See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/257018845

A multi-objective analysis for import quota policy making in a perishable fruit

and vegetable supply chain: A system dynamics approach

Article  in  Computers and Electronics in Agriculture · April 2013

DOI: 10.1016/j.compag.2013.01.010

CITATIONS

20

4 authors:

Ebrahim Teimoury Hessam Nedaei

Iran University of Science and Technology Iran University of Science and Technology
56 PUBLICATIONS   233 CITATIONS    4 PUBLICATIONS   26 CITATIONS   

SEE PROFILE SEE PROFILE

Sina Ansari Mostafa Sabbaghi

Northwestern University University at Buffalo, The State University of New York
2 PUBLICATIONS   34 CITATIONS    18 PUBLICATIONS   73 CITATIONS   

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Mostafa Sabbaghi on 05 October 2017.

The user has requested enhancement of the downloaded file.

 Computers and Electronics in Agriculture 93 (2013) 37–45

Contents lists available at SciVerse ScienceDirect

Computers and Electronics in Agriculture

journal homepage: www.elsevier.com/locate/compag

A multi-objective analysis for import quota policy making in a perishable

fruit and vegetable supply chain: A system dynamics approach
E. Teimoury a, H. Nedaei b,⇑, S. Ansari a, M. Sabbaghi c
a
Department of Industrial Engineering, Iran University of Science and Technology, Tehran, Iran
b
Department of Industrial Engineering, Sharif University of Technology, Tehran, Iran
c
Department of Industrial Engineering, K.N. Toosi University of Technology, Tehran, Iran

a r t i c l e i n f o a b s t r a c t

Article history: This study investigates a supply chain of perishable fruits and vegetables. A simulation model using sys-
Received 30 June 2012 tem dynamics approach is proposed to study the behaviors and relationships within the supply chain and
Received in revised form 24 January 2013 to determine the impact of the supply, demand and price interactions. The proposed model yields an
Accepted 27 January 2013
insight into the overall agricultural system, taking into account the influence of import quota policies.
The purpose of this study is to develop a multi-objective model identifying the best import quota policy
for Tehran Municipality Organization of Fruits and Vegetables. The multiple objectives are considered to
Keywords:
be price mean, price variation and markup. Key conclusions were reached on how to conduct the multi-
Perishable fruit and vegetable
Supply chain
objective analysis with the aim of import quota policy making. The multi-objective analysis provides not
Import quota a single solution but a complete non-dominated set of alternatives as potential compromise solutions
Price among multiple objectives. A close examination of surface and contour plots reveals informative trade-
Multi-objective modeling offs. The obtained Pareto frontier contour plots assist decision makers to develop quick import quota pol-
System dynamics application icies by presenting a range of choice available to decision makers and providing them with the trade-off
information among multiple objectives effectively.
Ó 2013 Elsevier B.V. All rights reserved.

1. Introduction
One of the main economic sectors of developing countries such
as Iran is agriculture. The potential of agricultural development in
Iran is so considerable that different types of agricultural products
were exported with the value of more than 3.7 billion dollars in
2010, which was more than four times the value in 2009. Further-
more, the total agricultural export quantity of IRAN was near
3.4 million tons in 2010 which was more than three times the va-
lue in 2009 (FAOSTAT, 2012). Fruits and vegetables exports consti-
tute 70% of the total agricultural product exports, which had been
steadily growing with the rate of 30% from 2009 to 2010 (Iran
Ministry of Agriculture, 2011). However, imports of fruits and veg-
etables were 2.5 billion dollars in 2010, which were as high as the
total exports (The Islamic Republic of Iran Customs Administration
(IRICA), 2011). According to the high influence of imports on Iran’s
agricultural market, it is indispensable for governmental decision
makers to set import quota policies more meticulously in fruit and
vegetable supply chains. This not only leads to a smoothed and rea-
sonable price for the final consumers, but also eventuates in sup-
porting the domestic producers more extensively.
⇑ Corresponding author. Tel.: +98 9191200975; fax: +98 2188192570.
E-mail address: h_nedaei@ie.sharif.edu (H. Nedaei).
Year 2012 has been named as the year of domestic production
in Iran in order to support Iranian producers. Even though the gov-
ernment attempts to support domestic agricultural producers with
reducing import quotas, this reduction leads to an extensive in-
crease in the prices of different products and final consumers’ dis-
satisfaction. Hence, Iranian government faces several challenges
including a decrease in people’s purchasing power and its follow-
ing dissatisfaction, and a reduction in domestic product’s sales.
Thus, restoring market balance with the aim of reducing price fluc-
tuations, and providing a long-term stability with the domestic
production is essential. This paper attempts to investigate the im-
port quota policy in a supply chain of perishable fruits and vegeta-
bles. This aim is achieved by proposing a simulation model in a
system dynamics framework.
The critical components of fruit and vegetable supply chains in-
clude supply, demand, and price for supply chain members includ-
ing farmers, wholesalers, and retailers. Price variability at all levels
of supply chain occurs due to dynamic factors such as substitute
goods price, inflation, import, export, production costs, customer
demands.
The purpose of this study is to develop a multi-objective model
identifying the best import quota policy from the government’s
point of view. Therefore, using a simulation model, the study inves-
tigates the impact of supply and demand interactions on the price of

