see that countries have quite dissimilar results, showing different
levels of implication in environmental innovation. The initial data provided by Eurostat is
divided by millions of people, after. By applying this transformation to the observations, the
data used for the regression eliminates any bias resulting from country differences in size and
5 economic development. The leaders of the field, according to the results of the CIS, are by far Germany, Luxembourg and Portugal. The lowest results are in Bulgaria and Poland, where a low number of enterprises engage in innovations with environmental benefits. As the graph shows some interesting results, such as the appearance of Portugal among the leaders, a more detailed description can be found in the data and methodology section.
The variety among countries can be the result of a number of factors, which stimulate or discourage the production of product and process innovations that would aid or simply avoid harming the natural environment.
As differences can be found in the political environment, economic situation, priorities or natural resources of countries, the same applies for their capacity to produce different types of innovations. Small countries have a disadvantage in the limited range of activities they can engage in, whereas large countries, supported by a large affluent population, have the possibility to diversify their range of manufacturing industries and activities (Nelson et al, 1993).
There are many aspects that build the uniqueness of a country and shape its technological innovation (Nelson et al, 1993), such as a country’s governmental policies, national investments, labor force, its accumulated stock of knowledge, spending on higher education, intellectual property protection or openness to international competition (Stern et al., 1999).
Numerous studies have tried to explain the differences between the innovation levels of countries by appealing to the culture argument and particularities of the values or beliefs that prevail among societies. Culture specific characteristics have the ability to shape people’s perspectives on innovation, as some countries choose to be more involved in its production while others simply prefer to adopt innovation. Moreover, these characteristics can influence a society or company’s freedom to explore, exercise or communicate ideas, its approach towards the uncertainty usually revealed by the process of producing or adopting innovations.
In the same way, people’s beliefs and ideas can also interfere in the development process of innovation (Kaasa and Vadi, 2008). As Kaasa and Vedi presents it, “the openness towards new experiences varies in different cultures” (Kaasa and Vadi, 2008, p. 584).For instance, G.
Judevičius (2008) tries to explain the innovation culture of countries by focusing on the
institutional elements of technological innovations as well as attempting to establish links
between cultural dimensions and these institutional elements. Kaasa and Vadi (2008) examine
the capability of innovation initiation in European countries in connection to the cultural
6 dimensions introduced by Geert Hofstede. Equally important is the study of Erumban and de Jong (2006), focused on the adoption of ICT determined by the same cultural indicators, the analysis made by Vecchi and Brennan (2009) on the innovation of manufacturing firms across countries, based on the „culture specific argument” or the article by van Everdingen and Waarts (2005) which explains the influence of national cultures on differences in innovation adoption rates across countries.
With regard to eco-innovation, most theoretical studies have concentrated on either defining the concept of eco-innovation and its specific characteristics (Rennings 1999, Ekins), finding the most appropriate indicators (Arundel. A and Kemp.R 2009, MEI report 2007, Speirs J. et.
al., European Environment Agency 2006,) or assessing the diversity of eco-innovation (Carrillo-Hermosilla, J. et al, 2010). The research done by Eurostat shows that there is a clear gap between countries when it comes to generating environmental friendly technologies and processes. Moreover, the Europe Innova report (2008) has explained that some of the barriers or drivers for such differences in innovation result from a socio-cultural value system that characterizes societies and communities. In turn, the socio-cultural value system is determined by socio-cultural factors, such as consumer behavior, human, cultural and social capital. The important factor here is the cultural capital, where cultural capital represents the “cultural background and basic value system that is shared by the individuals in a community and manifests in their attitudes and habits”, (Europe Innova, WP.10, 2007, p.5). The cultural capital is seen as a strong innovation promoter for some countries. In other countries, the same factor is seen as a barrier (Europe Innova, WP.10, 2007). Indicators such as the interest in science and technology, attitude towards risk from new technologies or attitude towards the natural environment were used to reach these results.
These variances can lead to the analysis on the relevance of cultural differences for the eco- innovation gap between countries. The research aim of the thesis is to test if the Hofstede cultural dimensions have any impact on the production of eco-innovation in European countries. For this purpose, the following main research question is proposed:
“Does national cultural influence the interest to engage in eco-innovation?
