Going Green: A Meta-Analysis on the
Effects of Green Product Innovation
Master thesis, MSc Supply Chain Management
2
Abstract
The need to reduce the burden on the environment is prevalent in business and society. As a consequence, green product innovation (GPI) is progressively grasping the attention of practitioners and researchers. Thus far, no consensus exists on how to conceptualize and operationalize GPI, leading to an immense diversity in definitions and measures. Furthermore, conflicting views exist on the relationship between GPI and firm performance that need to be resolved. Some studies found clear positive effects, whereas others revealed insignificant or even negative effects, leaving organizations puzzled with what actions to pursue. This prescribes the need to synthesize existing knowledge in a quantitative way. By means of a meta-analysis this study researches the relationship between GPI and firm performance. GPI is categorized into technological, marketing, internal integrative and external integrative capabilities. In total, 88 studies are included that show an overall clear positive relationship between GPI and firm performance. More specifically, it can be beneficial for an organization to invest in technological, internal integrative and external integrative capabilities. This can lead to increased business, operational and environmental performance. The amount of studies investigating marketing capabilities and reputation as performance measure was not sufficient to confirm positive relationships. Future research is therefore advised to devote more attention to these aspects. Overall, the results lead to two main contributions of this paper. First of all, a GPI framework is substantiated containing first and second order constructs. Secondly, a clear positive relationship between GPI and firm performance was found. These insights can help to evolve the GPI research field, and help organizations going green.
3
Content
Abstract ... 2 1. Introduction ... 4 2. Theoretical background ... 5 2.1 Defining GPI ... 5 2.2 Dimensions of GPI ... 7 2.2.1 Functional capabilities ... 7 2.2.2 Integrative capabilities ... 82.3 GPI and firm performance ... 9
2.3.1 Business performance ... 9 2.3.2 Operational performance ... 10 2.3.3 Reputation ... 10 2.3.4 Environmental performance ... 11 2.4 Research framework ... 11 3. Methodology ... 12 3.1 Research method ... 12 3.2 Sample selection... 13 3.3 Data collection ... 15 3.4 Coding ... 15 3.5 Meta-analytical procedures ... 16 4. Results ... 17 4.1 Descriptives ... 17 4.2 Main effects ... 18 4.3 Moderating effects ... 20 5. Discussion ... 20 5.1 Theoretical implications ... 20 5.2 Managerial implications ... 22
5.3 Limitations and future research ... 23
6. Conclusion ... 23
References ... 25
Appendix A: Coding scheme ... 33
4
1. Introduction
The necessity of reducing the burden on the natural environment is undebatable. Along with other green initiatives, the development of green products plays a vital role in ensuring the health of future generations (Robinson & Stubberud, 2015). As a consequence, scholars, businesses and policy makers have directed more and more attention towards the field of green product innovation (GPI). Going green is surely beneficial from an altruistic perspective, but businesses need to survive in the market as well. This leaves organizations plagued with two tenacious questions, namely how to go green, and whether it pays off.
As with any research field in development, an ongoing debate takes place on what GPI specifically entails. As a result, a plethora of different definitions of and perspectives on GPI have been applied, as well as GPI operationalizations (De Medeiros, Ribeiro, & Cortimiglia, 2014). It appears no established framework yet exists to guide research. Even though this research domain is relatively new, it grows exponentially (Dangelico, 2015), and with it the need for a unified framework based on systematic reviews.
Based on a literature review, Dangelico (2015) argues that conventional product innovation theory can provide a fruitful basis for a GPI framework. Although additional complexities underpin the GPI phenomenon, there is wide support that it is not radically different from conventional product innovation (Berchicci & Bodewes, 2005; Dangelico, Pontrandolfo, & Pujari, 2013; Driessen, Hillebrand, Kok, & Verhallen, 2013; Huang & Wu, 2010; Wong, 2012). The framework Dangelico (2015) proposes incorporates four types of capabilities crucial for GPI, namely technological, marketing, internal integrative and external integrative capabilities (based on Verona, 1999). This research attempts to substantiate this framework as a starting point for consensus in future GPI research.
Apart from the confusion surrounding the definition of GPI, uncertainty prevails as well when researching its effects on firm performance. Traditionally, going green is seen as a burden rather than an opportunity. Ever since Porter and Van der Linde (1995) first propagated a win-win perspective for organizations that green and financial gains can go hand in hand, research has attempted to support this hypothesis with mixed results. Some studies show positive gains resulting from GPI (Chang, 2011; Dangelico et al., 2013; Jackson, Gopalakrishna-Remani, Mishra, & Napier, 2016; Li, Jayaraman, Paulraj, & Shang, 2016; Zailani, Eltayeb, Hsu, & Tan, 2012), whilst others found insignificant results (González-Benito & González-(González-Benito, 2005; Leenders & Chandra, 2013), or even negative effects (Driessen et al., 2013; Wong, Lai, Shang, Lu, & Leung, 2012). There is a strong need for synthesizing these contradictory findings, as also suggested by Dangelico (2015) and De Medeiros, Ribeiro and Cortimiglia (2014).
5
generalizability can be obtained (Orlitzky & Benjamin, 2001). So far, however, a meta-analysis of GPI has remained absent.
The aims of this paper are to validate the GPI framework, as proposed by Dangelico (2015), and shed light on the true effect size of the different GPI capabilities on firm performance. This will be done via a meta-analysis. Similar to GPI, a wide variety of measures is used for firm performance. For the purpose of this research, these are categorized into business, operational, reputation and environmental performance. The research questions addressed in this thesis are: 1) What is the overall relationship between GPI and firm performance? 2) Is this relationship different across different GPI capabilities? And 3) Is this relationship different across different performance dimensions?
By answering these research questions, relevant contributions can be made to both the academic literature and the managerial world. First of all, this research is the first to reconcile knowledge on GPI by means of a meta-analytic study. This addresses a serious research gap, as meta-analyses are essential for compiling knowledge derived from multiple independent studies (Golicic & Smith, 2013). Secondly, this research seeks to substantiate the framework suggested by Dangelico (2015). Consensus on a GPI framework is crucial for further development in this research domain. Thirdly, our understanding of the relationship between GPI and firm performance will be enlightened. A positive significant result should provide managers with sufficient confidence to invest in GPI capabilities. The categorization of different GPI capabilities and performance dimensions also allows examination of more specific relationships. Doing this in a meta-analysis leads to a deeper understanding of the relationships (Golicic & Smith, 2013; Leuschner, Rogers, & Charvet, 2012). This way, the most relevant GPI capabilities can be identified, as well as clear expectations on the performance gains these can yield.
This paper is structured as follows. The next section discusses the theoretical background, leading to the development of hypotheses. Thereafter, an in-depth description follows of the research steps and methodology. Subsequently, the results of the meta-analysis are presented and interpreted. In the discussion, an elaboration follows on the theoretical and managerial implications of these results, as well as the research limitations. Finally, the conclusion reviews the key findings of this research.
