Professional Documents
Culture Documents
Institute of Engineering Technology, National Taipei University of Technology, 1, Sec. 3, Chung-Hsiao E. Road, Taipei 10643, Taiwan, ROC
Institute of Environmental Engineering & Management, National Taipei University of Technology, 1, Sec. 3, Chung-Hsiao E. Road, Taipei 10643, Taiwan, ROC
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 12 September 2007
Received in revised form 19 May 2008
Accepted 22 May 2008
Available online 21 July 2008
With increased outsourcing and environmental consciousness, this paper presents an analytic network
process (ANP) approach to incorporate the issue of hazardous substance management (HSM) into supplier selection. In this study, identication of criteria of HSM competence is categorized into four dimensions, a multi-criteria decision model is proposed. ANP is then applied to supplier selection and is
characterized by interdependencies among decision structure components. An illustrative example in an
electronics company is presented to demonstrate how to select a most appropriate supplier in accordance with the requirements of hazardous substance for environmental regulations.
2008 Elsevier Ltd. All rights reserved.
Keywords:
Supplier selection
Analytic network process
Hazardous substance management
Green supply chain management
1. Introduction
Green supply chain management (GSCM) is generally understood to involve screening suppliers based on their environmental performance and doing business only with those that meet
certain environmental regulations or standards [1]. Given growing
environmental concerns during the past decade, a consensus is
emerging that environmental pollution issues accompanying industrial development should be addressed together with supply
chain management, thus contributing to GSCM [2]. Since the Waste
Electrical and Electronic Equipment (WEEE), Restriction of Hazardous Substances (RoHS) and Eco-design for Energy using Products (EuP) directives were passed by the European Union (EU),
GSCM has been adopted as a strategy by leading electronics industry companies, including Dell, HP, IBM, Motorola, Sony, Panasonic, NEC, Fujitsu, and Toshiba [3]. Apparently, large electronics
rms have exerted pressure on their suppliers to achieve better
environmental performance, increasing the motivation for cooperation between suppliers and customers to achieve environmental objectives [4]. Supplier selection in GSCM is clearly a critical
activity in purchasing management [1,5,6], because rm environmental sustainability and ecological performance can be demonstrated by its suppliers [7].
256
analytic network process (ANP) [26,27], case based reasoning systems [28,29] and multiple objective programming [3032]. The application of the mathematical programming model to supplier
selection may have problems in including qualitative criteria, particularly for supplier partnership policies [33]. Furthermore, the
computational complexities inherent in multiple objective programming frequently prohibit consideration of many crucial attributes for supplier selection [25]. Either the weighting model of the
AHP or the ANP can be used since it is more useful for treating
qualitative factors than other models such as mathematical programming models [34]. Both Bayazit [26] and Gencer and Gurpinar
[27] utilized ANP in supplier selection to provide good insights in
terms of systematic feedback and interdependencies. ANP can capture the interdependencies between the criteria under consideration,
thus allowing for more systemic analysis [35]. Additionally, ANP can
be used as a decision analysis tool to solve multi-criteria supplier
selection problems that contain interdependencies [26] and provide
systematic feedback [27]. ANP modeling thus better ts the problem
examined in this study, and offers the advantage of providing a systematic approach to supplier selection for GSCM practice.
In view of the signicance of incorporating the HSM into supplier selection as well as the limitation of previous studies, we
propose a HSM-based supplier selection model by using the ANP
methodology. This study has two objectives: rst, to recognize the
criteria of supplier selection and evaluate with respect to HSM
competency in GSCM; and second, to develop a framework of the
supplier selection processes in GSCM using the ANP approach to
facilitate appropriate supplier selection.
The remainder of this paper is organized as follows. Section 2
discusses selection criteria for suppliers in GSCM in terms of HSM
competency. Next, Section 3 debates the ANP approach. Section 4
then presents an illustrative case for appropriate supplier selection
using the ANP model. Finally, Section 5 presents conclusions and
limitations.
2. Environmental criteria to hazardous substance
management in supplier selection
Based on integrating the categories and criteria identied from
the literature sources, an environmental framework was designed
for incorporating environmental criteria regarding the competency
of HSM into supplier selection in GSCM. The 19 criteria were
determined and categorized into ve main clusters as follows (see
Table 1).