0168-1699/$ - see front matter Ó 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.compag.2013.01.010
38 E. Teimoury et al. / Computers and Electronics in Agriculture 93 (2013) 37–45
perishable products in a fruit and vegetable supply chain through a
system dynamics approach (Forrester, 1990, 1999; Sterman, 2000).
Since the mathematical difficulties imposed by the supply and
demand interactions in a multi echelon supply chain prevent the
construction of closed form analytical models, we utilize computer
simulation for analysis. Owing to the fact that simulation models
can describe highly complex systems, system dynamics approach
is utilized to highlight the issues and relationships surrounding
the complex perishable fruit and vegetable supply chain.
This paper is organized as follows. Section 2 reviews relevant
literature. Section 3 discusses the logic of simulation model. Sec-
tion 4 presents the quantitative validation and analyses the model
by defining some scenarios for various conditions. Finally, Section 5
provides multi-objective analysis results and discussion for import
quota policy making.
2. Literature review
This section presents the current research in the fruit and veg-
etable supply chains, as well as, import policy making in agricul-
tural markets, system dynamics modeling and multi-objective
analysis.
2.1. Fruit and vegetable supply chains
The supply chain of agricultural products has been widely paid
attention in recent years (Ahumada and Villalobos, 2009; Boudahri
et al., 2011a, 2011b). Ahumada and Villalobos (2009) identify two
main types of agricultural supply chains: fresh agri-foods supply
chain and non-perishable agri-foods supply chain. They review
fresh products owing to their logistical complexity, their limited
shelf life, and the renewed interest of the public on the safety of
these products. The present study also analyses fresh agri-foods
supply chain, which includes highly perishable crops such as fresh
fruits and vegetables whose useful life can be measured in days.
Mergenthaler et al. (2009) investigate demand patterns of fresh
fruits and vegetables in Vietnam.
Several studies have applied mathematical modeling tech-
niques to optimize fruit and vegetable supply chain performance
indicators. In particular, Rong et al. (2011) provide a mixed-integer
linear programming model used for production and distribution
planning in a food supply chain. On the other hand, price variabil-
ity in fruit and vegetable supply chain has not been thoroughly
investigated. Nevertheless, Lambert and Miljkovic (2010) demon-
strate the influence of farm commodity prices, energy costs, and
other factors on food prices in the United States.
2.2. Import policy making in agricultural markets
Several recent studies have been done to investigate export–
import issues (Gorman and Mori, 1988; Wilson and Preszler, 1993;
Chen and Brooks, 1999; Hsu and Wann, 2004; Christou et al.,
2005). Rickard and Lei (2011) examine the economic implications
of tariffs regulations and apply their findings to global markets
for fresh apples and fresh oranges. Abbott and Paarlberg (1998)
scrutinize several aspects of tariff rate quotas as adopted during
tariffication of agricultural policies. However, they have not dealt
with the import quota policies with multi-objective approach.
2.3. System dynamics modeling
The main research on system dynamics modeling in supply
chain management focuses on information sharing, bullwhip ef-
fect, inventory planning/management, supply chain integration,
system performance, planning and forecasting demand and pro-
duction planning and scheduling (Tako and Robinson, 2012).
Kumar and Nigmatullin (2011) investigate the non-perishable
product food supply chain performance under a monopolistic envi-
ronment. They apply system dynamics approach to determine the
impact of demand variability and lead-time on supply chain per-
formance. Özbayrak et al. (2007) propose a modeling framework
to simulate the operation of a supply chain network. They focus
on measuring the supply chain performance metrics such as inven-
tory, WIP (Work In Progress) levels, backlogged orders and cus-
tomer satisfaction. Kishimoto and Yamauchi (1987) also design a
system dynamic model to investigate Japan citrus market.
Multi-objective analysis for import quota policy making in fruit
and vegetable supply chains has been less noticed in the pertinent
literature. Nonetheless, Sterman (2000) states some opinions for
modeling supply and demand in a single market, but there is no
discussion about import policy making in a supply chain.
2.4. Multi-objective analysis
Most of the literature in this section are focused on the applica-
tion of multi-objective analysis in planning issues in agriculture
such as Sarker and Ray (2009) formulating a crop-planning prob-
lem as a multi-objective optimization model. In this case, Ahumada
and Villalobos (2009) review the planning models in the agri-food
supply chain.
There are several studies investigate the price and demand
interactions in agricultural systems through multi-objective analy-
sis. From which, Romero (2000) determines an efficient cropping
pattern by considering the risk of the producers with a multi-
objective (max revenue, min variability) model.
Several other studies apply a multi-objective approach in agri-
cultural context rather than supply and demand interactions. For
example, Fuss et al. (2012) take into account the impact of climate
change on prices via increased volatility in crop yields. However,
the multi-objective import quota policy-making investigated in
this study is barely taken into account in the literature.
While past researches exclusively focus on a single market of