During my analysis, I will focus on 19 European countries, with data representing the
production of eco-innovation and national cultures based on the Community Innovation
Survey 2008 and the Hofstede cultural index respectively.
7 The layout of the thesis will start with the second section concentrating on the relevant fundamental concepts for the creation of eco-innovation, by mentioning some of the findings up to date of the field understudy. Based on these concepts presented by the literature, the hypotheses guiding the empirical analysis will be the closure of the section. A crucial aspect of my thesis is the eco-innovation measurements proposed so far, which I will introduce in chapter three. The following section will present the most important details of the data set as well as the econometrical methods put into practice. Sections five and six present and interpret the empirical results and limitations of the study followed by the conclusions in section 7.
2. Theoretical Background
In this section I will provide some theoretical explanations and concepts surrounding the creation of environmental innovations and cultural distances, to use as a guide for my analysis.
2.1. Eco-innovation
The simplest forms of defining the term of “eco-innovation” are provided by Ekins (2010, p.
269), stating that “environmental innovation means the changes that benefit the environment in some way”, and by Norberg - Bohm (1999), who define eco-innovation as the innovation that reduces the impact on the environment through waste minimization.
A more developed definition has been given by Rennings (2000) and Hellstrom (2007) who
point out that environmental innovation is the process of developing new ideas, behavior,
products and process that lead to a decrease in environmental burdens or to ecologically
specified sustainability targets. Applying a more detailed explanation, Blättel-Mink (1998)
note that eco-innovations may include the development and introduction of not only new
products, but also new markets and systems as well as the introduction of ecological
dimensions and economic strategies.
8 Rennings (2000) further emphasizes the need for sustainable development, not only on technological changes, but also on high long-term targets that have to be supported also by changes in infrastructures, lifestyles and institutions.
The articles of Ekins (2010) and Huppes et al. (2008) separate themselves from the other papers by stating that in order for enterprises to generate environmental innovation, they simultaneously need to obtain results in environmental improvements and economic performance.
While all these definitions are appropriate, there is one more important aspect that sets eco- innovation apart from the general assumptions of innovation. Along with the environmental benefits that are expected from the production of environmental innovations, it is also necessary to have support from adequate institutional and social structures (Hellstrom, 2007).
Rennings (2008) focuses more on this perspective and adds that unlike technological innovation, the nature of sustainable innovation is also organizational, social and institutional.
Organizational innovation occurs at the firm level and implies, for instance, changes in management instruments or also marketing strategies, defined as non-technological innovation by the OECD (2009). Social innovations refer to the shifts in lifestyles and consumer behavior that have the ability to further support the production of eco-innovation by enterprises. As a matter of fact, sustainable consumption patterns can represent an important factor for the movement of companies from the existent products and processes and the current use of innovation towards environmental friendly technologies, products or systems that not only boost economic performance but also have positive effects for the natural environment. The institutional nature of eco-innovation is explained by the support given to the enterprises generating such innovations, trough environmental policies, local networks and agencies or global governance (Rennings, 2008).
The idea of institutional support is consistent with that of Norgaard (1994) who suggests that unsustainable development can occur if only technological innovation takes places and it is not followed by an evolution of incentives and regulations. The work of Oosterhuis and Brink (2006) shows that environmental policies can promote the creation of eco-innovation and influence its “speed and direction”(Ekins, 2010, p.275).
Therefore, in order to prosper, producing eco-innovation has to be based on social structures
(Hellstrom, 2007). Using the “sub-system” of Freeman and Louça (2001), Ekins explains that
these environmental and economic performances are the results of economic activities
9 connected with environmental interest. These economic interests further on are driven by institutions, laws, norms and practices that define markets and economic structures. These institutions are in turn derived from polity, culture and social values.
As already mentioned, the findings of numerous studies have demonstrated that cultural features carry the ability to influence to some degree a country’s economic results. So far, in the various studies of the implications and effects of culture, one of the most useful measurements capturing culture diversity has been the national culture index provided by Geert Hofstede (2001).
