NA

TU

RA

L

co

Ns

TR

Uc

Te

d

C L A R K U N I V E R S I T YGeorge Perkins Marsh Institute

HUMAN

WoRKING PAPeR No. 2010-15

ocToBeR 2010

By Xiaoyong Fu1, Qinghua Zhu, and Joseph sarkis

A Grey-DeMATeL MeThoDoLoGy for Green SuppLier DeveLop-MenT proGrAM evALuATion

A Grey-DEMATEL Methodology for Green Supplier Development Program Evaluation

Xiaoyong Fu1, Qinghua Zhu1, and Joseph Sarkis2*

1 School of Management; Dalian University of Technology; Dalian, Liaoning Province 116024; PR of China

2* Corresponding Author: Graduate School of Management; Clark University; 950 Main Street; Worcester, MA 01610; 508-793-7659 tel.; 508-793-8822 fax; jsarkis@clarku.edu

Abstract: Supplier development is necessary as organizations and industries are competing on the capabilities of their supply chains and not necessarily their own individual organizations. Supplier development evaluation through formal modeling and decision support approaches is limited in the literature. Formal modeling tools to aid managers in evaluating green supplier development programs are virtually non-existent. This paper seeks to help fill the gap in this literature by presenting a managerial approach for organizations to help structure the influence relationships amongst green supplier development programs. We initially introduce the categories of green supplier development programs that may exist. In our study we utilize multi-functional managerial inputs on relationships amongst their green supplier development programs. These managerial inputs are evaluated using a grey-DEMATEL methodology. The grey-DEMATEL results in a prominence-causal diagram that provides managerial insights into the relationships amongst their green supplier development programs. Managerial and research implications of the proposed study, limitations and future research directions are also detailed.

Keywords: Environmental, Supply Chain Management, Supplier Development, Grey numbers, DEMATEL, Telecommunications.

WP2010-15

2

Introduction

Within green supplier management literature helping suppliers improve environmental performance has focused on ‘requiring’ suppliers to complete certain certifications or practices or focused on selection of existing green suppliers. Not all suppliers are in the position to effectively, on their own, improve their environmental or sustainability performance. Also not all suppliers are initially green. Similar to organizations aiding their suppliers improve on business and competitive performance, such as costs, quality and delivery, organizations may also aid their suppliers’ green development.

A critical aspect of green supply chain management literature that has been under-

researched is supplier development (Seuring & Müller, 2008a). Interestingly, even though significant research has been initiated and completed on green supply chain management (Srivastava, 2007; Seuring & Müller, 2008b; Ilgin & Gupta, 2010), investigation into sustainability and supplier development programs is virtually non-existent.

The development or application of formal tools and models has been very limited in general supplier development programs (Bai & Sarkis, 2010a). Some that have been developed consider how to further develop suppliers through, for example, adjusting supplier practices in response to requests for proposals (RFPs) (Narasimhan, Talluri, & Mahapatra, 2008b). These types of models are recommended by the literature to help in supplier improvement and management (Krause, Handfield, & Scannell, 1998; McGovern & Hicks, 2006). Formal tools and models for environmental supplier development to aid companies manage environmental performance of suppliers are even more limited.

As part of this formal modeling organizations need to determine, if they wish to green their supply chains, how they should develop and implement their programs. That is, some programs will be more valuable or foundational (influencing development of other supplier development programs). Thus, in recognition of this fact, we introduce a formal methodology to investigate the importance of organizational green supplier development programs (GSDP) and how they relate to each other. To help develop these relationships we introduce a unique grey-based Decision-Making Trial and Evaluation Laboratory (DEMATEL) approach to structure the program management environment.

One important purpose of this formal modeling approach is to aid organizations prioritize their investments in these programs, while also identifying critical relationships amongst the programs. The determination of which specific GSDP are most useful may free up resources that may enhance the returns by focusing on the most important programs that set the foundation and relationships to other environmental programs and goals. This formal modeling methodology is valuable for planning, design, implementation or maintenance of GSDP in organizations.

The formal modeling approach introduced in this paper, using grey-based DEMATEL approach, will go through a number of stages resulting in a relationship diagram (prominence and causal) amongst various GSDP on a prominence and cause/effect axis. The technique will

WP2010-15

3

involve the identification of GSDP, their interrelationship utilizing “grey” data input from management, making the grey data into crisp data, completing a series of DEMATEL steps, and eventually arriving at a final prominence-causal relationship diagram with some analysis.

Thus, the goal of this paper is to explicate how grey-based DEMATEL can be a valuable managerial tool to evaluate GSDP and their relationships to each other and importance for the organization. The paper begins with a short discussion and review of GSDP and some previous work on evaluating the relationships of these programs. In addition, some background on related tools for analysis of GSDP that do exist will help us identify what gaps occur in the literature. We then generally introduce the basic concepts associated with grey-based DEMATEL. We then apply the technique to a real-world situation that provides managerial insights and implications for the tools and its results. The results are presented with some discussion analyzing the results and implications of the technique with feedback from management. We finally conclude the paper with some limitations and future research directions in this emergent field and topic. Green Supply Chain Management and Development Programs

Green supply chain management has seen significant development in the past two decades ranging from initial practical sense-making and conceptual developments to more rigorous empirical and analytical studies (Seuring & Müller, 2008b). Green supply chain management is essentially the integration of natural environmental concerns into supply chain management. There are many activities and factors that may be incorporated into this organizational practice. Examples include selecting green suppliers, incorporating supplier input into greening organizational practices, expanding the environmental life cycle analysis of products into your supplier processes, helping develop and implement environmental management systems into your suppliers’ organizational structure, and many such practices.

It has been argued that the greening of supply chains is critical as organizations progress

from both competitive and environmental influences of industry (Zhu & Sarkis, 2004; Rao & Holt, 2005). Given the breadth of possible activities and functions within green supply chain management the necessary resources for the development of these programs may not exist, which is an especially profound barrier for smaller and medium sized enterprises. In fact, many times these programs are relegated to the ‘nice-to-have’ organizational decisions rather than the ‘must-haves’. The involvement and support of larger and more resource rich supply chain partners for supplier development can be crucial to a successful green supplier management program when resource poor partners exist. Given this situation, supplier development for green supply chains has been and will continue to be an important dimension. Even though significant formal modeling effort has focused on aspects of green supplier management (Sarkis, 2003; Piplani, Pujawan, & Ray, 2008; Lee, Kang, Hsu, & Hung, 2009; Bai & Sarkis, 2010b), the research in green supply chain management has not focused on how GSDP function, much less their existence or prevalence.

A little background shows that supplier development programs, in general, can be quite

extensive for organizations. The literature has relatively extensively studied supplier development programs in general (e.g. (Krause & Ellram, 1997; Krause et al., 1998; Lee & Humphreys, 2007; Narasimhan, Mahapatra, & Arlbjørn, 2008a; Wagner & Krause, 2008)).

WP2010-15

4

Utilizing general supplier development programs categorizations, GSDP may be grouped into at least three categories including: Green Knowledge Transfer and Communication; Investment and Resource Transfer; Management and Organizational Practices (Bai and Sarkis, 2010). A listing of possible green supplier development programs from the literature and grouped into these three major categories are shown in Table 1.

