You are on page 1of 17

International Journal of Production Research

ISSN: 0020-7543 (Print) 1366-588X (Online) Journal homepage: http://www.tandfonline.com/loi/tprs20

Development of a hybrid fresh food supply chain


risk assessment model

Dilupa Nakandala, Henry Lau & Li Zhao

To cite this article: Dilupa Nakandala, Henry Lau & Li Zhao (2016): Development of a hybrid
fresh food supply chain risk assessment model, International Journal of Production Research,
DOI: 10.1080/00207543.2016.1267413

To link to this article: http://dx.doi.org/10.1080/00207543.2016.1267413

Published online: 23 Dec 2016.

Submit your article to this journal

Article views: 6

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at


http://www.tandfonline.com/action/journalInformation?journalCode=tprs20

Download by: [Gazi Universitesi] Date: 27 December 2016, At: 19:57


International Journal of Production Research, 2016
http://dx.doi.org/10.1080/00207543.2016.1267413

Development of a hybrid fresh food supply chain risk assessment model


Dilupa Nakandala, Henry Lau* and Li Zhao

School of Business, University of Western Sydney, Penrith, Australia


(Received 10 December 2015; accepted 26 November 2016)

Supply chain managers and scholars recognise the importance of managing supply chain risk, especially in fresh food
supply chain due to the perishable nature and short life cycle of products. Supply chain risk management consists of sup-
ply chain risk assessment, risk evaluation and formulation and implementation of effective risk response strategies. The
commonly adopted qualitative methods such as risk assessment matrix to determine the level of risk have limitations.
This paper proposes a hybrid model comprising both fuzzy logic (FL) and hierarchical holographic modelling (HHM)
techniques where risk is first identified by the HHM method and then assessed using both qualitative risk assessment
model (named risk filtering, ranking and management Framework) and fuzzy-based risk assessment method (named FL
approach). The risk assessment results by the two different approaches are compared, and the overall risk level of each
risk is calculated using the Root Mean Square calculation before identifying response strategies. This novel approach
takes advantage of the benefits of both techniques and offsets their drawbacks in certain aspects. A case study in a fresh
food supply chain company has been conducted in order to validate the proposed integrated approach on the feasibility
of its functionality in a real environment.
Keywords: supply chain risk management; fuzzy logic; risk analysis; fresh food supply chain; risk assessment

1. Introduction
There is growing recognition that business is becoming riskier nowadays, and the number and the potential conse-
quences of risks are growing (Waters 2007). Risk management has become a vital topic and has gained considerable
attention in recent years in both academia and practice (e.g. Zsidisin 2003; Zsidisin and Ellram 2003; Peck 2005; Ellis
et al. 2010; Tummala and Schoenherr 2011; Wu, Chen, and Olson 2014; Wu, Luo, et al. 2016). Some researchers
focused on the classification of supply chain risk (e.g. Zsidisin 2003; Spekman and Davis 2004; Ritchie and Brindley
2007; Tang and Tomlin 2008), while a lot of others focused on supply chain risk management strategies and procedures
(e.g. Sinha, Whitman, and Malzahn 2004; Jüttner 2005; Wu, Blackhurst, and Chidambaram 2006; Ritchie and Brindley
2007). However, as it is an emerging area, lots of work still needs to be done (Rangel, Oliveira, and Leite 2015).
Supply chain leaders and scholars are all emphasising the importance of managing supply chain risk. But what is
supply chain risk? How can it be assessed? In order to meet the challenge of managing and mitigating supply chain risk,
we must first identify different dimensions of supply chain risk and assess the level of each risk. Several scholars who
investigated the procedure of supply chain risk management emphasised that the first step is to identify the sources of
risk. For example, Sinha, Whitman, and Malzahn (2004) pointed out that the first step in mitigating supplier risk is to
identify areas of risk and assess the risk level (RL). Wu, Chen, and Olson (2014) defined risk management as the
process of identification, analysis and decision-making of either acceptance or mitigation of uncertainty which also
supports that risk identification and analysis are the first two important steps of risk management.
This study focuses on the supply chain risk assessment in the fresh food industry. As the income and standards of
living rise and consumer preferences shift to freshness, quality and safety, the international trade in fresh produce has
been growing rapidly (Diop and Jaffee 2005). However, food supply chain risk increases with an increase in the global
supply of fresh food which puts great pressure on fresh food companies. For example, food safety incidents have been
frequently reported in the media. How to effectively manage fresh food supply chain risk, such as supply risk (low
quality, delivery failure, hot weather etc.) and demand risk (season food imbalance, new competitors) are really impor-
tant and urgent for fresh food companies.
Effective strategies to mitigate fresh food supply chain risk are important, but before implementing effective
mitigation strategies, fresh food companies need to identify the risk sources and evaluate the RL so that they can make

*Corresponding author. Email: H.Lau@uws.edu.au

© 2016 Informa UK Limited, trading as Taylor & Francis Group


2 D. Nakandala et al.

appropriate actions for risks with different levels. However, assessing the supply chain risk is really difficult, and
existing research mainly uses qualitative method, such as risk assessment matrix. The aim of this paper is to analyse the
risk factor of the fresh food supply chain so that practitioners can further develop and select proactive strategies to min-
imise the negative impact of the risk. A hybrid model including risk identification approach and risk assessment
approaches is proposed. One qualitative method named risk filtering, ranking and management Framework (RFRM)
framework and one quantitative method named fuzzy logic (FL) approach are selected to assess the RL and an overall
RL is calculated based on these two approaches. FL approach is selected as it has proved to be beneficial in terms of
vague, imprecise and uncertain contexts such as supply chain risk (Nakandala and Lau 2012).
The rest of the paper is organised as follows. Section 2 is a thorough literature review of the supply chain risk and
fresh food supply chain risk. In Section 3, we proposed a hybrid risk assessment model which incorporates the qualita-
tive and quantitative risk assessment approaches. In order to validate the proposed model, a case study is described in
Section 4 to verify the feasibility of its functionality in a real environment. Finally, we concluded in Section 5.

