Omega 42 (2014) 109123

Contents lists available at ScienceDirect

Omega
journal homepage: www.elsevier.com/locate/omega

Applications

A comprehensive environment friendly approach for supplier selection

Amit Kumar a, Vipul Jain a, Sameer Kumar b,n
a
b

Department of Mechanical Engineering, Indian Institute of Technology Delhi, Hauz Khas, New Delhi-110016, INDIA
Opus College of Business, University of St. Thomas, MN 55403-2005, USA

art ic l e i nf o

a b s t r a c t

Article history:
Received 17 November 2012
Accepted 11 April 2013
Processed by B. Lev
Available online 21 April 2013

As the green movement spreads across the globe, organizations are under pressure to reduce the
emissions across their supply chain. On the other hand, they need to cut supply costs to gain a
competitive edge. This paper proposes Green DEA (GDEA), a comprehensive approach based on Data
Envelopment Analysis (DEA) with carbon footprint monitoring. GDEA builds on an existing DEA model
with weight restrictions and dual role factors, and introduces carbon footprints as necessary dual role
factors with weight restrictions. Unlike other researches, GDEA incorporates heterogeneous suppliers
and also takes into account regional emission compliance standards and laws. GDEA encourages
suppliers to go green and cut down their carbon footprints and comply with emission norms along
with reducing costs in order to survive competition. Results from model validation in a well-known
automobile spare parts manufacturer in India are presented to verify the GDEA approach. Overall, GDEA
cuts across a huge and varied supplier base, caters to almost all businesses, is environment-friendly and
robust.
& 2013 Elsevier Ltd. All rights reserved.

Keywords:
Data envelopement analysis
Supply chain management
Carbon footprinting
Dual role factors
Emission norms

1. Introduction
Supplier selection is the process by which suppliers are
reviewed, evaluated, and chosen to become a part of the organization's supply chain [66]. This area has been highly researched and
is of extreme importance, especially in organizations where
purchasing has a signicant impact on the revenues. For such
organizations a comprehensive approach for decision making is
highly desirable in order to have the global edge in wake of
growing competition.
As the climate change movement gathers momentum, the need
of the hour is a comprehensive supplier selection strategy which
models supply costs as well as emissions. Traditional supplier
selection models focus on supplier's economic efciency while
ignoring the ecological efciency of the supplier. Such models
force suppliers to cut costs in order to survive competition. As
more and more environmental norms and compliance standards
are enforced into practice, organizations which are simply looking
at cutting supply chain costs are most likely to get stranded as
there is no guarantee that the suppliers would conform to these
norms. On the other hand, organizations should aim towards
spreading carbon awareness amongst the suppliers, along with
offering incentives to them to become environment-friendly.
Carbon footprints provide a precise, accurate and robust measure
n

Corresponding author. Tel.: +1 651 962 4350; fax: +1 651 962 4710.
E-mail addresses: skumar@stthomas.edu,
sameerkumar724@gmail.com (S. Kumar).
0305-0483/$ - see front matter & 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.omega.2013.04.003

that can t in as an environmental impact criterion for supplier

selection. Thus, the decision making approach in suppliers selection must take into account the environmental impact of the
suppliers vis--vis their carbon footprints.
Organizations are now demanding that their suppliers reduce
their carbon footprints while doing so themselves. The non-prot
Carbon Disclosure Project (CDP) in association with consultant
rm T. Kearney has cited in a report that some organizations have
even gone to the extent of dismissing suppliers who have no
carbon management plans of their own. A strong interest in
carbon footprinting and management has been indicated by the
responses therein. 63% Of these companies have a formal,
documented corporate climate change strategy, as per the report.
The remaining 37% have general guidelines, and 90% of all
companies surveyed have plans in place to reduce carbon emissions. Also seen in the report is the desire in these organizations
for their partners to turn green. A total of 89% of all respondents,
according to the survey, have an established strategy to work on
carbon related topics with suppliers [56].
Since there is no surety that suppliers will conform to environmental norms and compliance standards, even as more and more
of these are enforced into practice, the organizations most likely to
get stranded are the ones looking simply at cutting supply chain
costs. Rather, efforts should be made towards spreading carbon
awareness among suppliers by organizations, and may be even
offer incentives to them on being eco-friendly. This requires,
however, a collaborative approach on the part of both the buyer
and the supplier. The buyer too, should be prepared to take

110

A. Kumar et al. / Omega 42 (2014) 109123

extreme measures and threaten to de-select the supplier, in case

the supplier does not pay heed. On the other hand, the approach
for decision making in supplier selection should appreciate the
element of risk involved here.
This paper aims to propose Green DEA (GDEA), a common
integrated approach for supplier selection considering both the
objectives, cost cutting, and environmental efciency, in a exible
manner. The proposed model has a new capability to enforce
emission compliance norms and can be tailored to region specic
emission constraints. This is primarily a compliance issue which
none of the current supplier selection approaches address.
The GDEA approach is applied to a well-known automobile
spare parts manufacturer in India. The results demonstrate that
the GDEA model considers the environmental factors in a twopronged approach. Firstly, it lets the carbon footprints determine
the efciency of the supplier and secondly, the model penalizes a
supplier for not meeting emission standards. The approach provides the user with the exibility to adjust the impact of environmental factors in the overall supplier selection process.
The rest of the paper is organized as follows: Section 2
discusses the literature reviewed in brief. Section 3 gives a brief
introduction to carbon footprinting and highlights the steps
involved in estimation of carbon footprints. Section 4 presents
the proposed GDEA model with the mathematical formulation.
Section 5 demonstrates a numerical example and the results with
the GDEA approach. Section 6 presents research ndings, implications and managerial insights from the approach. Finally, the
concluding observations and remarks are given in Section 7.

2. Literature review
This section is structured as follows. Section 2.1 presents the
literature on various supplier selection approaches. Section 2.2
presents literature on DEA and DEA for supplier selection. Section
2.3 presents literature on Green Supply Chain Management and
green supplier selection approaches. Section 2.4 presents the
literature on carbon footprinting. Finally Section 2.5 presents
current gaps in literature and motivation for this research.
2.1. Literature on supplier selection approaches
Various mathematical approaches such as linear programming
[25,69], fuzzy programming [41], and genetic algorithms [78] have
been used to solve the supplier selection problem. A majority of
these approaches are based on the fact that supplier selection is a
multiple criteria problem, and the aim is to assign weights to all
the criteria in order to get an overall supplier rating.
Traditional approaches to supplier selection made use of
Analytical Hierarchy Process (AHP) in combination with a mathematical programming approach. Pi and Low [59], Haq and
Kannan [31], and Kahraman et al. [41] propose AHP models to
solve the supplier selection problem. Chan [10] proposed an AHP
based model for supplier selection. Chan and Chan [12] offered an
AHP based supplier selection model with a case study in advanced
technology industry. Chan et al. [14] proposed an AHP based
decision support system for supplier selection in airlines industry.
Chan and Chan [13] have applied an AHP based supplier selection
model in the apparel industry.
Chan and Kumar [15] used Fuzzy Extended AHP (FEAHP) for a
global supplier selection problem. Chan et al. [16] have also used a
similar fuzzy AHP approach for global supplier selection with an
example from manufacturing industry.
A variety of linear and integer programming approaches have
been proposed in literature. Ghodsypour and OBrien [24] proposed an integration of AHP and linear programming to consider

both tangible and intangible factors in choosing the best suppliers

and giving them optimal order quantities so that the total
purchasing value is maximized. Ghodsypour and OBrien [25]
presented a mixed integer non-linear programming approach to
solve the multiple sourcing problem. The approach takes into
account the total cost of logistics, including net price, storage,
transportation and ordering costs.
Verma and Pullman [82] analyze the differences between
managerial perceptions and actual choices in a supplier selection
process. An interesting outcome of this study is that though
managers perceive quality as the most important attribute in
selection, the actual choice is based more on cost and delivery
performance.
Humphreys et al. [36] presented the ndings from large
companies in Hong Kong in terms of supplier criteria requirement
and serves as a reference guideline for collaborative relationships
from the context of a supplier coming from a diverse background.
Jain et al. [39] proposed a supplier evaluation model using
evolutionary fuzzy based approach. This approach is primarily
dependent on linguistic nature of attributes associated with the
suppliers.
Bayazit [2] proposed an ANP approach to supplier selection.
Shyur and Shih [72] proposed a hybrid model for supplier selection in new task situations, which uses the Multi-Criteria Decision
Making (MCDM) in combination with Analytic Network Process
(ANP). These ANP models go one step ahead of the traditional AHP
approaches which tend to assume that the selection criteria are
independent of each other.
Kumar et al. [47] presented a Fuzzy Goal Programming
approach for Vendor Selection Problem (VSP). In this approach,
VSP is formulated as a Fuzzy Multi-Objective Integer Programming
problem with three important goals of cost minimization, quality
maximization and maximization of on-time delivery.
Ting and Cho [81] proposed an integrated approach using
Linear Programming (LP) which integrates AHP with linear
programming. The approach considers both tangible and
intangible factors in choosing best suppliers and giving them
optimal order quantities. The objective is to maximize total
purchasing value.
Chan et al. [11] presented a simulation study on Suppliers
Flexibility Levels (SFL). This helped a decision maker to identify
the information system automation level and variations in physical
characteristics of the alternative suppliers with respect to SFL.
Jain et al. [40] discussed supplier-related issues in modeling a
dynamic supply chain. It provided a review of main approaches to
these issues. The study provides a groundwork for diverse modeling approaches to be applied to the supplier selection problem.
Deng and Chan [23] presented the new fuzzy dempster MCDM
method for supplier selection. The approach combined the Fuzzy
Set Theory (FST) and Dempster Shafer Theory of evidence (DST).
Mansini et al. [55] developed an Integer Programming based
solution for supplier selection problem with quantity discounts
and truckload shipping. Experimental results provide valuable
insights into the impact of discounts and shipping on effective
sourcing strategies.
A major issue with all these approaches is that they assign
arbitrary weights which are subjective and based on surveys and
questionnaires. This leads to inaccurate results as it is very difcult
to accurately assign numbers to preferences. Also, these models do
not scale well as the number of performance criteria are increased.
Danese [21] analyzes the impact of supplier integration on
buyer's performance based on measures related to creating a fast
supply network structure. The ndings provide useful guidelines
for managers looking forward to supplier integration and also
nds relevance with managers looking at a comprehensive supplier selection approach.