Table 1
Criteria of supplier selection to HSM
Dimension
Criterion
Reference
257
258
259
for 19 criteria under ve main criteria clusters. The relative importance of criteria for supplier selection is determined based on
expert opinion as determined by sampling the companys GSCM
team. The expert group consists of ve people from the case
company who are responsible for the planning of green initiatives,
evaluating the suppliers performance and maintaining the list of
the approved suppliers with respect to the capability of management of hazardous substance. The case experience provided assistance in understanding how to establish the decision model for
supplier selection and selecting appropriate suppliers in GSCM. The
application and analysis of ANP methodology is presented in the
following steps.
4.1. Step 1: model construction and problem formation
The rst step in ANP method implementation is to construct the
decision structure of the supplier selection problem and to identify
the relevant criteria and alternatives developed based on the literature. This model has four levels (see Fig. 1). The second level
consists of ve main criteria clusters or dimensions, they are procurement management (D1), R&D management (D2); process
management (D3), incoming quality control (D4), and management
system (D5). There are 19 criteria under the above-mentioned ve
dimensions. The fourth level is the alternatives they are Supplier A,
Supplier B, and Supplier C in the illustrated case.
4.2. Step 2: pair-wise comparison of dimensions
During this step, the decision-maker is asked to respond to the
relative weighting of each dimension via a pair-wise comparison
matrix. A scale of 19 is used to compare the two components in
this comparison. A score of 1 indicates that the two components
have equal importance whereas a score of 9 indicates the overwhelming dominance of the considered component (row component) over the comparison component (column component). If the
impact of one component is weaker than that of its comparison
Supplier selection
Dimensions
Criteria
Procurement
Management (D1)
R&D
Management (D2)
Requirement of
green
purchasing (D11)
Capability of
green design
(D21)
Management for
hazardous
substances (D31)
Green materials
coding and
recording (D12)
Inventory of
hazardous
substances (D22)
Prevention of
mixed material
(D32)
Inventory of
substitute
material (D13)
Legal-complian
ce competency
(D23)
Process auditing
(D33)
Supplier
Management
(D14)
Alternatives
Process
Management (D3)
Pre-shipment
inspection (D34)
Incoming Quality
Control (D4)
Standard for
incoming
quality control
(D41)
Quality
management
system (D51)
Test equipment
(D42)
Environmental
management
system (D52)
Record of
incoming
quality control
(D43)
Hazardous
substance
management
system (D53)
Information
Systems (D54)
Warehouse
management
(D35)
Supplier A
Management
System (D5)
Supplier B
Supplier C
260
Table 2
Pair-wise comparison of dimensions
Dimension
D1
D2
D3
D4
D5
e-vector
D1
D2
D3
D4
D5
1.0000
1.1832
1.3229
1.5811
1.2910
0.8452
1.0000
1.4142
1.1547
0.7746
0.7559
0.7071
1.0000
0.7746
0.8165
0.6325
0.8660
1.2910
1.0000
0.5000
0.7746
1.2910
1.2247
2.0000
1.0000
0.1553
0.1920
0.2445
0.2425
0.1657
component, it will be scored from 1 to 1/9, with 1 indicating indifference and 1/9 indicating overwhelming dominance of the
column component over the row component. To make a reverse
comparison between already compared components, a reciprocal
value is automatically assigned within the matrix, so that in a matrix aijaji 1. This is complied with the method recommended by
Saaty [47]. In the formation of a pair-wise comparison matrix,
group decision-making may be applied to avoid decision-maker
bias toward particular providers. The authors thus adopted the
geometric mean of the individual judgments based on the recommendations of Dyer and Forman [59].
After the pair-wise comparison matrices are developed, a vector
of priorities (i.e. Eigen-vector or Eigenvector) in each matrix is
calculated and is then normalized to sum to 1.0 or 100%. This study
utilized a two-stage algorithm to calculate the e-vector, which rst
adds the value in each column of the matrix, and then separates
each entry in each column by the total of that column; the normalized matrix is acquired through meaningful comparison among
components. Subsequently, row sums are calculated and normalized to get the e-vectors of these relative importance weights as
shown in Table 2. And these e-vectors are used as Pj in Table 8 to
calculate the desirability indices for alternatives.
In this case, the process management (D3) criterion was found
to have the highest priority (0.2445) in selecting supplier, followed
by incoming quality control (D4) (0.2425), R&D management (D2)
(0.1920), management system (D5) (0.1670), and procurement
management (D1) (0.1657). Both process management and incoming quality control are considered to be the most important
dimensions in selecting an appropriate supplier in terms of HSM.