Table 1 about here

The number of programs in an organization can be extensive. Understanding the roles and relationships amongst these programs can provide management with additional informed insight of how to manage them. Sometimes there are foundational GSDP such as supplier auditing programs that may set the stage for helping suppliers improve their systems. These supplier environmental auditing programs may complementarily support other GSDP such as providing training for environmental improvement or knowledge sharing in weak areas. But a foundational program does not necessarily make it the most important supplier development program for an organization, if this supplier development program has already been established. Thus, the relationships and relative importance of supplier development programs may vary depending on the organization that is managing these programs. This is an example nuanced managerial issue that we seek to address in this paper.

We know of no specific formal planning model to help address this particular issue. That is why we introduce a technique in this paper that integrates the grey scale measures and DEMATEL to help managers plan, manage, and maintain their GSDP.

A Grey-Based DEMATEL Approach

In this section we introduce the grey numbers and an overview of various elements of the DEMATEL approach.

Grey System Theory

Grey system theory can be used to solve uncertainty problems in cases with discrete data and incomplete information (Deng, 1989). The major advantage is that it can generate satisfactory outcomes using a relatively small amount of data or with great variability in factors (Li, Tan, & Lee, 1997). Grey system theory provides an approach for analysis and modeling of systems with limited and incomplete information, and which may exhibit random uncertainty. Grey system theory has many successful applications in areas such as economics, agriculture, medicine, geography, earthquakes, industry, etc. In recent years, grey system theory has been an effective methodology that deals with uncertain and indeterminate problems.

We will now introduce some general notation and operations for grey systems. Let x

denote a closed and bounded set of real numbers. A grey number x! , is defined as an interval with known upper and lower bounds but unknown distribution information for x (Deng, 1989). That is, [ , ] [ ]x x x x x x x x! !" = " " = # |" $ $ " where x! and x! are the lower and upper bounds of x! , respectively.

WP2010-15

5

Some basic grey number mathematical operations are represented by the following relationships (expressions 1-4）:

1 2 1 2 1 2[ , ]x x x x x x! +! = + + (1)

1 2 1 2 1 2[ , ]x x x x x x! "! = " " (2)

1 2 1 2 1 2 1 2 1 2[min( , , , ),x x x x x x x x x x! "! = 1 2 1 2 1 2 1 2max( , , , )]x x x x x x x x (3)

1 2 1 1

2 2

1 1[ , ] [ , ]x x x x

x x! ÷! = " (4)

To deal with the problems of decision making in a grey environment, an effective grey aggregation method is required. Grey aggregation methods can be aided by de-“graying” to arrive at a crisp number. In this paper we adopt a variation of the CFCS (Converting Fuzzy data into Crisp Scores) defuzzication method for our grey aggregation procedure. This approach has been deemed to be more effective by researchers for arriving at crisp values (e.g. when compared to the Centroid method) (Opricovic & Tzeng, 2003; Wu & Lee, 2007).

Let us define p

ijx! as the grey number for an evaluator (decision maker) p, that will

evaluate the influence of GSDP i on a GSDP j. Also, p

ijx! and p

ijx! are respectively, the lower

and upper grey values by an evaluator p for the relationship evaluation between green supplier development program i to a green supplier development program j.

That is:

[ , ]

p p p

ij ij ijx x x! = ! ! .

The modified-CFCS method includes three-step procedure described as follows: （1）normalization

! max

min( min ) /p

p pij ij ij

jx x x! = ! " ! # (5)

! max

min( min ) /p

p pij ij ij

jx x x! = ! " ! # (6)

where max

minmax min

p p

ij ijjj

x x! = " # " (7)

(2) compute total normalized crisp value

! ! ! !( )! !( )

(1 ) ( )

1

p p p p

ij ij ij ijp

ij p p

ij ij

x x x xY

x x

! "! + ! #!=

"! +! (8)

WP2010-15

6

(3) compute crisp values

max

minmin

p p p

ij ij ijj

z x Y= ! + " (9)

DEMATEL

The DEMATEL method originated from the Geneva Research Centre of the Battelle Memorial Institute (Gabus & Fontela, 1973; Fontela & Gabus, 1976). The method is useful to help visualize the structure of complicated causal relationships through the use of matrices or digraphs. The matrices or digraphs portray relationships between system components with strengths of relationships amongst these relationships quantitatively portrayed. The DEMATEL method assumes a system contains a set of components C= {C1,C2, . . .,Cn}, with pairwise relations that can be evaluated. DEMATEL incorporates four generic stages:

1. Develop a pairwise direct-relation matrix between system components through evaluator or decision maker input.

2. Determine the initial influence matrix, through normalization of the direct-relation matrix.

3. Determine a total relation (influence) matrix. 4. Determine the cause/effect relationships (prominence-causal diagram) amongst the

components and relative strengths. Using these four generic stages for basic DEMATEL, we introduce the overall grey-based

DEMATEL methodology. This grey-based DEMATEL methodology is composed of the following major steps (see Figure 1 for a summary flow chart):

Figure 1 about here

For the first stage in the process, we have multiple sub-steps 1a-1d.

Step 1a: Define a grey pairwise influence comparison scale for the components. In discrete cases, the comparison scale can just be binary (0,1) with 0 for no influence, 1 otherwise. In our case, we will utilize a five level scale with the following scale items: N (no influence), VL (very low influence), L(low influence), H (high influence), and VH (very high influence). The grey scales for these linguistic values are defined in the case application. Step 1b: Develop the grey direct-relation matrix X by having evaluators introduce the grey pairwise influence relationships ( k

ijx! ) between the components in an n n! matrix. All the principal diagonal elements are initially set to a crisp value of zero (“N” = no influence). Step 1c: Make the grey direct-relation matrix into a crisp matrix Z using the modified-CFCS process as exemplified by expressions (5)-(9). The process will need to be completed for each of the evaluators' direct-relation matrices. If more than one evaluator exists go to step 1d, otherwise go to step 2.

WP2010-15

7

Step 1d: Evaluator weightings, for aggregation purposes, for each evaluator need to be determined. Either simple averaging (expression (10)) or weighted averaging (expression (11)) can be used to calculate an aggregate score. We will utilize expression (11) which requires determination of evaluator weightings, and may be defined by grey linguistic scale values for each evaluator p (

pw! ). Grey scaled evaluation weightings are required to crisp and sum to 1 as

shown in expression (11). A modified-CFCS method can be used to make grey scale evaluator weightings crisp.

1 21( ... )p

ij ij ij ijz z z z

p= + + + (10)

1 2

1 2...

p

ij ij ij p ijz w z w z w z= + + + such that

1

1

p

i

i

w=

=! (11)

Whereijz is the overall crisp evaluation for the relationship between supplier development

programs i and j，ij

pz is the crisp evaluation the relationship between supplier development

programs i and j by evaluator p，wp is the crisp evaluator weight assigned to evaluator p derived from the grey scale weight for each evaluator (

pw! ).

Step 2: On the basis of the overall crisp direct-relation matrix Z, the normalized direct-relation matrix N can be obtained through expressions (12) and (13):

N s Z= ! (12)

11

1, , 1, 2, , .

maxn

iji n

j

s i j n

z! !

=

= =

"… (13)

Step 3: The total relation matrix (T) is determined by expression (14) where I represents an n n! identity matrix.

2 3 1

1

( )i

i

T N N N N N I N

!"

=

= + + + = = "#! (14)

Step 4: Developing the causal influence and digraph diagram in DEMATEL requires three sub-steps.