2. Theoretical background
2.1 Defining GPI
Innovations are most commonly divided into product, process and organizational innovations (Rennings, 2000). GPI focuses on product innovations, which refer to “improvements to existing goods or the development of new goods or services” (Rennings, 2000: 322). When defining GPI, high consensus exists in the field on the meaning of the ‘product innovation’ aspect. However, the question how to incorporate the ‘green aspect’, is highly debated (Katsikeas, Leonidou, & Zeriti, 2016). As a consequence, many different definitions circulate in literature. In this section the scope and perspective of GPI in this study will be clarified on four matters, leading up to our definition.
6
environmental, social and economic aspects, also known as the triple bottom line (Kleindorfer, Singhal, & Van Wassenhove, 2005). GPI specifically focuses on the environmental element (Albino, Balice, & Dangelico, 2009). Green products are those that “use less resources, have lower impacts and risks to the environment and prevent waste generation already at the conception stage” (Commission of the European Communities, 2001).
Secondly, the distinction between green products and services deserves attention. The vast majority of studies researches GPI in a manufacturing context, largely neglecting the service industry (De Medeiros et al., 2014). However, in industrialized countries the provision of services has become more prevalent than products (Callaway & Dobrzykowski, 2009). Similar to physical products, services consume resources that can harm the environment (Leonidou, Leonidou, Fotiadis, & Aykol, 2015; Wong, Wong, & Boon-itt, 2013). Therefore, both products and services are inside the scope of GPI in this research. Third, we acknowledge that knowledge from conventional product innovation can be beneficial for our understanding of GPI. In a sense, GPI is not radically different from conventional product innovation (Dangelico et al., 2013; Driessen et al., 2013; Huang & Wu, 2010; Wong, 2012). Rather, it adds a layer of complexity (Huang & Wu, 2010; Pujari, Wright, & Peattie, 2003) by posing additional challenges to integrate environmental issues in the innovation process (Berchicci & Bodewes, 2005; Chen & Chang, 2013). New products need to be both “new” and “green” enough to be competitive (Wong, 2012). Fourth, in general, a division can be made between GPI definitions focused on the process or the results of GPI. Authors adopting the process perspective consider GPI as a distinctive capability within the firm (Chen & Chang, 2013; Dangelico et al., 2013; González-Benito & González-Benito, 2005; Hartmann & Germain, 2015; Katsikeas et al., 2016). A capability comprises a “complex bundle of skills and accumulated knowledge, exercised through organizational processes” which enables a firm to accomplish a given task (Day, 1994: 38). According to Danneels (2002), the two main tasks of product innovation are those relating to technology and those relating to customers. Basically, organizations need be capable to both make and sell a product innovation. The process perspective is in line with the Resource-Based View (RBV) and the Natural Resource-Based View (NRBV) that see unique resources and capabilities as the foundation for achieving a competitive advantage (see Section 2.3). In contrast, result-oriented authors suggest the actual ‘greenness’ of a product defines the level of GPI in an organization, such as the percentage of green patents (Aguilera-Caracuel & Ortiz-de-Mandojana, 2013; Driessen et al., 2013; Lin, Tan, & Geng, 2013; Triebswetter & Wackerbauer, 2008). In fact, this last interpretation is rather a performance measure than a true reflection of the complexities underpinning the GPI phenomenon.
7
Organizational capabilities regarding the making and selling of new or improved products or services with a reduced environmental impact.
2.2 Dimensions of GPI
The above definition of GPI illustrates a rather broad perspective. This would imply that different types of capabilities inherent to GPI can be distinguished. Since consensus on a GPI framework is missing, knowledge from conventional product innovation might provide a fruitful start for identifying these dimensions. A well-established framework in conventional product innovation is developed by Verona (1999). He distinguishes between four types of capabilities that an organization should master for effective innovation. The technological and marketing dimensions represent the required functional capabilities, whereas the internal integrative and external integrative dimensions embody integrative capabilities. Dangelico (2015) proposed an adaptation of this framework to a GPI context. This research attempts to substantiate this framework as a starting point for consensus in future research. In addition, the item-level inspection of GPI constructs performed in this research allows further adaptation and improvement of the framework. The dimensions or first order constructs are visualized in Figure 1. The second order constructs can be found in Appendix A.
2.2.1 Functional capabilities
The four capabilities focus on different dimensions of GPI. More specifically, the functional capabilities distinguish between the ability to design and manufacture a product (technological), and the ability to serve certain customers (marketing) (Danneels, 2002). Although one might expect these two functions to go hand in hand when introducing a new product, literature so far predominantly focused on technological capabilities. As a matter of fact, constructs labelled by the authors themselves as GPI or a
FIGURE 1 GPI framework
8
synonym were almost exclusively addressing technological capabilities. Indeed, GPI requires considerable knowledge about technology, materials and how to reduce the burden on the environment (Hartmann & Germain, 2015). Guiding processes in green product design are Design-For-Environment and Life-Cycle Analyses (Pujari, 2006). Essential is to take the complete product life cycle into account, ranging from raw materials acquisition, to parts processing, to production, to usage, and to recycling or disposal (Dangelico & Pujari, 2010).
In contrast, whilst many studies speculate about the impact of marketing capabilities on firm performance, few actually empirically examine this (Fraj-Andrés, Martinez-Salinas, & Matute-Vallejo, 2009; Leonidou, Katsikeas, & Morgan, 2013). This is remarkable, since understanding the target market and their needs is crucial for minimizing the risk of product failure (Pujari, 2006). A recent meta-analysis on conventional product innovation even concluded that marketing capabilities are relatively more important than technological capabilities (Eisend, Evanschitzky, & Calantone, 2016). This research will clarify whether this finding also holds in a GPI context. Richey, Musgrove, Gillison, & Gabler (2014) explain research so far focuses more on the demand-side than on the supply-side perspective of green marketing. By this they mean an emerging understanding of customer behavior, but a lack of insights in the strategic viability of marketing capabilities.
2.2.2 Integrative capabilities
Integrative capabilities distinguish between integration within (internal) and outside (external) the boundaries of the firm. This means organization, absorption and blending of functional knowledge amongst departments and organizations (Verona, 1999). Effective integration requires a long-term perspective on relationships, instead of a transactional view (Hartmann & Germain, 2015). Within the firm, the importance of cross-functional coordination is often stressed, especially for complex processes such as life-cycle analyses (Pujari, 2006). Multiple functional departments are involved in GPI, the most important ones being R&D, marketing and manufacturing (Genç & Di Benedetto, 2015). Ideally, green norms and values are deeply embedded in the culture of an organization (Leonidou, Fotiadis, Christodoulides, Spyropoulou, & Katsikeas, 2015). Clear green policies (Lirn, Lin, & Shang, 2014; Pujari, 2006), top management support (Huang, Hu, Liu, Yu, & Yu, 2016; Pujari et al., 2003; Roy & Khastagir, 2016) and performance systems (Tung, Baird, & Schoch, 2014; Zhu, Sarkis, & Lai, 2012) can foster such a green orientation.