agricultural products, this research extends the analysis to con-
sider a two-stage supply chain and its complex interactions. As
the authors realized, very little theoretical research has been done
on the multi-objective analysis for import quota policy making.
While many studies have been pursued with system dynamics ap-
proach, very few of them have tackled multiple objectives. In this
work, we have conducted multi-objective analysis using system
dynamics simulation for import quota policy making in an agricul-
tural supply chain.
3. Methods and materials
3.1. Problem characteristics
When a system is complex, a simulation model is normally
preferable, for several reasons as it is stated by Altiok and Melamed
(2007). The time spent in deriving a solution to an analytical model
may be excessive and the modeler may assert that an analytical
solution is hard to achieve, due to the apparent mathematical dif-
ficulties. Besides, the modeler may not even be able to formulate
an analytical model to capture the system’s behavioral aspects of
interest. In contrast, simulation modeling can capture virtually
any system, subject to any set of assumptions. This is basically
why simulation is utilized to model the complexity of the proposed
problem.
Fig. 1 illustrates the structure of agricultural products market in
Iran. While simulation models can handle more complex supply
 E. Teimoury et al. / Computers and Electronics in Agriculture 93 (2013) 37–45
Fig. 1. Market structure of agricultural products in Iran.
chain environments, such as the effect of middlemen, food indus-
tries and exporters, considering these complexities could poten-
tially mask the basic underlying economic tradeoffs in the supply
chain, which is the focus of this research. We suggest these addi-
tional complexities as topics for future studies.
Lines and arrows specify the relationships between supply
chain members. The dashed lines indicate the relationships that
are not considered in the present research. The agricultural pro-
ducers, which are farmers and gardeners, produce the fruits and
vegetables. The wholesalers purchase products and sell them to
various retailers. The retailers, which are greengrocers all over
the city, deal with the final consumers. Meanwhile, importers im-
port fruits and vegetables to fill the gap of product shortages
throughout the market. Furthermore, the food industries such as
fruit juice producers supply their raw material through wholesal-
ers or importers. Middlemen as the ones who buy things in order
to sell them to someone else play effectual and sometimes destruc-
tive role all over the market. In addition, Exporters export part of
the agricultural products in this structure. Finally, the government
develops various policies through monitoring the whole chain con-
tinuously. Here, we focus on the government’s import quota policy
making.
This study uses a system dynamics modeling approach applied
to the fruit and vegetable supply chain for perishable products. It
extends the supply–demand–price model of a single market (pre-
sented by Sterman, 2000) to a two-stage supply chain.
In this study, the interactions among wholesalers, retailers and
customers are modeled throughout a two-echelon supply chain.
Fig. 2 illustrates the proposed conceptual model for the study. In
this model, according to the investigated market structure, there
is a close relationship between farmers and wholesalers, which
leads us to consider them as one entity.
Wholesalers' Retailers' supply
supply and demand
- +
+
- + - +
Price (first market)
Price (second market)
Fig. 2. Conceptual model of the supply chain.
39
The wholesalers supply agricultural products in the first mar-
ket. The retailers receive the fruits and vegetables from the whole-
salers and supply them to the customers in the second market. The
retailers play roles of demander and supplier in first and second
markets respectively. The + (-) signs at the arrowheads indicate
that the effect is positively (negatively) related to the cause.
3.2. Simulation model
Using the conceptual model presented in Fig. 2, a system
dynamics simulation model is developed (see Fig. 3). Various levels
of the supply chain (shown in Fig. 2) are translated into corre-
sponding stock and flow variables shown in Fig. 3. This model is
based on continuous flow of price throughout the supply chain.
The supply chain consists of three stages: wholesalers, retailers,
and customers.
The time step is considered 1 day in the simulation. It is also
possible to run the simulation model based on time step less than
or greater than 1 day.
Model assumptions are as follows:
 The supply of retailers in second market assumes to be equal
to the minimum amount of their demand and the supply of
wholesalers in first market (although the waste has not been
considered in the present study, the model is designed flex-
ible enough to consider the waste in the supply chain).
 The product is perishable with the shelf life of one period (it
is a logical assumption due to fruits and vegetables’ holding
circumstances).
The notation is given in Table 1.
Fig. 3 represents the model of system dynamics simulation de-
signed in Vensim software. Additional details about the simulation
procedures are available from the authors.
Model presented in Fig. 3 is concerned with supply, demand
Customers' and their influences on price.
demand In system dynamics models, stock variables accumulate or inte-
-
grate their flow variables; the net flow into the stock is the rate of
change for the stock variable. This structure is represented in (1) as
an integral equation.
Z t
StockðtÞ ¼ ½InflowðsÞ  OutflowðsÞds þ Stockðt0 Þ ð1Þ
t0
40 E. Teimoury et al. / Computers and Electronics in Agriculture 93 (2013) 37–45