4.3. Step 3: pair-wise comparisons between dimensions/criteria
In this step, the decision-maker is asked to respond to a number
of pair-wise comparisons in which two components are compared
at a time in terms of an upper lever control criteria. The pair-wise
comparisons of the components at each level are performed regarding their relative inuence toward the control criterion. Pairwise comparison is thus performed between the applicable criteria
within a given dimension cluster. In this model, ve such pair-wise
comparison matrices are formed and the e-vectors acquired from
. One such pair-wise
these matrices are imported from Table 8 asAD
kja
comparison matrix for procurement management dimension is
listed in Table 3. For the pair-wise matrix, the question asked to the
decision-maker is, what is the relative impact on procurement
management by criterion a when compared to criterion b.
In Table 3, the relative importance of requirement of green
purchasing (D11) when compared to green materials coding and
Table 3
Pair-wise comparison for procurement management dimension
D11
D12
D13
D14
D11
D12
D13
D14
e-vector
1.0000
0.3333
1.1180
0.7071
3.0000
1.0000
1.8708
2.8284
0.8944
0.5345
1.0000
0.7746
1.4142
0.3536
1.2910
1.0000
0.3223
0.1186
0.2998
0.2593
Table 4
Pair-wise comparison for requirement of green purchasing under procurement
management dimension
D12
D13
D14
D12
D13
D14
e-vector
1.0000
1.5811
3.1623
0.6325
1.0000
1.7321
0.3162
0.5774
1.0000
0.1764
0.2925
0.5311
recording (D12) with respect to procurement management dimension is three. Table 3 also reveals that the criterion requirement of green purchasing (D11) has the maximum inuence
(0.3223) and green materials coding and recording (D12) has the
minimum inuence (0.1186) on the procurement management dimension. It implies that requirement of green purchasing (D11) has
the most inuence on the dimension of procurement management,
with a priority in selecting best supplier, which is followed by inventory of substitute materials (D13) (0.2998).
4.4. Step 4: pair-wise comparisons for criteria
This step involves performing pair-wise comparisons to consider the interdependencies among the criteria (third level). There
are 19 such pair-wise comparison matrices in this stage. Table 4 lists
one such comparison which represents the result of the procurement management dimension with requirement of green
purchasing as the control criterion over other criteria. The decision-maker is asked to answer the question for evaluating the
interdependencies related to when considering requirement of
green purchasing with respect to increasing procurement management, what is the relative impact of criterion a compared to
criterion b. Table 4 reveals that supplier management (D14) has
the strongest impact (0.5311) on the procurement management
dimension, with requirement of green purchasing (D11) as the
control criterion over others. Furthermore, green materials coding
and recording (D12) has the weakest impact (0.1764). It shows that
supplier management has the most inuence on the requirement of
green purchasing for the dimension of procurement management.
The e-vector from these matrices is thus used to form super matrices. The e-vector from Table 4 has been used in the 19 column of
the super matrix in Table 6.
4.5. Step 5: evaluation of alternatives
The nal set of pair-wise comparisons is made for the relative
impact of each of the alternatives on the criteria inuencing the
dimensions. The number of such pair-wise comparison matrices is
determined based on the number of criteria included in each dimension. The present case involves three alternatives (Supplier A,
B, and C) and 19 criteria, hence 19 such pair-wise comparison