9
Huang et al., 2016). More focus on these stakeholders would be beneficial, as GPI is faced with high market and technological uncertainty that cooperation with multiple actors might mitigate (Huang et al., 2016).
2.3 GPI and firm performance
Nowadays, most developed and developing countries have implemented regulations for being green (Lee, Rha, Choi, & Noh, 2013). Although this pressure leads to higher adoption of GPI (Chan, Yee, Dai, & Lim, 2015), many organizations do not go beyond compliance to the rules. Going green is traditionally seen as a burden, rather than an opportunity. Porter and Van der Linde (1995) were among the first to formulate a theory to break this pattern by propagating a win-win perspective. They postulate that innovations can make green and financial gains go hand in hand and lead to a competitive advantage. In order to achieve that, proactive GPI is substantially more effective than reactive GPI (Chen, Chang, & Wu, 2012; Li et al., 2016). This vision is backed up by the NRBV developed by Hart (1995). He extended the RBV with a natural component. The RBV theory predicts that resources and capabilities that are firm-specific, rare, non-imitate able and difficult to substitute can create a competitive advantage (Barney, 1991). This only occurs when these are synergistically combined (Katsikeas et al., 2016) and become deeply embedded in organizational processes and routines (Lynch, Keller, & Ozment, 2000). By introducing the NRBV theory, Hart (1995) pointed out that firms are constrained by the natural environment. A competitive advantage can be achieved by developing unique capabilities that reduce the firm’s burden on the environment. Important in this sense are preemptive commitments, such as setting new standards, gaining access to customers, locations or critical raw materials. This provides an organization with the focus necessary to develop either a cost leadership or differentiation advantage relative to competitors (as developed by Porter, 1980).
In essence, the NRBV shows the potential of GPI to have a positive effect on firm performance. Yet, surprisingly, no conclusive evidence has been found for this hypothesis. Conflicting findings are reported in research, leaving managers puzzled with what actions to pursue. Some studies show positive gains resulting from GPI (Chang, 2011; Dangelico et al., 2013; Jackson et al., 2016; Li et al., 2016; Zailani et al., 2012), whilst others found insignificant results (González-Benito & González-Benito, 2005; Leenders & Chandra, 2013), or even negative effects (Driessen et al., 2013; Wong et al., 2012). On top of that, authors have used a plethora of different operationalizations of firm performance. Therefore, based on an evaluation of the papers included in the meta-analysis, four main categories were detected to guide this research. These constitute business, operational, reputation and environmental performance. The first two performance dimensions are frequently used in supply chain and operations research (Golicic & Smith, 2013; Leuschner et al., 2012). The latter two are tailored to the GPI research field.
2.3.1 Business performance
10
always that developed yet. Wong et al. (2012) conclude that the financial investments and costs associated with setting up GPI are too high to be able to reap financial benefits. Moreover, they state that environmental practices are not always visible for customers, which does not inspire purchasing a green product. This would suggest that customers need to be properly informed about the greenness of a product, or stimulated otherwise to buy such a product. The effects of GPI on business performance are not immediately visible for organizations, since it requires time to overcome the initial investments (González-Benito & González-Benito, 2005). Yet, organizations with a long-term perspective may reap high financial gains. In line with the NRBV, we posit that business performance can be achieved if the right capabilities are in place. This hypothesis is also supported by a vast amount of research (Chang, 2011; Chiou, Kai, Lettice, & Ho, 2011; Graham & Potter, 2015; Hollos, Blome, & Foerstl, 2012; Kirchoff, Tate, & Mollenkopf, 2016; Leonidou, Fotiadis, et al., 2015; López-Gamero, Molina-Azorín, & Claver-Cortés, 2009; Walker, Ni, & Huo, 2014). Therefore, we do expect a positive relationship between GPI capabilities and business performance.
2.3.2 Operational performance
Operational performance refers to indicators as speed, quality, flexibility and costs. The effects of GPI on operational performance are somewhat debated too, which is mostly due to the cost-benefit trade-off (Kirchoff et al., 2016). Obviously, GPI requires initial investments that lead to an increase in costs (Liu, Dai, & Cheng, 2011). In addition, environmental practices need additional operations that may not be optimal in terms of cost and time (González-Benito & González-Benito, 2005). However, on the other hand GPI poses many advantages. First of all, GPI can lead to efficient use of resources and materials and a reduction of waste (Fraj-Andrés et al., 2009). It can also prevent environmental fees or fines (Jackson et al., 2016). Moreover, GPI increases conformance to specifications and durability of products (Yu et al., 2014). Contrary to the long suffered stigma that green products are of low quality, Gabler, Richey, & Rapp (2015) found them to be actually of better quality than their non-green alternatives. Hence, quality increases and cost decreases can simultaneously be obtained (Sroufe, 2003; Zailani et al., 2012). These arguments lead us to expect a positive relationship between GPI and operational performance. In addition to the direct positive effects, some authors found a mediating effect of operational performance on the relationship between GPI and business performance (Fraj-Andrés et al., 2009; Vijayvargy & Agarwal, 2014). Efficiency gains and cost reductions can enhance business performance as well.
2.3.3 Reputation
11
environment to stakeholders (Chen, Lai, & Wen, 2006; Fraj-Andrés et al., 2009). Reputation is not a typical performance measure of product innovation, but it can be an end-goal in itself in a green context. Although it may not have immediate financial payoffs (Driessen et al., 2013), it has the potential to lead to higher sales and increased market share (Eltayeb et al., 2011; Fraj-Andrés et al., 2009). Many authors detected a mediating effect of reputation on the relationship between GPI and business performance (Amores-Salvadó, Martín-de Castro, & Navas-López, 2014; Feng & Wang, 2014; Fraj-Andrés et al., 2009; Peng & Lin, 2008; Wu & Lin, 2016). A better reputation can lead to higher sales and satisfied customers, ultimately increasing business performance as well.
2.3.4 Environmental performance
Environmental performance is measured by the actual impact on the environment of products and services and the processes of making, using and disposing them. Examples are consumption of (hazardous) materials, air emission or prevention of environmental accidents. Many studies researched the impact of GPI on environmental performance, to see which capabilities need to be in place to actually reduce environmental impact. Their findings are rather consistent, as the vast majority detected a strong positive effect (Böttcher & Müller, 2015; Chen, Tang, & Jin, 2015; Choi & Hwang, 2015; De Sousa Jabbour, Jabbour, Latan, Teixeira, & De Oliveira, 2014; Huang & Li, 2015; Yang, Hong, & Modi, 2011; Zailani et al., 2012). This would imply the first part of the Porter and Van der Linde (1995) hypothesis is supported. From a pure altruistic point of view, one could say every organization should invest heavily in GPI. However, the question remains whether GPI also leads to advantages for the firm.