Step 4a: Determine row (Ri) and column (Dj) sums for each row i and column j from the total relation matrix (T). That is:

1

i

n

i ij

j

R t=

= !" (15)

1

jn

j ij

i

D t=

= !" (16)

The row values Ri are the overall direct and indirect effect of a green supplier development program i on other green supplier development programs. Similarly the column values Dj

WP2010-15

8

represent the overall direct and indirect effects of all the green supplier development programs on green supplier development program j.

Step 4b: Determine the overall importance or prominence (Pi) of a GSDP i and net effect (Ei) of GSDP i using expressions (17) and (18).

{ | }i i jP R D i j= + = (17)

{ | }i i jE R D i j= ! = (18)

The larger the value of Pi the greater the overall prominence (visibility/importance/influence) of GSDP i in terms of overall relationships with other GSDP. If Ei > 0 then GSDP is a net cause, foundation, for other green supplier development programs. If Ei < 0, then GSDP i is reliant on (net effect of) implementation or operation of other GSDPs (Tzeng, Chiang, & Li, 2007). These values may then be plotted onto a two-dimensional axis for each GSDP. Step 4c: A digraph relationship can be determined for each green supplier development program with respect to other supplier development programs using the total relation matrix T. To complete this step a threshold value ! should be determined by the evaluators, experts or the analysts (Liou, Tzeng, & Chang, 2007). If tij > ! , then supplier development program i influences or effects supplier development program j and a directed arrow is incorporated into the analysis. Case Study and Application

To help set the stage for demonstrating the grey scale DEMATEL process for GSDP, we utilize a case study from a Telecommunications Equipment provider located in China. Company A. This company is a major multinational company that produces, sells, and services various telecommunications products throughout China and Asia. It has come under increasing scrutiny by their customers to green their supply chains. In response to these pressures they have invested resources in various GSDP.

Participant-Evaluator Background

To apply the DEMATEL procedure we recruited managers from four functions within the company that had some role in interacting with suppliers. These managers included manager1 (procurement department), manager2 (research and development (R&D) department), manager3 (customer service department), and manager4 (marketing department). We provide some background on the managers that set the stage for insights into later steps of our procedure when assigning weighting and completing a sensitivity analysis.

Manager1 is young with 5 years work experience. He has a graduate degree in business

and works directly with 20 suppliers. He has only worked in the procurement department. Manager 1 has no project management or product development experience.

Manager2 has 12 years work experience also with a graduate degree. He has only been in

the R&D department. He works directly with 16 suppliers, managed 10 projects and 6 product families.

WP2010-15

9

Manager3 has 12 years work experience and a graduate degree. He has worked in both the

customer service and project management departments. He works directly with 7 suppliers, managed over 20 projects and 2 product families.

Manager4 has 9 years work experience and an undergraduate degree. His only assignments

were with the marketing department. He works directly with 3 suppliers (the fewest), managed 3 projects and 17 product families.

This group of managers provides a broad supply chain (both customers and suppliers)

focus with significant work experience and project management skills.

Application of the DEMATEL Process

In our initial step of the process we provided managers with the full listing of the Table 1 green supplier development programs. To arrive at a more manageable list we sat down with the managers and eliminated programs (e.g. some of the training programs) that they viewed as redundant. It was decided that only programs that the organization has implemented were explicitly considered. The managers felt that they would be most knowledgeable on the relationships of existing programs. Overall, 17 clearly elaborated and case applicable GSDP remained and are summarized in Table 2. We list the GSDP and assign a factor identifier with each program for ease of exposition in later stages.

Step 1a: To begin the evaluation we first assigned a grey linguistic scale for the assessments. A five level grey linguistic scale is used (see Table 3). We assumed a linear, non-overlapping scale for each linguistic term, with linguistic terms ranging from ‘no influence’ (N) to very high influence (VH).

Table 2 about here Table 3 about here

Step 1b: Separately, each manager was given a 17x17 linguistic/grey scale direct-relation matrix X for comparison of the supplier development programs. The diagonal elements of the grey scale direct-relation matrix, based on the DEMATEL procedure were assigned an “N” value. The remaining portions of the matrix were left blank to be filled in by the evaluators. A completed direct-relation matrix for manager1 within a linguistic scale assessment among the supplier development programs is shown in Table 4.

Table 4 about here

Step 1c: In this step we utilize the modified-CFCS method (expressions (5)-(9)) to develop a crisp value direct-relation matrix for each evaluator.

Utilizing Table 4 with the linguistic assessments by manager1, we will exemplify the normalization and crisping for GSDP a2 to a1. A linguistic value of VH or the grey scale value of 1

21x! = [0.75, 1] is currently assigned for this comparison by manager1. Essentially, it means

WP2010-15

10

that manager1 believes GSDP a2 has a very high influence on GSDP a1. Since the minimum value for each column j ( 1

minij

jx! ) is 0 for manager1 and the maximum value for each column j

( 1max

ijj

x! ) is 1 (from Table 4), our max

min! = 1. This characteristic, using expressions (5) and (6),

results in the normalized grey scale upper and lower values equivalent to the original grey scale upper and lower values (i.e. !1 1

ij ijx x! =! for all values of manager1’s normalized X matrix). This

situation may not always be true especially when moderate rates of influence predominate amongst the evaluators.

For our example problem given that !121x! =0.75 and !121x! =1 and using expression (8) we

get the following normalized crisp value for the influence relationship between GSDP a2 on GSDP a1 by manager1 ( 1

21Y ).

( )( )

2

21

0.75(1 0.75) (1 1)0.95

1 0.75 1Y

! + "= =

! +

To compute the final crisp value associated in the modified-CSCF approach, we utilize expression (9) for calculating 1

21z :

1

21z =0+0.95(1) = 0.95

Step 1d. In this sub-step we aggregate the crisp direct-relationship matrices, from step 1c, for each of the evaluators into one overall crisp direct-relationship matrix Z. We initially assign a linguistic weight and translate that weighting to a grey scale value weight on the linguistic grey values scale shown in Table 5.

Table 5 about here

The linguistic weight value for each evaluator is based primarily on the level of direct relationship they have with suppliers, control of supplier programs, as well as the extent of their overall experience. Given these items the managers agreed that the overall linguistic values for each manager are as shown in Table 6. Clearly, in this case example, the evaluators felt that the supplier management department (manager1) weight should be largest.

Table 6 about here

After these assignments have been made, we ‘crisp’ the evaluator weightings using the modified-CSCF approach. The crisp importance weight values for evaluators 1-4 (

pw ) are 0.397,

0.227, 0.227, and 0.150, respectively. Using the weighted aggregation formula (expression (11)) the overall crisp direct-relationship matrix Z is derived and shown in Table 7.

Table 7 about here

WP2010-15

11

Step 2: Using the overall direct relations matrix Z the normalized directed-relationship matrix N (Table 8) is acquired using the formulas (12)-(13). To arrive at this matrix N, the normalization value is s = 1/9.037 = 0.1107.

Table 8 about here

Step 3: The total direct-relationship matrix (T) is determined using expression (14) and shown in Table 9.

Table 9 about here

Step 4: This step, as mentioned in the overall stages discussion also has a number of sub-steps. We shall present each one separately.

Step 4a: Using expressions (15)-(16), we obtain the degree of the influence for each supplier development program (Table 10). A quick glance at these initial results (the R-sum and D-sum columns of Table 10) shows that GSDP a5 (informal supplier environmental evaluation and feedback), a8 (strong formal supplier environmental evaluation), and a11 (requiring ISO 14000 certification for suppliers) all were relatively strong direct influencers on each other and other GSDPs (all with values over 3.5). GSDP a8 and a11, along with a3 (giving green technological advice) all had strong effects on them from the other GSDP.

