Multi-Criteria Decision Models of ERP Selection

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 21

Article

A Comparative Analysis of Asia-Pacific Journal of Management


Research and Innovation

Multi-criteria Decision Models


12(3&4) 250–270
© 2017 Asia-Pacific
Institute of Management
for ERP Package Selection for SAGE Publications
sagepub.in/home.nav
Improving Supply Chain Performance DOI: 10.1177/2319510X16688988
http://apjmri.sagepub.com

Indranil Ghosh1
Sanjib Biswas1

Abstract
Advancement in information and communication technology (ICT) has turned information as one of the key organisational resources
for managing business operations. In this information age, one of the key strategic considerations for achieving competitive advantage
involves an information intensive supply chain management (SCM), integrating all the activities and partners beyond the enterprise
boundaries. Enterprise resource planning (ERP) forms the backbone of such extended enterprises. There has been a growing interest
of the organisations to invest in the ERP solutions. However, selection of an appropriate ERP package plays a critical role in achieving
the business objectives. In this article, we have identified the criteria for selecting an appropriate ERP package. Further, we have used
different multi-criteria decision analysis (MCDA) methods for ranking of the alternative ERP packages to select a suitable one. A
comparative analysis has been carried out. Some future scopes of work have also been highlighted.

Keywords
Enterprise resource planning (ERP), multi-criteria decision analysis (MCDA), supply chain management (SCM), Decision-making Trial
and Evaluation Laboratory (DEMATEL), exploratory factor analysis (EFA)

Introduction operations across the entire value chain. In order to mitigate


business risks while operating in the uncertain business
Over the last few decades, organisations across the globe
environment, collaborative supply chain management (SCM)
have witnessed a transformational change in business
operations. The definition of competitive advantage has has become one of the key enablers of success today.
undergone radical modification due to sublime growth in The advent of web 2.0 has put forward an unprecedented
technology and the changing nature of the global market. opportunity in front of the organisations to operate with
Over the years, the organisations have been confronted data-driven insight and take informed strategic decisions. It
with shorter product life cycle, challenges of disruptive enables the organisations to quickly respond to the changes
innovations, requirements for supreme quality and service by taking appropriate decisions on time and avail new
along with the pressure imposed by the fleeting lead time. opportunities while operating as an extended enterprise.
In order to operate fast while being flexible and economi- The electronic supply chain management (e-SCM) provides
cal, the focus of the organisations has been shifted from the a platform for the members of the supply chain to share the
efficiency of the individual operations to the effectiveness information on time to operate collaboratively in the global
of the entire system. In the present age, the competitive marketplace for achieving business objectives. This has
strategy essentially focuses on building the dynamic capa- been reflected in the substantial investments in the
bilities of a supply chain while operating in a volatile information technology (IT) made by the organisations
business environment. This stimulates the importance of over the years. The enterprise resource planning (ERP) is
having collaboration and integration of all the business an integrated system which provides support to all the core

1 Assistant Professor, Calcutta Business School, Diamond Harbour Road, Bishnupur, West Bengal, India.

Corresponding author:
Sanjib Biswas, Assistant Professor, Calcutta Business School, Diamond Harbour Road, Bishnupur 743503, 24 Parganas (South), West Bengal, India.
E-mail: sanjibb@acm.org
Ghosh and Biswas 251

business processes across the entire supply chain. It enables excellence. Balancing of the operations, enhancement of
the organisations to integrate various business processes productivity, quality of the products and services, and
across the extended enterprise for consolidation and reduction in the cost are no longer the order winners. These
optimisation of the resources (Mabert et al., 2003). In criteria have become mere order qualifiers. Ensuring
essence, ERP helps to share information and knowledge capability to promise what the customers want such as
among the members of the supply chain for improving the variety in the products and services, faster and convenient
efficiency and effectiveness of the business operations, delivery, innovative products and services at affordable
reduction of cost and better governance (Aladwani, 2001). cost have opened up a new frontier in business excellence.
It aims to bring synergy among different processes to The present age is witnessing the competition among the
harmonise cross-organisational applications and procedures supply chains instead of the organisations. Experts believe
(Wier, Hunton, & HassabElnaby, 2007). In effect, it that the survival of the organisations in the coming age
facilitates an uninterrupted flow of the functional shall depend on the ability of harmonisation, that is, with
information, which in turn enhances the capability of the the ability to operate with a network of the networks.
supply chain and improves its performance. Embracing the
Hence, it invokes transformational change in the perceived
benefits of ERP, organisations can elevate their competitive
meaning of SCM. The scope of SCM has been broadened
capabilities and withstand business risks imposed by the
up with the advent of the network economy (Arthur, 1996).
uncertain business environment.
Vertical excellence has been outplayed by the horizontal
In this article, we have highlighted the importance of
the ERP system in the context of supply chain perfor- synergy. The fundamental concept of the supply chain as a
mance and competitive advantage. We have summarised single entity consisting of all the functional areas such as
different criteria for selecting an appropriate ERP package. purchasing, manufacturing and distribution, responsible
By conducting an exploratory factor analysis (EFA), we for providing value-added products and services to the
have listed appropriate criteria for ERP package selection. customers (Christopher, 1992), has been elevated to a
Based on those criteria, we have applied multi-criteria much wider perspective. Supply chain in the present age
decision analysis (MCDA) techniques like Technique for represents the network of businesses and synchronisation
Order of Preference by Similarity to Ideal Solution of relationships (Lambert & Cooper, 2000) in order to meet
(TOPSIS), Decision-making Trial and Evaluation Labo- the demands of the ever-changing market in a customised
ratory (DEMATEL) and Measuring Attractiveness by a and transparent way (Jensen, 1999) at the pace at which the
Categorical Based Evaluation Technique (MACBETH) to business world is evolving (Brown & Eisenhardt, 1998). In
rank different commonly available ERP packages. Here, essence, SCM is the integration of the core business
we have presented a comparative analysis of the results processes across the entire chain. The crucial considerations
obtained from different MCDA techniques to arrive at an for this integration include: (i) identification of critical
appropriate decision. The future scope of work is also supply chain members; (ii) identification of the processes
included. to be linked; and (iii) identification of the nature and level
The remainder of this article is organised as follows. of integration (Lambert, 2006).
The next section discusses the importance of ERP vis-à-vis Fine (1998) argued that effective supply chain design
supply chain performance. The section after that reviews unifies the chains of different dynamic capabilities instead
the importance of ERP package selection and different of fostering collaboration among different organisations.
criteria for selection. The ‘Selection of criteria’ section This dynamic view of supply chain well reflects the
presents the results of EFA for selecting the appropriate evolving nature of SCM, consistent with the changing
selection criteria for ERP package selection. In section nature of the markets. In this context, the author compared
‘Ranking of ERP packages using DEMATEL and TOPSIS the rate of change in the products, processes, technologies
methods’, ranking of the commonly used ERP packages and organisational dynamics with the clock speed (referred
based on the selected criteria is carried out using TOPSIS
as industry clock speed). According to him, the prime
and DEMATEL methods. In ‘Ranking of the ERP packages
supply chain design commensurates with superlative SCM.
using MACBETH method’ section, the ranking is carried
Essentially, there are three perspectives for effective supply
out using MACBETH technique. The next section presents
chain design: (i) organisational supply chain that
the comparative analysis of the rankings using different
emphasises on the value-added activities being performed
methods. Finally, the last section concludes the article
while mentioning the future scope of work. across the chain; (ii) capable supply chain which identifies
key business capabilities and competence; and (iii)
technology supply chain which focuses on key technologies
Impact of ERP on Supply Chain (Fine, 1998). Kashyap (2011) recognised supply chain
Performance design as a core competency. The order-winning supply
The rise in technology and the changing nature of the chain strategy has two essential considerations: efficiency
competition have transformed the concept of business for supplying functional products by performing physical
252 Asia-Pacific Journal of Management Research and Innovation 12(3&4)