2.4 Research framework
Based on the literature review in the previous sections, a positive relationship between GPI capabilities and firm performance is expected. GPI capabilities have been further divided into technological, marketing, internal integrative and external integrative capabilities. These four dimensions represent different types of capabilities that are all expected to be positively related to firm performance. Firm
12
performance has been categorized as business, operational, reputation and environmental performance. GPI is expected to be positively related to all performance dimensions. Furthermore, the presence of three potential moderators is tested: industry, location and year. The effect of the industry is interesting, because of the predominant focus in GPI research on the manufacturing over the service industry (De Medeiros et al., 2014). The question remains whether differences exist in the effectiveness of GPI amongst industry type that justify this emphasis on manufacturing firms. In addition, the effect of location is of interest, because several authors stressed the importance of regional factors in adoption of GPI (Choi & Hwang, 2015; Peng & Lin, 2008; Zailani et al., 2012). The third moderator constitutes the year. It would be interesting to see if organizations have become better throughout the years in yielding performance gains from GPI. All relationships can be visually depicted as done in the research framework of Figure 2. Furthermore, Table 1 lists the three hypotheses that will be tested in this research.
TABLE 1
Overview of hypotheses
Hypothesis 1 GPI capabilities are positively related to firm performance
Hypothesis 2a Technological capabilities are positively related to firm performance
Hypothesis 2b Marketing capabilities are positively related to firm performance
Hypothesis 2c Internal integrative capabilities are positively related to firm performance
Hypothesis 2d External integrative capabilities are positively related to firm performance
Hypothesis 3a GPI capabilities are positively related to business performance
Hypothesis 3b GPI capabilities are positively related to operational performance
Hypothesis 3c GPI capabilities are positively related to reputation
Hypothesis 3d GPI capabilities are positively related to environmental performance
3. Methodology
3.1 Research method
13
estimated and the consistency of effects can be reviewed. More specifically, a meta-analysis combines the effects derived from multiple independent samples (Golicic & Smith, 2013). The fact that all samples are aggregated into one, means individual study biases can be largely accounted for. Goldsby & Autry (2011) elaborate on this and argue that a meta-analysis can reduce measurement and sampling error. Consequently, stronger conclusions and empirical generalizability can be derived from a meta-analysis (Orlitzky & Benjamin, 2001). In addition, the presence of moderators can be detected. Based on these arguments, a meta-analysis is deemed to be the most suitable research method for answering the research questions posed in this paper. By synthesizing data of existing surveys, the generic relationship between GPI and firm performance can be researched in order to resolve conflicting findings.
3.2 Sample selection
The literature search took place between February and April 2016 in three sequential steps. First, we started searching in the database of EBSCO Business Source Premier. The search was limited to peer-reviewed academic journals, but without further restrictions. Each search term entailed at least one “green” keyword (green, environmental, eco, ecological, sustainable) combined with a “practice” keyword ((product) innovation, product development, product design, new product, management, marketing, operations, supply chain management, performance). Different combinations of keywords were entered as search terms that needed to be present in the abstract of the article. This search produced an unfiltered database of 379 papers. Second, backward and forward searching of literature reviews covering GPI (Dangelico, 2015; De Medeiros et al., 2014) was used to detect articles that were either not present in EBSCO Business Source Premier or not found by the search terms. Four additional articles were found. Third, this snowballing approach, as in backward and forward searching, was applied to 36 other key articles present in the database. This process yielded another 87 articles, which totals an initial database of 470 articles.
Publication bias, also referred to as the file-drawer problem, is a common pitfall of meta-analyses (Field & Gillett, 2010). This problem originates due to the fact that significant studies are substantially more likely to be published than non-significant ones (Borenstein et al., 2009). As a consequence, the sample of a meta-analysis can be biased and possibly overestimates the true effect size (Borenstein et al., 2009). In order to counter the publication bias, relevant papers only published online or in conference proceedings were also included in the database. By including yet unpublished work, a better representation of the population is provided. To further address the risk of publication bias, the failsafe number was calculated. The failsafe number represents the amount of additional studies averaging null results that would have to be added to make the effect barely insignificant (p=0.05) (Rosenthal, 1979). Higher failsafe numbers proclaim to have a smaller risk of publication bias. The calculations can be found in Section 3.5.
14
exact items, but referred to other articles for the measures. Whenever we were not convinced about the items used to measure a construct, we excluded the article from the sample. Furthermore, the unit of analysis had to comprise the firm or project level, with a sample size higher than 40, retrievable data and an original dataset.
These inclusion criteria were step-wise applied to the original database of 470 articles. First, 283 papers that did not conduct empirical survey research were deleted. Next, 62 papers were removed because they did not investigate an effect on firm performance, or used items to measure firm performance that did not fit our definition. Subsequently, 28 of the remaining papers were excluded that did not investigate an effect of GPI, or used items to measure GPI that did not fit our definition. As a last step, 11 additional papers were deleted from the sample, since they contained an individual unit of analysis (2x); a sample size lower than 40 (1x); missing data that could not be retrieved (1x); or articles using the same dataset (5x) as or a subset of the dataset (2x) of another article, without adding new variables. This totals 86 articles that were included in the sample. A visual summary of the sample selection is presented in Figure 3.
Each sample is treated as a separate study, to maintain the assumption of independence among correlations (Hunter & Schmidt, 2004). All articles sharing one or more authors with another article were assessed for overlap in datasets, in order to ensure no sample was used twice. Whenever this was the case, we also reviewed whether the articles researched the exact same variables, or provided complementary information. As mentioned above, seven articles were deleted from the sample since they did not add new variables and therefore no new information. However, six other articles used the same dataset as another article, but did add new variables. Therefore, it was decided to combine these six articles into three studies (Chavez, Yu, Feng, & Wiengarten, 2016; Eltayeb et al., 2011; Wong, 2013, 2012; Yu et al., 2014; Zailani et al., 2012). Furthermore, four articles investigated multiple independent samples, which were split into nine separate studies (Jackson et al., 2016; Kuei, Chow, Madu, & Wu,
Literature search (470) Initial database built up by searching EBSCO Business Source Premier and snowballing of literature reviews and key
articles Survey research method (187) Exclusion of articles using qualitative research methods or explorative/ descriptive survey research Firm performance (125) Exclusion of articles that did
not have firm performance as
dependent variable, or did
not fit our definition
Green product innovation (97)
Exclusion of articles that did
not have green product innovation as independent variable, or did
not fit our definition
Usability (86)
Exclusion of articles that did
15
2013; Kuei, Madu, Chow, & Chen, 2015; López-Gamero et al., 2009). This totals 88 studies with a total sample size of 18,721 firms.
3.3 Data collection
Characteristics of the study, the constructs and the relationships were collected and assembled in a spreadsheet. Study characteristics include year, authors, journal, unit of analysis, industry, respondent types and sample size. Construct characteristics include the construct names, the items used in the survey to measure the constructs and the reliability coefficient. As reliability coefficient the Cronbach’s Alpha was chosen. Relationship characteristics include the effect sizes between the constructs. An effect size quantifies the relation between two variables (Borenstein et al., 2009). Multiple measures exist, but the most commonly used and recommended is the Pearson’s correlation coefficient (Field & Gillett, 2010). This measure will also be used in this paper. In order to guarantee a transparent chain of evidence, all information subtracted from the articles was coded in Atlas.ti.