Inflation

C S
RS
ES1
Markup1
CM1 Price1
B1 SP1 <Inflation>

ED1
IM1

RB1 Import DS
S1
ES2 RDS

Markup2 B2 SP2 <Inflation>

CM2 Price2

IM2
D
RB2 RD
S2 ED2

Information flow Information flow with delay

Stock variable
Flow variable

Fig. 3. Stock and flow diagram for two-echelon supply chain system.

where Inflow(s) represents the value of the inflow at any time s be- The mathematical formulations pertaining to effects of supply/
tween the initial time t0 and the current time t. demand balance on markup are provided as follows. The basic
The mathematical formulations related to the markup and modeling logic is based on the Sterman (2000) model for a single
prices are provided as follows. market.
Z t CMi ðtÞ ¼ delay½IMi ðtÞ; d  M i ðtÞ i ¼ 1; 2 ð5Þ
M i ðtÞ ¼ ½CM i ðsÞds þ Mi ð0Þ i ¼ 1; 2 ð2Þ
0
IMi ðtÞ ¼ M i ðtÞ  RBi ðtÞ i ¼ 1; 2 ð6Þ
P1 ðtÞ ¼ maxf0; Cð1 þ M 1 ðtÞÞg ð3Þ
si
P2 ðtÞ ¼ maxf0; P 1 ð1 þ M 2 ðtÞÞg ð4Þ RBi ðtÞ ¼ ½Bi ðtÞ i ¼ 1; 2 ð7Þ

Markup (profit rate) of each supply chain member is defined as DSðtÞ

a stock variable named Mi(t) according to (2). Price values in first B1 ðtÞ ¼ ð8Þ
SðtÞ þ Import
and second market are formulated as non-negative auxiliary vari-
ables in (3) and (4) respectively. DðtÞ
B2 ðtÞ ¼ ð9Þ
minfDSðtÞ; SðtÞg
Table 1 By (5), if there is a discrepancy between the desired (IMi(t)) and ac-
Notation list. tual state of markup (Mi(t)), corrective action is initiated to bring
Notation Term the state of the markup variable back in line with the desired value
according to (6).
P1, P2 Price in the first (second) market
M1, M2 The mark-up of the wholesaler (retailer) as the stock variable For better representation of the real world, inflation and its
CM1, CM2 Rate of change for markup of the wholesaler (retailer) as the effect on price values is taken into account. Because of several
flow variable delays in the informational transactions in the real world, the
IM1, IM2 Indicated markup of the wholesaler (retailer) sensitive to effect of demand/supply balance on price is considered with a
demand/supply balance
RB1, RB2 Effective term of demand/supply balance on markup of the
delay of 1 day. The delay parameters are approximate values
wholesaler (retailer) obtained based on information provided by experts in Tehran
B1, B2 Demand/supply balance term in the first (second) market Municipality Organization of Fruits and Vegetables. Delay type
SP1, SP2 Substitute goods price in the first (second) market 3, which is a system dynamics function that returns a third
S Supply of the wholesaler
order exponential delay of the input is applied due to its
DS The retailer’s demand/supply in the first/second market
D Demand of the final customers compatibility.
RS Reference value for S The effective term of demand/supply balance (RBi(t)) appeared
RDS Reference value for DS in (6), is defined in (7). By (8) and (9), Demand/supply balance
RD Reference value for D terms are defined as the related demand divided by the whole sup-
Import Import amount which is added to supply in the first market
ES1, ES2 Supply elasticity in the first (second) market
ply in the pertaining market. By (8), B1(t) is calculated by dividing
ED1, ED2 Demand elasticity in the first (second) market DS(t) by the total supply in the first market, which is summation of
s1, s2 Sensitivity of price to demand/supply balance in the first S(t) and Import.
(second) market In modeling of B2(t), as shown in (9), the final customers’ de-
C Cost per unit of product
mand (D(t)) is divided by the actual supply, which is the minimum
Inflation Inflation rate
of demand and supply in the first market.
 E. Teimoury et al. / Computers and Electronics in Agriculture 93 (2013) 37–45 41

The mathematical formulations regarding supply and demand Table 3

are provided as follows. Parameters estimation results (obtained by calibration in 2010).

" ES1 # Notation Term Parameters

P1 ðtÞ estimation
SðtÞ ¼ RSðtÞ  delay ;d ð10Þ
SP1 ðtÞ ES1 Supply elasticity of the wholesaler (market 1) 1.805
" ED1 # " ES2 # ES2 Supply elasticity of the retailer (market 2) 0.931
P1 ðtÞ P2 ðtÞ 0 ED1 Demand elasticity of the retailer (market 1) 1.567
DSðtÞ ¼ RDSðtÞ  delay ; d  delay ; d ð11Þ
SP1 ðtÞ SP2 ðtÞ ED2 Demand elasticity of the customer (market 2) 0.153
" ED2 # s1 Sensitivity of price to demand/supply balance 2
P2 ðtÞ (market 1)
DðtÞ ¼ RDðtÞ  delay ;d ð12Þ s2 Sensitivity of price to demand/supply balance 1.87
SP2 ðtÞ
(market 2)