A. Kumar et al. / Omega 42 (2014) 109123

111

Table 1
DEA approaches for supplier selection.
Reference

Approach used

Scope

[44]
[84]
[53]
[5]
[79]
[80]
[64]
[65]
[63,66]
[67]

DEA
Multi-objective programming (MOP) with DEA
DEA
DEA
Improvement on BCC model
Chance constrained DEA (CCDEA)
Imprecise DEA (IDEA)
Assurance region IDEA (AR-IDEA)
DEA
DEA with weight restrictions and dual role factors

Uses DEA to track performance of suppliers

Evaluating number of suppliers in procurement situation
Supplier evaluation for individual product
DEA based supplier selection to help purchasing in a dynamic market
Supplier performance evaluation and monitoring
Supplier selection with uncertain factors
Ranking suppliers in presence of imprecise data
Ranking suppliers in presence of weight restrictions and imprecise data
Supplier selection with non-homogeneous suppliers
Supplier selection in presence of weight restriction and dual role factors

2.2. Literature on DEA approaches for supplier selection

DEA provides a robust approach to supplier selection. The basic
DEA model that has been used is the BankerCharnesCooper
(BCC) model proposed by Banker et al. [1]. An explanation to the
basic DEA approaches is given in Cooper et al. [19]. Cook et al. [18]
discussed the dual role factors in DEA.
Weber [83] presented a DEA approach for supplier selection
based on multiple criteria and identied benchmark values which
can be used. Since then, a number of DEA approaches have been
proposed for supplier selection. The approaches reviewed are
summarized in Table 1.
Apart from supplier selection, DEA has been used for a variety
of other problems which involve multi-criteria decision making.
Samoilenko and Osei-Bryson [68] use DEA for monitoring
efciency-based performance of productivity-driven organizations.
2.3. Literature on green supply chain management (GSCM) and
green supplier selection approaches
Many organizations are moving towards green supply chain
management. Wharton Institute and Initiative for Global Environmental Leadership [38] presented a special report on the business
case for life-cycle analysis and building a green supply chain. The
article emphasized on the fact that life-cycle analysis, judiciously
applied, can help a company recognize lost opportunities for
increasing protability and simultaneously meet the climate goals.
Leadership Management International (LMI) Government Consulting [54] presented a white paper on GreenSCOR, a robust green
supply chain analytical tool. This tool uses Supply Chain Operations Reference (SCOR) model developed by Supply Chain Council
(SCC) as a foundation. This is essentially because SCOR is a proven
supply chain tool. GreenSCOR integrates environmental management into SCOR which provides a foundation for improving
operational activities while reducing the environmental impacts.
Kassinis and Soteriou [42] studied the impact of environmental
practices on performance of a supply chain. The paper explores the
impact of these practices on external prot chain as well. The
study is validated on data from European hospitality industry and
the ndings suggest that the impact of environmental practices is
indeed positive.
On similar lines, Pil and Rothenberg [60] explored the impact of
improving environmental performance on manufacturing
improvements. Empirical results indicate that attaining superior
environmental performance enables superior quality. The article
highlights the synergy between environmental improvement and
manufacturing improvement.
Noori and Chen [58] proposed a method for planning and
developing products with environmental attributes. The study
proves that successful development of such breakthrough products
creates a competitive advantage and also benets the environment.

Zhu and Sarkis [89] presented the relationship between Green

Supply Chain Management practices and its impact on performance in Chinese manufacturing enterprises. The study uses
hierarchical regression analysis and presents signicant ndings
and managerial implications for China as a developing country.
Kleindorfer et al. [43] presented a comprehensive review of
sustainable operations management practices covered in the rst
50 issues of Journal of Production and Operations Management.
The article concludes with some future research challenges in the
area of sustainable operations management.
Gonzlez-Benito and Gonzlez-Benito [26] study the relationship between environmental pro-activity and business performance on a set of 186 industrial companies. The outcomes are
split in terms of positive synergy between environmental efciency and business performance. However, there are enough
indicators from the study to prove that environmental management can bring competitive opportunities for companies.
Corbett and Klassen [20] argued that environmental excellence
can be the key to improving operations and is not a deterrent to
efciency. The idea is explained through the principle areas of
quality management and supply chain management. The environmental perspective extends the quality denition of customers to
stakeholders, defects to any form of waste. Similarly in supply chain,
there was more attention on reverse ows, product disposal which
would ultimately improve overall supply chain performance.
Brickman and Ungerman [6] discussed importance of climate
change management in business strategies especially purchasing.
The article presents ndings from a survey of more than 2000
global executives. Majority of respondents consider the importance of climate change in purchasing, yet they were unlikely to
report that it was considered in practice.
Lee and Klassen [51] discussed the drivers and enablers for
environmental management capabilities in small and mediumsize suppliers. The ndings indicate that buyers GSCM initiated
and then enabled the improvement of suppliers environmental
capabilities. The outcome was a synergistic linkage in buyer
supplier relationship, resource acquisition, and capability development. The article serves as a motivation for a buyer initiated
green supplier selection approach.
Zhu et al. [90] investigated the correlation of organizational
learning and extent of adoption of GSCM practices in Chinese
manufacturing rms. The study found signicant positive relationships between the two if a number of inuences such as regulations, marketing, cost pressures etc. can be controlled.
Ramudhin et al. [61-62] introduced a Mixed Integer Programming (MIP) formulation for Carbon Market SensitiveGreen
Supply Chain Network Design. Carbon trading considerations
are integrated in the supply chain network design phase itself.
The model analyzes trade-offs between cost and CO2 emissions
and also allows offsetting CO2 emissions through supply chain reengineering and carbon trading.

112

A. Kumar et al. / Omega 42 (2014) 109123

Diabat and Simchi-Levi [22] presented an optimization model

using MIP for green supply chain management, which integrates
environmental management (inclusive of carbon emission consequences) and its impact into supply chain. The ndings indicate that
carbon emission allowance and supply chain costs have an inverse
relationship. The supply chain managers should take this into
account while doing the long term planning to stay competitive.
Chen and Monahan [17] analyzed impacts of different pollution
control policies on various supply chain decisions such as production planning and inventory control. The study establishes that
traditional approach which suggests that environmental waste
reduction at the end of process also decreases the stock and
throughput levels is not true. Instead, an environmental compliance standard that limits total waste can increase the total
planned stock level, which can lead to higher environmental
waste before the standard enforcement as well as environmental
risks at other stages. This can be reversed by a voluntary control
approach which provides exibility to occasionally exceed the
standard.
Hofer et al. [32] presented an interesting study on competitive
drivers of a rm's environmental management (EM) activities in
the US manufacturing industries. The research investigates competitive interactions among leader and challenger rms in EM
domain. The ndings reveal that such rivalries exist and a rival's
past EM activity has a greater effect on a rm's EM activity for
more protable and smaller rms. This rivalries behavior of rms
serves as a motivation to build a supplier selection model which
makes use of EM criteria as a competitive edge and promote
healthier and greener competition.
Lai and Wong [49] presented an interesting study on Green
Logistics Management (GLM) done on Chinese manufacturing
exporters. It was found that GLM affects both environmental and
operational performance in a positive manner and the regulatory
pressure adds to the GLM-performance relationship.
Environmental regulations and legislations are inuencing
supply chains to adopt a lifecycle model. Guide et al. [27] and
Guide and Wassenhove [28-29] presented the idea of closed-loop
supply chains and the associated challenges. Closed-loop supply
chain management maximizes value creation over the entire lifecycle of a product. There is signicant interest in European Union
in closed-loop supply chains because of legislation [27,70].
Subramanian et al. [73] discussed various trade-offs among a
rm's compliance strategy. The strategy is realized in a marketbased program where a regulator auctions a xed number of
emission permits. The article concludes that in the presence of
competition, relatively clean rms can benet from a reduction in
number of available permits. This scenario is similar to an allowance
trading scenario and thus provides a motivation to build a generic
emission framework in the supplier selection paradigm.
There are very few approaches proposed in literature for green
supplier selection which take into account the environmental
impact of a supplier. Noci [57] presented a green supplier rating
system for assessment of a supplier's environmental performance.
The study provided a three phase approach for implementing a
green supplier selection procedure which takes into account the
corporate green strategies and the suppliers' key environmental
performance factors.
Bowen et al. [4] emphasized on the role of supply management
capabilities in green supply. The authors argued that green supply
is better implemented by focusing on development of internal
resources rather than external environmental pressures. The
article stresses on the fact that green supply can be built by a
proactive corporate environmental stance and a more strategic
purchasing approach.
Handeld et al. [30] proposed an AHP based model to evaluate
the relative importance of various environmental factors and assesses

suppliers based on these factors. Various case studies are presented

to demonstrate the strengths and weaknesses of the approach. The
paper also explored how AHP can be a part of comprehensive
Environmentally Conscious Purchasing (ECP) system.
Humphreys et al. [37] presented the various approaches organizations are adopting to respond to environmental management
requirements. The paper also presents a comprehensive list of
environmental criteria for green supplier selection.
Humphreys et al. [35] proposed a case based reasoning
approach for evaluating a suppliers environmental performance.
A comprehensive environmental evaluation framework has been
presented in this study.
Yan [88] presented a model which uses adaptive Genetic
Algorithm with AHP. AHP is applied to dynamically adjust green
supplier evaluation index weights. This was primarily to enhance
objectivity and efciency of system evaluation of green suppliers
and to reduce asymmetric information on green supply chain. Lee
et al. [50] proposed a green supplier selection model. The model
used fuzzy AHP to deal with the subjectivity attached to the
environmental factors that are considered as these are based on
questionnaires and surveys.
Hu et al. [33] presented an integrated approach for green
supplier selection using Multi Criteria Decision Making (MCDM).
The ndings from an electronics manufacturer in Taiwan indicated
that four most important criteria for supplier selection are carbon
governance, management systems of carbon information, training
related to carbon management, and carbon policy. The ndings
also indicated that these criteria affect the other green supplier
selection criteria for carbon management competencies.
Kuo et al. [48] presented a green supplier selection model
which integrates Articial Neural Networks (ANN) with DEA and
ANP. The model overcame traditional DEA drawbacks such as,
limitations of data accuracy and Decision Making Units (DMUs)
amounts constraint.
Kumar and Jain [45] proposed a Green Supplier Selection model
based on Data Envelopment Analysis and carbon footprinting but
without any data validation.
2.4. Carbon footprinting
Carbon footprinting is an accurate technique to deal with
subjectivity in environmental factors. It provides a precise and
accurate measure to judge a supplier's eco-efciency. The term
carbon footprint is commonly used to describe the carbon dioxide
(CO2) and other Greenhouse Gas (GHG) emissions for which an
individual or organization is responsible. Carbon footprints can
also be calculated for events or products [7]. There are other
denitions available for carbon footprints. Wiedmann and Minx
[85] explored the discrepancy between various usages of the term
and suggested a scientic denition based on commonly accepted
accounting principles and modeling approaches.
Sundarkarni et al. [77] presented a model for carbon footprint
estimation across a supply chain. The model used long-range
Lagrangian and Eulerian transport methods. This model is extremely useful where exact inputs for carbon footprinting are not
available for an organization.
2.5. Gaps in literature
Following are the gaps identied in literature and the motivation for our research:
1. A major issue with traditional approaches is that they assign
arbitrary weights which are subjective and based on surveys
and questionnaires. This leads to inaccurate results as it is very
difcult to accurately assign numbers to preferences.