functions and responsiveness for supplying innovative The internal operations and business decisions mostly
products by carrying out the market mediation functions get affected by the external factors while operating in a
while maintaining the agility (Cheung, Cheung, & Kwok, turbulent business environment. Thus, the critical success
2012; Fisher, 1997). of the organisations largely depends on their abilities to
With the development in information and communica- cope up the uncertainties prevailing over the global envi-
tion technology (ICT), SCM has witnessed a paradigm ronment. In this context, Tam et al. (2002) suggested that
shift. There has been a growing use of ICT in integrating the performance of ERP system should not be measured
the business operations like procurement, manufacturing stand-alone, rather that must be judged based on the activi-
and logistics to provide value-added products and services ties of the supply chain. In other words, SCM performance
to the end customers (Kovacs & Paganelli, 2003). Data is has a linkage with the effectiveness of the ERP system.
being generated from every operation across the supply The supply chain performance indicators play a signifi-
chain. Biswas and Sen (2016) summarised the types and cant role in governing the activities pertaining to SCM to
nature of the data generated by different operations in support the objectives of the supply chain and provide the
a typical supply chain. This necessitates the importance basis for long-term decisions and future course of actions
of the information in the context of a data-driven supply (Aramyan et al., 2007; Taylor, 2004). In the process, the
chain. Sharing of information among the members of a performances of the supply chain partners are integrated to
supply chain simplifies the operations, brings flexibility in achieve goal congruence for satisfying the needs of the
the operations, reduces cost, enhances the speed of the end customers while the feedbacks from the end customers
operations and, more importantly, enables the organisa- are being addressed and dynamic capabilities are being
tions to take appropriate decisions on time. ERP assists the assessed (Wisner, Tan, & Leong, 2008). Stavrulaki and
organisations in executing the strategic supply chain objec- Davis (2010) asserted that by setting key performance indi-
tives, which in turn enables to achieve competitive advan- cators cross-organisational activities can be defined, ana-
tage (Choy, Lee, & Lo, 2003). In their study, Tam, Yen and lysed and improved to achieve sustainable business growth.
Beaumont (2002) emphasised the integration of ERP and The basic objectives are to identify and eliminate waste,
SCM for effective communication and information flow reduce cycle time, improve quality and generate a supply
among the partners of a supply chain to overcome the bar- chain surplus (Gunasekaran & Ngai, 2004). In essence,
riers of natural boundaries. In fact, the integration of ERP performance measurement evaluates the efficiency and
and SCM is natural in nature, which requires strategic con- effectiveness of the system based on pre-defined metrics
siderations and managerial support (Tarn et al., 2002). (Gunasekaran & Kobu, 2007). The SCM performance is
Kumar and Van Hillegersberg (2000) presented ERP as an reflected on multiple dimensions. Thus, a single measure is
integrated, configurable information system, which inte- not sufficient enough to assess the performance (Asadi,
grates the cross-functional business processes and shares 2012). In this context, Wong and Wong (2008) mentioned
the information across different divisions and stages within the importance of treating the entire supply chain as a
and outside the boundaries of the organisation. This essen- whole entity while measuring its performance for accom-
tially improves the efficiency and flexibility. In essence,
modating multiple inputs and outputs to the system.
it provides a seamless integration of all the business
Kashyap (2011) mentioned that an ERP system is an
processes on a single information system platform. The
integral part of the SCM. It is evident from the empirical
organisations can leverage this cross-functional integra-
study on the Taiwanese IT firms carried out by Su and Yang
tion for achieving inclusive business growth and thus,
(2010) that ERP improves overall competence of SCM.
ERP plays as a critical success factor for the organisations
These benefits stem from operations, business processes
(Palaniswamy & Frank, 2000).
and management, and strategic planning and control.
However, Chang et al., (2008) mentioned that although
An ERP system enhances the visibility and transparency
the organisations are aware of the benefits of using an
across the supply chain. It enables to operate with more
ERP system, it is equally important to select the appropri-
customisation of the products and services and more stand-
ate partner for enhancing the efficiency and effectiveness
ardised processes and information. Essentially, ERP inter-
of SCM performance while operating in a complex busi-
mingles the back-office operations with the front-office
ness environment. They carried out a study on a transna-
activities. However, the performance of an ERP system is
tional textile firm in Taiwan to assess the competitive
often limited by the attributes like flexibility, functionality
advantage derived out of having an ERP system in place,
beyond transactions and modularity (Akkermans et al.,
from the perspective of SCM. The authors advocated
that integration of supply chain operations on the ERP 2003). Thus, one of the factors essential for successful
platform enables the partners to streamline business opera- implementation of ERP is the selection of suitable package.
tions for providing timely and value-added services to the Kashyap (2011) mentioned that not all ERP packages are
customers. suitable for all the organisations. The ERP package suitable
Ghosh and Biswas 253

for a particular organisation may not be suitable for the GFI, CFI and NFI values are greater than 0.9 and
other organisations, even if it has the same scale of opera- RMSEA value is less than 0.05. The value of AGFI is also
tions. It is imperative to make a wise choice of the package very close to 0.9. Hence, it is very apparent that ERP
which best suits the needs of the organisations. package selection plays a pivotal role in the supply chain
We have further tested a small structural model to process performance.
explore the nexus between ERP package selection and
supply chain process performance in the Indian industrial
sector to validate the related literature. The conceptual ERP Package Selection:
model is portrayed in Figure 1. Importance and Criteria
Standard partial least squares (PLS) model have been Selection of ERP package and implementation essentially
executed to determine path coefficients between ERP requires the integration of IT and business knowledge.
package selection, delivery lead time and quality. The ERP implementation includes a set of interrelated activities
results are narrated in Table 1. such as analysis of the requirement and the business pro-
Fitness statistics of the above model are presented in cesses, selection of the appropriate package, customisa-
Table 2. tion of the ERP functionalities aligned to the business

Figure 1. Structural Model

Source: Constructed using M-MACBETH software.

Table 1. Statistics of Structural Model

Comparison to
Path Estimate t-Value Critical Value Significance Inference
ERP package selection and 0.93 6.153 6.153>1.96 ** ERP package selection is positively associated
delivery lead time with delivery lead time
ERP package selection and quality 0.78 7.152 7.152>1.96 ** ERP package selection is positively associated
with quality.
Source: Authors’ own creation.
Note: ** Significant at 0.01 level.

Table 2. Fitness Measure

Chi-square/degree of
freedom GFI AGFI CFI NFI RMSEA
2.715 0.91 0.88 0.93 0.94 0.032
Source: Authors’ own creation.
254 Asia-Pacific Journal of Management Research and Innovation 12(3&4)

objectives and support services. Researchers and practi- implementation, application possibilities, flexibility, com-
tioners have argued that in many instances the organisa- patibility with other applications, user-friendliness, custo-
tions failed to reap the benefits of ERP due to inappropriate misation, etc.). Bueno and Salmeron (2008) mentioned that
choice of the package and service partners within the there is a strong interrelation between ERP selection crite-
constraints of time and budget. Berchet and Habchi ria and expectations out of the selected ERP package.
(2005) conducted a deeper study on ERP implementation. However, Ratkevičius, Ratkevičius and Skyrius (2012)
According to them, ERP system must enable the organisa- argued that there are no standard selection criteria for ERP
tions to achieve collaboration, efficiency, transparency and packages. They emphasised on the contextual require-
real time control, but strategic alignment is also required. ment and functionality. This is supported by the arguments
Over the years, researchers and practitioners have con- made by several other researchers as they felt that no ERP
ducted studies to find out the criteria to select an appropri- package alone can fulfil all the business objectives and
ate ERP package in different contexts. Table 3 exhibits the functional requirements (Hong & Kim, 2002; Teltumbde,
contributions of the researchers in finding out the influenc- 2000; Wei et al., 2005).
ing factors for ERP package selection.
In essence, the selection criteria are broadly related
to (i) system (technical features like modularity, function- Selection of Criteria
ality, etc. system requirements & implementation issues, In order to identify the key criteria for ERP package
security, maintenance, cost, etc.); (ii) support (vendor selection process, primary survey has been conducted in
support, reputation and quality, technical skills of the the Indian context through a detailed questionnaire. A total
employees, expert opinions, etc.); and (iii) use (ease of 1686 Indian Micro, Small and Medium Enterprises