2. Literature review
2.1 Supply chain risk
Risk exists in every aspect of our lives, and risk itself is not always bad as business exists to deal with risks (Wu,
Olson, and Luo 2014). Academics have studied risk definitions in a number of contexts. Lowrance (1976) defined it as
a measure of the probability and severity of adverse threats. In addition, Kaplan and Garrick (1981) further suggest the
definition of risk as a ‘set of triplets’, which involves the risk scenario, the likelihood of that scenario and the resulting
consequence. This ‘set of triplets’ provides a method to quantify the likelihood and resulting consequence of risk scenar-
ios. According to such a definition, any possible event can be compared and evaluated with a more objective standard.
The standard definition of risk by ISO31000 is ‘the effect of uncertainty on objectives’, and an effect is a positive or
negative deviation from what is expected. Risk management refers to a coordinated set of activities and methods that
are used to direct an organisation and to control the many risks that can affect its ability to achieve objectives
(ISO31000), and risk managers should ‘apply risk treatment options to ensure that the uncertainty of their agency meet-
ing its objectives will be avoided, reduced, removed or modified and/or retained’ (ISO31000:2009).
Supply chain risk is a relatively new concept compared to risk. Several researchers in the field of supply chain man-
agement have defined supply chain risks. Harland, Brenchley, and Walker (2003, 52), for instance, discussed several
definitions and concluded that supply chain risk is associated with the ‘chance of danger, damage, loss, injury or any
other undesired consequences’. (Wagner and Bode 2008) defined a supply chain disruption as the combination of (1) an
unintended, anomalous triggering event that materialises somewhere in the supply chain or its environment, and (2) a
consequential situation which significantly threatens normal business operations of the firms in the supply chain. These
definitions, however, focus on both operational and catastrophic supply chain risks. In a supply chain network, it is
important to manage the relationship and mutual trust with suppliers and customers in order to enhance value in the
marketplace as a whole. Therefore, any approach to managing risks from the angle of a supply chain network requires
the provision of insights related to how operations and processes are performed across various partners within the entire
network. Based on the supply chain perspective, risk management is concerned with the identification and assessment of
likely risks and their possible impact on operations which is a complex and difficult task for a single company (Jüttner
2005).
As for food and fresh food supply chain, the main focus of risk management is concerned with food safety, particu-
larly because of the increase in the global supply of products from various sources around the world. In recent years,
food safety incidents have been frequently reported in the media, creating great concerns for the company partners
within the food and fresh food supply chain. Apart from the risk of food safety issues, other food and fresh food supply
chain risk factors such as supply risk and demand risk are also included, all of which are to be thoroughly discussed in
this paper.

2.2 Supply chain risk categories


In attempting to differentiate supply chain risks from other business risks, many scholars have proposed different
typologies and/or taxonomies of supply chain risk (Svensson 2002; Christopher and Peck 2004; Jüttner, Peck, and
Martin 2003; Spekman and Davis 2004; Jüttner 2005; Tang 2006; Wagner and Bode 2008).
Most often, researchers categorised SCRs into two types. For example, Jüttner (2005) delineated two categories of
SCRs: internal risk and environmental risks. Internal risk includes supply risk, demand risk, process and control risks.
International Journal of Production Research 3

He distinguished between environmental, supply and demand risk sources on the one hand, and processes and control
mechanisms as a risk amplifier or absorber on the other. Similarly, Kleindorfer and Saad (2005) categorised these risks
into two types: (1) risks related to supply and demand coordination and uncertainty, and (2) disruption risks that are
caused by such events as natural disasters, terrorism and labour strikes. Also, Trkman and McCormack (2009) divided
uncertainty into endogenous uncertainty and exogenous uncertainty. They argued that endogenous uncertainty can be
reduced with a proper and proactive relationship with a supplier (using methods like information sharing, relationship
development, joint reviews, etc.). Exogenous uncertainty cannot be generally reduced. Tang (2006) classified SCR into
two dimensions: disruption risk and operational risk. Disruption risks are those caused by events such as bankruptcy,
natural disasters and terrorist attack. Operational risks are related to supply and demand coordination and uncertainty,
such as uncertain demand and uncertain supply. Disruption risks are rare but severe and hard to manage, while opera-
tional risk can be reduced through effective supply chain management activities.
Adapted from the existing SCR literature concerning SCR dimensions, this research classifies SCRs into three cate-
gories (Figure 1): internal risk, operational risk external to the firm and macro-level risk. Internal risk includes the risk
inside the firm such as process risk and control risk (Christopher and Peck 2004). Operational risk external to the firm
is the possibility of unplanned events that disrupt the normal flow of goods, service and information within a supply
chain, which in turn result in the inability to meet customer demand and cause negative consequences on supply chain
members (Zsidisin 2003). More specifically, operational risk external to the firm includes supply risk and demand risk,
while macro-level risk includes those risks caused by external catastrophic events, like natural disasters and wars.

2.3 Fresh food supply chain risk


Fresh food supply chain spans widely from the farm, processors, wholesalers, retailers to the fork as illustrated in
Figure 2. Fresh food supply chain is of particular fragile nowadays due to the geographic, economic and legislative
spread of participating entities (Dani and Deep 2010). As Dani and Deep (2010) pointed out, risk in fresh food supply
chain is catastrophic both to the customers and organisations. Hendricks and Singhal (2005) showed that the stock value
immediately dropped 6.6% following an announcement of supply chain disruption in food industries.
Risks within the food supply chain have been classified into different ways. For example, Christopher and Peck
(2004) categorised risks in food supply chain to process risk and control risk (which are internal to the firm), demand
and supply risks (which are external to the firm) and environment risk. Kleindorfer and Saad (2005) classified risks in
supply chains into risks arising from coordinating supply and demand and risks arising from disruptions. Agiwal and
Mohtadi (2008) classified the risks on a very broad level considering risks arising from either intentional or uninten-
tional causes. Dani and Deep (2010) categorised food supply chain risks to Type 1 risks (which are concerned with food
safety as well as maintaining a secure supply of food), and Type 2 risks which affect the supply chain but don’t have a
direct impact on food safety (include transportation strikes, loss of power, flooding, etc.).

Supply risk Demand risk


Internal risk
Initial Ultimate
... Supplier Focal firm Customer ...
Supplier customer

Operational risk external to the firm

Macro level risk

Figure 1. Category of supply chain risk.


Source: Adapted from Jüttner, Peck, and Martin (2003); Manuj and Mentzer (2008).
4 D. Nakandala et al.