A. Kumar et al. / Omega 42 (2014) 109123

2.

3.

4.

5.

This is the motivation for using DEA, which provides a

robust approach to supplier selection.
An integrated approach for supplier selection which incorporates the regional environmental laws is missing in literature.
This is the primary motivation for modeling environmental
constraints within the approach.
There is no comprehensive approach taking into account
carbon footprints as an environmental impact measure.
This is the motivation for using carbon footprints as a
comprehensive environmental impact criterion.
The approaches in literature for green supplier selection are
limited and not exible in terms of giving preferences to cost
efciency or environmental efciency.
This is the major motivation to make the proposed model
exible.
No integrated approach for supplier selection dealing with
non-homogeneity within the DEA framework is proposed.
This provided the motivation to account for nonhomogeneity within the integrated approach.

3. Carbon footprinting1
Carbon (CO2) emission results from a variety of activities that
humans undertake in their daily lives. Growing levels of human
attributed CO2 emissions are known to cause global warming.
Thus, today there is a pressing need to control these emissions in
order to arrest global warming. The term carbon footprinting
refers to the total amount of CO2 or green house gases emissions
an individual or organization is responsible for [7]. Full carbon foot
printing encompasses a wide range of emissions which include
direct emissions from organizational activities, emissions from the
use of electricity, and indirect emissions from products and
services [7].
In the context of supplier selection, carbon footprinting provides a concrete performance measure for the suppliers environmental efciency. Carbon footprinting is also undertaken to reduce
emissions over time, to identify and quantify the key emissions
sources, to report the footprint accurately to a third party, to fulll
requests from business/ retail customers or investors, to ascertain
level of emissions needed to offset, and to become carbon neutral.
Carbon footprinting essentially involves the following activities
[7]:

Calculating direct emissions and emissions from electricity

e.g. Onsite fuel usage, onsite electricity usage, transports which

individuals use.
Calculating basic carbon foot prints, converting the fuel, electricity and transport consumption gures into CO2 equivalents
by using the standard emission factors. The scope of this
activity should also include material consumption (cars, buildings, services, food etc.)

The exact steps for carbon footprinting are described next [7].
These steps are based on process analysis.
1. Dene the methodologyit is advisable to use a methodology
that is accurate, complete, widely accepted and understood
so that the results produced are more credible and comparable.
One commonly used methodology is the GHG protocol
produced by the World Resources Institute (WRI) and
the World Business Council for Sustainable Development
(WBCSD) [87].
1
This section has been prepared mostly with the help of experts in eld and
guidelines available on the Carbon Trust website http://www.carbontrust.com/.

113

2. Specify the boundary and scope of coveragethis is one of

the most important steps as footprint estimation would largely
depend on the boundary. Boundary denition may include/
exclude treatment of emissions from completely or partially
owned subsidiaries, treatment of emissions from leased assets,
such as from a van which is leased from a hired company. It
should also consider the types of emissions to be included CO2/
GHG, direct emissions from fuel onsite use and transport,
emissions from manufacturing process, electricity purchased,
organization's supply chain and other indirect activities.
Scott et al. [71] discuss the importance of carbon footprint
estimation boundaries.
Boundary denition should make sure that it does not need a
truncation of the supply chain, especially upstream truncation.
A lot of conventional carbon footprint estimation approaches
suffer from truncation errors. These are overcome by hybrid
approaches which combine Life-Cycle Assessment (LCA) with
InputOutput analysis. Suh et al. [75] outlines hybrid
approaches for dealing with truncation errors in supply chain
LifeCycle Assessment (LCA). Lenzen [52] discusses truncation
errors in detail and presents strategies to avoid such errors
using hybrid methods.
In the context of supplier selection, the scope may exclude any
supplier activities which do not cater to a buying organization.
As an example, consider a supplier supplying product A to
organization X and product B to organization Y. Thus, when
organization X has to make a supplier selection decision based
on carbon footprints, it can exclude product B from its scope.
However, this is just a guideline. Organizations that are very
conscious of their own carbon footprint might just extend the
scope of footprinting for the suppliers as well.
3. Collect emissions data and calculate the footprintthis step
involves collecting data based on onsite fuel consumption,
owned transport utilization, emissions from chemical reactions
in manufacturing processes or from land use or agricultural
activities, electricity consumption, employee travel by air, rail
and in vehicles not owned by the organization, and also
suppliers emissions. Data on energy consumption can be
translated into equivalent CO2 emissions data using standard
emission factors. These factors may be specic to a country. A
set of widely accepted emission factors are available from
Department for the Environment, Food and Rural Affairs
(DEFRA) and reproduced on the Carbon Trust website. Emissions of other greenhouse gases must be translated into
equivalent emissions data in tCO2e (tonnes of CO2 equivalent),
using the global warming potential factors published by DEFRA
and available from the Carbon Trust.
4. Verify resultsit is very important to lend credibility to carbon
foot printing by verifying it from a consultancy or accounting
rm. Greater levels of assurance or verication are more
onerous and expensive to achieve, but provide greater condence in the results. This also ensures that the footprints are
widely acceptable. In the context of supplier selection, organizations should employ a standard verication procedure for
carbon footprints calculated by the suppliers.
5. Disclose the footprintwhether the footprint is disclosed in
advertising material, a Corporate Social Responsibility (CSR)
report or other collateral, the supplier should ensure that the
data is presented transparently, providing full information
about the process followed and what the information means.

This procedure reects practices carried out under process

analysis. Recent developments in the eld of carbon footprinting
have hybridized process analysis with InputOutput Analysis. Such
approaches are referred to as hybrid IO-LCA approaches. Suh and

114

A. Kumar et al. / Omega 42 (2014) 109123

Huppes [74], and Suh and Nakamura [76] discuss the hybrid IOLCA approach in detail.
In supplier selection context, the boundary estimation with
only LCA is most likely to be affected by truncation error. This error
is systematic and not stochastic as one doesnt know how
incomplete a process analysis is. Thus, carbon footprint estimation
with a LCA only approach could not be used to make any
comparisons between suppliers on this basis.
Huang et al. [34] presented a GHG accounting approach based
on InputOutput models. The results can be used to develop a
complete upstream carbon footprint for screening purposes. The
article presents an analysis of the issues with application of cut-off
thresholds. This is extremely useful for decision makers as they are
well informed about expanding efforts in gaining progressively
greater accuracy for informed purchasing, investing, claiming
carbon credits, and policy-making. In supplier selection context,
this knowledge is very valuable and is highly recommended while
setting thresholds for data collection.
The end result is a carbon footprint in tCO2e. One important
point to note is that carbon footprints can be released as an annual
gure or a per product gure. A per product gure would be
suitable in case of supplier selection for one product. For multiple
products, the footprints can be combined or an annual footprint
can be used. This would be used in the supplier selection decision
making approach, which is presented in the next section.
3.1. Carbon footprinting and environmental standards in India
This section explains the implications of carbon footprinting
process in India along with the environmental regulations in place.
Apart from the standard carbon footprinting process described
previously, following emission factors to be used for footprint
estimation:
1. Electricity conversion consumption factors (CO2 per kWh) to be
acquired from CO2 emission factor database, version 08 (January, 2013), Central Electricity Authority (CEA), Government
of India.
2. Petrol and diesel consumption factors (CO2 per litre) from the
emission factors from across the sector tool section at the
GHG protocol website.
3. LPG consumption factor (CO2 per kg) from the emission factors
from across the sector tool section at the GHG protocol
website.
4. Conversion of other GHG to tCO2 can be done using the global
warming potential factors by DEFRA as these are not Geography
specic and depend only on chemical composition of the GHGs.
National standards have been developed for efuents and
emissions under the statutory power of Air (Prevention and
Control of Pollution) Act, 1981, by the Central Pollution Control
Board (CPCB), India. These standards have been approved and
notied by the Government of India, Ministry of Environment and
Forests, under Section 25 of the Environmental (Protection) Act,
1986. The latest amendment to these rules is called the Environment (Protection) (Third Amendment) Rules, 2012. Industry
specic standards as per latest amendment are available on CPCB
website at www.cpcp.nic.in [9].