Table 3. Criteria for ERP Package Selection

Criterion Authors
Cost Keil and Tiwana (2006); Kumar, Maheshwari and Kumar (2002); Kumar, Maheshwari and Kumar
(2003); Wei, Chien, & Wan (2005); Ayaǧ and Özdemir (2007); Bueno and Salmeron (2008);
Teltumbde (2000); Baki and Cakar (2005)
Reliability Keil and Tiwana (2006); Kumar et al. (2002); Kumar et al. (2003); Wei et al. (2005); Ayaǧ and
Özdemir (2007); Bueno and Salmeron (2008); Nah and Delgado (2006); Baki and Cakar (2005)
Functionality Keil and Tiwana (2006); Wei et al. (2005); Ayaǧ and Özdemir (2007); Han (2004); Nah and
Delgado (2006); Liao et al. (2007); Teltumbde (2000); Baki and Cakar (2005)
Technical factors Kumar et al. (2002); Kumar et al. (2003); Ayaǧ and Özdemir (2007); Han (2004); Rao (2000);
Teltumbde (2000); Baki and Cakar (2005)
Modular integration Kumar et al. (2002); Kumar et al. (2003); Bueno and Salmeron (2008); Baki and Cakar (2005);
Scalability Han (2004)
Ease of customisation Baki and Cakar (2005); Wu (2008)
Ease of use Keil and Tiwana (2006); Kumar et al. (2002); Kumar et al. (2003); Wei et al. (2005); Ayaǧ and
Özdemir (2007)
Set of applications Bueno and Salmeron (2008)
Ease of implementation Keil and Tiwana (2006); Teltumbde (2000); Baki and Cakar (2005); Bueno and Salmeron (2008)
Implementation time Wei et al. (2005); Bueno and Salmeron (2008); Baki and Cakar (2005)
Compatibility with allied Kumar et al. (2002); Kumar et al. (2003); Baki and Cakar (2005); Wu (2008)
organisational system
Availability of business practices Kumar et al. (2002); Kumar et al. (2003)
Fitment with organisational Kumar et al. (2002); Kumar et al. (2003); Bueno and Salmeron (2008); Nah and Delgado (2006);
processes Hong and Kim (2002)
Flexibility Wei et al. (2005); Ayaǧ and Özdemir (2007); Teltumbde (2000)
Vendor support Wei et al. (2005); Ayaǧ and Özdemir (2007); Han (2004); Verville and Halingten (2003)
Vendor reputation Keil and Tiwana (2006); Kumar et al. (2002); Kumar et al. (2003); Wei et al. (2005); Liao et al.
(2007); Verville and Halingten (2003); Teltumbde (2000)
Ghosh and Biswas 255

Criterion Authors
Vendor’s technical capability and Wei et al. (2005); Liao et al. (2007); Verville and Halingten (2003); Baki and Cakar (2005);
credentials Teltumbde (2000)
Hardware requirement Bueno and Salmeron (2008)
Administration Teltumbde (2000)
Consultancy support Baki and Cakar (2005)
Credibility of the system Bueno and Salmeron (2008); Nah and Delgado (2006)
Trust Bueno and Salmeron (2008); Rao (2000)
Knowledge/Skill Bueno and Salmeron (2008); Rao (2000)
Planning Bueno and Salmeron (2008)
Training and education Bueno and Salmeron (2008)
Organisational culture Bueno and Salmeron (2008)
Company size Bueno and Salmeron (2008)
Organisational strategy Bueno and Salmeron (2008); Liao et al. (2007); Teltumbde (2000)
Affordability Rao (2000)
Risk Teltumbde (2000)
Source: Authors’ own creation.

(MSME) organisations were approached out of which 364 Table 4. Sample Information
(21.59 per cent) organisations responded. Due to inadequate
Feature Percentage (%) Frequency
information and missing data, we excluded 41 (2.43 per
cent) samples which resulted into 323 (19.16 per cent) Workforce of the organisation
number of valid samples. Once the responses were received, *Less than or equal to 20 10.53 34
*Between 21 to 50 19.81 64
EFA was utilised to detect the key criteria (Table 4).
*Between 51 to 100 23.84 77
*Between 101 to 500 25.08 81
*More than 501 20.74 67
Exploratory Factor Analysis (EFA) Type of ownership
EFA attempts to discover a smaller subset of factors *New business 19.81 64
(Demsar et al., 2013) by analysing the correlation present *Franchised 34.67 112
*Family business 28.79 93
in the original data set to identify the highly correlated ones *Others 16.72 54
and put them under the umbrella of a new one. EFA is also Status of the firm
referred as dimension reduction technique. There are two *ISO 9001 Certified 69.35 224
steps in EFA process: factor extraction and factor rotation. *Planning to be ISO Certified 30.65 89
Principal component analysis (PCA), principal axis Annual Revenue
factoring (PAF), alpha factoring (AF), etc. are the methods *Rs. 50 Lakh to 5 Crore 14.55 47
*In between Rs. 5 Crore to 20 17.96 58
that are applied to carry out factor extraction process in
Crore 28.79 93
general. We have utilised PCA in our study to serve the *In between Rs. 20 Crore to 50 38.70 125
purpose. Theoretically, using PCA n number of factors or Crore
components can be extracted if number of original items *More than 50 Crore 21.05 68
are present in factor analysis. However, a smaller number Manufacturing Process 40.87 132
of factors, say k << n are extracted that account for the *Continuous 38.08 123
maximum variance in the data. Then factor rotation is *Batch/Repetitive
*Job-shop
usually performed for the betterment of solution by
checking the factor loadings. Varimax and equinox rotation Source: Authors’ own.
methods have been used here. Initially, Kaiser–Meyer– Since the KMO statistic value is greater than 0.8 and
Olkin (KMO) statistic and the Bartlett’s test of sphericity the significance associated with Bartlett’s test of spheri-
are computed to validate whether the data set is suitable city is less than 0.05, inference can be drawn that the data
for factor analysis or not. The results are displayed in set is appropriate for EFA. As discussed earlier, total six
Table 5. components have been extracted using PCA accounting
256 Asia-Pacific Journal of Management Research and Innovation 12(3&4)

Table 5. Test for Suitability for Factor Analysis

Kaiser–Meyer–Olkin (KMO) of sampling adequacy 0.875

Bartlett’s test of sphericity approx. chi-square 12316.423

df 1139
Significance (p-value) 0.000
Source: Authors’ own creation.

Table 6. Findings of EFA (Varimax Rotation)

Extraction Sums of Rotation Sums of Squared


Initial Eigenvalues Squared Loadings Loadings (Varimax)
% of Cumulative % of Cumulative % of Cumulative
Components Total Variance % Total Variance % Total Variance %
1. 5.63 19.58 19.58 5.63 19.58 19.58 5.74 19.15 19.15
2. 5.07 17.63 37.20 5.07 17.63 37.20 5.31 17.72 36.87
3. 4.56 15.86 53.06 4.56 15.86 53.06 4.75 15.85 52.72
4. 3.34 11.61 64.67 3.34 11.61 64.67 3.81 12.71 65.43
5. 2.56 8.90 73.57 2.56 8.90 73.57 2.89 9.64 75.08
6. 1.67 5.81 79.38 1.67 5.81 79.38 1.78 5.94 81.04
7. 0.88 3.06 19.58 5.63 19.58 19.58 5.74
8. 0.65 2.26 37.20 5.07 17.63 37.20 5.31
9. 0.57 1.98 53.06 4.56 15.86 53.06 4.75
10. 0.48 1.67 64.67 3.34 11.61 64.67 3.81
11. 0.39 1.36 73.57 2.56 8.90 73.57 2.89
12. 0.36 1.25 79.38 1.67 5.81 79.38 1.78
13. 0.32 1.11 19.58 5.63 19.58 19.58 5.74
14. 0.31 1.08 37.20 5.07 17.63 37.20 5.31
15. 0.29 1.01 53.06 4.56 15.86 53.06 4.75
16. 0.26 0.90 64.67 3.34 11.61 64.67 3.81
17. 0.24 0.83 73.57 2.56 8.90 73.57 2.89
18. 0.21 0.73 79.38 1.67 5.81 79.38 1.78
19. 0.2 0.70 19.58 5.63 19.58 19.58 5.74
20. 0.18 0.63 37.20 5.07 17.63 37.20 5.31
21. 0.16 0.56 53.06 4.56 15.86 53.06 4.75
22. 0.13 0.45 64.67 3.34 11.61 64.67 3.81
23. 0.11 0.38 73.57 2.56 8.90 73.57 2.89
24. 0.09 0.31 79.38 1.67 5.81 79.38 1.78
25. 0.06 0.21 19.58 5.63 19.58 19.58 5.74
26. 0.03 0.10 37.20 5.07 17.63 37.20 5.31
27. 0.01 0.03 53.06 4.56 15.86 53.06 4.75
Source: Authors’ own creation.

for nearly 80 per cent variation of the total data. Items We have outlined the summary of the extracted criteria
with loading values less than 0.5 were discarded from and their respective constituent items for package selection
further analysis. Tables 6 and 7 exhibit the findings in the Indian context in Table 8.
of EFA.
Ghosh and Biswas 257

Table 7. Findings of Equimax Rotation

Rotation Sums of Squared Loadings (Equimax)


Total % of Variance Cumulative %
5.74 18.95 18.95
5.31 17.42 36.37
4.75 15.98 52.35
3.81 12.54 64.90
2.89 9.51 74.41
1.78 5.91 80.31
Source: Authors’ own creation.