Product information flow


Delivery information flow
Cost information flow

Temporal pattern
IMPORT HOUSES
of planned fresh
represents produce supply
Expected prices
Demand forecast
represents Specification RETAILERS
Price offers Independents
Grocery
Specialty buyers
Seed varieties
Fertilizer options
ARGO-SUPPLIERS Crop protein
options PRODUCERS PROCESSORS WHOLESALERS
Seeds, Fertiliser, Pesticide,
Weedicides, etc. Prices of inputs
Technical support
RESTAURANTS
Independent
Time-based requirements of inputs
Quick service
Institutional

EXPORT HOUSES

INTERNATIONAL
BUYERS

KNOWLEDGE PARTNERS REGULATORY BODIES


LOGISTICS PARTNERS FINANCIAL PARTNERS
Agricultural extension services Process compliance
Warehousing Banks
R&D Institutes Content specification
3PL Specialized funding agencies
Engineering/ equipment firms Traceability
Freighting and shipping Insurance firms
Trade consultants Labelling

Price information flow


Demand information flow
Specification information flow

Figure 2. Generic Fresh food and vegetable supply chain.

Research in food supply chain risk is relatively limited comparing to supply chain risk literature. Hornibrook,
McCarthy, and Fearne (2005)’s survey research explored the pre-pack consumers’ perceptions of risk associated with
beef in the Republic of Ireland. The results confirmed that food safety and health issues are still the main concerns for
beef purchasers, and perceptions of risk are reduced through loyalty to supermarkets. Zingg, Cousin, et al. (2013) exam-
ined food-safety perceptions separately for every step of the total meat supply chain using a survey method. The results
revealed a clear distinction between risk perception at the production stage and risk perception at home in the total meat
supply chain, in that people perceived significantly less risk at home. However, people’s risk perceptions of the single
stages in the total meat supply chain were overall slightly above average. Lagerkvist et al. (2013) investigated the food
risks in urban and peri-urban supply chains. They quantified the subjective risk judgements with regard to food safety
hazards and examine the extent of discrepancies in perceived risk relating to vegetables in domestic urban markets
among the three groups. Differences were found between respondent categories in terms of both specific source risks
and overall risks. Differences were also found with respect to the socio-demographic and structural determinants of the
levels of perceived risks.
Research in fresh food supply chain risk assessment is rare. However, there are some attempts. For example, Lang
and Ding (2008) classified the logistics outsourcing risks in food supply chain into technology risks, cooperation risks,
management risks, environmental risks and financial risks. They assessed the risks using the risk matrix method and
offered practical methods to control these risks. Diabat, Govindan, and Panicker (2012) created a model to analyse the
various risks involved in a food supply chain with the help of interpretive structural modelling. The risks in food supply
chain are clustered into five categories and risk mitigation is discussed with the help of a case study in a food products
manufacturing firm.
International Journal of Production Research 5

3. Hybrid fresh food supply chain risk assessment model


This research proposed a hybrid risk assessment model as shown in Figure 3. First, risk is identified by the hierarchical
holographic modelling (HHM) method (Haimes 2004). Then, qualitative risk assessment model (named RFRM) and
fuzzy-based risk assessment method (the FL approach) are used to assess the RLs. The risk assessment results by the
two different approaches are compared to see if there is some discrepancy. In case there is some discrepancy, a value in
between the two numbers computed from RFFM and FL may serve the purpose. Root Mean Square (RMS) value is
adopted. RMS values are arithmetically greater than the average values since the larger numbers are multiplied by a lar-
ger factor so they carry more weight. This is preferred by companies who are normally inclined to be more conservative
and protective regarding the chance of any occurrence of risk and a higher RL number will give them an early alert for
precaution.

3.1 Risk identification


As Wu, Olson, and Dolgui (2015) pointed out, risks arise from many sources, some natural, some malicious caused by
humans and some due to interaction between human activity and nature. The risk identification stage of the supply chain
risk management process is critical to the success of managing supply chain risks as it ensures that all significant
activities within the organisation have been identified and all the risks following from these activities defined (Wu,
Blackhurst, and Chidambaram 2006; Neiger, Rotaru, and Churilov 2009). According to Sinha, Whitman, and Malzahn
(2004), risk analysis is a practice with methods and tools for identifying risks in a process. By identifying a risk, deci-
sion-makers become aware of events that may cause disturbances. To assess SCR exposures, the company must identify
not only direct risks to its operations, but also the potential causes or sources of those risks at every significant links
along the supply chain (Norrman and Jansson 2004).
This research adopts the HHM developed by Haimes (1991) to identify the fresh food supply chain risks. HHM can
be regarded as a holistic methodology which aims at capturing and representing the essence of the inherent diverse
characteristics and attributes of a system including its multiple aspects, perspectives, facets, views, dimensions and hier-
archies (Lambert et al. 2001). A hierarchical holographic model can be developed and coordinated to capture the
essence of multiple dimensions and perspectives. Managers were asked to identify potential risk scenarios in the fresh
food industry. Based on the information provided by managers from various companies, major aspects of the risks are

Fresh food supply


chain risk
identification: HHM

Qualitative risk Fuzzy-based risk


assessment tool: assessment tool: Fuzzy
RFRM logic approach

Risk level of each Risk level of each


risk: R1 risk:R2

Difference?

R1 2 R 22
Overall risk level=
2

Figure 3. A hybrid risk assessment model.


6 D. Nakandala et al.

identified. These distinctive attributes provide a holographic view of a business system. Thus, HHM is capable of
identifying a large set of risk scenarios, and it is useful for a scalable organisation to analyse its complex and hierarchi-
cal business environment.

3.2 Risk assessment


Based on the risk identification results, risk assessment continues to identify potential losses, establishing the extent of
losses, understanding the likelihood of potential losses, assigning significance to potential losses, and appraising overall
risk (Yates and Stone 1992). There are two main risk assessment methods – qualitative and quantitative. This research
adopts RFRM framework which is a qualitative risk assessment method and FL approach which is a quantitative
method. Fuzzy set theory has proved to be beneficial in terms of vague, imprecise and uncertain contexts and it
resembles human reasoning as it adopts approximate information and uncertainty to come up with general decisions.
Traditional approaches adopt crisp and precise values during the process of computation and lack the capability to deal
with vagueness and imprecision which are common in natural human thinking. Fuzzy systems rely on experts’ knowl-
edge and enable the expression of fuzzy ‘inputs’ in linguistic terms. They can be used effectively when some data are
difficult to measure accurately or difficult to quantify numerically, for example, the RL (Nakandala and Lau 2012).