4. Proposed GDEA model

The model proposed by Saen [67] which considers both dual
role factors and weight restrictions is presented below. The model
does not demand exact weights from the decision maker. Suppliers undergoing the selection process are referred to as Decision

Making Units (DMUs) in DEA terminology. The relative weight

restrictions are applicable only in respect of DMU0 being assessed,
leaving free the relative virtual values of the comparative DMUs.
First, the notation used is explained.
4.1. Notation
Parameters
J
r
i
f
yrj
xij
yr0
xi0
wfj
ci, di

1,,n collection of suppliers (DMUs)

1,,s the set of outputs
1,,m the set of inputs
1,,F the set of dual role factors
rth output of jth DMU
ith input of jth DMU
rth outputs of DMU0 under investigation
ith inputs of DMU0 under investigation
fth dual role factor of jth DMU
user specied constants for weight restrictions

Decision variables
weight of rth output
weight of ith input
output weight of fth dual role factor
input weight of fth dual role factor

ur
vi
f
f

The DEA formulation with weight restriction and dual role

factors is given in Saen [67].
s

r1

f 1

f 1

Max ur yr0 f wf 0 f wf 0

subject to:
m

vi xi0 1

i1
s

r1

f 1

f 1

i1

ur yrj f wf j f wf j vi xij 0

ci

vi xi0 vi xi0 0

i1

vi xi0 di

vi xi0

i1

!
0

vi

ur

f ; f 0

The objective function (1) is inclusive of the output weights ur,

as well as weights of dual role factors f and f on output side and
input side respectively. Constraint (2) is the standard DEA constraint
to linearize the objective function (1). Constraint (3) is the basic
DEA constraint inclusive of weights from dual role factors. Constraints (4) and (5) are virtual input and virtual output weight
restrictions applied to DMU0 respectively. Constraints (6)(8)
impose lower bounds on the weights. The model presents a unied
approach to deal with weight restrictions and dual role factors.
We propose three modications in the GDEA model described
above:
1. Consider carbon footprints as essential dual role factors with
weight restrictions.

A. Kumar et al. / Omega 42 (2014) 109123

2. Introduce emission constraints to model country specic environmental norms and penalize suppliers for not meeting the norms.
3. Alter the formulation to account for heterogeneous suppliers.
These modications are explained in detail next.
In the proposed model, carbon footprints are considered as
dual role factors in the above approach and following generic
constraint (9) is introduced to model emission constraints.
w10 ECAP

Since, carbon footprint is a necessary dual role factor, let f 1,

denote carbon footprints.
Here, ECAP is the maximum emission quantity in TCO2e which
every DMU can emit. The above constraint puts an upper limit on
the emission quantity. The maximum emission cap can be DMU
specic, or based on the geographic location of the DMU. For sake
of simplicity, we consider a common emission cap. This can
however be extended to a DMU specic emission cap.
Constraint (9) has a small aw. In order to satisfy this constraint,
the input weight of carbon footprint 1 might be forced to zero. Since
we want the carbon footprints to go down in order to have a better
eco-efciency, we want these to have a higher input weight which is
quite contradictory to what the constraint will enforce. Thus, we need
another constraint to ensure that carbon footprints behave more as an
input factor. Constraint (10) does exactly that and along with
Constraint (11) ensures that the emission cap is satised by the carbon
footprints. Here, the parameter would be the minimum carbon
footprint weight as an input factor in the DEA model. More the value
of , more the efciency would be dependent upon carbon footprints.
Thus, this parameter must be set carefully. This is different from
modeling carbon footprints as an input factor as it is desired by the
model to give out that inference.
1 1

10

Since, the above emission constraint cannot be satised by all

DMUs, we introduce an allowance variable ej for each DMU which
goes into the objective function as a negative term. So Constraint
(9) becomes.
w10 e0 ECAP

11

Constraints (10) and (11) together provide an effective

approach to encourage suppliers with higher emissions to cut
down on the emissions in order to survive the competition. Also,
Constraint (11) provides a generic framework to model environmental regulations discussed in Section 2.4.
An important point to note here is that if the expected behavior
is to minimize the carbon footprints, then it should be an input
factor strictly. However, Constraint (11) ensures this behavior and
lets the model decide the input and output weight for carbon
footprints, rather than keeping footprints as a strict input factor.
Another modication to above constraint would be to put a
total emission cap on all DMUs, but that would be incorrect as the
total emission cap should only apply to the selected suppliers, and
not all DMUs.
The objective becomes,
s

r1

f 1

f 1

Max ur yr0 f wf 0 f wf 0 Pe0

12

here, P is a penalty which enforces a stringent emission norm on

the supplier, i.e. if the supplier does not meet the emission norm
then its relative efciency is bound to go down.
The model still does not account for heterogeneity, i.e., it
cannot handle the case when some suppliers do not comprehensively consume common inputs to supply common outputs. Saen
[63] presented an approach using series mean technique and AHP
to supply the missing values of inputs and outputs for such

115

suppliers. Saen [66] presented another approach to supply the

missing values with expert opinions. However, both of these
approaches are for DEA models without dual role factors and
weight restrictions. We discuss both the approaches briey and
their implications to the proposed model.
For the suppliers lacking one or some feature(s) (input, output
or dual role factor), the contribution with respect to the lacking
factor is considered as missing value(s). These values are inserted
in a two step process:
1. Calculate the series mean of the input, output or dual role
factor vector which lacks at least one DMU.
2. Calculate a correcting factor using AHP based on the mean of
other inputs, outputs or dual role factors (unless the vector
pertained to the DMU has a missing value) with respect to
input and output values that a DMU has a missing value. Based
on this a correcting factor is calculated which applies to the
series mean to provide the missing value. For this, a hierarchy is
constructed in which the second level comprises all input and
output factors, excluding the factor(s) that certain DMU lacks.
In the third (nal) level of the hierarchy, two alternatives are
being compared with respect to the attribute of the higher
levels. These two alternatives are (i) the DMU which lacks the
factor(s) and (ii) the mean of other DMUs. The data for mean of
other DMUs is obtained by taking the mean of each factor of all
DMUs except the one which has a missing value.
The missing values obtained from the above procedure are
placed in the DEA model above. However, this translates the said
DEA model into a Chance Constrained DEA (CCDEA), as the above
estimation assumes that data is normally distributed and makes
use of series means. This is essentially to minimize the risk as the
missing value calculated above is a mean value. The variance
needs to be computed from the concerned column of missing data.
Then, an appropriate probability level can be applied to formulate
a deterministic version of CCDEA.
The idea for the formulation is presented briey. We dene the
following subsets:
Subset denition
o1: set of all combinations (r, j) where the value of rth output
for jth DMU is given
o2: set of all combinations (r, j) where the value of rth output
for jth DMU is missing
i1: set of all combinations (i, j) where the value of ith input for
jth DMU is given
i2: set of all combinations (i, j) where the value of ith input for
jth DMU is missing
f1: set of all combinations (f, j) where the value of fth dual role
factor for jth DMU is given
f2: set of all combinations (f, j) where the value of fth dual role
factor for jth DMU is missing
Next, we dene a modied set of inputs and outputs to account
for the missing values as follows:
yrj yrj for r; jo1
Missing output value supplied from series mean and AHP
technique for (r, j)o2
xij xij f or i; ji1
Missing input value supplied from series mean and AHP
technique for (i, j)i2
wf j wf j for f ; jf1

116

A. Kumar et al. / Omega 42 (2014) 109123

Missing output value supplied from series mean and AHP

technique for (f, j)f2
Then, Constraint (3) would transform into the following CCDEA
constraint.
s

r1

f 1

Start

ur yrj f wf j

Collect data for all suppliers including

carbon footprints

m
f wf j vi xij
i1
f 1

v
u s
F
u
u ur 2 Varyrj  f 2 Varwf j 
ur 1
f 1
u
1:645u
i
F h
m
u

2
t f Varwf j vi 2 Varxij vi xij

mean and AHP technique

i1

f 1

Insert Missing values if any by series

13

Decide parameters P and

The complete CCDEA model with emission constraints is as

follows:

Calculate efficiency by proposed CCDEA

r1

f 1

f 1

Max ur yr0 f wf 0 f wf 0 Pe0

14

model

subject to:
m

vi xi0 1:645

i1

q
vi 2 Varxi0  1

Stop

15

Fig. 1. Flow chart of proposed GDEA model.

s

ur yrj f wf j

r1

f 1

f 1

i1

f wf j vi xij
v
u s
F
u
u ur 2 Varyrj  f 2 Varwf j 
ur 1
f 1
u
1:645u
F
m
u
t f 2 Varwf j  vi 2 Varxij vi xij
i1

f 1

j
!

ci

16

vi xi0 vi xi0 1:645

i1

vi xi0 di

vi xi0

i1

q
vi 2 Varxi0  0

17

q
vi 2 Varxi0  0

18

!
1:645

w10 e0 ECAP

19

1 1

20

vi

21

ur

22

f ; f 0
ej 0

23
24

here, objective function (14) is same as objective function (12)

explained earlier. Constraints (15), (17) and (18) correspond to Constraints (2), (4) and (5) respectively in Saen's[67] model with an
additional term to include variances of missing values which have
been estimated. Constraint (16) is same as Constraint (13). Constraints
(19) and (20) correspond to Constraints (10) and (11) explained earlier.
Constraints (21)(23) are same as Constraints (6)(8) respectively in
Saen's [67] model. Constraint (24) is included to keep the emission
quantity beyond the emission cap strictly positive.
The other approach to heterogeneous supplier selection as
suggested by Saen [66] is to use estimations in the form of

intervals. These intervals are based on expert opinions. The exact

data are viewed as a special case of interval data with the lower
and upper bounds being equal.
A slight modication is suggested if a supplier does not have
carbon footprinting done. Although using the series mean and
AHP technique, the missing footprint can be estimated, it would
not encourage the supplier to go for carbon footprinting. In that
case, a small penalty in the form of weight restrictions or negative
efciency can be levied on the supplier, to encourage the supplier
to go for carbon footprinting in order to survive competition or
else face elimination. This penalty may be similar to the penalty
applied in objective function (14) and would go against a supplier's
selection in case the supplier does not have carbon footprinting in
place. Fig. 1 shows the steps involved in the whole process.
Following binary variable is introduced:
g j 1;
0;

if a DMUj does not have carbon footprinting done

otherwise:

Then, objective function (14) would change as follows:

s

r1

f 1

f 1

Max ur yr0 f wf 0 f wf 0 Pe0 Fg 0

25

where F would enforce the penalty in terms of efciency percentage.