Table 8. Description of the Criteria

Criteria Description Items


Vendor support Vendor support is highly desirable during the entire 1. Service level agreement
process of ERP implementation. Their active collaboration 2. Professionalism
acts as a key driver of ERP implementation success. 3. Stability
4. Customer base
External factor Package selection is highly influenced by various external 1. Market feedback
factors. Package’s market perception, consultants’ 2. Reports of consultants
feedback, etc. should be considered before final selection. 3. Number of successful implementations in other
organisations.
Technical features Depending upon the needs of the organisation, 1. Real time enquiries and reporting functions
functionality requirement of ERP package is decided. 2. Modularity
However, there are few pre-requisite technical features 3. Multilanguage and multicurrency support
which are essential for any ERP packages. Packages 4. Customizable
must be carefully rated against these features for their 5. Ease of usage
candidature of selection. 6. Integration capability with different platforms
and data
ERP cost As the majority of ERP packages is extremely costly 1. Implementation cost
themselves, packages must be chosen that account for the 2. Maintenance cost
least cost of implementation, maintenance, extension, etc. 3. Adoption and extension cost.
activities.
Security The packages’ selection process is heavily affected by 1. Audit trailing ability
security features. As ERP shares a common database, 2. Password protection
proper authorisation and security aspects are extremely 3. Access control
critical for smooth daily operations.
Ease of Once the selection process is over, the Implementation 1. Data backup capability
implementation process commences. So packages that have comparatively 2. Interface with other packages
higher flexibility in various phases of the overall 3. Integration with other utilities.
implementation process, should be chosen in order to
result in commercial success.
Source: Authors’ own creation.

We have outlined the summary of the extracted criteria interactions to evaluate their importance in the package
and their respective constituent items for package selection selection process. We have adopted MCDA framework to
in the Indian context in Table 8. serve the purposes. DEMATEL method and TOPSIS have
Factor loadings of the constituent items under individual been used separately to evaluate the criteria importance or
factors and corresponding reliability scores have been weight values and final ranking of packages, respectively.
reported in the following Table 9. The working principles of the methods and findings obtained
from them have been explained later in this section.
Ranking of ERP Packages Using Due to ever-increasing complexity of performance
measurements, which is one of the most important pro-
DEMATEL and TOPSIS Methods cesses in the management literature, and as its measure-
Once the criteria of package selection process in the Indian ment is critical for judging the success or failure of a firm,
context have been identified, the next step is to study their MCDA techniques have recently been in the limelight of
258 Asia-Pacific Journal of Management Research and Innovation 12(3&4)

Table 9. Statistics of EFA homogeneous groups in a preference


order.
Factor Cronbach’s Ranking: Rank the alternatives of A from best to
Constructs Items Loadings Alpha
worst.
Vendor 1. Service level 0.8526 0.8971
support agreement 0.8138
In the process of ranking, it is very critical to carefully
2. Professionalism 0.7769
3. Stability 0.7540 assess and assign the weights of the criteria based on which
4. Customer base alternatives are ranked. As stated, we have deployed
External 1. Market feedback 0.8642 0.8742 DEMATEL for the problem considered.
factor 2. Reports of 0.8027
consultants 0.7866
3. Number Findings of DEMATEL Methodology
of successful
implementations in It was developed by the Science and Human Affairs
other organisations. Programme of the Battelle Memorial Institute of Geneva
Technical 1. Real time enquiries 0.8539 0.8446 between 1972 and 1976 to critically study and explore
features and reporting 0.8326 complicated real life problems through assessing the causal
functions 0.7741
interrelations between factors/criteria related to the said
2. Modularity 0.7358
3. Multilanguage and 0.6826 problem by construction of a network relations map (NRM;
multicurrency support 0.6537 Zhou, Huang, & Zhang, 2011). It basically is a multivariate
4. Customizable data analysis tool that applies classical matrix algebra to
5. Ease of usage examine complex causal models to distinguish between
6. Integration cause and effect group of factors (Hsu, 2012). It also
capability with
provides digraphs for visualisation of contextual
different platforms
and data relationships and to assist decision-makers in strategy
ERP cost 1. Implementation 0.7781 0.8029 formulation. It has been used to identify the critical success
cost 0.7435 factors of product development in different context as well
2. Maintenance cost 0.7261 (Fekri, Aliahmadi, & Fathian, 2009; Shinno & Hashizume,
3. Adaption and 2002). The computational steps of DEMATEL in the
extension cost.
context of ERP package selection are elucidated below:
Security 1. Audit trailing ability 0.8163 0.7847
2. Password 0.7589 Step 1. Calculate the average matrix: Experts were
protection 0.7317 asked to indicate the degree of direct influence between
3. Access control any two package selection criteria by an integer score
Ease of 1. Data backup 0.8210 0.8342 ranging from 0 to 4, representing no influence, low
implementation capability 0.7972 influence, medium influence, high influence and very
2. Interface with 0.7683 high influence, respectively in order to construct the
other packages
3. Integration with
initial direct influence matrices. A non-negative matrix
other utilities ^ X k = [x kij] h can be obtained for each expert review where
Source: Authors’ own creation.
xij denotes the measure to which the kth (1 ≤ k ≤ H ) expert
evaluates direct effect of criteria i on criteria j. An average
research. MCDA techniques are tailor-made to cater a sys- matrix (A= [aij]) incorporating opinions from every expert
tematic and deterministic approach to tackle complex real can be computed as
world decision-making problems composed of several
1 (1)
intertwining and incommensurate criteria. Roy (1990) a ij = | H xk 
H k = 1 ij
argued that solving MCDA problems and searching for an
optimal solution are clearly two distinct measures; the
prime focus of MCDA is to assist decision-makers (DMs), Step 2. Compute the normalised initial direct-relation
evaluate the complex judgements and to carefully analyse matrix: The normalised initial direct-relation matrix (D) is
data involved in their problems, and advance towards an determined by normalising the average matrix (A) in a
acceptable solution. The entire process is subdivided into manner so that the elements of D can take values between
three parts, a set of alternatives, A, is evaluated to produce 0 and 1. It is constructed as
a final decision result:
D = A # S, (2)
Choice: Choose the best alternative from A.
Sorting: Sort the alternatives of A into relatively where S is measured as
Ghosh and Biswas 259