3.2.1 Qualitative risk assessment: RFRM method


Once a large set of risk scenarios is defined by HHM, different risk scenarios can be filtered and ranked according to
their importance to prioritise risk mitigation activities. Haimes, Kaplan, and Lambert (2002) created a Risk Filtering,
Ranking, and Management (RFRM) framework that aims at providing priorities in the scenario analysis. It is a tool that
provides a systematic risk scoring for either screening hazards or prioritising risks for risk management (Kradjian 2006).
The two dimensions: probability of occurrence and severity of consequences typically use qualitative categorical
descriptors, as shown in Table 1. Based on the probability and severity of each risk, the risk assessment matrix is
formed for each firm. The risk assessment matrix helps to prioritise risks into different levels (i.e. no, moderate, signifi-
cant, severe) and planning for actions that could mitigate the corresponding risk, which has been widely adopted as a
simple and effective approach to risk management (Cox 2008).

3.2.2 Fuzzy-based risk assessment: FL approach


The FL approach is adopted in this paper to assess the overall RL based on fuzzy sets of input risk factors. The concept
of FL was first proposed by Zadeh (1965), who introduced fuzzy sets to represent knowledge that is vague or imprecise,
as ‘fuzzy’. Zadeh’s classical FL concept is applied to a rule-based system and then forms a FL rule-based system. Each
fuzzy rule contains an antecedent part which incorporates several preconditions, and a consequent part prescribing the
corresponding value (Kandel and Langholz 1992). The main factors and the associated terms which represent the lin-
guistic variables can be identified by means of fuzzy sets. For example, a set including linguistic values (Low, Medium,
High) may be used. FL systems are knowledge-based or rule-based constructed from human knowledge in the form of
fuzzy IF-THEN rules (Wang 1992).
The FL RL assessment system consists of four parts: data collection and knowledge acquisition, fuzzy inference
engine, fuzzification and defuzzification (Figure 4). A fuzzy rule base consists of a set of fuzzy IF-THEN rules. The rule
base is the foundation of a FL system. All other components in the system follow these rules in order to predict outputs
in a reasonable manner.

Table 1. Risk assessment matrix.

Likelihood of occurrence Severity of consequence

0 Impossible Negligible
1 Rare Very low
2 Unlikely Low
3 Moderate Medium
4 Likely High
5 Almost Certain Very high
International Journal of Production Research 7

The fuzzy rule base in the study estimates overall RL in three different aspects using three fuzzy (linguistic)
variables, i.e. macro-level risk, supply chain operation risk and internal risk. For each aspect of the risks, we use two
fuzzy variables, likelihood (L) and consequence of severity (X). L and X values of each input risk factors are put before
the ‘THEN’ statement (antecedent) in the fuzzy rule, while output RL in different aspects is found in the statements after
‘THEN’ (consequent). The magnitudes of the linguistic variables of the input and output are described in terms of words
(linguistic values) such as high, low or very low. A complete fuzzy rule is written as follows:
IF L is very low AND X is negligible, THEN RL is very low
Since all rules in the fuzzy rule base are formed by linguistic values, there is a process to convert from quantitative
(crisp) values of input data-sets to linguistic values. This process is called fuzzification. Fuzzification is the first process
of core fuzzy system that converts the crisp data collected from the data collection procedure into fuzzy sets. A fuzzy
set is characterised by a membership function (Nakandala and Lau 2012). In this study, both linguistic values of input
and output variables are derived from the quantitative (crisp) scale of 1–5, and the membership function is a triangular
membership function (shown in Figure 5).

3.2.2.1 Input 1: Macro-level risk (MR). MR is a fuzzy set with mri denoting the elements in the data-set, and μMR(mri)
denotes the membership function.

X
n
l MR ðmri Þ
MR ¼ (1)
i¼1
mri

MR = {VL, L, M, H, VH}, where very low (VL), low (L), medium (M), high (H) and very high (VH).

3.2.2.2 Input 2: Operational risk external to the firm (OR). OR is a fuzzy set with ori denoting the elements in the
data-set, and μOR(ori) denotes the membership function.

X
n
l OR ðori Þ
OR ¼ (2)
i¼1
ori

OR = {VL, L, M, H, VH}. OR consists of five linguistic values, i.e. very low (VL), low (L), medium (M), high (H)
and very high (VH).

3.2.2.3 Input 3: Internal Risk (IR). IR is a fuzzy set with iri denoting the elements in the data-set, and μIR(iri) denotes
the membership function.

Fuzzy system

Data collection Fuzzification


Risk input data set
Past experiences Triange membership
Previous records function

Knowledge Fuzzy rule base


acquisition Fuzzy inference engine
Expert interview

Defuzzification Risk assessment result

Figure 4. The conceptual model of the FL RL assessment system.


8 D. Nakandala et al.

R (ri )
R: risk

VL L M H VH
1

R
0 1 2 3 4 5

Figure 5. The input fuzzy set μR(ri).

Xn
lIR ðiri Þ
IR ¼ (3)
i¼1
iri

IR = {VL, L, M, H, VH}. IR consists of five linguistic values, i.e. very low (VL), low (L), medium (M), high (H)
and very high (VH).

3.2.2.4 Output: Overall RL. The output of RL is a fuzzy set with the fuzzy elements of rli and μRL(rli) is the member-
ship function. For simplicity, we assume here μRL(rli) is a triangular membership function and it is graphically repre-
sented as in Figure 5. The fuzzy set is:

Xn
lRL ðrli Þ
RL ¼ (4)
i¼1
rli

RL consists of five linguistic values, i.e. very low (VL), low (L), medium (M), high (H) and very high (VH).
RL = {VL, L, M, H, VH}.
In order to convert the input fuzzy sets into the output fuzzy sets, fuzzy inference engine is the next process of the
core fuzzy system. Fuzzy inference process includes the rule block information, rule composition, rule firing, implication
and aggregation (Lau and Dwight 2011; Nakandala and Lau 2012). As shown in Figure 4, fuzzy inference process
adopts the knowledge acquired in the earlier procedure by managers’ internal meeting and historical data mining, and
changes them into fuzzy rule blocks which are the sets of IF-THEN statements. The quality of the fuzzy rules has direct
impact on the output and managerial decisions.
Table 2 shows the rules of this study three dimensionally for the three inputs macro-level risk, risk external to the
firm and internal risk. By collecting information from many supermarket managers and a thorough data mining of
the historical database, the differences of the views were overcome and the consistent whole rule sets were developed.
The output RL is decided based on the rules.