Table10 about here

Step 4b: Using expressions (17) and (18) we calculate the overall prominence (Pi) and net effect (Ei) for the GDSPs. Not surprisingly, the most prominent GSDPs make sure that a formal supplier evaluation system is in place and that the suppliers have ISO 14001 certification. Both of these programs seem to be most prominent with strong cause and effect linkages. That is, having a basic system in place for planning and monitoring both internally and externally seems to be critical for Company A’s GSDP. This telecommunications company favors suppliers managing their own environmental program (through ISO 14001 certification) but, they do view it as important for having a GSDP that continuously monitors these supplier programs.

Step 4c: The graphical representation (the prominence-causal diagram) and digraphical relationships are now constructed. This step will allow a clearer visualization of the structure and relationships amongst the GSDPs. One of the first activities of this sub-step is to plot the various GSDPs on a two-axes the prominence horizontal axis (R+D) and the net cause/effect vertical axis (R-D). We do this to help us observe general patterns and relationships amongst all the programs simultaneously and in pairs. For example, we see that a16 have very little influence/effect on the other programs, and is more of an effect or influenced by others. That is, the company does not feel it is necessary to develop specific criteria for suppliers for entry into a supplier development program, and its development will not greatly influence the development of other programs.

The development of the digraphs (arrows) in Figure 2 shows the interrelationships amongst each of the individual GSDPs. Since the number of relationships can include all the possibilities, we only mapped those relationships that are over the threshold ! . Due to the large number of GSDP programs, we decided to maintain a high threshold value. We calculated this

WP2010-15

12

value by taking the mean and standard deviation of the values tij from the T matrix. Thus, ! = 0.226. All the relationships that meet or exceed the threshold are bolded in the T matrix (Table 9). We then plot these dyadic relationships. Two-way significant relationships are represented by dashed lines, while one-way relationships are represented by solid lines. In doing this plotting of the arrow the concentration of relationships exist among the more prominent GSDPs.

Figure 2 about here

Discussion of Results

The methodology is valuable in making a number of managerial and organizational observations, some of which we have touched upon in our earlier discussion. Not surprisingly, in our first major observation, the major foundational element (cause) that is required is top management support (a13) for GSDP, which has the highest cause score on our DEMATEL prominence-causal diagram. It also has a strong influence on the most prominent GDSP of requiring suppliers to have ISO 14001 certification. The GSDPs that are most prominent, furthest to the right on the graph, are also most strongly connected, as evidenced by the most significant digraph relationships. The literature is relatively clear on this issue (e.g. (Zhu, Sarkis, Cordeiro, & Lai, 2008)) without the necessary resources allocation and support of top management, the rest of the programs are bound to fail. Unfortunately, none of the particular GSDPs can greatly influence the support of top management. It is not clear that a failure of programs may cause top management support to wane. We did not seem to capture this dimension directly, but it may be implicit that all programs have some, albeit very minor, influence on top management support. The level of prominence may not be so critical in Company A’s case since top management support has existed. If there was a need to convince Company A top management of the importance of these programs or to get their ‘buy-in’ then some pilot GSDP and their outcomes may be necessary to influence top management support.

GSDP a4, the eco-design related advice to suppliers seemed not to be too influential or important to the management team. In fact, we had earlier eliminated eco-design partnerships from inclusion (due to specificity and redundancy) in our analysis. This overall result seems to reinforce the lack of importance to the cross-functional set of managers, who are also relatively unclear with its role to other programs. Interestingly, management did not feel that criteria to incorporate suppliers into GSDP was important, nor linked to other programs. This result is surprising since in many organizations certain suppliers may be included into GSDP based on their environmental or continuity criticality dimensions or other criteria (Trowbridge, 2001; Rossi, Charon, Wing, & Ewell, 2006). Supplier participation overall seemed to be not as critical to this organization, as evidenced by the a17 (supplier participation in green procurement/production), a10 (environmental information sharing) and a9 (joint/team problem solving) GSDPs. This immature level of collaboration in green supply chain practices typically means that the strategic development of GSDP is still lacking in the organization (Sarkis, 2003). The importance of monitoring and a unidirectional approach to green supply chain management in this organization is evidenced by the relative influence and importance of GSDP a6 (supplier assessment programs). Thus, cooperation and collaboration is not viewed as important, which could be due to the level of technological complexity and/or lack of pressures for the telecommunications company to make sure that its products and their design have to meet

WP2010-15

13

specific environmental requirements (e.g. waste electrical and electronic equipment (WEEE) or the Restriction of the use of certain hazardous substances in electrical and electronic equipment (ROHS) -like requirements).

We can also observe general clusters into cause and effect groups. Generally the GSDPs that are part of the effect cluster include a1,a3,a7,a9,a10,a11,a12,a16,and a17; the cause cluster includes a2,a4,a5,a6,a8, a13, a14 and a15. Overall characteristics of many of these programs have been discussed, but a hands-off focus rather than collaborative GSDP programs seems to be a prevalent characteristic of the programs that are less influential and have net effects rather than causes. As the organization matures and evolves, it may be expected that additional programs with a strategic relational focus may be implemented. Currently, a more traditional observe and control approach is utilized through observations from the GSDP prominence, influences and relationships.

A Sensitivity Analysis

We provide some sensitivity analysis to the grey-based DEMATEL methodology. This sensitivity analysis will allow us to determine whether the potential biases of a particular manager may have influenced the results obtained in our case study. It will also be helpful to see how a sensitivity analysis may also be completed in this methodology from a methodological generalizability aspect. A sensitivity analysis may have been completed in a variety of ways, such as changing the level of influence amongst various GSDPs or the level of weighting given to a particular manager. We decided in this exemplary sensitivity analysis to focus on the latter approach of adjusting the weights of the most ‘influential’ manager (manager1) and provide some insights into the results.

To complete this sensitivity analysis we decided to focus on the weightings given to

manager1. Initially this manager had a ‘very high’ linguistic weighting (i.e. [0.7,1.0] grey scale weighting as shown in Table 6). We decided to alter this weighting to ‘high’ [0.5, 0.9], ‘medium’ [0.4,0.7], and ‘low’ [0.3,0.5] importance weight levels for evaluator manager1 respectively, leaving the weights for each of the other managers the same. After repeating the methodology for three additional runs (for each of the new manager1 weightings), we result in three additional prominence-causal DEMATEL diagrams (see figure 3).

Figure 3 about here

Overall, it seems that the results are not overly sensitive to the changes in the importance weight of manager1. But, if we take a careful look at some of the individual GSDPs we see some significant alterations (sensitivity). One specific example is GSDP a4 (ecodesign feedback). Initially it represented one of the least influential (large effect side), less prominence programs (Figure 2/3a). Looking at figure 3d, it rises to the level and prominence of many of the other programs (e.g. a1). What seems to be happening in this circumstance is that the level of importance from the evaluators shifts to the R&D manager (manager2). Not surprisingly, this manager is more concerned with the supplier involvement in product or service design. Also, this manager has more experience with the suppliers, thus building up collaborative and trusting relationships with suppliers is expected to have occurred, pulling up some of the more

WP2010-15

14

collaborative and longer-term supplier development programs. We also see this situation with GSDP a9 which increased in prominence and influence (albeit still below others). GSDP a9 and a3 are almost at the same levels with a slight shift towards less ‘advice’ giving and more sharing. Again, this may be symptomatic of the other managers (with greater experience and fewer suppliers to manage) developing the more collaborative relationships and relying on suppliers to be more involved in GSDP.