5. Industrial application
The GDEA approach is applied to a well-known automobile
spare parts manufacturer in India. The company is an ISO
9001.14001/18001 certied company and was established in
1967. It is an established manufacturer of high quality rubber
parts catering to the automotive industry for the last four decades.
The company's main plant is located in Delhi with two other
plants, one each in the state of Punjab and the state of Haryana.
The product range includes Engine Mountings, Transmission
Mountings, Silent Block Bushes, Suspension Bushes, Bellows,
Buffers and various other Rubber molded components. The company brand name enjoys immense popularity and condence
among its customers and the products are being used in India by

A. Kumar et al. / Omega 42 (2014) 109123

e. Carbon Footprint (Mtons CO2)the overall carbon footprint

in tCO2e. Calculations for the same are described in 3rd
bullet of this sub-section.
f. Industry Position and Rating (Likert 15)the position in
industry (including production leadership and reputation)
along with nancial position and credit rating of each
supplier. The industry position, nancial position and credit
ratings are merged into a single rating number. This is then
provided a rating on Likert scale.
g. Grade and Finishing (Likert 15)the grade for steel tubes
supplied; nish quality and the ability to meet quality
specications consistently. All these are merged into a single
rating number. This is then provided a rating on Likert scale.
h. Past Business and communication (Likert 15)the amount
of past business which has been done with each supplier
along with the communication system (with information on
progress data of orders). This is put into 5 past business and
communication buckets based on a Likert scale of 15.
i. Performance (Likert 15)the performance history of each
supplier. The past performance is measured in terms of
prot achieved per unit net price offered and then put into
performance buckets based on a Likert scale of 15.

all major O.E.M. and Tier 1 suppliers. The company's lines of

business include Manufacturing, Export and Wholesale with 6000
employees including labor.
The current supplier selection process in the company is giving
weightage to a few criteria and using a linear combination of these
criteria to come up with nal supplier rating. However, this
process has come under immense criticism as the process of
setting weights is in itself very ad-hoc at the rm and over the
period of last two years, multiple decision makers in purchasing
have not been able to converge to a single solution. This inghting
also is one of the primary reasons the company was looking for a
solution that can be built with collaborative approach.
The results from the existing supplier selection model have not
been presented in this paper due to legal concerns. However, in
some sections the preferred suppliers as per current approach
have been pointed out in some sections while comparing the
results from the proposed model.
Data collection steps and ndings are presented in the next few
sections.
5.1. Data collection
The data collection proceeded in following steps:
1. The manufacturer was presented with a list of criteria for
supplier selection which was primarily compiled with the help
of criteria given in Weber [83] and Humphreys et al. [35]. The
manufacturer tallied this with the current supplier selection
criteria present in the organization and came up with a
comprehensive list of supplier selection criteria. Table 2 lists
the criteria (as column headings) and their values with respect
to the competing suppliers.
A brief description of the criteria/attributes in Table 2 is as follows:
a. Net Price (INR/kg)the net price (including discounts)
offered by each supplier.
b. Distance (km)the distance based on geographical location
of the supplier.
c. Shelf Life (Months)the shelf life or longevity of product
supplied.
d. Lead Time (Days)average delivery time take by the supplier to deliver an order.

117

It may be noted that some attributes have been merged into a

single attribute. This is primarily done to improve DEA performance. The discriminatory power of DEA goes down as the
number of attributes increase.
2. The preferences and relative restrictions on each criterion were
calculated with the help of management. The restrictions on
DMU weights are determined as follows:
a. Net Price should have the maximum priority amongst input
weights.
b. The relative importance of Net Price (virtual weight restriction in DEA) should be between 0.5 (ci in Eq. (17)) and 3 (di
in Eq. (18)).
c. The relative importance of Lead Time should be between
0.1 and 2.5.
d. The relative importance of Performance should be less
than 3.
e. The relative importance of Quality should be between
0.2 and 2.5.

Table 2
Related attributes for 18 suppliers.
DMU Inputs

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18

Dual role

Outputs

Net Price
(INR/kg)

Dist.
(km)

Shelf Life
(Mths)

Lead Time
(days)

Carbon Foot-print
(MTons CO2)

Industry Position and

Rating (15)

Grade and Finish Past Bus. and

(15)
Comm. (15)

Perf.
(15)

83
76
77
85
84
84
77
75
77
81
82
82
76
77
75
85
77
75

1659
1499
1355
1440
459
1749
146
1371
637
266
1774
1408
138
607
1545
212
53
456

7
7
4
10
5
6
5
12
5
8
7
12
5
10
10
6
5
10

15
22
22
15
22
18
18
15
25
25
22
19
20
21
22
22
16
19

552
312
540
324
516
324
492
300
588
408
492
528
312
432
528
456
492
372

4
2
3
3
4
4
3
4
3
3
4
2
5
3
4
2
3
4

3
5
3
4
3
2
2
4
3
3
3
4
5
3
3
3
2
4

4
5
2
2
4
2
4
3
4
4
2
3
5
2
5
5
4
3

4
3
2
3
1
1
4
2
1
5
2
1
3
2
3
3
1
2

Abbreviations: INR/kg, Indian Rupees per kilogram; Dist., Distance; Mths, Months; MTons, Metric Tons; Bus., Business; Comm., Communication; Perf., Performance; (15),
on Likert scale from 1 to 5 (1 being worst, 5 being the best).

118

A. Kumar et al. / Omega 42 (2014) 109123

f. The relative importance of Supplier Prot (Output from

Carbon Trading Model) should be between 0.1 and 2.
All the above managerial preferences were modeled as absolute
or virtual weight restrictions in DEA model.
3. A reputed consulting rm in the eld of climate change was
appointed to undertake the carbon foot-printing process for all
the suppliers. The rm was updated with specic guidelines for
carbon foot-printing for supplier selection in setting up the
estimation boundaries. The guidelines were taken from Kumar
and Jain [46], Lenzen [52], Suh et al. [75], Suh and Huppes [74]
and Woensel et al. [86].
The rm took into account the factors and regulations mentioned
in Section 3.1. Primary and secondary carbon emissions were
considered and the hybrid approach given by Suh et al. [75] and
Lenzen [52] which combine the life cycle approach with input
output analysis. Such an approach is effective in dealing with
truncations errors within the supply chain.
As a standard procedure the consulting rm has the entire carbon
footprinting process veried by established accountancy rm. The
details of audit and verication process have been made available
to the concerned spare parts manufacturer.
An important point to note here is that carbon footprinting with
respect to the supplier selection process has to be a consistent
procedure across all suppliers. As the hybrid carbon footprint
calculator was developed independent of any supplier, the consistency of approach is veried once the entire process has been
veried from an established third party.
The rm was also assigned the task of calculating an individual
cap on the carbon footprints of each supplier based on an ideal
supplier. The annual carbon footprint of each supplier is given in
dual role factor column of Table 2. The individual cap calculated
as per standards given for Iron and Steel Industry in Environment (Protection) (Third Amendment) Rules, 2012 and was
xed at 375 Metric Tonnes of CO2 per year for each supplier.
The proposed model is run with the following parameter settings.

P 1/ECAP
This would ensure that the efciency rating of suppliers having
carbon footprints reasonably within ECAP, would be reasonable. If the carbon footprints of a supplier exceeds 2  ECAP,
then Constraint (19) would enforce e0 to be ECAP and the total
penalty in objective function (14) would be 1. That would imply
that supplier emitting more than twice the emission cap would
have zero or negative efciency rating, which would effectively
mean that they would not be selected.

0.005
This parameter in inequality (20) has been carefully set by
studying the input role of carbon footprints. A general rule would
be to set the parameter so as not to drive the efciencies to a
negative value. An unrestricted DEA run can be done to provide a
reasonable upper bound on . The lowest value of input dual role
weight from unrestricted DEA run can serve as an upper bound on
. The higher this parameter is the bigger role carbon footprints
play in a supplier's efciency rating.
5.2. Results
The results are shown in Table 3. The efciency and ratings
used for comparison are from the model proposed by Saen [67],
where carbon footprints have not been considered. Suppliers with
the same efciency rating have been ranked on the basis of their
carbon footprints (lesser the better). Another model for fair
comparison has been included in Table 3 which is AHPTOPSIS

Table 3
Efciency scores with the GDEA model and comparison.
DMU

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18

GDEA model

Saen's model

AHPTOPSIS

Efciency

Ranking

Efciency

Ranking

Score

Ranking

0.363
0.988
0.203
0.55
0.18
0.47
0.75
1
0.107
0.93
0.17
0.116
1
0.366
0.674
0.417
0.54
1

13
4
14
8
15
10
6
1
18
5
16
17
1
12
7
11
9
1

0.743
0.898
0.568
0.551
0.49
0.473
1
1
0.576
0.639
0.429
0.455
1
0.495
1
0.595
0.8
1

8
6
12
13
15
16
1
1
11
9
18
17
1
14
1
10
7
1

0.487
0.380
0.357
0.753
0.628
0.843
0.310
0.413
0.282
0.562
0.628
0.557
0.360
0.402
0.254
0.633
0.308
0.369

8
11
14
2
5
1
15
9
17
6
4
7
13
10
18
3
16
12

which is a popular Multi Attribute Decision Making (MADM)

approach [3]. See [8] for AHPTOPSIS results and comparison.
It may be noted that the prior model by Saen [67] would be
referred as Saen's Model throughout the current paper.
5.3. Observations and conclusions from results
Few arguments are presented in support of the results. Before
the comparisons, we look at results from AHPTOPSIS approach.
The AHPTOPSIS approach nds suppliers 4 and 6 as most
efcient. Both of these suppliers have very low efciency ratings
in the other two models. At a glance, these suppliers do well on
carbon footprints, but, not so well on all other criteria. The
approach was unanimously rejected by management. Thus, the
approach is not considered in further comparisons. The approach
considered for comparison is Saen's approach as the direct
comparison with a model with without carbon footprints is visible.
Further validation of the model is provided with a section on
sensitivity analysis.
Section 5.3.1 presents observable deviations from Saen's model.
Section 5.3.2 presents arguments based on statistical reasoning.
Finally, Section 5.3.3 presents a sensitivity analysis of GDEA model
with respect to the emissions cap.
5.3.1. Observable deviations
There are a few interesting observations:

Note the efciency of supplier 7. Volume-wise supplier 7 is

largest manufacturer of steel tubes in India and was a preferred

supplier prior to GDEA. With the GDEA model, supplier 7 drops
to 6th position and with a lower efciency rating. The GDEA
model has penalized the high carbon footprints in two ways.
One with the carbon footprints as dual role factors weighing
more on the input side and the other is the penalty for not
meeting the emission cap.
Supplier 10 which is a relatively clean supplier jumps from a 9th
position to 5th position in the GDEA model.