By (10), the actual amount of wholesalers’ supply, S(t), is ob-

tained by considering the effect of the price (P1(t)) and substitute
goods price (SP1(t)) in the first market. ES1 adjusts the intensity 4. Scenario-based analysis and validation
of the mentioned effect. This formulation is consistent with the for-
mulation of a single market by Sterman (2000). We applied the model in Tehran’s fruits and vegetables market.
The effect of price on supply and demand is considered with a Cherry was considered as a perishable fruit for analysis. The spe-
delay shown as d parameter in the formulation. While d parame- cific characteristics of the system are as follows. Inflation rate per
ters are shown with the same notation, the values are different year is 18% based on the information presented by IRI Central Bank
for each equation. (2012). The Iran’s total cherry production is 255,500 tons in 2010
Eq. (11) is the most related equation to the supply chain. (FAOSTAT, 2012). Besides, 3292 tons of the production has been
According to the model assumptions, the demand of the retailer exported. The remaining amount is supplied in local market of
in the first market equals the supply of the retailer in the second Iran; but this study investigates the market of Tehran (the capital
market. DS(t) that models both supply and demand of the retailers of Iran). Therefore, we estimate the total supply in the Tehran’s
is calculated by multiplying the impact of both markets’ prices by market using population and consumption coefficients. Finally,
the reference value (RDS(t)). Thus, the demand and the supply of the average demand of Tehran’s consumers is estimated to be
the retailers in the first and second market are affected by the 3500 tons per week. Average production cost (per kilogram) is
prices in both markets. 1.6 dollars. Substitute goods price in the first and second market
Similar to (10), the Eq. (12) shows the effect of prices on the are 2 and 2.5 dollars, respectively. The detailed supply trend in
customers’ demand. Final customers’ demand elasticity, ED2, ad- the market is obtained from expert judgment in the validation
justs the intensity of the mentioned effect. periods equals to 14 weeks from May 20 to August 19, 2010.
The mathematical formulations related to reference values for The quantitative validation is conducted in two steps. First, the
supply and demand are provided as follows. parameter values are estimated using 2010 data. The calibration
and parameter estimation results are shown in Tables 2 and 3,
RSðtÞ ¼ smooth½DSðtÞ; s ð13Þ
respectively. Then, the simulation model runs and the results are
RDSðtÞ ¼ smooth½DðtÞ; s ð14Þ compared with 2011 data (see Fig. 4).
While supply elasticity and sensitivity of price to demand/sup-
The smooth function that is commonly used to model averaging
ply balance are some positive values, demand elasticity is a nega-
in a process, is used in Eqs. (13) and (14) with the aim of averaging
tive value according to the modeling logic. The calibration has been
the demand in each market. This models the general behavior of
done using AnyLogic™ software (www.anylogic.com).
wholesalers and retailers on adjusting their supply according to
Fig. 4 illustrates the comparison of the simulation outputs
the demand. By (13), RS(t) which is the reference value for the
against the actual data of the wholesaler’s and retailer’s price for
wholesalers’ supply, is obtained by averaging the retailers’ demand
the cherry in 2011. The comparison shows that the model is quan-
(DS(t)). By (14), the same equation is modeled in the second
titatively validated.
market.
In addition to the quantitative validation, the model is also val-
idated qualitatively. In this case, some realistic scenarios are de-
fined based on what happens in the real world. The results
Table 2 gained from these scenarios provide several insights. These scenar-
Calibration results for the price of two markets in 2010. ios can also be used for the validation of the proposed model. We
Date Market 1 price ($) Market 2 price ($) have checked the model’s validity by conducting validation tests
suggested by the system dynamics literature (Sushil, 1993; Barlas,
Calibration Actual Calibration Actual
results dataa results dataa 1994; Sterman, 2000). The qualitative validation test is done by the
method called ‘‘behavior reproduction test’’ (Sushil, 1993). Before
May 20 2.854 3.262 7.940 8.236
May 27 2.125 2.627 4.603 5.379 testing the behavior accuracy, it is necessary to test the validity
June 3 1.87 1.935 3.322 4.468 of the structure (Barlas, 1994). The structure test has been done
June 10 1.796 1.923 2.819 3.960 by assessing the validity of the model equations individually by
June 17 1.778 1.923 2.633 3.638 comparing equations with the related literature and available
June 24 1.779 1.923 2.582 3.046
July 1 1.783 1.923 2.566 2.728
knowledge of the real system.
July 8 1.789 1.846 2.551 2.692 The following scenarios are based on numerical examples to
July 15 1.796 1.615 2.526 2.429 conduct the qualitative validation.
July 22 1.802 1.531 2.478 2.393
July 29 1.803 1.531 2.409 2.408
August 5 1.797 1.764 2.338 2.705 4.1. Scenario 1 – increase in the final customers’ demand
August 12 1.842 1.899 2.457 2.825
August 19 2.073 2.157 3.225 3.656
Sudden increase in fruit price in both levels of supply chain
a
Source: Central Bank of the Islamic Republic of Iran (2012). is occurred when sudden increase in demand of fruit happens
42 E. Teimoury et al. / Computers and Electronics in Agriculture 93 (2013) 37–45

Fig. 4. Actual vs. simulation output of the wholesaler’s and retailer’s price for the cherry in 2011.