One additional observation in the sensitivity solutions is the growth of causal importance of program a14 (building top management commitment at suppliers organization) and decrease in a6 (developing a supplier assessment program). The increase in program a14 essentially has the role of using the buyer to help induce acceptance of GSDP and green supply chain management by organizations by encouraging top management support in their suppliers. This encouragement can usually be effective with longer term relationships that have been developed. Also a decrease in a6 may be due to the increasing trust from these longer term relationships, thus less need for assessment programs.

So, in this sensitivity analysis we did observe some biases from manager1 due to the relative inexperience (compared to other managers) and possibly the more traditional adversarial practices of purchasing/supply chain departments and their suppliers.

Summary and Conclusion

In this paper we introduced and described how grey-based DEMATEL can be a valuable managerial tool to evaluate green supplier development programs (GSDP) and their relationships to each other and importance for the organization. We introduced a practical application/case study of this technique in a telecommunications provider. The technique involved multiple managers providing insights into the relationships amongst the GSDPs. For this organization we found that even though they have implemented a number of GSDPs, their focus was relatively adversarial and distant (hands-off) and less collaborative with its suppliers. We also found, through a sensitivity analysis, that this may not be the case for all functions and managers that deal with suppliers in this company. These insights were achieved from the initial and sensitivity analysis portion of the study.

The managerial implications and usefulness of the technique are clear. Managers are able

to determine what GSDP within their organization need more attention (high cause, high importance) and which ones may be given less priority. They can also complete sensitivity analyses in ways that allow them to determine the stability of their observations. Over time, they may also use it to help plan the direction of their organization in this area by determining how particular GSDPs influence other ones. For example, if they wish to further develop existing GSDPs or other programs the digraphs provide clear relationships on which GSDPs should be emphasized to insure greater success of programs. Investments and prioritization in more foundational and prominent programs should be emphasized.

Researchers can also find this tool valuable for either broad based study, or as in our case,

for single in-depth cases. For example, comparative analyses across case study organizations may help identify particular characteristic of the relationships of GSDP. We observed, in this

WP2010-15

15

case, that a more traditional relationship was evident for the company. Other companies may not have similar general relationships. The prominence-causal diagrams may be used to identify the various structures and relationships of different companies whether based on size, industry, product family or other characteristics typically used for comparative analyses of cases. Longitudinally, the results may also change depending on the strategic emphasis of the organization and these observations may be completed over time with the same case company.

These are just some examples of implications, research and managerial that can be derived from this methodology. But, even with the insights provided in this methodology, there are still limitations, which may set the stage for future research. One major limitation is the evaluation effort required with this technique. For this study evaluators had to each complete over 272 comparisons. Fatigue is easily a possibility which may cause some reliability problems. One way to overcome this situation is to group the factors and complete a hierarchical evaluation of the relationships (similar to the analytical hierarchy process). That is, a smaller set of sub-groups of similar factors could be compared and at a higher level general groups may be compared to determine their relationships. Some information is lost, but analysis may be improved with fewer factors. The evaluation of the ‘importance’ of the factors in these types of studies is based on the connectedness and level of influence of factors on each other. This influence may only be observational and not necessarily be an importance characteristic of the factor. For example, even a program that does not have a strong causal relationship to other programs may be critical to the company due to a strategic direction the company wishes to pursue. This information is not necessarily captured by the methodology. Incorporation and inclusion of other ranking approaches (e.g. the analytical hierarchy process or rough set theory) may provide ways of overcoming this limitation. In addition we utilized a given grey-scale value for a linguistic variable. These values were assumed. Careful consideration and investigation of grey scale/linguistic assignments and dispersion of the grey scale could be investigated to determine the sensitivity of the solution.

All these limitations represent future directions for research. In addition, future research into a comparative analysis of this across multiple companies may be of interest to determine the generalizability of GSDP relationships. The application to other supply chain and environmental sustainability issues may also be an interesting direction to take this research. We also identified the relationships of existing programs, expanding to planned programs and their fit with existing programs may be completed with this technique. Linkage and observations to strategic programs overall is another direction for research.

Overall, as we have seen this approach is flexible and useful for application in a broad

variety of managerial and decision environments. We believe this research sets the stage for significant work in foreseen and unforeseen directions. This paper provides an important step into further research on greening the supply chain with formal modeling.

WP2010-15

16

References Bai, C., & Sarkis, J. (2010a). Green supplier development: Analytical evaluation using rough set theory. Journal of Cleaner Production, In Press, Corrected Proof. Bai, C., & Sarkis, J. (2010b). Integrating sustainability into supplier selection with grey system and rough set methodologies. International Journal of Production Economics, 124(1), 252-264. Das, A., Narasimhan, R., & Talluri, S. (2006). Supplier integration--Finding an optimal configuration. Journal of Operations Management, 24(5), 563-582. Deng, J.-L. (1989). Introduction to Grey System Theory. Journal of Grey Systems, 1(1), 1-24. Dunn, S. C., & Young, R. R. (2004). Supplier Assistance Within Supplier Development Initiatives. The Journal of Supply Chain Management, 40(3), 19-29. Fontela, E., & Gabus, A. (1976). The DEMATEL observer. DEMATEL 1976 report. Geneva, Switzerland: Battelle Geneva Research Center. Gabus, A., & Fontela, E. (1973). Perceptions of the world problematique: Communication procedure, communicating with those bearing collective responsibility. DEMATEL Report No. 1. Geneva, Switzerland: Battelle Geneva Research Centre. Ilgin, M. A., & Gupta, S. M. (2010). Environmentally conscious manufacturing and product recovery: a review of the state of the art. Journal of Environmental Management, 91(3), 563-591. Krause, D. R., & Ellram, L. M. (1997). Critical elements of supplier development: The buying-firm perspective. European Journal of Purchasing and Supply Management, 3(1), 21-31. Krause, D. R., Handfield, R. B., & Scannell, T. V. (1998). An empirical investigation of supplier development: reactive and strategic processes. Journal of Operations Management, 17(1), 39-58. Lawson, B., Cousins, P. D., Handfield, R. B., & Petersen, K. J. (2009). Strategic Purchasing, Supply Management Practices and Buyer Performance Improvement: An Empirical Study of UK Manufacturing Organizations. International Journal of Production Research, 47(10), 2649-2667. Lee, A. H. I., Kang, H.-Y., Hsu, C.-F., & Hung, H.-C. (2009). A green supplier selection model for high-tech industry. Expert Systems with Applications, 36(4), 7917-7927. Lee, P. K. C., & Humphreys, P. K. (2007). The role of Guanxi in supply management practices. International Journal of Production Economics, 106(2), 450-467. Li, P., Tan, T. C., & Lee, J. Y. (1997). Grey Relational Analysis of Amine Inhibition of Mild Steel Corrosion in Acids. Corrosion, 53(3), 186-194.