Several studies did not report Cronbach’s Alpha values, correlation coefficients, or both. If present, the Composite Reliability was used as a substitute for the Cronbach’s Alpha. Although these measurements tend to yield somewhat higher values, the differences are negligible for a meta-analysis (Peterson & Kim, 2013). Whenever both values were missing, the generally accepted minimum Cronbach’s Alpha of 0.7 was assumed. Single item constructs were assigned a value of 1. As a first attempt to retrieve missing correlation coefficients, the corresponding authors of the articles were contacted with a request to send the correlation matrix. Some missing data could be retrieved by this approach, but other studies still required the use of substitute values. For articles reporting regression analyses, this value constitutes the standardized beta coefficient. This generally results in fairly accurate effect-size estimates, and is preferred over the alternative of excluding the study from the meta-analysis (Peterson & Brown, 2005). For articles reporting structural equation modelling, the correlation coefficients were traced back by summing the product of all paths directly or indirectly linking the variables (Kenny, 1979: 30).
3.4 Coding
All constructs were coded with either one of the four GPI dimensions, one of the four firm performance dimensions, or excluded from the study. According to the research framework, GPI was divided into technological, marketing, internal integrative and external integrative capabilities. Firm performance was divided into business, operational, reputation and environmental performance. First, a trial round was conducted to test the initial coding scheme (based on Dangelico, 2015). Inspection of a subset of the items made clear this scheme was not complete. Debates amongst the authors resulted in a more extensive coding scheme which is presented in appendix A. In this scheme it is also visible which constructs were considered to be irrelevant and outside the scope of this study. Amongst others, this entails non-green or non-product innovation practices such as (green) logistics, purchasing or process innovation.
16
only included for testing hypothesis 1. The coding was done in three rounds with discussion sessions in between to resolve disagreements. This process yielded an inter-rater reliability of 92%. In Appendix B an overview of the codes, sample size and average uncorrected effect sizes of all studies is provided.
3.5 Meta-analytical procedures
Two different ways to conceptualize a meta-analysis are fixed-effects and random-effects models (Field & Gillett, 2010). Fixed-effect models assume that all studies included in a meta-analysis use samples from the same population with a homogenous effect size. Random-effects models assume that the samples are derived from different populations with heterogeneous effect sizes. Although both models are being used by scholars, it appears that fixed-effects are rarely, if ever, a sound assumption (Hunter & Schmidt, 2000). The field of GPI is no exception in this matter, and therefore a random-effects model will be used. The two most commonly used meta-analyses methods considering random-effects are developed by Hunter and Schmidt (2004) and Hedges (Hedges & Olkin, 1985; Hedges & Vevea, 1998). Researchers comparing these methods through a Monte-Carlo simulation, concluded that the Hunter and Schmidt method (2004) most accurately yields estimates of the population correlation (Field, 2001; Hall & Brannick, 2002). In addition, this method has been the choice in the field of operations management (Mackelprang & Nair, 2010; Nair, 2006), supply chain management (Leuschner et al., 2012) and product innovation (Chen, Damanpour, & Reilly, 2010). Therefore, the Hunter and Schmidt method (2004) was used in this study as well.
Some studies included multiple constructs that were coded the same. If this was the case, both the correlation and reliability coefficients were averaged to obtain one value per study. The process steps and related calculations of the meta-analysis are presented in Table 2. In short, steps 1-4 correct the effect size for measurement error, steps 5-10 detect the potential presence of moderators and step 11 tests the reliability. First, these calculations were applied to test the main effects. Second, these calculations were applied to test the moderating effects.
TABLE 2
Meta-analytical calculations
Steps Formula Variables
1. Mean attenuated
effect size (𝑟̅) 𝑟̅ =
∑𝑘𝑖=1𝑛𝑖𝑟𝑜𝑖 ∑𝑘𝑖=1𝑛𝑖
𝑛𝑖 =sample size of study i
𝑟𝑜𝑖 =observed correlation coefficient of study i 2. Attenuation factor (𝐴𝑖) 𝐴𝑖 = √𝑎𝑥𝑖√𝑎𝑦𝑖 𝑎𝑥 =reliability coefficient GPI of study i
𝑎𝑦=reliability coefficient performance of study i 3. Corrected effect size
(𝑟𝑐𝑖)
𝑟𝑐𝑖 = 𝑟𝑜𝑖 𝐴𝑖
𝑟𝑜𝑖 =observed correlation coefficient of study i 𝐴𝑖 =attenuation factor
4. Mean corrected effect
size (𝑟̅𝑐) 𝑟̅𝑐 =
∑𝑘𝑖=1𝑛𝑖𝐴𝑖2𝑟𝑐𝑖 ∑𝑘𝑖=1𝑛𝑖𝐴2𝑖
𝑛𝑖 =sample size of study i 𝐴𝑖2=attenuation factor squared 𝑟𝑐𝑖 =corrected effect size of study i 5. Sampling error
variance (𝑉𝑖) 𝑉𝑖 =
(1 − (𝑟)̅2)2 𝑛𝑖− 1
𝑟̅ =mean attenuated effect size 𝑛𝑖 =sample size of study i 6. Sample-size weighted
average of sampling error variance (𝑉) 𝑉 =∑ 𝑛𝑖𝐴𝑖 2𝑉 𝑖 𝑘 𝑖=1 ∑𝑘𝑖=1𝑛𝑖𝐴𝑖2
17
4. Results
4.1 Descriptives
The 86 articles in the sample selection stem from a variety of journals. Most were published in Journal of Cleaner Production (20%), followed by Journal of Business Ethics (10%), International Journal of Production Economics (9%) and Business Strategy and the Environment (6%). The remaining articles were selected from 32 different journals, which can be found in Appendix B. Three articles are published conference proceedings. The oldest articles were published in 2003, and the most recent ones in 2016. Figure 4 visualizes the articles selected per year. A clear trend is visible, showing GPI exponentially received more attention by scholars from 2011 till 2016. The drop in 2016 is due to the fact data collection finished in April 2016. Based on this visualization, a rough division can be made between the 7. Sample effect size
variance (𝑆2) 𝑆2=
∑𝑘𝑖=1𝑛𝑖(𝑟𝑐𝑖− 𝑟̅)2 ∑𝑘𝑖=1𝑛𝑖
𝑛𝑖 =sample size of study i 𝑟𝑐𝑖 =correct effect size of study i 𝑟̅ =mean attenuated effect size 8. Between-studies
variance (𝑇2)
𝑇2 = 𝑆2− 𝑉 𝑆2=sample effect size variance
𝑉 =sample-size weighted average of sampling error variance
9. Credibility interval 90% 𝑟̅
𝑐± 1.45√𝑇2 𝑟̅𝑐 =mean corrected effect size 𝑇2=between-studies variance
10. Moderator ratio 𝑉
𝑆2
𝑉 =sample-size weighted average of sampling error variance
𝑆2=sample effect size variance 11. Failsafe number ( 𝑁
2.706) ∗ (𝑁(𝑍̅)
2− 2.706) 𝑁 =number of studies
𝑍̅ =standardized value of mean corrected effect size 0 5 10 15 20 25 30 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 A m o u n t o f ar ticl e s sel e cte d Year (up till April 2016)
Articles per year
FIGURE 4
18
earlier articles introducing GPI (2003-2010), the upcoming years (2011-2014) and the most recent years (2015-2016). This division is also used for the moderator analysis in Section 4.3.