5.3.2. Rank sum test

This subsection deals with statistical comparison between two
groups of efciency values, i.e., with Saen's model in the absence

A. Kumar et al. / Omega 42 (2014) 109123

119

overall rating studied. Multiple scenarios were run with following

congurations:

of carbon footprints as dual-role factor, and with the GDEA model

in the presence of carbon footprints as dual-role factor. The
Wilcoxon Rank-Sum test is used to identify whether differences
between two groups are signicant.
Using T, the null hypothesis that the two groups have the same
population at a level of signicance can be checked. In this
example, there is T 1.788. If 0.05 (5%) is chosen, then it holds
that T0.025 1.96. Since T 1.788o 1.96 T0.025, the null hypothesis
at the signicance level 5% cannot be rejected. Consequently, the
differences among the efciency scores obtained by Saen's model
and GDEA model are not statistically signicant.

Shelf Life missing for DMU 8.

Past Communication rating missing for supplier 7 and 15.
Performance ratings missing for supplier 18.
Industry rating missing for supplier 3 and 10.
Lead Time missing for supplier 4.

In all the scenarios the method of missing values presented in

previous section was used to calculate the missing values. As
indicated in Section 4, the process involves two steps: calculating
the series mean from known values, and then applying an AHP
based correction factor to the series mean. The values from these
two steps, along with the nal missing values are shown in Table 5.
The calculated missing values are quite close to the actual missing
values as evident from Table 5. Also, when all the above ve
scenarios were run with GDEA (Chance Constrained), the overall
rankings remained same as original GDEA with complete values.
Hence, the approach is considered suitable for missing values.

5.3.3. Sensitivity analysis with emission cap

Next we validate the model by running it with different values
of emission cap and trying to capture the results. Although, 375
was set as emission cap, the model is run with emission caps of
350, 400, 500 and 600 to study the model behavior as shown in
Table 4.
Note that in Table 4, as the emission cap is tightened to 350, the
efciency scores drop. Note the efciency scores for DMUs 2, 4, 6
8 and 13. All these DMUs are compliant DMUs even to the tightest
compliance level (350), thus, their efciency scores are not
affected.
Another important point to note is that efciencies go up as the
emission cap is relaxed. For DMUs except 2 and 10, the efciency
at emission cap of 600 is approaching the efciency with the prior
model. These are the DMUs (2 and 10) which are relatively cleaner
and gain with the GDEA model. This validates the two-pronged
theory of the model as follows:

6. Research ndings, implications and managerial insights

The ndings were a revelation to the company. Prior to this
study, the management considered suppliers 1, 2 and 7 as their
Table 5
Missing values.
Scenario DMU Missing eld

Even though DMUs 1, 3, 5, 9, 11, 12, 14 and 17 are totally

compliant at a relaxed emission cap, they are still less efcient

in the GDEA model as these are one of the more polluting
DMUs and inefcient elsewise as well.
A tighter emission cap would force down the efciency even for
a relatively efcient but less clean DMU.

8
7

15

18

5.4. Validation of missing values approach

3
4

In the current data set, none of the values were found missing.
However, to validate the missing values approach, some values
were intentionally deleted from the data set and their results on

10
4

Series
mean

Shelf Life
7.18
Past
2.19
communication
Past
2.19
communication
Performance
3.52
rating
Industrial
3.375
rating
Industrial
3.375
rating
Lead Time
20.17

Calculated
missing
value

Actual
missing
value

3.2
1.32

10.38
3.51

12
4

0.73

2.92

0.25

3.27

0.21

3.165

0.38

2.995

Correction
factor

3.21

16.96

15

Table 4
Efciency scores (EFF) with different emission caps.
DMU

EFF ECAP 350

Rank

EFF ECAP 400

Rank

EFF ECAP 500

Rank

EFF ECAP 600

Rank

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18

0.278
0.988
0.12
0.55
0.101
0.47
0.675
1
0.018
0.867
0.094
0.035
1
0.3
0.593
0.348
0.465
0.95

13
3
14
8
15
9
6
1
18
5
16
17
1
12
7
11
10
4

0.436
0.988
0.275
0.55
0.248
0.47
0.816
1
0.186
0.984
0.235
0.186
1
0.423
0.744
0.478
0.606
1

12
4
14
9
15
11
6
1
17
5
16
17
1
13
7
10
8
1

0.657
0.988
0.491
0.55
0.454
0.47
1
1
0.421
1
0.42
0.398
1
0.487
0.955
0.59
0.79
1

9
6
12
11
15
14
1
1
16
1
17
18
1
13
7
10
8
1

0.74
0.988
0.555
0.55
0.48
0.47
1
1
0.562
1
0.42
0.442
1
0.487
1
0.59
0.79
1

9
7
12
13
15
16
1
1
11
1
18
17
1
14
1
10
8
1

120

A. Kumar et al. / Omega 42 (2014) 109123

preferred suppliers because of long term relationships and the

brand value of these suppliers. The results from the study show
that supplier 1 has a low efciency rating when evaluated on
various parameters in accordance with DMs priorities, while
supplier 2 with a ranking of 4 may not be the most efcient
supplier.
An important criterion of any research based study is the user
acceptance. In the current scenario, multiple DMs are involved,
and hence, an overall acceptability criterion is used for user
validation.
The user validation criteria for the approach were developed as
follows:

Table 6
Validation scores from DMs.
Criteria

Saen's approach

GDEA

DM1 DM2 DM3 DM4 DM5 DM1 DM2 DM3 DM4 DM5
Feasibility
Preference
accounting
Competency
Sustainability
Total
Grand total

3
3

3
4

2
3

2
2

3
3

4
3

5
5

4
5

2
4

5
5

5
3
14

2
3
12

3
3
11

3
4
11

4
5
15

4
4
15

3
4
17

3
4
16

5
4
15

4
4
18

63

81

The DMs set a target rate of 80% overall satisfaction with results

based on following criteria:

Feasibility and Applicabilitydoes the supplier selection
approach select suppliers which are feasible and practical
for a long-term procurement relationship?
Preference Accountingdoes the supplier selection
approach select suppliers which incorporate all the preferences supplied by the DM?
Competencydoes the supplier selection approach select
suppliers competitively superior to others?
Sustainabilitydoes the supplier selection approach select
suppliers which are sustainable and efcient on the
environmental front?
All ve DMs (CFO, CIO, Purchasing Head, and two senior
managers in the Purchasing department) were presented with
the results and were asked to rate the results on a 5-level Likert
scale (1 being Strongly Disagree and 5 being Strongly Agree)
on all the four criteria listed above. The scores from these were
translated to a percentage satisfaction.

The DMs overall scoring is presented in Table 6.

As shown in Table 6, the DMs overall satisfaction score was
calculated based on sum total from all the responses. The overall
score was 81% for results from GDEA model and 63% from prior
approach. This is primarily because the GDEA approach scores well
in terms of all four criteria above with all the DMs on an overall
basis. One of the major reasons for this was that the approach
brought out some incorrect decisions which the rm had been
taking for the last few years, such as keeping supplier 7 as
preferred supplier. The approach was also able to demonstrate to
the DMs why supplier 7 should not be a preferred supplier.
With the above validation approach, the method combines a
comprehensive quantitative approach with the qualitative view of
managers. Most of the GDEA results are easily explainable to the
management before they do the validation. The sensitivity analysis
adds another level of validation to the approach itself.
An important point to note here is that success of such an
approach would not only depend on the approach itself, but also
the quantitative ability of the DMs to analyze and interpret the
results of a quantitative approach and use it for effective decision
making.
Some of the implications from the research are listed below.

This type of study can be extremely relevant in supplier

selection scenarios where decision making is a collective

process, which is more often the case in medium and large
scale manufacturing companies.
The approach is validated in terms of feasibility, applicability
and sustainability as well, which is an added advantage of the
approach.

The overall satisfaction of the manufacturer based on user

validation criteria presents the positive outlook towards the GDEA

approach. The total deployment of approach at the company is

currently being designed. An overall change in purchasing procedure would take place which would be as follows. The changes
described below are already on the company's roadmap which
indicates that the company is seriously considering use of GDEA in
its supplier selection practices:
1. Based on company inventory policies (re-order points and
stocking norms), the procurement cycle for each SKU of
inbound raw material was decided (The study was conducted
only for steel tubes). One more decisive factor in determining
the procurement cycle was lead times. As one does not know
which supplier is to be selected at the start of cycle, maximum
lead time among all suppliers was chosen to be lead time input
for deciding the procurement cycle.
2. At the start of procurement cycle, the GDEA based tool would
be run to determine the most efcient supplier. Buy in from all
DMs would be required to validate the results before placing
the procurement request with the chosen supplier. This would
be similar to the user validation process in previous section
with a target rate of 80% overall satisfaction with all DMs.
3. In case the buy-in is not realized, the DMs would consult again
on priorities. Sometimes the priorities among DMs are so far
apart that achieving a target rate of 80% acceptance might not
be possible. In such a case some DMs need to adjust their
priorities in order to achieve 80% target.
4. The company would track all the suppliers on the basis of their
efciencies across multiple procurement cycles and would
iteratively update the past performance data for all suppliers.
5. The supplier data update and data validation process is xed to
be a monthly audit cycle.
During the pilot project with steel tubes as the raw material in
the scope, some of the observations and conclusions from the DMs
and managers were:
1. Ease of usethe DMs were not bafed with providing a lot of
comparative inputoutput restrictions or relative weight
restrictions to synchronize with their priorities.
2. Improved Efciencyvisibility into trade-offs which the DMs
were not considering earlier, allowed them to choose a supplier
which would add to overall efciency. Overall procurement
costs went down by 5% (only for steel tubes). The lead time had
increased, but, the overall quality of nal product had a denite
positive impact. This was evident from the fact that the
manufacturer found that it was churning out more export
quality suspension parts than usual with an overall increase
of 7%. Export items are major revenue generators for the
company. This is a direct impact of improved steel grade from
the procurement process.

A. Kumar et al. / Omega 42 (2014) 109123

121

Table A1
AHP comparison matrix.