WP2010-15

17

Liou, J. J. H., Tzeng, G.-H., & Chang, H.-C. (2007). Airline safety measurement using a hybrid model Journal of Air Transport Management, 13(4), 243-249. McGovern, T., & Hicks, C. (2006). Specifications and supplier development in the UK electrical transmission and distribution equipment industry. International Journal of Production Economics, 104(1), 164-178. Modi, S. B., & Mabert, V. A. (2007). Supplier development: Improving supplier performance through knowledge transfer. Journal of Operations Management, 25(1), 42-64. Narasimhan, R., Mahapatra, S., & Arlbjørn, J. (2008a). Impact of relational norms, supplier development and trust on supplier performance. Operations Management Research, 1(1), 24-30. Narasimhan, R., Talluri, S., & Mahapatra, S. (2008b). Effective response to RFQs and supplier development: A supplier's perspective. International Journal of Production Economics, 115(2), 461-470. Opricovic, S., & Tzeng, G. H. (2003). Defuzzification within a multicriteria decision model. International Journal of Uncertainty. Fuzziness and Knowledge-Based Systems, 11(5), 635–652. Piplani, R., Pujawan, N., & Ray, S. (2008). Sustainable supply chain management. International Journal of Production Economics, 111(2), 193-194. Rao, P., & Holt, D. (2005). Do green supply chains lead to competitiveness and economic performance? International Journal of Operations & Production Management, 25(9), 898 - 916. Rossi, M., Charon, S., Wing, G., & Ewell, J. (2006). Design for the Next Generation: Incorporating Cradle-to-Cradle Design into Herman Miller Products. Journal of Industrial Ecology, 10(4), 193-210. Sarkis, J. (1999). How Green is the Supply Chain? Practice and Research. SSRN eLibrary. Sarkis, J. (2003). A Strategic Decision Making Framework for Green Supply Chain Management. Journal of Cleaner Production, 11(4), 397-409. Seuring, S., & Müller, M. (2008a). Core issues in sustainable supply chain management - a Delphi study. Business Strategy and the Environment, 17(8), 455-466. Seuring, S., & Müller, M. (2008b). From a literature review to a conceptual framework for sustainable supply chain management. Journal of Cleaner Production, 16(15), 1699-1710. Srivastava, S. K. (2007). Green supply-chain management: A state-of-the-art literature review. International Journal of Management Reviews, 9(1), 53-80.

WP2010-15

18

Trowbridge, P. (2001). A case study of green supply-chain management at Advanced Micro Devices. Greener Management International, 35, 121-134. Tzeng, G. H., Chiang, C. H., & Li, C. W. (2007). Evaluating intertwined effects in e-learning programs: a novel hybrid MCDM model based on factor analysis and DEMATEL. Expert Systems with Applications 32, 1028-1044. Wagner, S. M., & Krause, D. R. (2008). Supplier development: communication approaches, activities and goals. International Journal of Production Research, 47(12), 3161-3177. Williams, S. J. (2006). Managing and developing suppliers : can SCM be adopted by SMES? International Journal of Production Research, 44(15-16), 3831-3846. Wu, W.-W., & Lee, Y.-T. (2007). Developing global managers' competencies using the fuzzy DEMATEL method. Expert Systems with Applications, 32(2), 499-507. Zhu, Q., & Sarkis, J. (2004). Relationships between operational practices and performance among early adopters of green supply chain management practices in Chinese manufacturing enterprises. Journal of Operations Management, 22(3), 265-289. Zhu, Q., Sarkis, J., Cordeiro, J. J., & Lai, K.-H. (2008). Firm-level correlates of emergent green supply chain management practices in the Chinese context. Omega, 36(4), 577-591.

WP2010-15

19

Figure .1. Graphical Representation of the Grey-DEMATEL methodology for evaluation of Green Supplier Development Programs.

Definegreyscalesforevaluation

Constructagreydirect‐relationmatrixXforeachevaluator.

Morethanoneevaluator?

Determinegreyscaleratingforeachevaluator

Usemodified‐CFCSmethodtocrispevaluatorratings.

Useweightedaveragingtodetermineoverallcrispmatrix

MakeeachevaluatormatrixacrispmatrixZ.

Determinenormalizeddirect‐relationmatrix(N).

Constructprominence‐causaldiagraph

Grey‐basedDEMATELMethodology

Yes

Normalizegreydirect‐relationmatrixXforeachevaluator.

No

Determinetotalrelationmatrix(T).

MethodologyRequirementsGreenSupplierDevelopmentPrograms

EvaluationTeamInputsfromEvaluationTeam

GreyScalesforEvaluationTeam

WP2010-15

20

Table 1 A comprehensive listing and categorization of environmentally-oriented (green) supplier development practices and activities. Green Knowledge Transfer and Communication

Training supplier employees on environmental issues Train suppliers in stakeholder expectations Train users in environmental capabilities Train suppliers on environmental and cost controls Giving green manufacturing related advice and awareness raising for suppliers Giving green technological advice to suppliers Giving ecodesign product development related advice to suppliers (e.g. processes, project management) Conduct training and education programs for supplier personnel Supplier environmental evaluation and feedback Develop supplier environmental assessment programs Providing feedback about supplier environmental performance Strong formal supplier environmental evaluation Setting environmental improvement targets for suppliers Auditing suppliers Joint and team problem solving on environmental issues Information sharing on environmental topics Ongoing communication with supplier community via supplier environmental councils

Investment and Resource Transfer Invest in improvement of transaction processes Reduce supplier environmental costs Solve supplier environmental technical problems Finance supplier major capital environmental expenditures Transferring supplier employees with environmental expertise to buying firm Transferring employees with environmental expertise to suppliers Investment in supplier capacity building Supplier rewards and incentives for environmental performance

Management and Organizational Practices Requiring ISO 14000 certification for suppliers Long-term contracts with environmental dimensions incorporated Introduce a cross-functional supply chain team with environmental presence Building top management commitment/support within buyer organization for green supply practices Building top management commitment/support for supplier organization for green supply practices Organization management has formal long-term plans improve supplier performance Formal process for supplier development Identification of high-performing critical suppliers for cost reduction and other improvement opportunities Criteria established about when to enter suppliers into green supplier development programs Formal process to identify supplier cost reduction targets The participation level of suppliers in the eco-design stage. Having suppliers participate in the process of green procurement and production.

Sources: (Krause et al., 1998; Sarkis, 1999; Dunn & Young, 2004; Das, Narasimhan, & Talluri, 2006; Williams, 2006; Modi & Mabert, 2007; Narasimhan et al., 2008a; Wagner & Krause, 2008; Lawson, Cousins, Handfield, & Petersen, 2009)

WP2010-15

21

Table 2 Final case supplier development programs

GSDP Green supplier development programs a1 Training supplier employees on environmental issues a2 Train suppliers in stakeholder expectations a3 Giving green technological advice to suppliers a4 Giving eco-design product development related advice to suppliers (e.g. processes, project management) a5 Supplier environmental evaluation and feedback a6 Develop supplier environmental assessment programs a7 Providing feedback about supplier environmental performance a8 Strong formal supplier environmental evaluation a9 Joint and team problem solving on environmental issues a10 Information sharing on environmental topics a11 Requiring ISO 14000 certification for suppliers a12 Long-term contracts with environmental dimensions incorporated a13 Building top management commitment/support within buyer organization for green supply practices a14 Building top management commitment/support for supplier organization for green supply practices a15 Formal process for supplier development a16 Criteria established about when to enter suppliers into green supplier development programs

a17 Having suppliers participate in the process of green procurement and production

WP2010-15

22

Table 3 The grey linguistic scale for the respondents’ assessments

Linguistic terms Grey numbers

No influence(N) [0,0] Very low influence(VL) [0,0.25]

Low influence(L) [0.25,0.5] High influence(H) [0.5,0.75]

Very high influence(VH) [0.75,1]

WP2010-15

23

Table 4 The Linguistic scale direct-relation matrix for supplier development programs by manager1.