3.2.3 Overall risk assessment


Based on the two risk assessment results by the two different approaches, further comparison is done to see if there is
difference between the two results. The RFFM and FL approaches should be able to suggest the RLs in a fairly reliable
way. In case there is some discrepancy, a value in between the two numbers computed from RFFM and FL may serve
the purpose. RMS value is adopted as RMS values are arithmetically greater than the average values since the larger
numbers are multiplied by a larger factor so they carry more weight. This is preferred by companies who are normally
inclined to be more conservative and protective regarding the chance of any occurrence of risk and a higher RL number
will give them an early alert for precaution.
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
R21 þ R22
R¼ (5)
2
International Journal of Production Research 9

Table 2. The rule tables for the fuzzy inference engine.

Input OR
Input IR VL L M H VH

Input ER = VL
VL VL VL L M H
L VL L L M H
M L L M M H
H M M M H H
VH H H H H VH
Input ER = L
VL VL L L M M
L L L L M M
M L L M M H
H M M M H H
VH M M H H VH
Input ER = M
VL L L M M M
L L M M M H
M M M M H H
H M H H H H
VH H H H H VH
Input ER = H
VL L M M M H
L M M M H H
M M M H H H
H M H H H VH
VH H H H VH VH
Input ER = VH
VL M M M H VH
L M M H H VH
M M H H H VH
H H H VH VH VH
VH H VH VH VH VH

As shown in Equation (5), we calculate the overall RL of this company as the square root of the arithmetic mean of the
squares of the two risk values. In brief, our hybrid risk assessment model can help managers avoid the risk of the dis-
crepant judgement using only one method, and our model provides a very clear RL for decision-makers to take appro-
priate actions for risks with different levels.

4. A case study
4.1 Background
The case study focuses on companies selling the fresh food, such as fruits and vegetables in Australia. Supermarkets,
like Coles and Woolworths, and some fresh food retailers like Adams’ Apples are the dominant fresh food companies in
the market. In response to the new entry of a competitive international supermarket chain Aldi, a Germany-based inter-
national discount chain with an emphasis on private label products, Coles, Woolworths and other independent supermar-
ket chains have expanded their private label (home brand) range. To manage their brand image and risk exposure, these
supermarket chains have introduced a series of private standards on food safety and quality. We created a company
named company A to emulate the situations of a firm which may not be aware of the potential risk threats existed such
as unexpected disruption of supply.
Even though ‘buy local’ is encouraged in recent years, the food chain of company A is becoming progressively
more globalised as foods are sourced from global markets, which created a range of opportunities and risks. Food costs
change as the supply of raw materials from all over the world changes, some due to disasters such as hurricane, and
some due to diseases such as bird flu. Poor quality of the supplier products such as is sometimes happening, for exam-
ple, fruits and vegetables of low level of freshness has not been uncommon are not freshly provided. The delivery
would probably delay due to bad weather or labour strikes which may cause serious problems because of the perishable
10 D. Nakandala et al.

nature of fresh food. Also, customers’ preferences change from time to time and demand for products varies, especially
for bakery products. In order to provide safe and fresh food to customers and avoid potential risks, it is important to
reveal its potential risk scenarios. The proposed hybrid risk assessment model is applied to identify all the major risk
scenarios, assess these risks using RFRM approach and FL approach and determine strategies and specific action plans.

4.2 HHM for risk identification


Mangers were asked to identify potential risk scenarios in fresh food industry. Based on the information provided by
managers from various companies, three major aspects of the risks are identified:

4.2.1 Macro level risks


(a) Natural disaster (1a) – natural disaster risks include loss of power and flooding. As Dani and Deep (2010) pointed
out, they can be equally detrimental to the organisations viability, but are usually resolved without government
involvement. For example, Walmart shut down 126 stores during hurricane Katrina.
(b) Diseases (1b) – diseases like bird flu.
(c) Government regulation (1c) – for example, change of the food safety standards.
(d) Labour strikes (1d) – for example, a truck driver strike.

4.2.2 Operational risks external to the firm


(a) Volatile demand (2a) – demand uncertainty and volatility can cause demand information mislead which in turn
could cause ‘bullwhip effect’ (Svensson 2002), which is characterised by an amplification of demand volatility
in the upstream direction of the supply chain.
(b) Customer tastes changes (2b) – customer taste changes especially for the bakery products.
(c) Supply quality risk (2c) – poor quality of the supplied products, especially unsafe products could cause serious
problems.
(d) Supplier delivery delay (2d) – unable to deliver the foods in time causes lots of problems, especially because of
the perishable characteristics of fresh food.

4.2.3 Internal risks


(a) Internal operations risk (3a) – Disruptions to internal assets, reliability of supporting communication and infras-
tructures can cause food quality problems.
(b) Excessive inventory (3b) – excessive inventory due to inadequate planning could lead to unsold and expired
products.

4.3 Risk assessment: RFRM


Managers in the company were asked to determine the level that is associated with each potential risk scenario. In order
to quantify the supply chain risks that the fresh food company may suffer, the company needs to collect the data about
the occurrence of problems such as food safety regulation changes, late deliveries of the products and customers’ com-
plaints. Table 3 shows the results of the risk assessment using the RFRM method. Figure 6 illustrates the assessment of
various risk scenarios.
Table 3 and Figure 6 show the likelihood and severity of each risk, and we can get a very clear map of which risk
is high risk and provide evidence for managers to facilitate different supply chain risk management strategies. However,
it is more convenient to have a single overall RL so that managers can have a very clear idea of how risky their com-
pany is. We further calculate the overall RL based on the risk results for each risk. The range of the RFRM rating is the
likelihoods (from 0 to 5) multiple by the consequences of the risks (from 0 to 5), but the RL of the FL approach is
from 0 to 5. Here in this paper, we transfer the RFRM rating to a reading from 0 (negligible) to 5 (very high) by extract
the square root so that we can compare this result with the FL results. Then we put the same weight on each risk and
get an average RL as the overall RL, as shown in Equation (6). Thus, the overall RL of this company is 2.84 out of 5.
International Journal of Production Research 11

Table 3. Results of risk assessment: RFRM method.