As I mentioned in the introduction, Kaasa and Vadi (2010), focusing on European countries, analyze the relationship between culture and innovation initiation. In their context, they use innovation initiation as the first phase of the innovation process, the moment when ideas are generated, that will later on, in the next phase (innovation implementation), be adopted and exploited. They argue that “culture affects innovation because it shapes the patterns dealing with novelty, individual initiatives and collective actions, and understandings and behaviors in regard to risks as well as opportunities” (Kaasa and Vadi, 2010, p.2). Their hypotheses are based on Hofstede’s cultural dimensions, with the difference that Kaasa and Vadi measure culture by appealing to the European Social Survey and innovation by using patenting intensity. The novelty of the study stands particularly in the use of the ESS survey, which allows for a regional level analysis and not just a national one. Moreover, it has the ability to account for the same cultural dimensions created by Hofstede, but on a sample representing a broader society. The results of Kaasa and Vadi’s study show a negative relationship between power distance, uncertainty avoidance and masculinity with patenting intensity, supporting their hypotheses. In the case of individualism, the study reveals a weak positive connection to patenting intensity.
Unlike the previous study, Erumban and de Jong (2006) focus on the effects on ICT adoption
of Hofstede’s cultural dimensions. When it comes to power distance and uncertainty
avoidance, the results also show a negative relationship with the adoption of ICT. On the
other hand, the authors obtained weak results for individualism and no support for the
relationship between masculinity and ICT adoption. Everdingen and Waarts (2005) also
support the idea that cultural differences, large enough even among the European countries,
are one of the factors contributing to the variances in innovation adoptions. In their study,
Everdingen and Waarts (2005) represented the adoption of innovation as the adoption of
10 Enterprise Resource Planning (ERP) software by medium-sized companies. Just like Erumban and de Jong, they have reached the same results with regard to power distance and uncertainty avoidance. However, they found support for a negative relationship between masculinity and low levels of ERP adoption.
These examples of studies on the effects of cultural dimensions and a variety of aspects of the innovation field are a useful guide for the aim of the thesis as to what type of relationships can result between eco-innovation and Hofstede’s cultural dimensions (2001).
2.2. The Hofstede Cultural Dimensions
By relying on a survey among IBM employees in 40 countries, Hofstede (2001) found that
“four statistically-independent dimensions explained the inter-country variation in employee responses to his survey questions” (Drogendijk and Slangen, 2006, p. 363). Further on, I will briefly explain each of the dimensions.
In his work, Hofstede uses Mulder’s (1977, p.90) definition of power distance (PDI) as “the degree of inequality in power between a less powerful individual (I) and the powerful Other (O), in which I and O belong to the same (loosely and tightly knit) social system” (Hofstede, 2000, p.83). The hierarchy in the low PDI countries is more a formal concept and, as Hofstede describes it (Hofstede, 2001, p. 102), the “inequality of roles is established for convenience”.
Some of the differences between countries, as a feature of the power distance dimension, are
higher income and education levels or more technology use for the low-PDI countries. On the
other hand, Hofstede (2001) also found that these low-PDI societies seem more skeptical with
regard to the benefits resulting from the use of technology. In a study, Hofstede (2001) finds
that in countries with larger power distance, innovators rely more on the confirmation and
support of the higher ranks for generating ideas or producing new technologies. Also, a higher
willingness to communicate, translated in an easier flow of information, in a low PDI society,
is a positive aspect that is important for the process of creating innovations as well as the
cooperation of citizens with the authorities. An example in this respect is the waste recycling
done by the population in low power distance countries, as opposed to high power distance
countries, where inhabitants wait for the authorities to take action for environmental
protection (Hofstede, 2001).
11 Individualism (INV) focuses on the level to which people act or perform activities on their own and for their own advantage as oppose to countries described by collectivism, where people heavily rely on other members of the group or the information they might provide (Vecchi and Brennan, 2008). In the case of individualism versus collectivism, seen as a dual dimension, Hofstede’s study discovered that in less individualistic countries innovators tended to be more involved in the creation process and collaborate more with other members of an organization than to work on their own. Moreover, the findings of Vechi and Brennan (2008) showed that collectivist societies had a higher degree of innovation input, a more intensive use of technology for operational activity and ERP as well as more resources invested in R&D.
“Masculinity (MAS) 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, p. 297). One valuable aspect in the more feminine cultures is the higher importance given to the quality of life and the preservation of the natural environment (Hofstede, 1983).