1 Findings of TOPSIS Method


S= (3)
max 1 # i # n | j = 1 a ij
n
Introduced by Hwang and Yoon (1981), The TOPSIS
method presents a useful framework to rank the alterna-
Step 3. Compute the total-relation matrix: The total-
tives in multiple-criteria decision-making problem. The
relation matrix (T) is computed as
basic principle of the TOPSIS is quite simple and intuitive:
T = D + D 2 + D 3 + ..... + D p = D (I + D + it measures alternatives’ distances to predefined ideal and
anti-ideal points first and then, aggregate the separate dis-
D 2 + . ... . + D p - 1) [(I - D) (I - D) -1]
tance information to reach overall evaluation results. The
salient features of TOPSIS and its successful applications
= D (I - D p) (I - D) -1 .(4) in the improvement of manufacturing competitiveness and
strategic vendor selection are reported in the works of
When p " 3, D p = 0 then T = D ^ I - Dh-1(5) Shyura and Shih (2006), respectively. In our study, experts
from industrial sectors have been appointed to rate 10 ERP
where I is the identity matrix. packages, predominantly used in the Indian context,
If r and c are (n × 1) and (1 × n) vectors representing the against 6 key criteria for ranking purpose. The names of the
sum of rows and columns of matrix T, then ri, sum of ith packages have not been revealed here for confidentiality of
row in T, measures the direct and indirect effects triggered information.
by criteria i on rest of the criteria. On the contrary, Cj, sum The procedural steps of TOPSIS are summarised as
of jth column, is the sum of influences that criteria j has Step 1. For a MCDM problem having n alternatives and
received from the rest. When i = j, the sum (ri + ci) represents m criteria, a matrix (X) can be constructed based on the
the total influence given and received by the criteria in the priority scores assigned to each alternative on each attribute
system. Basically the magnitude of the expression (ri+ci) and is denoted by
represents the degree of influence of criteria i in the system. (6)
Whereas the net contributing effect of criteria i in the X = ^ X ij hnxm
system is measured by the expression, (ri – ci). If (ri – ci) is The importance weight values (wj) of the attributes are
positive, then it belongs to cause group of criteria, while a determined without violating the following constraint:
negative (ri – ci) value implies that the same criteria is being
dominated by others in the system. | m  (7)
j=1
w j = 1, where j = 1, 2,3, ......m.
Step 4. Construction of digraph: In order to generate the
digraph for visualisation of interactions among the criteria, Step 2. The normalised decision matrix is obtained
a threshold value is computed as the average of all the subsequently applying the following equation:
elements of the total-relation matrix (T). Computation of
threshold value is essential to filter out some negligible
rij = x ij / (| k = 1 x 2kj) - 0.5 j = 1, 2, ...m; i = 1, 2, ....n.(8)
n
effects. The effects which are greater than the threshold
value are displayed in the digraph. The obtained total-
Step 3. Next, the weighted normalised decision matrix is
relation matrix (T) is shown in Table 10.
calculated as
Total influence levels (ri + ci), net influence levels (ri –
ci), ranking based on total influence levels and groups
based on net influence levels of individually identified six Vij = w j rij; j = 1, 2, ...., m; i = 1, 2, ....., n.(9)
criteria based on total-relation matrix (T) are mentioned in
Table 11. Criteria having positive net influence levels have Step 4. The Positive Ideal Solution (PIS) and Negative
been categorised into cause group, while others with Ideal Solution (NIS) values are determined by equations x
negative net influences into effect group. Magnitude of and y, respectively:
interactions between any two criteria is actually represented A + = (v 1+, v 2+, (ev n+) = {(max i{v ij}
by the corresponding entries of the total-relation matrix. (10)
| j ! B), (min i {v ij} | j ! C)}
Interaction levels greater than the threshold value are
considered to be the significant ones and are indicated by
an arrow in the digraph as shown in the Figure 2. Importance A - = (v 1-, v 2-, (v n-) = {(min i{v ij}
(11)
or weight values of individual criteria are fixed according | j ! B), (max i {v ij} | j ! C)}
to the values of total influence level (ri + ci) for ranking the
packages using TOPSIS method. Criteria are ranked based Step 5. The separation measures of each alternative from
on total influence levels (ri + ci) as well. Ties are broken by the PIS and NIS in terms of Euclidean distance is calculated
as
considering the net influence levels (ri – ci).
260 Asia-Pacific Journal of Management Research and Innovation 12(3&4)

Table 10. Total-relation Matrix

Vendor External Technical Ease of


Criteria Support Factor Features ERP Cost Security Implementation
Vendor support 0 0.4 0.4 0.4 0.2 0.4
External factor 0.1875 0 0 0.2 0 0.4
Technical 0.375 0.6 0 0.8 0.6 0.6
features
ERP cost 0.375 0.2 0.6 0 0.4 0.4
Security 0.1875 0.4 0.2 0.4 0 0.2
Ease of 0.1875 0.4 0.2 0.6 0.2 0
implementation
Source: Authors’ own creation.

Table 11. Total Effect, Net Effect, Rank and Group of Individual Criteria

Criteria ri + ci ri – ci Rank Group


Vendor support 3.1125 0.4875 3 Cause
External factor 2.7875 –1.2125 5 Effect
Technical features 4.375 1.575 1 Cause
ERP cost 4.375 –0.425 2 Effect
Security 2.7875 –0.0125 4 Effect
Ease of 2.5875 –0.4125 6 Effect
implementation
Source: Authors’ own creation.

Figure 2. Digraph

S +i = $ | j = 1 (v ij - v +j ) 2 . ; i = 1, ...., n(12)
m 0.5 preferred and least preferred feasible solutions. The ranking
of the packages based on the TOPSIS method is given in
Table 12.
S -i = $ | j = 1 (v ij - v -j ) 2 . ; i = 1, ...., n(13)
m 0.5

Step 6. Finally, the relative closeness of a particular Ranking of the ERP Packages Using
alternative (Ti) to the ideal simulator, can be expressed MACBETH Method
by
s -i MACBETH is one of the MCDA approaches, which is
Ti = + (14) based on multi-attribute value theory. MACBETH is an
(s i + s -i ) interactive pairwise comparison approach to form a quan-
The set of alternative is sorted in a descending order based titative value model based on qualitative value judgements
on the relative closeness values, indicating the most (Bana e Costa & Vansnick, 1999). It consists of a set of
Ghosh and Biswas 261

Table 12. Ranking of the ERP Packages Using TOPSIS Method alternative with respect to two reference levels. The upper
reference level is treated as good and is assigned a value of
Packages Aggregate Score Ranking
100, whereas the lower reference level is considered as
Pkg 1 0.5177 1 neutral and is assigned a score of 0. However, these do not
Pkg 2 0.4816 4 necessarily ascertain any conclusion on the best or the
Pkg 3 0.4932 2 worst performance of the alternatives for a particular
criterion.
Pkg 4 0.4547 8
Step 2. In this step, the ordinal performance scale is
Pkg 5 0.4727 6 converted into a proportional cardinal scale for comparing
Pkg 6 0.4768 5 the performances of all the alternatives with respect to all
Pkg 7 0.4371 10 the decision criteria. In order to carry out such operation,
Pkg 8 0.4883 3
one (n x n) matrix is constructed for each criterion where n
is the number of performance levels assigned to that crite-
Pkg 9 0.4639 7 rion where the performance levels follow the descending
Pkg 10 0.4430 9 order according to their importance from left to right and
Source: Authors’ own creation. from top to bottom. Next the decision-maker needs to put
qualitative judgements regarding the difference of attrac-
tiveness between two alternatives at a time. In MACBETH
techniques that are applied to investigate the performance method, there are seven semantic scales such as no, very
of a number of alternatives, having multiple criteria and weak, weak, moderate, strong, very strong and extreme on
conflicting objectives based on a set of decision criteria continuum to map the difference of attractiveness where,
expressed in a qualitative form (Bana e Costa & Oliveira, no represents the indifference of attractiveness between the
2012). In this technique, two criteria are compared based alternatives and extreme signifies that one alternative is
on the difference of attractiveness at a time with qualitative extremely attractive over the other. However, it is to be
judgements. This method not only allows the decision- noted that if two alternatives are equally attractive as per
makers to check the consistency of the judgements but also the consideration, they are assigned with no. This (n x n)
allows improving the judgements in the case of inconsist- matrix is also called the matrix of judgement. The judge-
ency. Consistent judgements are then transformed into ments are then checked for consistency. The tool
MACBETH scale based on linear programming models. M-MACBETH automatically verifies the consistency and
Finally, based on the overall weighted scores, the alterna- notifies if there is any consistency suggesting possible
tives are ranked according to the degree of attractiveness. alterations. After that the consistent judgements are con-
M-MACBETH is a tool developed using linear program- verted into proportional cardinal scales. The mathematical
ming algorithm, which supports the MACBETH method. explanation is given below.
Notation:

Steps of the MACBETH Method Li: Performance levels; i = 1, 2, 3… n


L+: Upper reference level for the criterion Cj
Karande and Chakraborty (2014) applied the MACBETH
L–: Lower reference level for the criterion Cj
method to propose a facility selection model. The basic
Cj: Any particular criterion; j = 1, 2, 3 …. n
steps of the aforesaid method as described below are based
Ak : Any particular alternative or option.
on the summarisation made by Karande and Chakraborty
Wj: Weight for the criterion Cj
(2014) following the concept given by Bana e Costa and
V (Li): Value of the performance level Li, that is,
Vansnick (1999) and Bana e Costa, Corrêa, De Corte and
MACBETH score; V (Li) ϵ [0,100]
Vansnick (2002).
H: Strength, where H ϵ {0, 1, 2, 3, 4, 4, 5, 6}
Step 1. The first step involves the formation of a value
tree based on the decided relevant set of decision criteria. Let us consider five performance levels L1, L2, L3, L4
The numbers of alternatives are entered as options into the and L5. Let us consider L1 as good and L5 as neutral and the
system. After that the possible performance levels of the preference of importance as L1> L2> L3> L4> L5. Therefore,
alternatives for each criterion are entered on an ordinal V (L1) = 100 and V (L5) = 0 and V (L1) > V (L2) > V (L3) >
scale. In this context, the decision-maker needs to assign a V (L4) > V (L5). The next step requires comparison of the
set of reference levels with respect to each such criterion. strengths of performance of the levels. For n number of
The minimum number of reference levels is two such as reference levels, there can be maximum n (n-1)/2 number
the upper reference level and the lower reference level. of comparisons. If the decision-maker does not assign any
Each alternative is assigned with a performance score, strength, it is notified by positive in the cell of the matrix
which represents the relative attractiveness of the under consideration.
262 Asia-Pacific Journal of Management Research and Innovation 12(3&4)

Let us consider the performance levels for a particular V (A k) = | j = 1 w j $ V ^C j h,


n

criterion that are arranged in a matrix as depicted by Table (20)


where | j = 1 w j = 1 and V ^ L+h = 100,V ^ L-h = 0
n
13. In the next step, the comparison of the strengths of
performance is carried out based on the scales as mentioned
earlier.
Results: Ranking based on MACBETH
V ^L 1h - V ^L 2h = Ha,(15) Method
Here, we use the MACBETH method for ranking of some
where α denotes the coefficient necessary to satisfy V (Li) commonly used ERP packages based on pre-qualified
ϵ [0,100]. selection criteria. For carrying out the analysis, we have
The MACBETH scores are calculated by solving the used the software M-MACBETH, version 2.4.0 as a tool.
related equations for all the performance levels. The criteria are listed in the Table 14.
From Table 13, we can obtain the equations as Based on these criteria, a value tree is constructed which
is shown in Figure 3. The performances of all the alternative
V ^L 1h - V ^L 2h = 5a(16) ERP packages are expressed in the form of ordinal, that is,
qualitative form. In our case, we have 10 alternative
V ^L 2h - V ^L 3h = 4a(17) packages to compare. Thus, the number of options is 10.
Among the aforesaid criteria, all the criteria are beneficial
in nature except the criterion of ERP cost, which has a non-
V ^L 3h - V ^L 4h = 3a(18) beneficial nature, that is, if the cost of the package is high,
organisations treat this as a less attractive one. Karande and
Chakraborty (2014) followed degree of seven ordinal
V ^L 4h - V ^L 5h = 2a(19) performance levels such as very good (VG), good (G),
medium good (MG), fair (F), medium poor (MP), poor (P)
On solving, we get, α = 7.14, V (L2) = 64.26, V (L3) = 35.7, and very poor (VP) to compare the performance of the
V (L4) = 14.28, and we know that V (L1) = 100 and alternatives as used by Li, Yamaguchi and Nagai (2007)
V (L5) = 0. while solving a supplier selection problem. Here, we also
In this way, the quantifications for all the criteria are use this scale in our analysis. For all the beneficial criteria,
carried out. the performance levels are organised in the descending
Step 3. In this step the performances of different order of their attractiveness, that is, VG is treated as upper
alternatives are assessed with respect to different criteria. reference level and VP is considered as lower reference
This table is called table of performances. In this case, the level, whereas for the non-beneficial criterion it follows the
relative criteria weights are set based on the judgements of opposite order.
the decision-makers or the experts. The next step involves the mapping of the difference of
Step 4. In this step, the overall weighting table is attractiveness among the ordinal performance measures
constructed and finally the table of scores is obtained, between two alternatives at a time by the decision-makers.
which ranks different alternatives based on the overall The differences of attractiveness of the performance levels
scores. Here, the logic of additive value model is followed. based on the aforesaid semantic scales for the criteria, ERP
As per this model, the overall global score of a particular cost and technical features are shown in the Figures 4–7
alternative Ak is given by respectively. It is seen from the Figure 4 that the
performance level VP is very weakly attractive to P, in case

Table 13. Strengths of the Performance Levels for the


Criterion Cj
Table 14. List of Criteria
Performance
Levels L1 L2 L3 L4 L5 Criterion Description

L1 (Good) No Very Positive Positive Positive C1 ERP cost


strong C2 Technical features
L2 No Strong Positive Positive C3 Security
L3 No Moderate Positive C4 External factor
L4 No Weak C5 Vendor support
L5 (Neutral) No C6 Ease of implementation
Source: Authors’ own creation. Source: Authors’ own creation.
Ghosh and Biswas 263

Figure 3. Value Tree Representing the Criteria

Source: Constructed using M-MACBETH software.

of ERP cost criterion. The consistency check for the attractiveness of each criterion with respect to another is
judgements is conducted next, and it is found to be judged using the above-mentioned seven-point scale. The
consistent. Also, based on the differences of attractiveness, judgements are then checked for consistency. After that, in
the ordinal performance levels are converted into order to judge the performances of all the alternative ERP
proportionate cardinal MACBETH scales following linear packages with respect to the criteria, corresponding criteria
programming models. For example, from Figure 6 the weights are calculated, and finally the overall weighting is
converted MACBETH scales of the performance levels for carried out based on the opinions of the decision-makers.
the criterion of technical features can be calculated as Figure 9 exhibits the weighting references and Figure
follows: 10 presents the overall weighting judgements for all the
V(VG) – V(G) = α, V(G) – V(MG) = α, V(MG) – V(F) alternatives based on the available criteria. A consistency
= α, V(F) – V(MP) = α, V(MP) – V(P) = α and V(P) – check is performed and the result was found to be con-
V(VP) = α, where, V(VG) = 100 and V(VP) = 0 for the sistent. M-MACBETH software calculates the overall
criterion of ERP cost. On solving, we get α = 16.67, V(P) = attractiveness scores for all the alternative ERP packages
16.67, V(MP) = 33.33, V(F) = 50, V(MG) = 66.67 and as calculated by using the equations mentioned earlier.
V(G) = 83.33. Figure 11 exhibits the overall attractiveness scores for the
alternatives.
The same methodology is followed to quantify the
Table 15 shows the ranking of the ERP packages based
ordinal performance levels for all the other criteria.
on the overall quantified cardinal scores. However,
In our study, we have 10 alternative ERP packages to
criterion-wise ranking of the ERP packages is also shown
compare and rank them based on their performance scores
in Figure 12. It can be seen that package number 1 is ranked
for selecting the appropriate one. In order to set relative
first, whereas the package number 7 is at the last position.
weights for the different criteria to construct the table of Therefore, the ranking follows the order as Pkg. 1–3–8–
performances for the alternative ERP packages, a team 2–6–4–5–9–10–7. After that sensitivity analysis was
of six decision-makers was formed. The team consisted of carried out to check the stability of the results and the
industry experts and subject experts from academia. In our analysis where the criteria weights were changed for testing
case, we considered the average of mapping scores whether such deliberated constrained changes put any
suggested by different experts of the team as the significant impacts on the obtained results. In our example,
performance of an alternative ERP package with respect to C1 has the maximum weight. The effect of variation in the
a particular criterion. Figure 8 represents the table of weight of C1 on the overall rankings of the alternative ERP
performances, which shows all the alternative ERP packages is tested in the form of sensitivity analysis. Here,
packages and their performance requirements with respect the weight of the criterion C1, that is, the ERP cost varies
to the criteria for evaluation. M-MACBETH calculates consistently, while adjusting the difference over the other
the criteria weights in the same way quantification of the criteria weights. Figure 13 shows the results of the
ordinal data is done. In this table of performances, sensitivity analysis with respect to the criterion C1. It can
the decision-makers set the performance level requirement be seen that Packages 1 and 2 remain the first two choices
with respect to each criterion for the alternatives. The while having no impact on Package 10 as the last choice.
264 Asia-Pacific Journal of Management Research and Innovation 12(3&4)

Figure 4. Comparison of Performance Levels for the Criterion ERP Cost

Source: Constructed using M-MACBETH software.