(e.g. in the days before the New Year holiday in which people (Fig. 5B). Thus, retailers have to reduce their markup (Fig. 5C) and
tend to purchase more fruits for the ceremonies). The reference finally, price will reach a stable state.
value for final customers’ demand (RD(t)) is increased by
50 tons from time 60 to 80. Fig. 5 illustrates the results of sce- 4.2. Scenario 2 – decrease in the wholesalers’ supply
nario 1.
According to Fig. 5, the increase in customer’s demand not only With sudden decrease in supply, for example due to climatic
leads to an increase to retailers’ supply and demand, but also an in- changes such as drought, which is probable in Iran, sudden in-
crease in wholesalers’ supply (Fig. 5A). This eventuates in product’s crease is observed in the prices in both levels of supply chain. On
price augmentation. Even though price increases after a rise in de- the other hand, over the specified period, because of such increase
mand on the 60th day (Fig. 5D), with a delay, customers’ demand in the price, the pertaining demand will decrease in order to con-
decreases when this price exceeds the price of the substitute goods trol the impact of the rise in the price. This scenario considers a
supply or demand (tons)
supply or demand (tons)

570
570
2
1 2
520 520 1 1
2
2 1 12 1 12 12
12 12 1 2 2 12 2 12 1
1

470 470
0 30 60 90 120 150 0 30 60 90 120 150
Time (Day) Time (Day)
Wholesalers' supply 1 1 1 1 Retailers' supply 1 1 1 1
Retailers' demand 2 2 2 2 2 Customers' demand 2 2 2 2

4
price ($ per kg)
markup (percent)

0.4

2 2 2 2 2
0.3 2.5 2 2 2
2 1 1
2 1 1 1 1 1 1
1
2 12 1 12 12
1 2 12
0.2 1 1
0 30 60 90 120 150 0 30 60 90 120 150
Time (Day) Time (Day)
Wholesalers' markup 1 1 1 1 1 Price (Market 1) 1 1 1 1 1
Retailers' markup 2 2 2 2 2 Price (Market 2) 2 2 2 2 2 2

Fig. 5. Scenario 1 outputs (increase in the final customers’ demand). From top to bottom and from left to right: A, B, C, D.
 E. Teimoury et al. / Computers and Electronics in Agriculture 93 (2013) 37–45 43

90% decrease in the reference value for the supply of wholesalers reached and then closing the border to further imports (Reinert
(RS(t)) from time 60 to 80. Fig. 6 illustrates the results of scenario et al., 2008).
2. Governments often use import quotas to protect domestic pro-
Fig. 6 indicates that after a reduction in supply on 60th day ducers from import competition or to encourage domestic produc-
(Fig. 6A), although wholesalers sell the product with higher tion. Different states corresponding to the amounts of import are
markup (Fig. 6C), the retailers cannot increase their markup x1, x2 2 {0, 100, . . ., 500} regarding to the historical data presented
immediately. This occurs because they decrease their supply by the government and different states relating to import change
with a delay. However, when the price rises with a delay time (the time in which the import quota changes from x1 to x2)
(Fig. 6D), the demand decreases (Fig. 6B) and thus the wholesal- are obviously importChangeTime 2 {1, 2, . . ., 12}, indicating the
ers no longer are able to increase their markup so as the retailers month number of a year.
(Fig. 6C). These patterns continue until the price values become In reality, several events might occur simultaneously in the
stable. market. Prior to providing an import quota policy, other parame-
Followings are other situations defined as practical scenarios for ters are estimated according to the expert judgment and history
analysis: data. As a result of the future analysis, a scenario is defined that
sets a 40% decrease in the supply of wholesaler on the 4th month
 A decrease in substitute goods prices. and 40% increase in the customers’ demand on the 8th month.
 An increase or decrease in the production costs: a sudden The supportive role of the government in supply chain manage-
increase in production cost leads to a decrease in demand in ment of fruits and vegetables has two sides: supporting domestic
each level. producers and supporting final consumers. The first goal is
 A changes in supply or demand elasticity. achieved by maximizing the wholesalers’ markup (which conse-
 A changes in the sensitivity of price to the demand–supply quently maximizes the farmers’ markup). The second goal is
balance. achieved by minimizing the price mean and variation in the second
market. These different goals are analyzed in a multi-objective
Elasticity and sensitivity parameters were analyzed in sensitiv- approach.
ity analysis and it revealed that changes in these parameters would The multi-objective formulation would be as follows:
intensify changes in the values of other variables. The detailed re-
X
n1
sults for these scenarios are not presented due to space limitations, Min Z 1 ¼ ðpricetþ1  pricet Þ2 ð15Þ
but are available from the authors. t¼1
The defined scenarios not only provide detailed analyses of the !
X
n
model, but also help to indicate that the model is valid based on Min Z 2 ¼ pricet =n ð16Þ
the behavior reproduction method. In reality, a combination of t¼1
!
scenarios could create various trends in supply and demand that X
n