Scenario Likelihoods Consequences Rating

1a 1 5 5 (Moderate)
1b 1 4 4 (Moderate)
1c 2 4 8 (Moderate)
1d 2 3 6 (Moderate)
2a 4 3 12 (Significant)
2b 3 3 9(Significant)
2c 3 4 12(Significant)
2d 3 4 12(Significant)
3a 2 3 6(Moderate)
3b 3 3 9(Significant)

moderate risk Significant risk high risk


High(5)
1a

4
1b 1c 2c 2d
Consequences

2a
3
1d 3a 3b 2b

Low(0) 1 2 3 4 High(5)
Likelihoods

Figure 6. Risk assessment for fresh food retailer A.

pffiffiffiffiffiffiffi pffiffiffiffiffiffiffi pffiffiffiffiffiffiffi pffiffiffiffiffiffiffi


R1a þ R1b þ R1c þ . . . þ R3b
RLRFRM ¼ (6)
10

4.4 Risk assessment: FL approach


The FL approach based on three inputs is used in this paper in order to assess the RL of the fresh food company. Each
of the input risks, namely the macro-level risk (MR), operational risk external to the firm (OR) and the internal risk
(IR) is estimated based on the perception of managers in the Company and also include industry experts outside the
company. Based on the perception of managers in the company and also industry experts outside the company, the
macro-level risk (MR) is considered as 1.85 out of 5. The operational risk external to the firm (OR) is 3.6, and the IR
is 2.3.
The crisp input values are fuzzified based on the membership functions for each input mentioned in Section 3.3.
The variable ER cuts VL predicate at 0.15 and L predicate at 0.85 (Figure 7). Therefore, the membership function val-
ues for the Low and Medium macro-level risk are 0.2 and 0.8, respectively. Similarly, the variable OR cuts M predicate
at 0.4 and H predicate at 0.6, and the variable IR cuts L predicate at 0.7 and M predicate at 0.3.
Based on the membership values of the three input variables, eight rules are generated from the fuzzy rule blocks
(as shown in Table 2), which are shown in Table 4. The composition results are shown in Table 5, where the minimum
membership function values are chosen.
The composition results are then implicated using the Mamdani Operator as the implication operator to determine
the output fuzzy set, and rules creating the same output results are clustered together. These for implication results are
then aggregated to generate the final fuzzy set, using the Union operator as the aggregation operator. The aggregated
result is shown in Figure 8.
12 D. Nakandala et al.

MR (mri ) MR: Macro level risk

VL L M H VH
1

0.85

015

MR
0 1 1.85 2 3 4 5

Figure 7. The membership value of μMR(1.85).

Table 4. Rule generation based on the membership function of L and X.

Rule NO ‘IF’ ‘THEN’ Overall Risk level IS …

1 The macro-level risk IS very low AND Low


The operational risk external to the firm IS medium AND
The internal risk IS low
2 The macro-level risk IS very low AND Medium
The operational risk external to the firm IS medium AND
The internal risk IS medium
3 The macro-level risk IS very low AND Medium
The operational risk external to the firm IS high AND
The internal risk IS low
4 The macro-level risk IS very low AND Medium
The operational risk external to the firm IS high AND
The internal risk IS medium
5 The macro-level risk IS low AND Low
The operational risk external to the firm IS medium AND
The internal risk IS low
6 The macro-level risk IS low AND Medium
The operational risk external to the firm IS medium AND
The internal risk IS medium
7 The macro-level risk IS low AND Medium
The operational risk external to the firm IS high AND
The internal risk IS low
8 The macro-level risk IS low AND Medium
The operational risk external to the firm IS high AND
The internal risk IS medium

In the defuzzification process, we use the most commonly used method, Centre of Area (COA) method, to convert
the fuzzy values into crisp values. The general equation of COA method is shown in Equation (7), where w denotes the
weight, n denotes the Centre of gravity and A denotes the area of each individual implication.

PN
j¼1 wj Cj Aj
RL ¼ PN (7)
j¼1 wj Aj
Using the COA Equation (7) and the parameters of each polygon summarised in Table 6, the crisp value of the
supply delivery delay risk can be calculated as below.

X
Rule8
wCA ¼ 9:9125 (8)
Rule1
International Journal of Production Research 13

Table 5. The composition results for the IF part of rules.

Rule Composition result

1 0.15^0.4^0.7 = 0.15
2 0.15^04^0.3 = 0.15
3 0.15^0.6^0.7 = 0.15
4 0.15^0.6^0.3 = 0.15
5 0.85^0.4^0.7 = 0.4
6 0.85^0.4^0.3 = 0.3
7 0.85^0.6^0.7 = 0.6
8 0.85^0.6^0.3 = 0.3

Figure 8. The aggregation results.

Table 6. The defuzzification process.

Polygon Area (A) Center of gravity (C) Weight (w) A×C×w A×w

Rule 1 0.2775 2 1 0.555 0.2775


Rule 2 0.2775 3 1 0.8325 0.2775
Rule 3 0.2775 3 1 0.8325 0.2775
Rule 4 0.2775 3 1 0.8325 0.2775
Rule 5 0.64 2 1 1.28 0.64
Rule 6 0.51 3 1 1.53 0.51
Rule 7 0.84 3 1 2.52 0.84
Rule 8 0.51 3 1 1.53 0.51

X
Rule8
 ¼ 3:61
wA (9)
Rule1

PRule8
wCA
RLFL ¼ PRule1
Rule8 
¼ 2:75 (10)
wA Rule1

Results of defuzzification show that RL equals 2.75, which means the overall RL of the company is in the region
between low and medium.