Figure 5. MACBETH Scale for the Criterion ERP Cost

Source: Constructed using M-MACBETH software.

Figure 6. Comparison of Performance Levels for the Criterion Technical Features

Source: Constructed using M-MACBETH software.


Ghosh and Biswas 265

Figure 7. MACBETH Scale for the Criterion Figure 8. MACBETH Table of Performance
Technical Features

Source: Constructed using M-MACBETH software. Source: Constructed using M-MACBETH software.

Figure 9. Weighting References

Source: Constructed using M-MACBETH software.

Figure 10. Overall Weighting Judgements

Source: Constructed using M-MACBETH software.


266 Asia-Pacific Journal of Management Research and Innovation 12(3&4)

Figure 11. Overall Attractive Scores of the Alternative ERP Packages

Source: Constructed using M-MACBETH software.

Table 15. Overall Ranking of the ERP Packages Figure 12. Criterion-wise Ranking
Package No. Overall Score Ranking
1 94.79 1
2 85.18 4
3 93.76 2
4 83.83 6
5 78.95 7
6 84.39 5
7 71.12 10
8 90.48 3
9 73.61 8
10 71.47 9
Source: Authors’ own creation.

Source: Constructed using M-MACBETH software.


Similarly, the sensitivity analyses for all the other
criteria are carried out. It can be seen that the results
obtained out of MACBETH analysis are quite consistent
and acceptable, which withstand the variations in the C2 is increased to ± 4 per cent, the model will no longer
weights assigned to different criteria. have enough information to determine the relative
After that a robustness analysis was also conducted. In attractiveness between the packages 2 and 6, but it does not
the real world, decision-making is most often required to have an impact on the best choice given by the model.
deal with incomplete or uncertain information. The robustness Thus, it can be concluded that the results are consistent and
analysis reveals the effect of the data uncertainty on the the model is robust.
results given by a particular model (Bana e Costa, De
Corte, & Vansnick, 2012). Figures 14 and 15 represent the
Comparative Analysis
result of the robustness analysis.
From Figure 15, it is evident that if in the local It is seen from the above analyses that the results are
information table, the cardinal scale value for the criterion comparable in nature and show consistency in the findings.
Ghosh and Biswas 267

Figure 13. Sensitivity Analysis with Respect to the Criterion C1

Source: Constructed using M-MACBETH software.

To carry out the comparative analysis of the obtained EFA to identify the criteria for selecting an appropriate
rankings by DEMATEL–TOPSIS and MACBETH ERP package. Advanced MCDA techniques like
frameworks, Spearman’s rank correlation test has been DEMATEL–TOPSIS and MACBETH have been per-
applied to two sets of rank alternatives. The result of formed to rank the alternative ERP packages. We have
Spearman’s rank correlation test is shown in Table 16. done a comparative analysis of the results obtained from
As the correlation coefficient is greater than 0.9 and these MCDA methods. As it has been found, the results are
highly significant, it can be concluded that rankings consistent in nature. Since the impact of proper ERP
obtained from two different MCDA frameworks are package selection on both commercial success of ERP and
consistent. Hence, the usage of both these methods for ERP supply chain performance is quite immense, the applica-
package selection is well justified. tion of the presented research framework is justified. It is
very hard to find research studies reporting the usage of
EFA, DEMATEL–TOPSIS and MACBETH simultane-
Conclusion ously for ERP package selection. However, future research
This article has highlighted the importance of selecting an studies can be made on the application of different MCDA
appropriate ERP package in the context of SCM perfor- techniques to rank the performance indicators of a supply
mance. First, we have discussed the impact of ERP on the chain. Further studies can be made to analyse and estimate
performance of a supply chain. Then, a survey of the exist- the impact of ERP on supply chain performance. Analysis
ing literatures has been made to review the factors involved can be made on the applications of ERP in optimising
in the ERP package selection. Then, we have conducted an supply chain performance. Finally, MACBETH is still
not a widely used technique. This technique can further
be applied to solve other multi-criteria based strategic
Table 16. Spearman’s Rank Test decision-making problems pertaining to SCM or any other
areas of management.
Spearman’s Rank Correlation Coefficient Significance
0.964 **
Source: Authors’ own creation.
Note: ** Significant at 0.01 (2 tailed).
268 Asia-Pacific Journal of Management Research and Innovation 12(3&4)

Figure 14. Robustness Analysis (At Cardinal Scale ± 0% for All Criteria)

Source: Constructed using M-MACBETH software.

Figure 15. Robustness Analysis (At Cardinal Scale ± 4% for the Criterion C2)

Source: Constructed using M-MACBETH software.


Note: ▲ Represents ‘dominance’; + represents ‘additive dominance’.
Ghosh and Biswas 269