Uncertainty Avoidance (UAI) is defined as “the extent to which the members of a culture feel threatened by uncertain or unknown situations” (Hofstede, 2001, p. 162). The UAI dimension is based on the analysis of rule orientation, employment stability and stress. The research done by Hofstede (2001) shows that more structured bureaucratic organizations had higher results in innovative work as the employees tended to be more flexible intellectually, at the same time feeling more constrained by the rules and regulations. Even though low uncertainty avoidance countries engaged easier in the production of innovations, the opposite type of countries presents a more serious involvement. In a comparison with Schwartz’s research in human values, Hofstede (2001) finds correlation between the former’s harmony category and the uncertainty avoidance dimension. If “harmony stands for unity with nature, protecting environment and world of beauty” (Hofstede, 2001, p.159) and is more appropriate for high UA as compared to low UA, the expectation would be that countries with a high score are more concerned with environmental protection.
One of the most enthusiastic critics of Hofstede’s work is Schwartz S. H. (1994). He claims
that the purpose of the survey used as the basis of creating the cultural index was not designed
12 to determine cultural dimensions, thus related questions may have been omitted, or that the questioned employees were not a representative sample of their countries, since from the time the data has been obtained some cultural changes might have occurred. In this aspect, Hofstede, in the revised version of his work, explained that the cultural dimensions still have the same value and meaning as culture does not change easily over time (Hofstede, 2001). A final criticism of Schwartz (1994) is the perception of values that can differ from the perspective of each culture.
In their analysis on the way MNE’s choose their establishment mode, Drogendijk and Slangen (2005) have found that the measurements based on Hofstede or Schwartz’s work are equally useful and they disagree with the misconception that “Hofstede’s work is outdated or inaccurately reflecting national cultures” (Drogendijk and Slangen, 2005, p. 362) or that it should be replaced by Schwartz’s more recent findings.
2.3. Theoretical hypotheses
In order to formulate the hypotheses of my analysis, I have used the above presented literature. My main guide were the findings of Hofstede (1983, 2001) on the basic features of each cultural dimension in connection with the way national cultures engage in innovation, behave during the innovation process as well as their approach to the natural environment.
Hypothesis 1: National cultures defined by low power distance (high power distance) present higher eco-innovation (lower eco-innovation).
Hypothesis 2: National cultures defined by low individualism score (high individualism) present a higher eco-innovation (lower eco-innovation).
Hypothesis 3: National cultures identified by a low masculinity score (high masculinity score) present a higher eco-innovation (lower eco-innovation).
Hypothesis 4: National cultures defined by low uncertainty avoidance (high uncertainty
avoidance) present lower eco-innovation (higher eco-innovation).
13 3. Proposed measures of eco-innovation
At the present moment, the methods of obtaining indicators of eco-innovation are not clearly defined, as the main obstacle is the lack of an alignment between innovations and the environmental aspects (EEA-Eco-innovation indicators, 2006). In other words, the difficulty stands in the attempt to observe if there are any positive environmental results from the introduction of innovations. Nevertheless, some methods towards this goal have been proposed and are seen as the basis for future measurements of eco-innovation.
The environmental industry offers little help in obtaining eco-indicators as it provides statistical analyses on the produced goods or services in the industry, but offers only partial information related to eco-innovation and environmental benefits. Moreover, the data from the environmental industry concentrates on the economic aspect, but the link between innovation and positive results or improvements for the natural environment is weakly reflected. In this respect, the European Environment Agency proposes (2006) to capture the degree of eco-innovation through an eco-efficiency analysis, by assessing the improvements in environmental behavior obtained by different agents. These improvements should indicate the degree of eco-innovation or structural changes, as well in production as in consumption patterns. The benefit of this option is that it captures the environmental aspect of a product or services. However it fails to show the changes in the capacity to generate innovations at national level (EEA, Eco-innovation indicators 2006).
Some of the data sources proposed by the literature are patents, environmental R&D or surveys.
In the case of patents, a harmonization between the European, North American and Japanese patent office has been made and even though they provide a rich database that can be used for the majority of innovation studies, the insights it offers for the sustainable technology are rather ambiguous and difficult to separate from those innovations unrelated to the natural environment.