From these articles, 88 studies were derived with a total sample size of 18,721. Except for three studies exploring GPI projects (Dangelico et al., 2013; Genç & Di Benedetto, 2015; Pujari, 2006), the unit of analysis comprised the firm level. These studies only investigated one project per firm. Furthermore, a wide variety of industries is represented in the sample, ranging from automotive to shipping to hotels. The majority of the studies researched manufacturing firms (75%), and others the service industry (9%) or both (8%). Moreover, some studies included multiple different industry types (58%), whereas others focused on one single industry (34%). For seven studies, the industry was unknown. Finally, companies were investigated from all over the world. Apart from five cross-country studies, the studies focused on one single country. Figure 5 visualizes the representation per continent. As can be seen, the vast majority of research is done in Asia, mainly China and Taiwan. Besides, all three South American studies originate from Brazil.
4.2 Main effects
The average effect size between GPI capabilities and firm performance shows a clear significant positive relation (r̅ = 0.37), especially after correction for measurement error (r̅c = 0.42). Combined with the high failsafe number (488.55), this leads to the conclusion hypothesis 1 can be accepted; GPI capabilities are indeed positively related to firm performance. The wide range of effect sizes (0.04:0.82) can be explained by the presence of moderators. This can be concluded since the credibility interval is rather big (0.109 : 0.733), and the moderator ratio (0.112) does not even come near the threshold of 0.75. The results of the main effects can be found in Table 3.
Separately, all four GPI capabilities show significant positive relations with firm performance. There are small but noteworthy differences amongst these effects. The strongest effect size was found for
52% 29%
10% 4% 3% 2%
Studies per continent
Asia Europe North America South America Cross-continental Australia FIGURE 5
19
technological capabilities (r̅c = 0.47), followed by internal integrative (r̅c = 0.39), external integrative (r̅c = 0.36) and marketing capabilities (r̅c = 0.30). The failsafe numbers are high enough to conclude these results are reliable, except for marketing capabilities (-5.81). This is due to the fact that only eight studies investigated the effects of marketing capabilities on firm performance. Hence, hypotheses 2a, 2c and 2d can be accepted, whereas hypothesis 2b should be rejected; technological, internal integrative and external integrative capabilities are indeed positively related to firm performance. In addition, all relationships appear to be affected by moderators, except for marketing capabilities. This can be concluded since the credibility interval is very narrow (0.239:0.352) and the moderator ratio (0.762) exceeds the threshold of 0.75.
Separately, all four firm performance dimensions show significant positively relations with GPI. There are small but noteworthy differences amongst these effects. The strongest effect size was found for environmental performance (r̅c = 0.50), followed by business performance (r̅c = 0.44), operational performance (r̅c = 0.36) and reputation (r̅c = 0.35). The failsafe numbers are high enough to conclude these results are reliable, except for reputation (-5.08). This is due to the fact that only ten studies investigated the effects of GPI on reputation. Hence, hypothesis 3a, 3b and 3d can be accepted, whereas hypothesis 3c should be rejected; business, operational and environmental performance are indeed positively related to GPI. In addition, all relationships appear to be affected by moderators. This can be concluded since the credibility intervals are very wide or even include zero (reputation), and the moderator ratios do not even come near the threshold of 0.75.
TABLE 3
Main effects between GPI and firm performance
Hypothesis Number of studies Sample size Range Uncorrec-ted effect size (𝐫̅) Corrected effect size (𝐫̅𝐜) Credibility interval 90% Modera-tor ratio Failsafe number 1. GPI capabilities and
20
4.3 Moderating effects
The previous section detected the presence of moderators affecting the relationship between GPI and firm performance. As a first attempt to identify what these moderators are, three contextual factors were investigated, namely industry, location and year. First of all, the effect size of organizations in a manufacturing industry (r̅c = 0.43) is practically the same as organizations in the service industry (r̅c = 0.44). Secondly, a superior corrected effect size was found in North America (r̅c = 0.55) when comparing to Asia (r̅c = 0.43) and especially Europe (r̅c = 0.36). Despite of this difference, this moderator only explains a small part of the variance. The wide credibility intervals and low moderator ratios indicate the presence of more moderators. Third, the earlier studies ranging from 2003 to 2010, show the highest corrected effect size (r̅c = 0.48), rather similar to the most recent studies in 2015 and 2016 (r̅c = 0.45), and higher than the studies ranging from 2011-2014 (r̅c = 0.35). Again, this moderator only explains a small part of the variance. The results of the moderating effects can be found in Table 4.
TABLE 4
Moderating effects on GPI and firm performance
Moderator Number of studies Sample size Range Uncorrec-ted effect size (𝐫̅) Corrected effect size (𝐫̅𝐜) Credibility interval 90% Modera-tor ratio Failsafe number Industry: Manufacturing 66 15,212 0.04 : 0.82 0.37 0.43* 0.100 :0.754 0.096 268.45 Industry: Service 8 1,670 0.14 : 0.80 0.38 0.44* 0.128 : 0.751 0.111 -2.74 Location: Asia 46 9,046 0.04 : 0.82 0.38 0.43* 0.128 : 0.735 0.122 120.71 Location: Europe 25 5,717 0.09 : 0.72 0.31 0.36* 0.141 : 0.576 0.222 7.44
Location: North America 9 2,804 0.11 : 0.72 0.50 0.55* 0.178 : 0.926 0.040 2.55
Years: 2003-2010 13 3,307 0.09 : 0.82 0.42 0.48* 0.089 : 0.880 0.056 4.48 Years: 2011-2014 34 6,584 0.04 : 0.80 0.31 0.35* 0.093 : 0.598 0.186 21.38 Years: 2015-2016 41 8,830 0.10 : 0.77 0.40 0.45* 0.170 : 0.740 0.127 108.64 * p<.001
5. Discussion
5.1 Theoretical implications
21
integrative and external integrative capabilities. Yet, GPI appears to add an extra level of complexity that needs to be taken into account (Huang & Wu, 2010; Pujari et al., 2003). New products need to be both “new” and “green” enough to be competitive (Wong, 2012). Therefore, some adaptations and additions were made to the original framework to incorporate the green criteria. Scholars can benefit from the proposed framework to position their research along one or more of these relevant dimensions.