GSDP a1 a2 a3 a4 a5 a6 a7 a8 a9 a10 a11 a12 a13 a14 a15 a16 a17

a1 N VH VH H VH L VH VH H VL L N VL H VL N N a2 VH N VH VH VL N N H L L VH VH H VH N N VH a3 VH VH N VL VH VH VH VH VH VH VH VH L VH VH VL VL a4 L VH H N N N L VL N N H VL VL VL L L H a5 VH H VH H N L VH VH VH VH L H VL VH VL VL N a6 N VH VH N VH N VH L H H VH VH VH H VH VH N a7 VL N VH N VH VL N VH VH VH VH VH L VH N N N a8 VH L VH H VH VL VH N VH VH VH VH VH VH H L VL a9 N N VH N H N VH H N VH VL VL N L N N VH a10 L N VH N L L VH VH VH N VH VH VL H N N L a11 VH VH VH L L H VH VH L L N H VL VH VH VL N a12 L VH VH L H VH H H VL VL H N L H L L VL a13 N L L L VL VH VL L N N VH L N N H VL N a14 VH VH VH VL L VL L L H VL H H N N N N N a15 L H VH N H H H H VL VL VH L VH VL N VH L a16 N N N N N L N N N N H N N N H N N a17 N N N VH N N N N H H N N VL H N N N

WP2010-15

24

Table 5 Linguistic scales for the importance weight of evaluators

Linguistic variable Grey values

Very low [0,0.3]

Low [0.3,0.5]

Medium [0.4,0.7]

High [0.5,0.9]

Very high [0.7,1.0]

Table 6 Grey scale importance weight of evaluators.

Evaluator Linguistic

Weighting

Grey Scale

Value Weights Manager1(Supplier Management department)

ententertainmententdepartment)

Very High [0.7,1]

Manager2(R&D department) Medium [0.4,0.7]

Manager3(Customer Service department) Medium [0.4,0.7]

Manager4(Marketing department) Low [0.3,0.5]

WP2010-15

25

Table 7 The overall crisp direct-relationship matrix Z for the case.

GSDP a1 a2 a3 a4 a5 a6 a7 a8 a9 a10 a11 a12 a13 a14 a15 a16 a17 a1 0.000 0.689 0.622 0.503 0.681 0.443 0.551 0.543 0.419 0.181 0.433 0.159 0.034 0.258 0.084 0.025 0.025

a2 0.749 0.000 0.600 0.600 0.257 0.229 0.229 0.484 0.368 0.368 0.671 0.603 0.272 0.402 0.096 0.096 0.492

a3 0.530 0.533 0.000 0.179 0.527 0.595 0.528 0.531 0.593 0.527 0.592 0.524 0.153 0.391 0.533 0.176 0.176

a4 0.469 0.625 0.384 0.000 0.300 0.232 0.439 0.382 0.303 0.232 0.603 0.246 0.119 0.113 0.374 0.374 0.493

a5 0.749 0.435 0.664 0.424 0.000 0.562 0.810 0.950 0.617 0.546 0.739 0.547 0.048 0.470 0.237 0.265 0.232

a6 0.375 0.261 0.596 0.166 0.662 0.000 0.680 0.671 0.490 0.490 0.671 0.603 0.405 0.351 0.612 0.680 0.303

a7 0.189 0.170 0.603 0.174 0.594 0.252 0.000 0.739 0.748 0.677 0.739 0.671 0.167 0.538 0.323 0.303 0.303

a8 0.617 0.380 0.693 0.500 0.891 0.255 0.680 0.000 0.680 0.732 0.792 0.739 0.476 0.606 0.558 0.255 0.184

a9 0.093 0.099 0.550 0.242 0.498 0.232 0.609 0.484 0.000 0.752 0.186 0.327 0.041 0.313 0.232 0.303 0.609

a10 0.232 0.099 0.596 0.166 0.308 0.371 0.606 0.656 0.681 0.000 0.535 0.598 0.048 0.283 0.167 0.096 0.306

a11 0.748 0.688 0.619 0.381 0.442 0.629 0.680 0.724 0.379 0.439 0.000 0.552 0.323 0.742 0.823 0.323 0.303

a12 0.300 0.547 0.619 0.381 0.543 0.609 0.490 0.484 0.260 0.305 0.484 0.000 0.309 0.561 0.442 0.504 0.323

a13 0.011 0.167 0.305 0.305 0.181 0.609 0.252 0.433 0.311 0.303 0.724 0.501 0.000 0.371 0.629 0.391 0.371

a14 0.473 0.547 0.551 0.194 0.442 0.323 0.442 0.433 0.568 0.323 0.605 0.552 0.312 0.000 0.371 0.303 0.371

a15 0.288 0.357 0.475 0.098 0.543 0.490 0.490 0.552 0.252 0.252 0.814 0.365 0.547 0.181 0.000 0.680 0.442

a16 0.025 0.028 0.106 0.106 0.104 0.235 0.096 0.159 0.096 0.025 0.416 0.221 0.099 0.161 0.419 0.000 0.161

a17 0.096 0.099 0.106 0.483 0.164 0.164 0.164 0.227 0.493 0.493 0.227 0.227 0.119 0.419 0.232 0.085 0.000

WP2010-15

26

Table 8 The normalized direct-relation matrix (N).

GSDP a1 a2 a3 a4 a5 a6 a7 a8 a9 a10 a11 a12 a13 a14 a15 a16 a17 a1 0.000 0.076 0.069 0.056 0.075 0.049 0.061 0.060 0.046 0.020 0.048 0.018 0.004 0.029 0.009 0.003 0.003

a2 0.083 0.000 0.066 0.066 0.028 0.025 0.025 0.054 0.041 0.041 0.074 0.067 0.030 0.045 0.011 0.011 0.054

a3 0.059 0.059 0.000 0.020 0.058 0.066 0.058 0.059 0.066 0.058 0.066 0.058 0.017 0.043 0.059 0.019 0.019

a4 0.052 0.069 0.043 0.000 0.033 0.026 0.049 0.042 0.034 0.026 0.067 0.027 0.013 0.013 0.041 0.041 0.055

a5 0.083 0.048 0.073 0.047 0.000 0.062 0.090 0.105 0.068 0.060 0.082 0.061 0.005 0.052 0.026 0.029 0.026

a6 0.041 0.029 0.066 0.018 0.073 0.000 0.075 0.074 0.054 0.054 0.074 0.067 0.045 0.039 0.068 0.075 0.034

a7 0.021 0.019 0.067 0.019 0.066 0.028 0.000 0.082 0.083 0.075 0.082 0.074 0.018 0.060 0.036 0.034 0.034

a8 0.068 0.042 0.077 0.055 0.099 0.028 0.075 0.000 0.075 0.081 0.088 0.082 0.053 0.067 0.062 0.028 0.020

a9 0.010 0.011 0.061 0.027 0.055 0.026 0.067 0.054 0.000 0.083 0.021 0.036 0.005 0.035 0.026 0.034 0.067

a10 0.026 0.011 0.066 0.018 0.034 0.041 0.067 0.073 0.075 0.000 0.059 0.066 0.005 0.031 0.018 0.011 0.034

a11 0.083 0.076 0.068 0.042 0.049 0.070 0.075 0.080 0.042 0.049 0.000 0.061 0.036 0.082 0.091 0.036 0.034

a12 0.033 0.060 0.068 0.042 0.060 0.067 0.054 0.054 0.029 0.034 0.054 0.000 0.034 0.062 0.049 0.056 0.036

a13 0.001 0.018 0.034 0.034 0.020 0.067 0.028 0.048 0.034 0.034 0.080 0.055 0.000 0.041 0.070 0.043 0.041

a14 0.052 0.060 0.061 0.021 0.049 0.036 0.049 0.048 0.063 0.036 0.067 0.061 0.034 0.000 0.041 0.034 0.041

a15 0.032 0.039 0.053 0.011 0.060 0.054 0.054 0.061 0.028 0.028 0.090 0.040 0.060 0.020 0.000 0.075 0.049

a16 0.003 0.003 0.012 0.012 0.011 0.026 0.011 0.018 0.011 0.003 0.046 0.024 0.011 0.018 0.046 0.000 0.018

a17 0.011 0.011 0.012 0.053 0.018 0.018 0.018 0.025 0.055 0.055 0.025 0.025 0.013 0.046 0.026 0.009 0.000

WP2010-15

27

Table 9 The total-relation matrix (T).