Some useful information can be provided by the expenditure for environmental R&D or
environmental protection. Such data can expose the amount of money dedicated on processes
or technologies that prevent or reduce pollution. Unfortunately, it says little of the novelty
they bring to the market. In other words, from this source of information one cannot
distinguish if the technology or equipment is new to the market or only an old standardized
14 one. Also, the data are available at industry level for only a small number of countries, while most is presented in an aggregate form, typically included with other environmental areas along with other activities that cannot be classified elsewhere (EEA – Eco-innovation indicators, 2006).
Up to now, the most useful data on eco-innovation, that also represents the basis of the present analysis, has been provided by the survey employed by EU. Starting with 1993 and following the methodological and definition recommendations of the Oslo Manual (Eurostat and OECD, 2005), the EU has conducted a harmonized survey questionnaire under the name the Community Innovation Survey. The purpose of the CIS is to monitor the innovation activity within the European countries, typically covering a three-year period (Eurostat Pocketbooks, 2011, Parva, 2007).
One of the benefits of the survey is the relevance of its questions in connection to environmental friendly technology. The CIS survey has the ability to offer a more comprehensive measurement of innovation by including expenditure on innovation, capturing output indicators representative for innovation and identifying possible sources of knowledge for innovation or barriers (Eurostat, 2009). Nevertheless, the survey has some important limitations. It fails to make a clear distinction between the methods used by enterprises to introduce environmental friendly innovations, by either producing it or adopting it, the module is not compulsory, thus distorting the reality of the actual situation of eco-innovation in Europe. The fact that the eco-innovation module of the survey is optional for enterprises in filling it can only aggravate the problem.
4. Data and Methodology
4.1. Sample
The analysis is based on the Community Innovation Survey 2008 (CIS), conducted by
Eurostat in collaboration with the EU Member States. Relying on 2008 as a reference year
and running over a three year period, 2006-2008, the purpose of this standard core
questionnaire is to collect data on a number of innovation related variables in enterprises
(Eurostat 2011). A relevant aspect of the CIS 2008 that makes it useful for the present study
15 and also sets it apart from the previous CISs is the introduction of an optional module on eco- innovation.
Some of the results of the module can be seen in the graph mentioned in the introduction. The graph represents enterprises that introduced innovations with environmental benefits. To be more specific, the graph shows the number of enterprises whose innovations have reduced energy per unit of output. Germany is among the leading countries in innovation and performs just as well in activities for eco-innovation, as can be seen in the graph. A large portion of the companies in Germany are active in the renewable energy field (EIO, 2011, Germany). An interesting appearance is the presence of Portugal in the third place. An explanation might be the R&D spending on environment and energy that have been gradually increasing, notable changes in its energy policy over the last five years (EIO, 2011, Portugal). Moreover, the Eco- Innovation Observatory report (2011) notes that the results obtained by Portugal might be a
„reflection of a market transformation, lead by consumer demand, policy instruments and also by current economic conditions” (EIO, 2011,Portugal, p. 7) or the introduction of the National Law on Waste Management in 2006 (EIO, 2011). The report (EIO, 2011) adds that more tight regulations and economic environment may have induced the development or implementation of more environmental friendly innovations and activities by companies in Portugal. Unusual results are also in the case of Netherlands, typically a country that performs above average in the innovation field while environmental and sustainable topics are well integrated in its national policies. The report (EIO, 2011, Netherlands) mentions that in the case of the previous CIS5 survey (2006), the low score was related to the low response on material and energy reduction innovation activities (EIO, 2011, Netherlands). Furthermore, this could be related to the service sector in Netherlands. The same factors might also explain to some extent the results of Netherlands for the current CIS 2008. The lowest results are obtained by Bulgaria and Poland. In the case of the former, the CIS 2008 has actually registered a decline in the number of enterprises engaging in eco-innovation from the previous survey (EIO, 2011, Bulgaria). In the case of Poland, an analysis related to innovations benefiting the environment was done for the first time in the CIS 2008 (EIO, 2011, Poland). This could suggest the low concern for the environment or for the interest concerning eco-innovation.