The results of this study confirm that multiple GPI dimensions can be distinguished that show a positive relationship with firm performance. However, assessment of previous research revealed a tendency to downplay GPI to solely technological capabilities. Constructs labelled with ‘green product innovation’ specifically or a synonym, almost exclusively referred to technological capabilities. However, this dimension does not fully grasp the concept of GPI. Internal integrative, and to some extent external integrative capabilities received attention as well, but were not always regarded to be in the scope of GPI. The lack of interest in marketing capabilities is remarkable. Up to now, not a sufficient amount of studies investigated the effects of marketing capabilities to prove its positive relationship with firm performance. Richey, Musgrove, Gillison, & Gabler (2014) explain research so far focuses more on the demand-side than on the supply-side perspective of green marketing. By this they mean an emerging understanding of customer behavior, but a lack of insights in the strategic viability of marketing capabilities. This study advocates acknowledgment that more complexities underpin the GPI concept, and especially more research on marketing capabilities would be beneficial.
Furthermore, the meta-analysis findings show a convincing positive relationship between GPI and various firm performance measures. GPI does what it aims to do (i.e. higher environmental performance) but goes beyond that by posing additional advantages for companies. Support was found for higher business as well as operational performance. This provides evidence for the win-win paradigm as first propagated by Porter and Van der Linde (1995), stating that both the environment and the organization benefit from GPI. For the vast majority of researchers, firm performance is reflected by tangible, objective measures. This makes sense, but it may be expected that GPI also leads to more intangible performance improvements. The results of this meta-analyses pointed towards reputational gains, but this could not be confirmed due to a lack of studies addressing this. More attention to intangible performance dimensions (e.g. reputation, learning spill-overs) is recommended.
Apart from the main effects, three moderators were tested. First, a distinction was made between organizations in the service and the manufacturing industry. It appears that the relationship between GPI and firm performance is equally strong in both industry types. This is in line with Leonidou, Fotiadis, et al. (2015) and Wong et al. (2013), who also found positive linkages in the service industry. Remarkable in this sense is the extent to which GPI research neglects services, as already pointed out by De Medeiros et al. (2014). The results of this study point to the fact that GPI can be equally beneficial for service-providing firms, even though our understanding of how this works is limited. Hence, more research on this area would be helpful.
22
Union is leading in environmental friendly legislation (Amann, Roehrich, Eßig, & Harland, 2014). Moreover, firms located in developed countries tend to give higher priorities to environmental issues (Jackson et al., 2016). Yet, Golicic & Smith (2013) found a similar result. In line with their interpretation, it is deemed plausible that Europe has reached some sort of a threshold in GPI adoption, whereas Asia and North America are still growing. An enormous rise in sustainable technologies is visible in newly industrializing countries (Walz, 2010). Following this train of thought, it is possible that organizations going green in Asia and North America still reap first-mover advantages. Organizations pioneering in GPI can ask for higher prices and improve their image (Chen, 2008). Without legislation, it is substantially easier for a firm to differentiate itself on the green market. As is already known, proactive GPI is more effective than reactive GPI (Chen et al., 2012; Li et al., 2016). Organizations in Asia and North America might be either proactive or non-active, whilst European companies need to be reactive due to legislation.
The third moderator is the year of the study. It appears that the early years of GPI (2003-2010) and the recent years (2015-2016) had a higher effect than the upcoming years (2011-2014). The fact that organizations in the early years reaped high performance gains can be explained in the same way as done for the continent. It is possible that early adopters gain a first-mover advantage and can easily differentiate itself on the green market, an advantage that diminishes as the market matures. Furthermore, it is plausible that organizations are learning and become better at developing GPI capabilities. Hence, the rise in the most recent years.
5.2 Managerial implications
In addition to the theoretical implications, this study has several managerial implications. Obviously, GPI requires initial investments that lead to an increase in costs (Liu et al., 2011). Deciding what to invest in is a delicate matter that requires sufficient confidence in positive outcomes. So far, research showed contradictory effects on firm performance, possibly hindering the willingness to invest. Yet, by synthesizing these results, a convincing positive effect is visible. This study should provide managers with the confidence GPI can help improve a multitude of performance dimensions. Besides reducing the impact on the environment, it enhances operational performance such as quality, flexibility and costs. In addition, it improves business performance in general, including financial and market aspects. Furthermore, different GPI dimensions have been investigated. It appears multiple aspects are worth looking into, as no big differences amongst them have been found. On one hand, going green can be beneficial at the functional level, such as in R&D, operational and presumably marketing departments. On the other hand, the value of crossing intra- and inter-firm boundaries has been shown. This comprises integrative practices, communication and collaboration between different departments or with external partners.
23
to differentiate relative to competitors. Therefore, managers are advised to pay attention to their environment and adapt to the needs of their specific stakeholders. Lastly, organizations delivering a service instead of a product are equally recommended to assess possibilities to go green.
5.3 Limitations and future research
Every research poses limitations and this study is no exception. First of all, the danger of publication bias should be acknowledged. It is possible our sample is positively biased in a way that it overestimates the actual population effect size. Despite an extensive literature search, it is also imaginable some papers were missed. However, Pagell and Kristal (2011) surprisingly discovered that the file drawer in the supply chain management field is close to empty. Since GPI to a great extent overlaps the SCM research domain, the threat may not be as big as it seems. Yet, to reduce the publication bias risk, relevant papers only published online or in conference proceedings were also included in the sample. Moreover, the failsafe numbers were calculated and turned out rather high, except for two hypotheses that had to be rejected. Future research can resolve what the effects of marketing capabilities on firm performance are, as well as the effects of GPI on reputation.
Furthermore, our research only addresses direct relationships between GPI and firm performance. However, it could be the case that full or partial mediating effects are at play. Several authors found the relationship between GPI and business performance to be mediated by operational performance (Fraj-Andrés et al., 2009; Vijayvargy & Agarwal, 2014) or reputation (Amores-Salvadó et al., 2014; Feng & Wang, 2014; Fraj-Andrés et al., 2009; Peng & Lin, 2008; Wu & Lin, 2016). This should be kept in mind when interpreting our results. It would be interesting if future research would address the causalities at work.
In addition, some difficulties were encountered with missing data. Out of the 86 articles included in the sample, 23 did not report correlation coefficients and 13 missed Cronbach’s Alpha values. Some authors send us these values upon our request, and for the others substitute values were used. Although this is in line with practices advocated in literature (Kenny, 1979; Peterson & Brown, 2005; Peterson & Kim, 2013) using anything other than the exact values is suboptimal. This is not only true for a meta-analysis, but also for other researchers assessing the reliability and validity of a survey article. Therefore, we recommend future survey research to ensure it reports on reliability measures, correlation coefficients and items used to measure constructs.
Lastly, the meta-analysis results detected the presence of moderators affecting the relationship between GPI and firm performance. This means non-significant and negative results found by previous literature can be attributed to heterogeneous factors. Further analyses showed that location and year do explain some of the variance, but other unknown moderators are at play as well. Future research is advised to assess what these variables are and what role they play.