finally influence the price. Next section seeks to conduct a multi- Max Z 3 ¼ markupt =n ð17Þ
t¼1
objective analysis for import quota policy making in the situation
in which some of the previously mentioned scenarios happens in While objective (15) aims to obtain a smoothed price in the
the market. market by minimizing the price variation, objective (16) minimizes
the mean of price. As in Karimi-Nasab and Konstantaras (2012) and
5. Multi-objective results, analysis and discussion Karimi-Nasab and Ghomi (2012), we take into account the varia-
tion in form of sum of the squared difference of subsequent prices.
An import quota sets a maximum permissible volume or value Objective (17) deals with maximizing the mean of the wholesaler’s
of a product that can enter a country during a particular time. The markup. This objective indicates the supporting role of the govern-
import quota limit is normally declared before the beginning of ment for agricultural producers.
this time. Import quotas can be administered on a ‘‘first-come, Fig. 7 shows the surface plot of price variation vs. price mean
first-served basis’’ by allowing imports until the quota ceiling is and markup. The surface plot is developed by changing the
supply or demand (tons)

supply or demand (tons)

545 540
1
2 2 12 12
505 1 12 12 2
502.5 1
1 1 12 1 12
2 1 12
1 2
465 2 465
0 30 60 90 120 150 0 30 60 90 120 150
Time (Day) Time (Day)
Wholesalers' supply 1 1 1 1 1 Retailers' supply 1 1 1 1 1
Retailers' demand 2 2 2 2 2 Customers' demand 2 2 2 2 2

4
0.4
price ($ per kg)
markup (percent)

1
2 2 2 2
2 2 2 2
0.3 2.5 1
1 1 1 1 1 1
2 12 1 1 1
2 12 2 1
1
2
0.2 1 1
0 30 60 90 120 150 0 30 60 90 120 150
Time (Day) Time (Day)
Wholesalers' markup 1 1 1 1 Price (Market 1) 1 1 1 1 1
Retailers' markup 2 2 2 2 Price (Market 2) 2 2 2 2 2 2

Fig. 6. Scenario 2 outputs (decrease in the wholesalers’ supply). From top to bottom and from left to right: A, B, C, D.
44 E. Teimoury et al. / Computers and Electronics in Agriculture 93 (2013) 37–45

upper/lower bound of two objectives and finding the value of the
third objective corresponding to the best state. Applying the meth-
odology proposed by Messac et al. (2003), the Pareto-frontier is
built based on the 81 non-dominated solutions found.
The surface plot (Fig. 7) along with the government’s three-
dimensional utility function as preference information is used to
select a subset of Pareto solution for the import quota policy. The
utility function is a mathematical expression that attempts to
model the decision-makers’ preferences. In this analysis, the gov-
ernment’s utility function represents the relative importance of
the three objectives.
Nevertheless, the trade-off between different objectives could
be better presented by contour plot. The contour plots shown in
Figs. 8–10 give a representation of the trade-off solutions. Fig. 8
depicts the contour plot of markup vs. price mean and price
variation.
The contour plot shows the trade-off between price mean and
different values for the price mean indicate the high sensitivity
of price mean in the multi-objective analysis. This clearly shows
that the multi-objective analysis is a nonlinear non-convex prob-
lem due to the fact that the contour plot is neither linear nor
convex.
The intersection of the government’s two-dimensional utility
function and the contour plot is indicative of the best import quota
policy. The obtained Pareto frontier contour plot is proposed to
support decision makers for quick import quota policy making in
the case of a change in government’s priorities.
3.125 10 20
3.100
25
3.075
25

price mean ($)

price variation. The values specified on the lines indicate the per- 3.050
taining markup values. Fig. 8 demonstrates that if the decision ma-
3.025
ker expects to have a lower price variation, he ought to accept a
higher price mean. This behavior changes through different values 3.000
for the markup. Considering the line with the markup value of 15%,
2.975 15
a threshold is observed for the price variation. While a price vari-
5
ation beneath 0.5 causes a steep gradient in price mean, setting 2.950
the price variation value over 0.5 will not have such an effect. 5

Higher values of markup such as 25% make the decision maker to 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
accept higher values for price mean and variation. Besides, it limits price variation
the decision makers’ scope of changing priorities of objectives.
Fig. 8. Contour plot of markup vs. price mean and price variation.
Fig. 9 illustrates that if the decision maker expects to have a
lower price mean, he ought to accept a lower markup. For specific
price variation value of 0.2, if the decision maker expects a lower
value for price mean to satisfy final customers, a threshold is ob-
30
served for the price mean. Lowering the price mean up to 3 (the 0.4
threshold) will be achieved without setting a lower value for mark-
25 0.6
up (the government is willing to support the producers by allowing
higher markups). However, setting the price mean value beneath 3 20
markup (%)

causes an unwanted reduction in markup.