References Cheung, C.F., Cheung, C.M., & Kwok, S.K. (2012). A knowledge
based customization system for supply chain integration.
Akkermans, H.A., Bogerd, P., Yiicesan, E., & van Wassenhove,
Expert Systems with Applications, 39(4), 3906–3924.
L.N. (2003). The impact of RP on supply chain management:
Choy, K.L., Lee, W.B., & Lo, V. (2003). Design of an intelligent
Exploratory findings from a European delphi study. European
supplier relationship management system: A hybrid case
Journal of Operational Research, 146(2), 284–301.
based neural network approach. Expert Systems with
Aladwani, A.M. (2001). Change management strategies for succ- Applications, 24(2), 225–237.
essful ERP implementation. Business Process Management Christopher, M. (1992). Logistics and supply chain management.
Journal, 7(3), 266–275. London: Pitmans.
Aramyan, L.H., Alfons G.J.M., Lansink, O., van der Vorst, Demsar, U., Harris, P., Brunsdon, C., Foteringham, S.A., &
J.G.A.J., & van Kooten, O. (2007). Performance measurement Mcloone, S. (2013). Principal component analysis on spatial
in agri-food supply chains: A case study. Supply Chain data: An overview. Annals of the Association of American
Management: An International Journal, 12(4), 304–315. Geographers, 103(1), 106–128.
Arthur, W.B. (1996). Increasing returns and the new world of Fekri, R., Aliahmadi, A., & Fathian, M. (2009). Identifying
business. Harvard Business Review, 74(4), 100–109. the cost and effect factors of agile NPD process with fuzzy
Asadi, N. (2012). Performance indicators in internal logistic DEMATEL method: The case of Iranian company. Journal
systems. Proceedings of the International Conference of Intelligent Manufacturing, 20(6), 637–648.
on Innovation and Information Management, IPCSIT, Fine, C.H. (1998). Clockspeed: Winning industry control in the
Singapore, 36, 48–52. age of temporary advantage. Boulder, CO: Perseus Books.
Ayağ, Z., & Özdemir, R.G. (2007). An intelligent approach to Fisher, M.L. (1997). What is the right supply chain for your
ERP software selection through fuzzy ANP. International products? Harvard Business Review, 75(2), 105–116.
Journal of Production Research, 45(10), 2169–2194. Gunasekaran, A., & Kobu, B. (2007). Performance measures
Baki, B., & Çakar, K. (2005). Determining the ERP package- and metrics in logistics and supply chain management: A
selecting criteria: The case of Turkish manufacturing review of recent literature (1995–2004) for research and
companies. Business Process Management Journal, 11(1), applications. International Journal of Production Research,
45(12), 2819–2840.
75–86.
Gunasekaran, A., & Ngai, E.W.T. (2004). Virtual supply chain
Bana e Costa, C.A., Corrêa, E.C., De Corte, J.M., & Vansnick,
management. Production Planning & Control, 15(6),
J.C. (2002). Facilitating bid evaluation in public call for
584–595.
tenders: A socio-technical approach. Omega, 30(3), 227–242.
Han, S.W. (2004). ERP-enterprise resource planning: A cost-
Bana e Costa, C.A., De Corte, J.M., & Vansnick, J.C. (2012). based business case and implementation assessment. Human
Macbeth. International Journal of Information Technology & Factors and Ergonomics in Manufacturing, 14(3), 239–256.
Decision Making, 11(2), 359–387. Hong, K., & Kim, Y. (2002). The critical success factors for ERP
Bana e Costa, C.A., & Oliveira, M.D. (2012). A multicriteria implementation. Information and Management, 40(1), 25–40.
decision analysis model for faculty evaluation. Omega, 40(4), Hsu, C.C. (2012). Evaluation criteria for blog design and analysis
424–436. of causal relationships using factor analysis and DEMATEL.
Bana e Costa, C.A., & Vansnick, J.C. (1999). The MACBETH Expert Systems with Applications, 39(1), 187–193.
approach: Basic ideas, software and an application. In N. Hwang, C.L., & Yoon, K. (1981). Multiple attribute deci-
Meskens & M. Roubens (Eds), Advances in Decision Analysis sion making: Methods and applications. New York, NY:
(pp. 131–157). Dordrecht: Kluwer Academic Publishers. Springer-Verlag.
Berchet, C., & Habchi, G. (2005). The implementation and Jensen, R. (1999). The dream society. New York, NY: McGraw-
deployment of an ERP system: An industrial case study. Hill.
Computers in Industry, 56(6), 588–605. Karande, P., & Chakraborty, S. (2014). A facility layout selection
Biswas, S., & Sen, J. (2016). A proposed framework of next model using MACBETH method. Proceedings of the 2014
generation supply chain management using big data analytics. International Conference on Industrial Engineering and
Proceedings of National Conference on Emerging Trends in Operations Management, Bali, Indonesia, 17–26.
Business and Management: Issues and Challenges, Kolkata, Kashyap, A. (2011). Impact of ERP implementation on supply
INDIA. Retrieved 28 May 2016, from http://ssrn.com/ chain management. International Journal of Computer
abstract=2755828 Applications in Engineering Sciences, 1(4), 474–479.
Keil, M., & Tiwana, A. (2006). Relative importance of evaluation
Brown, S.L., & Eisenhardt, K.M. (1998). Competing on the edge:
criteria for enterprise systems: A conjoint study. Information
Strategy as structured chaos. Boston, MA: Harvard Business
Systems Journal, 16(3), 237–262.
School Press.
Kovacs, G.L., & Paganelli, P. (2003). A planning and management
Bueno, S., & Salmeron, J.L. (2008). Fuzzy modelling enterprise infrastructure for large, complex, distributed projects: Beyond
resource planning tool selection. Computer Standards and ERP and SCM. Computers in Industry, 51(2), 165–183.
Interfaces, 30(3), 137–147. Kumar, V., Maheshwari, B., & Kumar, U. (2002). Enterprise
Chang, I.C., Hwang, H.G., Liaw, H.C., Hung, M.C., Chen, S.L., resource planning systems adoption process: A survey of
& Yen, D.C. (2008). A neural network evaluation model Canadian organizations. International Journal of Production
for ERP performance from SCM perspective to enhance Research, 40(3), 509–523.
enterprise competitive advantage. Expert Systems with Kumar, V., Maheshwari, B., & Kumar, U. (2003). An investigation of
Applications, 35(4), 1809–1816. critical management issues in ERP implementation: Empirical
270 Asia-Pacific Journal of Management Research and Innovation 12(3&4)

evidence from Canadian organizations. Technovation, Stavrulaki, E., & Davis, M. (2010). Aligning products with supply
23(10), 793–807. chain processes and strategy. The International Journal of
Kumar, K., & Van Hillegersberg, J. (2000). ERP experiences and Logistics Management, 21(1), 127–151.
evolution. Communications of the ACM, 43(4), 23–26. Su, Y., & Yang, C. (2010). A structural equation model for
Lambert, D.L. (2006). Supply chain management: Processes, analyzing the impact of ERP on SCM. Expert Systems with
partnerships, performance (2nd ed.). FL: Hartley Press. Applications, 37(1), 456–469.
Lambert, D.M., & Cooper, M.C. (2000). Issues in supply Tam, J.M., Yen, D.C., & Beaumont, M. (2002). Exploring
chain management. Industrial Marketing Management, 29(1), the rationales for ERP and SCM integration. Industrial
65–83. Management & Data Systems, 102(1), 26–34.
Tarn, J.M., Razi, M.A., Yen, D.C., & Xu, Z. (2002). ERP and
Li, G.D., Yamaguchi, D., & Nagai, M. (2007). A grey-based
SCM systems. International Journal of Manufacturing
decision-making approach to the supplier selection problem.
Technology and Management, 4(5), 420–439.
Mathematical and Computer Modelling, 46(3/4), 573–581.
Taylor, D.A. (2004). Supply chains: A manager’s guide. Boston,
Liao, X.W., Li, Y., Lu, B. (2007). A model for selecting an MA: Addison-Wesley.
ERP system based on linguistic information processing. Teltumbde, A. (2000). A framework for evaluating ERP projects.
Information Systems, 32(7), 1005–1017. International Journal of Productions Research, 38(17),
Mabert, V.A., Soni, A., & Venkataramanan, M.A. (2003). 4507–4520.
Enterprise Resource Planning: Managing the Implementation Verville, J., & Halingten, A. (2003). A six-stage model of the
Process. European Journal of Operational Research, 146(2), buying process for ERP software. Industrial Marketing
302–314. Management, 32(7), 585–594.
Nah, F.F.H., & Delgado, S. (2006). Critical success factors for Wei, C.C., Chien, C.F., & Wang, M.J.J. (2005). An AHP-based
enterprise resource planning implementation and upgrade approach to ERP system selection. International Journal of
[Special issue]. Journal of Computer Information Systems, Production Economics, 96(1), 47–62.
46(5), 99–113. Wier, B., Hunton, J., & HassabElnaby, H.R. (2007). Enterprise
Palaniswamy, R., & Frank, T. (2000). Enhancing manufacturing resource planning systems and non-financial performance
performance with ERP systems. Information Systems incentives: The joint impact on corporate performance.
Management, 17(3), 43–55. International Journal of Accounting Information Systems, 8(3),
Rao, S.S. (2000). Enterprise resource planning: Business needs 165–190.
Wisner, J.D., Tan, K.C., & Leong, G.K. (2008). Principles of
and technologies. Industrial Management and Data Systems,
supply chain management: A balanced approach. Mason,
100(1/2), 81–88.
OH: South-Western Cengage Learning.
Ratkevičius, D., Ratkevičius, C., & Skyrius, R. (2012). ERP Wong, W.P., & Wong, K.Y. (2008). A review on benchmarking
selection criteria: Theoretical and practical views. Ekonomika, of supply chain performance measures. Benchmarking: An
91(2), 97–116. International Journal, 15(1), 25–51.
Roy, B. (1996). Multicriteria methodology for decision aiding. Wu, J. (2008). Critical success factors for ERP system
Dordrecht: Kluwer. implementation. In L.D. Xu, A.M. Tjoa, S.S. Chaudhry (Eds),.
Shinno, H., & Hashizume, H. (2002). Structured method for Research and practical issues of enterprise information
identifying success factors in new product development of systems II volume 1 (vol. 254, pp. 739–745), Boston,
machine tools. Annals of the CIRP, 51(1), 281–284. MA: Springer.
Shyura, H.J., & Shih, H.S. (2006). A hybrid MCDM model for Zhou, Q., Huang, W., & Zhang, Y. (2011). Identifying critical
strategic vendor selection. Mathematical and Computer success factors in emergency management using a fuzzy
Modelling, 44(7/8), 749–761. DEMATEL method. Safety Science, 49(2), 243–252.

You might also like