4.5 Final RL
Based on the results of the two different approaches, we further compare the difference of these two results and assess
the final RL.
14 D. Nakandala et al.

rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
R21 þ R22 2:752 þ 2:842
R¼ ¼ ¼ 2:795 (11)
2 2
As shown in Equation 11, we calculate the overall RL of this company as the square root of the arithmetic mean of
the squares of the two risk values, which is 2.795. This number provides a numerical value of the general RL of the
entire food supply chain network.
Our case study clearly indicates how our hybrid supply chain risk assessment model is used in a real situation, and
how it can help managers to identify and evaluate risk and have a general idea of how risky their company is facing.
The RL of the company in our case study is 2.795 out of 5, which is medium level. It is more convenient to have a sin-
gle overall RL for the company. Decision-makers can be better aware of any changes of the company’s RL, hence
choose appropriate mitigation strategies for supply chain risk management in a more rational manner (Tang and Lau
2011).
Our hybrid model also provides a deeper understanding of the details of the risk scenarios contributing to this over-
all RL through the processes of these two approaches. In Table 3, three risk scenarios with rating 12 have been high-
lighted, indicating that these scenarios are considered as significant and needed to be discerned. These three scenarios
include 2a, 2c and 2d which are volatile demand, supply quality risk and supplier delivery delay, respectively. As to the
FL approach, the three input of risks include macro-level risk (MR), operational risk external to the firm (OR) and inter-
nal risk (IR) with values of 1.85, 3.6 and 2.3, respectively. The most significant input is 3.6 which is the OR. This is in
line with the assessment based on HHM. In this case, it is highly recommended that the company should look into the
4 scenarios related to operational risks and find ways to examine the actual scenarios for possible improvements. It can
be seen here that although the recommended hybrid approach only provides one overall RL of the fresh food supply
chain of the related company, the details of the sources of risks points which contribute to the final RL can be traced
through the procedures of these approaches. This is very useful for companies which are keen to look for risk manage-
ment strategies to ease the RL for the benefit of long term success.

5. Conclusions
The importance of supply chain risk management is known to supply chain practitioners and researchers. For those con-
cerned on fresh food supply chains, supply chain risk management becomes more important due to the short product life
cycle. In order to manage and mitigate supply chain risks, we must first identify the different dimensions of supply
chain risks and evaluate the RL for each risk to identify the overall RL before implementing effective mitigation strate-
gies. This paper proposed a hybrid risk assessment model which includes risk identification and risk evaluation pro-
cesses. HHM, RFRM and FLs methods are combined to form a more comprehensive risk management model that can
be applied in real industrial settings. A case study in fresh food industry has been conducted to validate the practicality
and feasibility of this approach.
Based on the case-study results, the RLs obtained are 2.84 out of 5 (RFRM method) and 2.75 out of 5 (FL
approach), respectively. Using the RMS calculation, the overall RL is 2.795 out of 5. This number provides a numerical
value of the general RL of the entire food supply chain network. Our model also indicates that three risk scenarios
which belong to the Operational risk external to the firm: volatile demand, supply quality risk and supplier delivery
delay are highlighted as the significant risk scenarios which need to be examined thoroughly. Our case study indicates
that the details of the sources of risks points which contribute to the final RL can be traced through the procedures of
these approaches. This is very useful for companies which are keen to look for ways to ease the RL for the benefit of
long-term success.
Our hybrid approach allows the senior management to generate risk analysis in a multiple-view fashion. Not only
the general RL of the entire food supply chain is calculated, but also the details of the risk sources which contribute to
the overall RL can be traced. Although the example is in fresh food industry, the proposed approach can be applied in
different industries for risk management practices. Future research can examine the model in other areas such as supplier
selection as suggested by Wu and Olson (2008).

Disclosure statement
No potential conflict of interest was reported by the authors.
International Journal of Production Research 15

References

Agiwal, S., and H. Mohtadi. 2008. “Risk Mitigating Strategies in the Food Supply Chain.” Annual Meeting, American Agricultural
Economics Association, Orlando, FL, July 27–29.
Christopher, M., and H. Peck. 2004. “Building the Resilient Supply Chain.” The International Journal of Logistics Management
15 (2): 1–14.
Cox, L. A. 2008. “What’s Wrong with Risk Matrices.” Risk Analysis: An International Journal 28 (2): 497–512.
Dani, S., and A. Deep. 2010. “Fragile Food Supply Chains: Reacting to Risks.” International Journal of Logistics Research and
Applications 13 (5): 395–410.
Diabat, A., K. Govindan, and V. V. Panicker. 2012. “Supply Chain Risk Management and Its Mitigation in a Food Industry.”
International Journal of Production Research 50 (11): 3039–3050.
Diop, Ndiame, and S. Jaffee. 2005. “Fruits and Vegetables: Global Trade and Competition in Fresh and Processed Product Markets.
In Global Agricultural Trade and Developing Countries, edited by M. A. Aksoy & J. C. Beghin, 237–257. Washington, DC:
World Bank.
Ellis, S. C., et al. 2010. “Buyer Perceptions of Supply Disruption Risk: A Behavioral View and Empirical Assessment.” Journal of
Operations Management 28 (1): 34–46.
Haimes, Y. Y. 1991. “Total Risk Management.” Risk Analysis 11 (2): 169–171.
Haimes, Y. Y. 2004. Risk Modeling, Assessment, and Management. New York: Wiley.
Haimes, Y. Y., S. Kaplan, and J. H. Lambert. 2002. “Risk Filtering, Ranking, and Management Framework using Hierarchical
Holographic Modeling.” Risk Analysis 22 (2): 383–397.
Harland, C., R. Brenchley, and H. Walker. 2003. “Risk in Supply Networks.” Journal of Purchasing and Supply Management 9 (2):
51–62.
Hendricks, K. B., and V. R. Singhal. 2005. “An Empirical Analysis of the Effect of Supply Chain Disruptions on Long-Run Stock
Price Performance and Equity Risk of the Firm.” Production and Operations Management 14 (1): 35–52.
Hornibrook, S. A., M. McCarthy, and A. Fearne. 2005. “Consumers’ Perception of Risk: The Case of Beef Purchases in Irish Super-
markets.” International Journal of Retail & Distribution Management 33 (10): 701–715.
Jüttner, U. 2005. “Supply Chain Risk Management: Understanding the Business Requirements from a Practitioner Perspective.” The
International Journal of Logistics Management 16 (1): 120–141.
Jüttner, U., H. Peck, and C. Martin. 2003. “Supply Chain Risk Management: Outlining an Agenda for Future Research.” International
Journal of Logistics Research and Applications 6 (4): 197–210.
Kandel, A., and G. Langholz. 1992. Hybrid Architectures for Intelligent Systems. Boca Raton, FL: CRC Press.
Kaplan, S., and B. J. Garrick. 1981. “On the Quantitative Definition of Risk.” Risk Analysis 1 (1): 11–27.
Kleindorfer, P. R., and G. H. Saad. 2005. “Managing Disruption Risks in Supply Chains.” Production and Operations Management
14 (1): 53–68.
Kradjian, K. 2006. “Regulatory Risk Assessment in Early Development: A Practical How to.” Regulatory Affairs Focus 11 (3):
22–27.
Lagerkvist, C. J., et al. 2013. “Food Health Risk Perceptions among Consumers, Farmers, and Traders of Leafy Vegetables in
Nairobi.” Food Policy 38: 92–104.
Lambert, J. H., Y. Y. Haimes, D. Li, R. Schooff, and V. Tulsiani. 2001. “Identification, Ranking, and Management of Risks in a
Major System Acquisition.” Reliability Engineering and System Safety 72 (3): 315–325.
Lang, J., and J. Ding. 2008. “Study on Logistics Outsourcing Risk Management Based on Food Supply Chain.” IEEE International
Conference on Service Operations and Logistics, and Informatics l (1):1421–1426.
Lau, H. C. W., and R. A. Dwight. 2011. “A Fuzzy-based Decision Support Model for Engineering Asset Condition Monitoring – A
Case Study of Examination of Water Pipelines.” Expert Systems with Application 38 (10): 13342–13350.
Lowrance, W. W. 1976. Of Acceptable Risk. Los Altos, CA: William Haufmann.
Manuj, I., and J. T. Mentzer. 2008. “Global Supply Chain Risk Management.” Journal of Business Logistics 29 (1): 133–155.
Nakandala, D., and H. C. W. Lau. 2012. “A Novel Approach to Determining Change of Caloric Intake Requirement Based on Fuzzy
Logic Methodology.” Knowledge-Based Systems 36: 51–58.
Neiger, D., K. Rotaru, and L. Churilov. 2009. “Supply Chain Risk Identification with Value-focused Process Engineering.” Journal of
Operations Management 27 (2): 154–168.
Norrman, A., and U. Jansson. 2004. “Ericsson’s proactive supply chain risk management approach after a serious sub-supplier
accident.” International Journal of Physical Distribution & Logistics Management 34 (5): 434–456.
Peck, H. 2005. “Drivers of Supply Chain Vulnerability: An Integrated Framework.” International Journal of Physical Distribution &
Logistics Management 35 (4): 210–232.
Rangel, D. A., T. Oliveira, and M. Leite. 2015. “Supply Chain Risk Classification: Discussion and Proposal.” International Journal
of Production Research 53 (22): 6868–6887.
Ritchie, B., and C. Brindley. 2007. “Supply Chain Risk Management and Performance.” International Journal of Operations &
Production Management 27 (3): 303–322.
Sinha, P. R., L. E. Whitman, and D. Malzahn. 2004. “Methodology to Mitigate Supplier Risk in an Aerospace Supply Chain.” Supply
Chain Management: An International Journal 9 (2): 154–168.
16 D. Nakandala et al.