Net Price
Distance
Shelf Life
Lead Time
Carbon Footprint
Industry position and rating
Grade and nish
Past business and
communication
Performance
Final AHP weight

Net
Price

Distance Shelf
Life

Lead
Time

Carbon
Footprint

Industry
Position
and Rating

Grade and
Finish

Past Business
and
Communication

Performance

1.000
0.143
0.143
0.143
3.000
0.143
0.143
0.143

7.000
1.000
0.333
5.000
3.000
7.000
9.000
0.333

7.000
3.000
1.000
3.000
3.000
3.000
3.000
0.333

7.000
0.200
0.333
1.000
3.000
3.000
0.333
0.333

0.333
0.333
0.333
0.333
1.000
0.333
3.000
0.333

7.000
0.143
0.333
0.333
3.000
1.000
3.000
0.200

7.000
0.111
0.333
3.000
0.333
0.333
1.000
0.200

7.000
3.000
3.000
3.000
3.000
5.000
5.000
1.000

0.333
0.111
0.111
0.111
3.000
0.333
3.000
0.111

3.000
0.241

9.000
0.025

9.000
0.025

9.000
0.077

0.333
0.179

3.000
0.069

0.333
0.160

9.000
0.018

1.000
0.206

3. Less Inghtingconicts within management and DMs have

reduced allowing for a more smooth and streamlined decision
process.
4. Greener Supply Chainreduced carbon impact from the supplier reduces the overall rm's carbon footprint.

7. Concluding remarks and scope for future work

As the climate change movement gathers momentum, there is
a pressing need to assess suppliers based on their environmental
performance along with other criteria for supplier selection. Thus,
the need of the hour is a comprehensive supplier selection
strategy which models supply costs as well as emissions.
This paper proposed a unied GDEA approach to model the
supplier selection problem with carbon footprints of suppliers as a
necessary dual role factor. The GDEA approach caters to nonhomogeneous suppliers as well, and encourages suppliers to cut
their emissions while also going for carbon footprinting (if they
have not done it yet).
An important point to note is that the GDEA approach is based
on encouraging the suppliers to reduce their carbon footprints in
order to survive the competition and become green. The GDEA
approach does not tell how a supplier should become green, but
provides a reason as to why they should do so. A brief section on
carbon footprints (Section 3) is included to give the reader an idea
of the methodology involved.
Another important point to note is that the GDEA model is not
an enhancement over DEA, but an approach to model environmental factors within a DEA framework.
In the GDEA model, carbon footprint is assumed as a necessary
dual role factor, which might not be true in all cases. Also, weight
restriction assumptions may not always hold. In such cases, the
model needs to be tailored to exclude the assumptions. Finally the
model would need to be modied to model the emission constraints as per region-specic emission laws.
An application to a well-known automobile spare parts manufacture is presented to verify the GDEA approach. The results
prove that the model considers the environmental factors in a
two-pronged approach. Firstly, it lets the footprints determine the
efciency and secondly, the model penalizes a supplier for not
meeting emission norms.
The proposed GDEA approach provides a basic framework for
incorporating region specic emission norms and compliance
standards. However, there is immense scope for environmental
modeling within this framework which would be governed by the
environmental regulations within the region.
The approach considered in this paper provides a common
framework for future research in terms of a green supplier

Table A2
TOPSIS output.
DMU

Score

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18

0.486911
0.380279
0.357252
0.75277
0.627567
0.8426
0.310047
0.413475
0.281569
0.561672
0.627924
0.556747
0.360375
0.401755
0.253529
0.63325
0.307834
0.36886

selection strategy. A suggested research in this area would be to

apply the model to other multi criteria decision making problems
as well.

Appendix. AHP TOPSIS results and comparison

This section explains the AHPTOPSIS methodology and presents the results of using AHPTOPSIS technique, which is used for
comparison in this paper.
AHPTOPSIS is an effective decision making approach which
combines the power of Analytic Hierarchy Process (AHP) with an
easy to use ranking techniqueTechnique for Order Preference by
Similarity to Ideal Solution method (TOPSIS). The methodology of
TOPSIS is very similar to a DEA. DEA nds the efciency of a DMU
by comparing it with the most efcient DMU in the group.
Similarly TOPSIS calculates the ranking based on ideal solution
for an attribute. The AHPTOPSIS method is applied in two steps.
Step 1: calculate weights for each criterion based on AHP.
Step 2: input weights from AHP to TOPSIS method to
rank DMUs.
The overall advantage of AHPTOPSIS approach lies in ease of
use, straight-forward computation process, understandability and
the fact that it can take into account all types of criteria (subjective
and objective) [3].

122

A. Kumar et al. / Omega 42 (2014) 109123

The next two tables present the results of using AHPTOPSIS

technique with our current data set.
Table A1 gives the AHP matrix used for pair-wise comparisons
in accordance with the DMs views.
Final AHP weights were fed into the TOPSIS software Triptych by
Statistical Design Institute, which gave the output shown in Table A2.

References
[1] Banker RD, Charnes A, Cooper WW. Some methods for estimating technical
and scale inefciencies in data envelopment analysis. Management Science
1984;30(9):10781092.
[2] Bayazit O. Use of analytic network process in vendor selection decisions.
International Journal of Benchmarking 2006;13(5):566579.
[3] Bhutia PW, Phipon R. Application of AHP and TOPSIS method for supplier
selection problem. IOSR Journal of Engineering 2012;2(10):4350.
[4] Bowen FE, Cousins PD, Lamming RC, Farukt. AC. The role of supply management capabilities is green supply. Production and Operations Management
2001;10(2):174189.
[5] Braglia M, Petroni A. A quality assurance-oriented methodology for handling
trade-offs in supplier selection. International Journal of Physical Distribution
and Logistics Management 2000;30(2):96111.
[6] Brickman C, Ungerman D. Climate change and supply change management.
McKinsey Quarterly (Operations) 2008.
[7] Carbon Trust. Carbon footprinting: an introduction for organizations. http://
www.carbondecisions.ie/resources/footprint_for_organisations.pdf; 2007.
[8] Central Electricity Authority (CEA), Government of India. CO2 emission factor
database. Version 08. http://www.cea.nic.in/reports/planning/cdm_co2/
cdm_co2.htm; 2013.
[9] Central Pollution Control Board India. Industry specic environmental standards. http://www.cpcb.nic.in/Industry_Specic_Standards.php; 2012.
[10] Chan FTS. Interactive selection model for supplier selection process: an
analytical hierarchy process approach. International Journal of Production
Research 2003;41(15):35493579.
[11] Chan FTS, Bhagwat R, Wadhwa S. Study on suppliers exibility in supply
chains: is real-time control necessary. International Journal of Production
Research 2009;47(4):965987.
[12] Chan FTS, Chan HK. Development of the supplier selection modela case
study in the advanced technology industry. Proceedings of The Institution of
Mechanical Engineers Part B: Journal of Engineering Manufacture 2004;218
(12):18071824.
[13] Chan FTS, Chan HK. An AHP model for selection of suppliers in the fast
changing fashion market. International Journal of Advanced Manufacturing
Technology 2010;51(912):11951207.
[14] Chan FTS, Chan HK, Ip RWL, Lau HCW. A decision support system for supplier
selection in the airline industry. Proceedings of the Institution of Mechanical
Engineers Part B: Journal of Engineering Manufacture 2007;221(4):741758.
[15] Chan FTS, Kumar N. Global supplier development considering risk factors
using fuzzy extended AHP-based approach. OMEGAInternational Journal of
Management Science 2007;35(4):417431.
[16] Chan FTS, Kumar N, Tiwari MK, Lau HCW, Choy KL. Global supplier selection: a
fuzzy-AHP approach. International Journal of Production Research 2008;46
(14):38253857.
[17] Chen C, Monahan GE. Environmental safety stock: the impacts of regulatory
and voluntary control policies on production planning, inventory control, and
environmental performance. European Journal of Operational Research
2010;207(3):12801292.
[18] Cook WD, Green RH, Zhu J. Dual-role factors in data envelopment analysis. IIE
Transactions 2006;38(2):105115.
[19] Cooper WW, Seiford LM, Tone K. Introduction to data envelopement analysis
and its uses. New York: Springer; 2006.
[20] Corbett CJ, Klassen RD. Extending the horizons: environmental excellence as
key to improving operations. Manufacturing and Service Operations Management 2006;8:522.
[21] Danese P. Supplier integration and company performance: a congurational
view. OMEGAInternational Journal of Management Science 2013;41
(6):10291041.
[22] Diabat A, Simchi-Levi D. A carbon-capped supply chain network problem. IEEE
International Conference on Industrial Engineering and Engineering Management 2009:523527.
[23] Deng Y, Chan FTS. A new fuzzy dempster MCDM method and its application in
supplier selection. Expert Systems With Applications 2011;38(8):98549861.
[24] Ghodsypour SH, OBrien C. A decision support system for supplier selection
using an integrated analytic hierarchy process and linear programming.
International Journal of Production Economics 1998;56(2):199212.
[25] Ghodsypour SH, OBrien C. The total cost of logistics in supplier selection,
under conditions of multiple sourcing, multiple criteria and capacity
constraint. International Journal of Production Economics 2001;73(1):
1527.
[26] Gonzlez-Benito J, Gonzlez-Benito O. Environmental proactivity and business
performance: an empirical analysis. OMEGAInternational Journal of Management Sciences 2005;33(1):115.