6. Conclusion
24
25
References
Agan, Y., Acar, M. F., & Borodin, A. 2013. Drivers of environmental processes and their impact on performance: a study of Turkish SMEs. Journal of Cleaner Production, 51: 23–33.
Aguilera-Caracuel, J., & Ortiz-de-Mandojana, N. 2013. Green innovation and financial performance: An institutional approach. Organization & Environment, 26(4): 365–385.
Albino, V., Balice, A., & Dangelico, R. M. 2009. Environmental strategies and green product development: An overview on sustainability-driven companies. Business Strategy and the Environment, 18: 83– 96.
Amann, M., Roehrich, J. K., Eßig, M., & Harland, C. 2014. Driving sustainable supply chain management in the public sector: The importance of public procurement in the European Union. Supply Chain
Management: An International Journal, 19: 351–366.
Amores-Salvadó, J., Martín-de Castro, G., & Navas-López, J. E. 2014. Green corporate image: Moderating the connection between environmental product innovation and firm performance. Journal of
Cleaner Production, 83: 356–365.
Ar, I. M. 2012. The impact of green product innovation on firm performance and competitive capability: The moderating role of managerial environmental concern. Procedia - Social and Behavioral
Sciences, 62: 854–864.
Barney, J. 1991. Firm resources and sustained competitive advantage. Journal of Management, 17(1): 99–120.
Berchicci, L., & Bodewes, W. 2005. Bridging environmental issues with new product development.
Business Strategy and the Environment, 14: 272–285.
Borenstein, M., Hedges, L. V., Higgins, J. P. T., & Rothstein, H. R. 2009. Introduction to meta-analysis. Chichester: John Wiley & Sons.
Böttcher, C. F., & Müller, M. 2015. Drivers, practices and outcomes of low-carbon operations:
Approaches of German automotive auppliers to cutting carbon emissions. Business Strategy and
the Environment, 24: 477–498.
Callaway, S. K., & Dobrzykowski, D. D. 2009. Service-oriented entrepreneurship: Service-dominant logic in green design and healthcare. Service Science, 1(4): 225–240.
Chan, H. K., Yee, R. W. Y., Dai, J., & Lim, M. K. 2015. The moderating effect of environmental dynamism on green product innovation and performance. International Journal of Production Economics. http://dx.doi.org/10.1016/j.ijpe.2015.12.006.
Chan, R. Y. K., & Ma, K. H. Y. 2016. Environmental orientation of exporting SMEs from an emerging economy: Its antecedents and consequences. Management International Review.
http://dx.doi.org/10.1007/s11575-016-0280-0.
Chang, C. 2011. The influence of corporate environmental ethics on competitive advantage: The mediation role of green innovation. Journal of Business Ethics, 104(3): 361–370.
Chavez, R., Yu, W., Feng, M., & Wiengarten, F. 2016. The effect of customer-centric green supply chain management on operational performance and customer satisfaction. Business Strategy and the
Environment, 25: 205–220.
Chen, C., & Chang, Y. 2013. The determinants of green product development performance: Green dynamic capabilities, green transformational leadership, and green creativity. Journal of Business
Ethics, 116: 107–119.
Chen, J., Damanpour, F., & Reilly, R. R. 2010. Understanding antecedents of new product development speed: A meta-analysis. Journal of Operations Management, 28: 17–33.
Chen, Y. 2008. The driver of green innovation and green image? Green core competence. Journal of
26
Chen, Y. J., Wu, Y. J., & Wu, T. 2015. Moderating effect of environmental supply chain collaboration: Evidence from Taiwan. International Journal of Physical Distribution & Logistics Management, 45(9/10): 959–978.
Chen, Y., Lai, S., & Wen, C. 2006. The influence of green innovation performance on corporate advantage in Taiwan. Journal of Business Ethics, 67(4): 331–339.
Chen, Y., Tang, G., & Jin, J. 2015. Linking market orientation and environmental performance: The influence of environmental strategy, employee’s environmental involvement, and environmental product quality. Journal of Business Ethics, 127: 479–500.
Chen, Y.-S., Chang, C., & Wu, F.-S. 2012. Origins of green innovations: The differences between proactive and reactive green innovations. Management Decision, 50(3): 368–398.
Chen, Y.-S., Lin, Y.-H., Lin, C.-Y., & Chang, C.-W. 2015. Enhancing green absorptive capacity, green dynamic capacities and green service innovation to improve firm performance: An analysis of structural equation modeling (SEM). Sustainability, 7: 15674–15692.
Cheng, C. C. J., Yang, C., & Sheu, C. 2014. The link between eco-innovation and business performance: A Taiwanese industry context. Journal of Cleaner Production, 64: 81–90.
Chiou, T., Kai, H., Lettice, F., & Ho, S. 2011. The influence of greening the suppliers and green innovation on environmental performance and competitive advantage in Taiwan. Transportation Research
Part E, 47(6): 822–836.
Choi, D., & Hwang, T. 2015. The impact of green supply chain management practices on firm
performance: The role of collaborative capability. Operations Management Research, 8: 69–83. Commission of the European Communities. 2001. Green paper on integrated product policy.
Dangelico, R. M. 2015. Green product innovation: Where we are and where we are going. Business
Strategy and the Environment.
Dangelico, R. M., & Pontrandolfo, P. 2015. Being green and competitive: The impact of environmental actions and collaborations on firm performance. Business Strategy and the Environment, 24: 413– 430.
Dangelico, R. M., Pontrandolfo, P., & Pujari, D. 2013. Developing sustainable new products in the textile and upholstered furniture industries: Role of external integrative capabilities. Journal of Product
Innovation Management, 30(4): 642–658.
Dangelico, R. M., & Pujari, D. 2010. Green product innovation: Mainstreaming why and how companies integrate environmental sustainability. Journal of Business Ethics, 95(3): 471–486.
Danneels, E. 2002. The dynamics of product innovation and firm competences. Strategic Management
Journal, 23: 1095–1121.
Day, G. S. 1994. The capabilities of market-driven organizations. Journal of Marketing, 58(4): 37–52. De Medeiros, J. F., Ribeiro, J. L. D., & Cortimiglia, M. N. 2014. Success factors for environmentally
sustainable product innovation: A systematic literature review. Journal of Cleaner Production, 65: 76–86.
De Sousa Jabbour, A. B. L., Jabbour, C. J. C., Latan, H., Teixeira, A. A., & De Oliveira, J. H. C. 2014. Quality management, environmental management maturity, green supply chain practices and green performance of Brazilian companies with ISO 14001 certification: Direct and indirect effects.
Transportation Research Part E, 67: 39–51.
Driessen, P. H., Hillebrand, B., Kok, R. H. W., & Verhallen, T. M. M. 2013. Green new product development: The pivotal role of product greenness. IEEE Transactions on Engineering
Management, 60(2): 315–326.
Dubey, R., Gunasekaran, A., & Ali, S. S. 2015. Exploring the relationship between leadership, operational practices, institutional pressures and environmental performance: A framework for green supply chain. International Journal of Production Economics, 160: 120–132.