Spekman, R. E., and E. W. Davis. 2004. “Risky Business: Expanding the Discussion on Risk and the Extended Enterprise.”
International Journal of Physical Distribution & Logistics Management 34 (5): 414–433.
Svensson, G. 2002. “A Conceptual Framework of Vulnerability in Firms’ Inbound and Outbound Logistics Flows.” International
Journal of Physical Distribution & Logistics Management 32 (2): 110–134.
Tang, C. S. 2006. “Perspectives in Supply Chain Risk Management.” International Journal of Production Economics 103: 451–488.
Tang, C. X. H., and H. C. W. Lau. 2011. “A Rule-based System Embedded with Fuzzy Logic for Risk Estimation.” Eighth
International Conference on FSKD, Shanghai.
Tang, C. S., and B. Tomlin. 2008. “The Power of Flexibility for Mitigating Supply Chain Risks.” International Journal of Production
Economics 116 (1): 12–27.
Trkman, P., and K. McCormack. 2009. “Supply Chain Risk in Turbulent Environments: A Conceptual Model for Managing Supply
Chain Network Risk.” International Journal of Production Economics 119 (2): 247–258.
Tummala, R., and T. Schoenherr. 2011. “Assessing and Managing Risks Using the Supply Chain Risk Management Process
(SCRMP).” Supply Chain Management: An International Journal 16 (6): 474–483.
Wagner, S. M., and C. Bode. 2008. “An Empirical Examination of Supply Chain Performance along Several Dimensions of Risk.”
Journal of Business Logistics 29 (1): 307–325.
Wang, L. X. 1992. A Course in Fuzzy Systems and Control. Englewood Cliffs, NJ: Prentice-Hall.
Waters, D. 2007. Supply Chain Risk Management: Vulnerability and Resilience in Logistics. London: Kogan Page.
Wu, T., J. Blackhurst, and V. Chidambaram. 2006. “A Model for Inbound Supply Risk Analysis.” Computers in Industry 57: 350–365.
Wu, D., S.-H. Chen, and D. L. Olson. 2014. “Business Intelligence in Risk Management: Some Recent Progress.” Information
Science 256: 1–7.
Wu, D., and D. L. Olson. 2008. “Supply Chain Risk, Simulation, and Vendor Selection.” International Journal of Production
Economics 114 (2): 646–655.
Wu, D., D. L. Olson, and C. Luo. 2014. “A Decision Support Approach for Accounts Receivable Risk Management.” IEEE
Transactions on Systems, Man, and Cybernetics: Systems 44 (12): 1624–1632.
Wu, D., D. L. Olson, and A. Dolgui. 2015. “Decision Making in Enterprise Risk Management: A Review and Introduction to Special
Issue.” Omega 57: 1–4.
Wu, D., C. Luo, et al. 2016. “Bi-level Programing Merger Evaluation and Application to Banking Operations.” Production and
Operations Management 25 (3): 498–515.
Yates, J. F., and E. R. Stone. 1992. Risk appraisal. Risk-taking Behavior [M], 49–85. New York: John Wiley & Sons.
Zadeh, L. A. 1965. “Fuzzy Sets.” Information and Control 8: 338–353.
Zingg, A., M. E. Cousin, et al. 2013. “Public Risk Perception in the Total Meat Supply Chain.” Journal of Risk Research 16 (8):
1005–1020.
Zsidisin, G. A. 2003. “A Grounded Definition of Supply Risk.” Journal of Purchasing and Supply Management 9 (5–6): 217–224.
Zsidisin, G. A., and L. M. Ellram. 2003. “An Agency Theory Investigation of Supply Risk Management.” The Journal of Supply
Chain Management 39 (3): 15–27.

You might also like