Sources do not provide a clear reference for whom the module is optional, either the
enterprises to answer the questions or the countries to include it in their CISs. Nevertheless, I
believe that the module is non-compulsory for companies since the data source provides
16 results on innovation with environmental benefits for all the countries participating in the survey. The non-compulsory aspect of the eco-innovation module can lead to a low response rate from the enterprises, thus not capturing the real situation of eco-innovation. Moreover, low response rates on the optional module show the reduced concern of companies for the health of the natural environment. The same can be said of the countries those companies originate from. A country with solid environmental regulations and policies would attract the attention of firms, making them more aware of this aspect. Thus, in turn, the firms would not ignore the eco-innovation module, despite its optional nature.
By adding the eco-innovation module has resulted the data on the number of enterprises with innovations with environmental benefits. The questions used, clearly link innovation with results for the environment, by asking the respondents if they have introduced an innovation with one or more environmental benefits, what types of drivers (such as current regulations, expected regulations, grants or other financial incentives, expected demand or voluntary codes of practice) stimulated them to introduce environmental innovation, while the last question asks the respondents if any procedures are applied to identify the environmental impacts (Arundel and Kemp, 2009).
The dataset that I will employ in my model consists of 114 observations, without missing values. Six types of environmental benefits are mentioned, that can result during the use of innovation by an enterprise and three types of benefits can result during the use of an innovation by the end user. Two sub questions are provided as to offer clear distinction between the two. The distinction is made as some environmental benefits can take place when the innovation is used within the enterprise, while other environmental benefits occur as the innovation is used by the end user or final consumer. The downside of the survey is the way that the question is formulated. The word “introduction” is used, even for the question dedicated to enterprises. The question makes is difficult to separate between enterprises that have actually produced an innovation and those that only use the innovation, but was generated by another company. This aspect also makes the dependent variables less reliable as an indicator of generated eco-innovation, since it is hard to establish if the number of enterprises with environmental benefits innovations includes the same innovation counted not only in the company that produced it but also in the company that adopt it.
Moreover, the questionnaire provides a binary set of answers to the eco-innovation module,
meaning respondents can select between a ”yes” or ”no” answer. The environmental benefits
17 produced during the use of an innovation are the reduction of material or energy use per unit of output, reduced CO2 by the enterprise, replaced materials with less polluting or hazardous substitutes, reduced soil, water, noise, or air pollution, and recycled waste, water, or materials.
I will attempt to use the results of the CIS 2008 as an indicator of eco-innovation.
The main statistical unit of the survey is the enterprise and the target population is the total population of enterprises of 10 or more employees. Most countries relied on a stratified sample survey, while a few countries used census or a combination of the two. A stratified sample is obtained by dividing a heterogeneous population into sub-populations or non- overlapping groups from which sample are then extracted. The advantage of this sampling method is its higher precision, especially if the strata or groups present great differences.
Moreover, this type of sampling procedure provides better coverage of the population than a simple random sampling, but at the same time it is more difficult to organize it and analyze the results. The subdivisions of the population are the size of the enterprises and their main activity (Eurostat – CIS, 2009). The census procedure refers to a systematic method of acquiring and recording information about populations in different, area, in this case about enterprises.
The data used for the thesis will focus on 19 European countries. For each country, I will have six categories, depending on the environmental benefit, previously presented, that the introduction of a product, process or organizational and marketing innovation has.
4.2. Dependent Variable
The dependent variable of my model will be represented by the number of enterprises that employ innovations with environmental benefits per millions of people, in 19 European countries. According to the results of the CIS 2008 survey, the environmental advantages obtained by innovations are divided in six categories:
• the reduction of material per unit of output
• the reduction of energy per unit of output
• reduced CO2 'footprint' (total CO2 production),
• replaced materials with less polluting or hazardous substitutes,
18
• reduced soil, water, noise, or air pollution, and
• recycled waste, water, or materials.
In addition, in order to consider the size of each country, the number of enterprises will be divided by the population number of each country, measured in millions. The data available for population has been scaled from its original form, making it more convenient to use in the regression analysis (Hill et al, 2000).
The data for the dependent variable was collected from the Eurostat online database.