GSDP a1 a2 a3 a4 a5 a6 a7 a8 a9 a10 a11 a12 a13 a14 a15 a16 a17 a1 0.116 0.178 0.213 0.141 0.203 0.156 0.200 0.210 0.176 0.144 0.205 0.152 0.066 0.141 0.117 0.086 0.091

a2 0.199 0.117 0.220 0.159 0.169 0.145 0.177 0.212 0.179 0.169 0.238 0.204 0.096 0.165 0.128 0.099 0.146

a3 0.190 0.181 0.178 0.124 0.215 0.195 0.227 0.239 0.219 0.203 0.253 0.216 0.093 0.178 0.186 0.122 0.125

a4 0.154 0.163 0.175 0.082 0.153 0.127 0.175 0.180 0.153 0.138 0.209 0.149 0.071 0.118 0.140 0.117 0.133

a5 0.235 0.194 0.276 0.167 0.188 0.213 0.284 0.310 0.250 0.230 0.298 0.245 0.094 0.210 0.178 0.145 0.146

a6 0.184 0.164 0.254 0.131 0.242 0.147 0.257 0.269 0.223 0.212 0.280 0.239 0.127 0.187 0.209 0.184 0.147

a7 0.155 0.145 0.241 0.123 0.222 0.162 0.174 0.260 0.237 0.221 0.267 0.233 0.094 0.196 0.167 0.135 0.139

a8 0.231 0.198 0.292 0.181 0.288 0.196 0.284 0.229 0.267 0.258 0.320 0.276 0.143 0.233 0.221 0.154 0.152

a9 0.110 0.103 0.190 0.103 0.171 0.124 0.192 0.189 0.123 0.192 0.164 0.157 0.061 0.137 0.123 0.108 0.143

a10 0.134 0.114 0.208 0.103 0.166 0.149 0.205 0.219 0.203 0.125 0.211 0.196 0.068 0.145 0.127 0.095 0.119

a11 0.239 0.224 0.277 0.166 0.239 0.226 0.275 0.294 0.230 0.222 0.233 0.252 0.128 0.240 0.242 0.158 0.159

a12 0.164 0.181 0.236 0.142 0.211 0.194 0.217 0.229 0.182 0.175 0.240 0.158 0.109 0.192 0.176 0.154 0.138

a13 0.109 0.119 0.173 0.115 0.147 0.173 0.164 0.192 0.158 0.150 0.231 0.183 0.065 0.150 0.177 0.129 0.126

a14 0.175 0.175 0.222 0.119 0.194 0.160 0.205 0.216 0.206 0.172 0.241 0.208 0.104 0.128 0.162 0.127 0.138

a15 0.154 0.153 0.212 0.108 0.202 0.177 0.208 0.227 0.172 0.163 0.263 0.189 0.130 0.149 0.126 0.167 0.144

a16 0.050 0.049 0.073 0.048 0.067 0.074 0.071 0.081 0.065 0.054 0.112 0.080 0.040 0.066 0.094 0.039 0.055

a17 0.078 0.075 0.102 0.104 0.098 0.085 0.106 0.118 0.133 0.128 0.122 0.107 0.051 0.113 0.092 0.062 0.057

WP2010-15

28

Table 10 The degree of prominence and net cause/effect values for each green supplier development program.

GSDP R sum D sum R+D R-D

a1 2.595 2.680 5.275 ‐0.085 a2 2.822 2.532 5.354 0.290 a3 3.143 3.543 6.686 ‐0.400 a4 2.437 2.116 4.553 ‐1.107 a5 3.663 3.174 6.836 0.489 a6 3.453 2.702 6.155 0.752 a7 3.171 3.423 6.594 ‐0.252 a8 3.924 3.674 7.598 0.250 a9 2.388 3.175 5.563 ‐0.787

a10 2.588 2.956 5.544 ‐0.368 a11 3.805 3.886 7.691 ‐0.081 a12 3.097 3.244 6.341 ‐0.147 a13 2.563 1.540 4.103 1.023 a14 2.950 2.749 5.699 0.202 a15 2.945 2.663 5.608 0.283 a16 1.119 2.081 3.200 ‐0.962 a17 1.633 2.159 3.792 ‐0.527

WP2010-15

29

Figure .2.The prominence-causal DEMATEL graph with digraph relationships for most influential relationships.

!"#$

!"#%

!%#$

%#%

%#$

"#%

"#$

%#% "#% &#% '#% (#% $#% )#% *#% +#% ,#%

-"

-&

-'

-(

-$

-)

-*

-+

-,

-"%

-""-"&

-"'

-"(

-"$

-")

-"*

R-D

R+D

WP2010-15

30

Figure .3. Sensitivity Analysis DEMATEL prominence-causal graphs for various Manager1 weights.

!

!"#$

!"#%

!%#$

%#%

%#$

"#%

"#$

%#% "#% &#% '#% (#% $#% )#% *#% +#% ,#%-"

-&

-'

-(

-$

-)

-*

-+

-,

-"%

-""-"&

-"'

-"(

-"$

-")

-"*

R-D

R+D

Figure 3a: Manager1 weight = [0.7, 1]

!

!"#$

!"#%

!%#$

%#%

%#$

"#%

"#$

%#% "#% &#% '#% (#% $#% )#% *#% +#% ,#%

R-D

-"

-&

-'

-(

-$

-)

-*

-+

-,

-"%

-""

-"&

-"'

-"(

-"$

-")

-"*

R+D

Figure 3b: Manager1 weight = [0. 5, 0.9 ]

WP2010-15

31

Figure 3 (cont.) Sensitivity Analysis DEMATEL prominence-casual diagrams for various Manager1 weights.

!

!"#$

!"#%

!%#$

%#%

%#$

"#%

"#$

%#% "#% &#% '#% (#% $#% )#% *#% +#% ,#%-"

-&

-'

-(

-$-)

-*

-+

-,

-"%

-""

-"&

-"'

-"(

-"$

-")

-"*

R-D

R+D

Figure 3c: Manager1 weight = [0. 4, 0.7]

!

!"#$

!"#%

!%#$

%#%

%#$

"#%

"#$

&#%

%#% "#% &#% '#% (#% $#% )#% *#% +#%,"

,&

,',(

,$,)

,*

,+

,-,"%

,""

,"&

,"'

,"(

,"$

,")

,"*

R-D

R+D

Figure 3 d: Manager1 weight = [0. 3, 0.5]