[27] Guide Jr. VDR, Harrison TP, Wassenhove LNV. The challenge of closed-loop
supply chains. INFORMS Interfaces 2003;33(6):36.
[28] Guide Jr. VDR, Wassenhove LNV. Closed-loop supply chains: an introduction to
the feature issue (part 1). Production and Operations Management 2006;15
(3):345350.
[29] Guide Jr. VDR, Wassenhove LNV. Closed-loop supply chains: an introduction to
the feature issue (part 2). Production and Operations Management 2006;15
(4):
471472.
[30] Handeld R, Walton SV, Sroufe R, Melnyk SA. Applying environmental criteria
to supplier assessment: a study in the application of the analytical hierarchy
process. European Journal of Operational Research 2002;141(1):7087.
[31] Haq N, Kannan G. Design of an integrated supplier selection and multi-echelon
distribution inventory model in a built-to-order supply chain environment.
International Journal of Production Research 2006;44(10):19631985.
[32] Hofer C, Canter DE, Dai J. The competitive determinants of a rm's environmental management activities: evidence from US manufacturing industries.
Journal of Operations Management 2012;30(12):6984.
[33] Hu AH, Hsu CW, Chen SH. Incorporating carbon management into supplier
selection in the green supply chain: evidence from an electronics manufacturer in Taiwan. In: Proceedings of 16th annual international sustainable
development research conference. Hong Kong; 2010.
[34] Huang AY, Lenzen M, Weber C, Murray J, Matthews HS. The role of input
output analysis for the screening of corporate carbon footprints. Economic
Systems Research 2009;21(3):217242.
[35] Humphreys PK, McIvor R, Chan FTS. Using case-based reasoning to evaluate
supplier environmental management performance. Expert Systems With
Applications 2003;25(2):141153.
[36] Humphreys PK, Shiu WK, Chan FTS. Collaborative buyersupplier relationships
in Hong Kong manufacturing rms. Supply Chain Management: An International Journal 2001;6(3,4):152162.
[37] Humphreys PK, Wong YK, Chan FTS. Integrating environmental criteria into
supplier selection process. Journal of Materials Processing Technology 2003;138:
349356.
[38] Initiative for Global Environmental Leadership (IGEL) & Knowledge@Wharton.
Green evolution: managing the risks, reaping the benets. Internet site:
http://environment.wharton.upenn.edu; 2010.
[39] Jain V, Tiwari MK, Chan FTS. Evaluation of the supplier performance using an
evolutionary fuzzy-based approach. Journal of Manufacturing Technology
Management: Special Issue on Logistics and Supply Chain Management With
Articial Intelligence TechniquesPart 1 2004;15(8):735744.
[40] Jain V, Wadhwa S, Deshmukh SG. Select supplier-related issues in modelling a
dynamic supply chain: potential, challenges and direction for future research.
International Journal of Production Research 2009;47(11):30133039.
[41] Kahraman C, Cebeci U, Ulukan Z. Multi-criteria supplier selection using fuzzy
AHP. Logistics Information Management 2003;16(6):382394.
[42] Kassinis GI, Soteriou AC. Greening the service prot chain: the impact of
environmental management practices. Production and Operations Management 2003;12(3):386403.
[43] Kleindorfer PR, Singhal K, Wassenhove LNV. Sustainable operations management. Production and Operations Management 2005;14(4):482492.
[44] Kleinsorge IK, Schary PB, Tanner RD. Data envelopment analysis for monitoring customersupplier relationships. Journal of Accounting and Public Policy
1992;11(4):357372.
[45] Kumar A, Jain V. Supplier selection: a Green approach with carbon
footprint monitoring. In: Proceedings of 8th international conference on
supply chain management and information systems. Hong Kong; 2010. p.
6229.
[46] Kumar A, Jain V. A green approach to supplier selection. In: Proceedings of
2010 annual INFORMS meeting. Austin, Texas; 2010.
[47] Kumar M, Vrat P, Shankar R. A fuzzy programming approach for vendor
selection problem in a supply chain. International Journal of Production
Economics 2006;101:273285.
[48] Kuo RJ, Wang YC, Tien FC. Integration of articial network and MADA methods
for green supplier selection. International Journal of Cleaner Production
2010;18:11611170.
[49] Lai K, Wong CWY. Green logistics management and performance: some
empirical evidence from Chinese manufacturing exporters. OMEGAInternational Journal of Management Science 2012;40(3):267282.
[50] Lee HI, Kang H, Hsu C, Hung H. A green supplier selection model for high-tech
industry. International Journal of Expert Systems with Applications 2009;36
(4):79177927.
[51] Lee S, Klassen RD. Drivers and enablers that foster environmental management capabilities in small- and medium-sized suppliers in supply chains.
Production and Operations Management 2008;17(6):573586.
[52] Lenzen M. Errors in conventional and inputoutput-based life-cycle inventories. Journal of Industrial Ecology 2000;4(4):127148.
[53] Liu J, Ding FY, Lall V. Using data envelopment analysis to compare suppliers
for supplier selection and performance improvement. International Journal of
Supply Chain Management 2000;5(3):143150.
[54] LMI Government Consulting. GreenSCOR: developing a green supply chain
analytical tool. Internet site: http://www.lmi.org/logistics/Documents/green
Supply/GreenSCOR%20White%20Paper.pdf; 2003.
[55] Mansini R, Savelsbergh MWP, Tocchella B. The supplier selection problem
with quantity discounts and truckload shipping. OMEGAInternational Journal of Management Science 2012;40(4):445455.

A. Kumar et al. / Omega 42 (2014) 109123

[56] Murphy S. Report: companies going green, and expecting suppliers to follow
suit. Internet site: http://palisadesmarketing.blogspot.com/2010/02/reportcompanies-going-green-and.html; 2010.
[57] Noci G. Designing green vendor rating systems for the assessment of a
supplier's environmental performance. European Journal of Purchasing and
Supply Management 1997;3(2):103114.
[58] Noori H, Chen C. Applying scenario-driven strategy to integrate environmental
management and product design. Production and Operations Management
2003;12(3):353368.
[59] Pi WN, Low C. Supplier evaluation and selection via Taguchi loss functions and
an AHP. International Journal of Advanced Manufacturing Technology 2006;27
(56):625630.
[60] Pil FK, Rothenberg S. Environmental performance as a driver of superior
quality. Production and Operations Management 2003;12(3):404415.
[61] Ramudhin A, Chaabane A, Paquet M. Carbon market sensitive sustainable
supply chain network design. International Journal of Management Science
and Engineering Management 2009;5(1):3038.
[62] Ramudhin A, Chaabane A, Kharoune M, Paquet M.. Carbon market sensitive
green supply chain network design. In: Proceedings of the 2008 IEEE IEEM;
2008. p. 10937.
[63] Saen RF. A new mathematical approach for suppliers selection: accounting for
non-homogeneity is important. Journal of Applied Mathematics and Computation 2007;185(1):8495.
[64] Saen RF. Suppliers selection in the presence of both cardinal and ordinal data.
European Journal of Operational Research 2007;183(2):741747.
[65] Saen RF. Supplier selection by the new AR-IDEA model. International Journal
of Advanced Manufacturing Technology 2007;39(1112):10611070.
[66] Saen RF. A new approach for selecting slightly non-homogeneous vendors.
Journal of Advances in Management Research 2009;6(2):144153.
[67] Saen RF. Restricting weights in supplier selection decisions in the presence of
dual role factors. Journal of Applied Mathematical Modelling 2010;34
(10):28202830.
[68] Samoilenko S, Osei-Bryson K. Using data envelopment analysis (DEA) for
monitoring efciency-based performance of productivity-driven organizations: design and implementation of a decision support system. OMEGA
International Journal of Management Science 2013;41(1):131142.
[69] Sanayei A, Mousavi SF, Abdi MR, Mohaghar A. An integrated group decisionmaking process for supplier selection and order allocation using multiattribute utility theory and linear programming. Journal of the Franklin
Institute 2008;345(7):731747.
[70] Savaskan RC, Bhattacharya S, Wassenhove LNV. Closed-loop supply chain
models with product remanufacturing. Management Science 2004;50
(2):239252.
[71] Scott MH, Weber CL, Hendrickson CT. The importance of carbon footprint
estimation boundaries. Environmental Science and Technology 2008;42
(16):58395842.
[72] Shyur HJ, Shih HS. A hybrid MCDM model for strategic vendor selection.
Mathematical and Computer Modelling 2006;44(78):749761.

123

[73] Subramanian R, Gupta S, Talbot B. Compliance strategies under permits for

emissions. Production and Operations Management 2007;16(6):763779.
[74] Suh S, Huppes G. Methods for life cycle inventory of a product. Journal of
Cleaner Production 2005;13:687697.
[75] Suh S, Lenzen M, Treloar GJ, Hondo H, Horvath A, Huppes G, Jolliet O, Klann U,
Krewitt W, Moriguchi Y, Munksgaard J, Norris G. System boundary selection in
life-cycle inventories using hybrid approaches. Environmental Science and
Technology 2004;38(3):657664.
[76] Suh S, Nakamura S. Five years in the area of inputoutput and hybrid LCA.
International Journal of Life Cycle Assessment 2007;12:351352.
[77] Sundarkarni B, de Souza R, Goh M, Wagner SM, Manikandan S. Modeling
carbon footprints across the supply chain. International Journal of Production
Economics 2010;128:4350.
[78] Syarif A, Yun Y, Gen M. Study on multi-stage logistic chain network: a
spanning tree-based genetic algorithm approach. Journal of Computers and
Industrial Engineering 2002;43:299314.
[79] Talluri S, Sarkis J. A model for performance monitoring of suppliers. International Journal of Production Research 2002;40(16):42574269.
[80] Talluri S, Narasimhan R, Nair A. Vendor performance with supply risk: a
chance constrained DEA approach. International Journal of Production Economics 2006;100(2):212222.
[81] Ting SC, Cho DI. An integrated approach for supplier selection and purchasing
decisions. International Journal of Supply Chain Management 2008;13:
116127.
[82] Verma R, Pullman ME. An analysis of the supplier selection process. OMEGA
International Journal of Management Science 1998;26(6):739750.
[83] Weber A. A data envelopment analysis approach to measuring vendor
performance. Journal of Supply Chain Management 1996;1(1):2839.
[84] Weber A, Current J, Desai A. An optimization approach to determining the
number of vendors to employ. International Journal of Supply Chain Management 2000;5(2):9098.
[85] Wiedmann T, Minx J. A denition of carbon footprint. ISA-UK research report.
2007.
[86] Woensel TV, Creten R, Vandaele N. Managing the environmental externalities
of trafc logistics: The issue of emissions. Production and Operations Management 2001;10(2):207223.
[87] WRI, WBCSD. The greenhouse gas protocol. Internet site: http://www.
ghgprotocol.org/les/ghg-protocol-revised.pdf; 2004.
[88] Yan GE. Research on green suppliers evaluation based on AHP and genetic
algorithm. In: Proceedings of international conference on signal processing
systems; 2009. p. 6159.
[89] Zhu K, Sarkis J. Relationships between operational practices and performance
among early adopters of green supply chain management practices in Chinese
manufacturing enterprises. Journal of Operations Management 2004;22
(3):265289.
[90] Zhu Q, Sarkis J, Cordeiro JJ, Lai K. Firm-level correlates of emergent green
supply chain management practices in the Chinese context. OMEGAInternational Journal of Management Science 2008;36(4):577591.