4.3. Independent Variables
For my model, I chose to represent the independent variables with the cultural dimensions index, that I have previously described, (power distance, individualism, masculinity, uncertainty avoidance) computed by Geert Hofstede (1983, 2001). The national culture index is available for 50 countries and 3 regions. Nonetheless, the focus is on 19 European countries. As Hofstede (2001) stated, culture maintains the same characteristics over large periods of time, thus the cultural dimension index is not available over time. Moreover, as mentioned earlier, Drogendijk and Slangen, (2005) support with their findings the fact that the index remains a reliable assessment of national culture. Consequently the research will be based on a cross-section dataset and, combined with the data from the CIS 2008, I will focus on the year 2008.
The scores for each of the countries of the analysis can be found in Table 5 of the Appendix.
The data is collected from the webpage of Geert Hofstede.
4.4. Control Variables
An appropriate solution to avoid any bias between the parameters estimation, due to the lack of other factors that might have an influence on eco-innovation, is to add control variables.
As mentioned in the literature, Hofstede (2001) has presented high income levels as a high
power distance feature, Norgaard (1994), while Oosterhuis and Brink (2006) explain the
importance of institutions and environmental policies to support eco-innovation. Thus, I
choose to capture these aspects with the following control variables.
19 GDPperCapita
One factor that has an influence on innovation is the available income of countries, expecting high innovation levels for high income countries. Higher income will not only permit higher investments in R&D, but will also allow end consumers to make decisions based on more possibilities. Thus, the population affords to create a demand for environmental friendly products, a need that end users in low income countries would not develop. Hofstede (2001) presented high income levels as prevailing in high power distance countries. Moreover, he discovered that the degree of individualism of a country is related to its wealth: “wealthy countries are more Individualist and poor countries more Collectivist” (Hofstede 1983, p. 81).
The problem with the connection between income and cultural dimensions and the GDP per capita variable is the possible risk of multicollinearity.
The income level will be represented in my model by using world nominal GDP per capita measured in US dollars which will be transformed employing the natural logarithm (lgGDPcapita). The transformation is used to modify extreme values into values closer to a normal distribution and is usually appropriate for positive monetary values. Thus, for this reason I choose to use the natural logarithm of the nominal GDP per capita.
The data was taken from the World Bank webpage.
Environmental Tax Revenues
From the literature previously presented, it is clear that for the production of eco-innovation, environmental policies and instruments have a high degree of importance and influence.
Moreover, the Eco-Innovation Observatory, based on the results of the CIS 2008, stated that
“nearly every fourth (23%) innovating firm in the EU introduced environmental innovation in
response to existing regulations or taxes on pollution” (Eco-Innovation Observatory report,
2011, p. 64). Therefore, my model will capture this aspect by employing the revenues from
environmental taxes, where environmental taxes refer to a tax base on any physical unit (or a
proxy of it) that have a negative impact on the natural environment (OECD Glossary,2007). I
expect environmental taxes to stimulate companies to find environmental friendly
technologies rather than paying taxes.
20 The initial numbers presented by the environmental tax revenues, expressed in millions of Euro, are also modified to take into consideration the size of countries. Thus, I divide environmental tax revenues by millions of people of each country.
As the previous control variable, the environmental tax revenues variable will also be transform with the help of the natural logarithm, since its values are positive and monetary.
The data, extracted from Eurostat online database, will reflect energy taxes, transport taxes and taxes on pollution or resources.
Education and Eco-labeling
In a study on the socio-cultural factors influencing eco-innovation, it has been agreed that potential users of environmental technologies should be aware of the costs and benefits of such products. Moreover, they explain that users should be informed and educated in order to understand how their actions are linked to natural resources and the environment. One example given on the methods of informing the public is the use of eco-labels on products in order to show their quality (Europe Innova report, 2008). In addition, given the technological aspect behind eco-innovation and behind more complex equipments or innovations, I think that users should show interest beyond recognizing or understand eco-labeling. They should also be interested in how eco-innovation or technologies with environmental benefits work.
This would create higher and more challenging demands for eco-innovation on enterprises.
The demand of the market is an actual determinant of eco-innovation. The Eco-innovation Observatory concluded, based on the results of CIS 2008, that 16% of firms introduced an environmental innovation as a response to the “current or expected market demand for environmental innovations from the customers” ( Eco-innovation Observatory, 2011, p. 64).