matrices were constructed in this step.
Table 5 lists one example of such a pair-wise comparison matrix,
in which the impacts of alternatives are evaluated based on the
criterion requirements of green purchasing (D11). In comparing the
three suppliers, we asked which supplier is more preferable with
respect to determining the best alternative under the requirements
Table 5
Pair-wise comparison of importance of requirements of green purchasing on the
alternatives
Supplier A
Supplier B
Supplier C
Supplier A
Supplier B
Supplier C
e-vector
1.0000
0.2236
0.3333
4.4721
1.0000
3.4641
3.0000
0.2887
1.0000
0.6141
0.1076
0.2783
261
Table 6
Super matrix M before convergence
D2
D1
D1 D11
D12
D13
D14
D2 D21
D22
D23
D3 D31
D32
D33
D34
D35
D4 D41
D42
D43
D5 D51
D52
D53
D54
D11
D12
D13
D14
0
0.1764
0.2925
0.5311
0.3944
0
0.3447
0.2609
0.4808
0.3445
0
0.1747
0.5902
0.2304
0.1794
0
D21
D3
D22
D23
D4
D31
D32
D33
D34
D35
0
0.4478
0.2448
0.1286
0.1787
0.3466
0
0.3119
0.0911
0.2504
0.3682
0.4203
0
0.0779
0.1336
0.3139
0.1925
0.1369
0
0.3567
0.4003
0.2038
0.1553
0.2406
0
D41
D5
D42
D43
D51
D52
D53
D54
0
0.3333
0.3333
0.3333
0.2403
0
0.3237
0.4359
0.1955
0.3792
0
0.4253
0.1522
0.3153
0.5325
0
0
0.5858 0.5505
0.2612 0
0.4495
0.7388 0.4142 0
0
0.7388 0.7101
0.5505 0
0.2899
0.4495 0.2612 0
Di
J
X
I
Pj AD
kj Akj Sikj
j1
Table 7
Super matrix M after convergence
D2
D1
D1 D11
D12
D13
D14
D2 D21
D22
D23
D3 D31
D32
D33
D34
D35
D4 D41
D42
D43
D5 D51
D52
D53
D54
D11
D12
D13
D14
0.3330
0.1922
0.2109
0.2638
0.3330
0.1922
0.2109
0.2638
0.3330
0.1922
0.2109
0.2638
0.3330
0.1922
0.2109
0.2638
D21
D3
D22
D23
D4
D31
D32
D33
D34
D35
0.2646
0.2556
0.1877
0.1146
0.1772
0.2646
0.2556
0.1877
0.1146
0.1772
0.2646
0.2556
0.1877
0.1146
0.1772
0.2646
0.2556
0.1877
0.1146
0.1772
0.2646
0.2556
0.1877
0.1146
0.1772
D41
D5
D42
D43
D51
D52
D53
D54
0.1626
0.2560
0.2913
0.2897
0.1626
0.2560
0.2913
0.2897
0.1626
0.2560
0.2913
0.2897
0.1626
0.2560
0.2913
0.2897
262
Table 8
Desirability indices
Dimensions
D1
Pj
Enablers
0.1553
D11
0.1553
D12
0.1553
D13
0.1553
D14
0.1920
D21
D2
0.1920
D22
0.1920
D23
0.2445
D31
D3
0.2445
D32
0.2445
D33
0.2445
D34
0.2445
D35
0.2425
D41
D4
0.2425
D42
0.2425
D43
0.1657
D51
D5
0.1657
D52
0.1657
D53
0.1657
D54
Desirability indices Di
Normalized desirability indices DiN
AD
kj
AIkj
S1kj
S2kj
S3kj
Supplier A
Supplier B
Supplier C
0.3223
0.1186
0.2998
0.2593
0.3070
0.1655
0.5275
0.4078
0.3315
0.1167
0.0558
0.0882
0.3364
0.4140
0.2496
0.2469
0.1457
0.2071
0.4003
0.3330
0.1922
0.2109
0.2638
0.3610
0.2632
0.3758
0.2646
0.2556
0.1877
0.1146
0.1772
0.4204
0.3097
0.2699
0.1626
0.2560
0.2913
0.2897
0.6141
0.1402
0.1333
0.6582
0.5249
0.6144
0.6144
0.1739
0.1565
0.1801
0.5691
0.1827
0.5336
0.3813
0.1696
0.1419
0.1827
0.3633
0.5214
0.1076
0.5751
0.4933
0.1122
0.2296
0.1661
0.2195
0.4368
0.4557
0.3049
0.1643
0.4692
0.2332
0.3888
0.4152
0.5490
0.4692
0.4075
0.2823
0.2783
0.2847
0.3734
0.2296
0.2455
0.2195
0.1661
0.3893
0.3878
0.5150
0.2666
0.3481
0.2332
0.2299
0.4152
0.3091
0.3481
0.2292
0.1963
0.0102
0.0005
0.0013
0.0070
0.0112
0.0051
0.0234
0.0046
0.0032
0.0010
0.0009
0.0007
0.0183
0.0119
0.0028
0.0009
0.0011
0.0036
0.0100
0.1177
0.4060
0.0018
0.0020
0.0048
0.0012
0.0049
0.0014
0.0084
0.0115
0.0094
0.0016
0.0003
0.0018
0.0080
0.0121
0.0068
0.0037
0.0029
0.0041
0.0054
0.0921
0.3177
0.0046
0.0010
0.0037
0.0024
0.0052
0.0018
0.0063
0.0103
0.0080
0.0028
0.0004
0.0013
0.0080
0.0071
0.0068
0.0021
0.0022
0.0023
0.0038
0.0801
0.2763
5. Conclusion
The GSCM based conceptual framework and operational model
to incorporate HSM into supplier selection have been presented. By
identifying the related criteria of HSM activities for the proposed
framework, an ANP methodology was applied to an electronics
company.
263
264