I choose to represent the education variable by using the number of students of tertiary level of each country, divided by 1000. The chances of users being interested in the technical or more complex aspects of environmental friendly technologies or products are higher than among those with secondary education.
The data for eco-labeling provides the number of eco-label licenses per country. This form of
labeling is assigned to products or services with reduced environmental impacts. Based on
this data I create a dummy variable, where 1 will represent the presence of eco-labels licenses
in a given countries, no matter the number, while 0 represents the non-existence of eco-labels.
21 Even if in some countries the number of eco-label licenses is really low, the attempt towards informing consumer exists. Education and access to environmental information should complement each other. If there is a high education level, but no attempt to inform consumers, then it hardly stimulates eco-innovation since the population is not aware of environmental friendly products and technologies. On the other hand, if eco-labeling exists, or the attempt to inform consumers is there, but the education level is low, the population cannot fully understand the benefits and implication of technologies or products that are not harmful to the environment. Thus, to capture this aspect, I employ an interaction variable based on education and eco-labeling.
Both data can be found in the online database of Eurostat.
4.5. The econometric model
As presented, the model is based on a cross-section dataset, consisting of a basic regression composed of the four independent variables mentioned above. Later on the model is also tested under the effects of the control variables.
The dataset that I will use consists of 19 European countries, each having six different categories for the dependent variables, according to the environmental benefits that I already explained, while the observations for the independent variables will remain the same. The dataset will contain 114 observations, with no missing values.
Firstly, I will employ the natural logarithm to transform the GDP per capita variable as well as the environmental tax revenue variable, since their values are both positive and in monetary values. By using this method, the more extreme values will be closer to a normal distribution.
The model is tested to establish if heteroskedasticity is present or not. In other words, it is important to observe if all observations have the same variance or if they differ, thus disregarding one of the least squares assumptions (Hill et al, 2008). I test the model by applying the White test for heteroskedasticity. If the results of the test reject the null hypothesis of homoskedasticity, I can conclude that my model violates the assumption of constant variance and a measure to correct it should be applied.
The heteroskedasticity problem can be solved by using the robust standard errors, as the errors
become valid in large samples for homoskedastic as well as heteroskedastic errors (Hill et al,
2008). The estimator for the standard errors can help in eliminating the problem of computing
22 incorrect interval estimates or incorrect values for test statistics, if heteroskedasticity exists.
The down side is that the standard errors do not attendant to one of the implications of heteroskedasticity, meaning that the least squares estimator is no longer best.
After testing for heteroskedasticity, I can conclude that two of regressions of my model do not suffer of heteroskedasticity, while the third one rejects the null hypothesis of homoskedasticity. The outcomes of the test can be observed in the Appendix, Table 8.
Also, I intend to employ a test for normality in order to asses if the regression errors of my model are normally distributed. In order to do so, I will use the automatic normality test of Stata 1 (sktest), which is similar to the Jarque-Bera test (Adkins and Hill, 2008). The outcomes of the test show that the normality of the regression errors can be a problem. This difficulty can be minimized by employing the robust standard errors. The results can be found in the Appendix, Table 8.
Further on, I test the model for multicollinearity, following the model of Erumban and de Jong (2006) described in the literature section, who found this problem as a consequence of the power distance and individualism variables highly correlated with the GDP per capita control variable. In order to detect the existence of multicollinearity, or an “approximate linear relationship among some of the regressors” (Kenney, 1998, p. 184), I make use of the correlation matrix, a correlation coefficients between all pairs of the explanatory variables (Kennedy, 1998). The problem posed by multicollinearity is not that it violates one of the least squares assumptions, but the fact that the data might not contain enough information about the individual effects of the independent variables (Hill et al, 2008).
As a guide to assess which model is the most appropriate for my analysis, I can test each of them in order to detect the presence of any misspecifications such as omitted variables, the inclusion of irrelevant ones, inappropriate functional form or violation of any of the multiple regression model assumptions. I choose to apply the RESET test (Regression Specification Error Test), which is designed to detect omitted variables and incorrect functional form. The basic idea of the test is that if it concludes that the original model is inadequate, the model can be improved by artificially introducing powers of the predictions of the model. If the null hypothesis of no misspecification is rejected at a 5% significance level, then I can improve my model by adding omitted variables, thus better explaining the dependent variables or
1