Fig. 10 demonstrates that if the decision maker expects to have 15
a higher markup, he ought to accept a higher price variation which 0.6
is not satisfying. This behavior is observed in different values of 10
price mean. For specific price mean value of 3, it is observed that
the trade-off is meaningful for only some parts of Pareto frontier 5
line. That is, for markup values between 10 and 20 and price vari-
0.2
ation values between 0.6 and 0.8, changing the preference between
markup and price variation results in deterioration in both objec- 2.950 2.975 3.000 3.025 3.050 3.075 3.100 3.125
tives. Therefore, it is not logical for the decision maker to accept price mean ($)
this range for markup and price variation when the price mean is Fig. 9. Contour plot of price variation vs. markup and price mean.
set to be 3. In addition, different patterns of the contour lines for

1.0 3.00
3.00
0.9
0.8
3.10
price variation

1.0 0.7
price variation 0.6
0.5 0.5
3.05
0.4
2.95
0.0 30
0.3
20 3.00
3.10 0.2 3.05
3.05 10 markup (%)
3.00
2.95 0 5 10 15 20 25 30
price mean ($) markup (%)

Fig. 7. Surface plot of price variation vs. price mean and markup. Fig. 10. Contour plot of price mean vs. price variation and markup.
 E. Teimoury et al. / Computers and Electronics in Agriculture 93 (2013) 37–45
6. Conclusion
This study proposed a simulation model to investigate the rela-
tionships in fruits and vegetables supply chains using system
dynamics approach. It is useful as a high-level tool to analyze the
interactions of supply, demand and price. The model was validated
by means of behavior reproduction test. Our observation shows
that applying the best import policy helps to reduce the undesired
price changes in case of different scenarios for demand and supply.
Thus, it may prove useful to policy makers especially governmental
authorities dealing with a wide spectrum of strategic fruit and veg-
etable supply chain management issues. A multi-objective analysis
was conducted to empower the system dynamics model through-
out trade-off solutions for import quota policies obtained in the
form of Pareto Frontier. The specified thresholds observed in this
study provide an insight for governmental decision makers to have
better understanding of the trade-offs between three objectives:
price mean, price variation and markup.
Combination of agent-based simulation and system dynamics is
suggested for future research. In addition, this model can be ex-
panded to take into account further market characteristics such
as the interventions of middlemen. Even though the model is use-
ful to study relationships between supply chain members, the
model does not take into consideration detailed inventory consid-
erations, transportation and warehousing. These are topics for fu-
ture research.
Acknowledgments
The authors are grateful to the anonymous referees for their
insightful suggestions and comments. Authors thank Dr. Karimi-
Nasab for his consultations and also the agricultural experts in
Tehran Municipality Organization of Fruits and Vegetables for their
supports.
References
Abbott, P.C., Paarlberg, P.L., 1998. Tariff rate quotas: structural and stability impacts
in growing markets. Agricultural Economics 19, 257–267.
Ahumada, O., Villalobos, J.R., 2009. Application of planning models in the agri-food
supply chain: a review. European Journal of Operational Research 196, 1–20.
Altiok, T., Melamed, B., 2007. Simulation modeling and analysis with Arena.
Academic Press.
Barlas, Y., 1994. Model validation in system dynamics. In: Proceedings of the 1994
International System Dynamics Conference, pp. 1–10.
Boudahri, F., Sari, Z., Bennekrouf, M., 2011. Renovation of a distribution network of
poultry products: application city of Tlemcen (ALGERIA). In: 2011 4th
International Conference on Logistics (LOGISTIQUA). Presented at the 2011
4th International Conference on Logistics (LOGISTIQUA), IEEE, pp. 115–121.
Boudahri, F., Sari, Z., Maliki, F., Bennekrouf, M., 2011. Design and optimization of the
supply chain of agri-foods: application distribution network of chicken meat.
In: 2011 International Conference on Communications, Computing and Control
Applications (CCCA). Presented at the 2011 International Conference on
Communications, Computing and Control Applications (CCCA), IEEE, pp. 1–6.
View publication stats
45
Chen, K., Brooks, H., 1999. State trading and non-price discriminatory trade: the
case of Japan’s wheat imports. Agribusiness 15, 41–51.
Christou, A.P., Kilmer, R.L., Sterns, J.A., Feleke, S.T., 2005. Derived demand for
disaggregated cheese products imported into Japan. Agribusiness 21, 1–16.
FAOSTAT, 2012. Food and Agriculture Organization of the United Nations (FAO).
Forrester, J.W., 1990. Principles of Systems. Productivity Press.
Forrester, J.W., 1999. Industrial Dynamics. Pegasus Communications.
Fuss, S., Havlik, P., Szolgayova, J., Obersteiner, M., Schmid, E., 2012. AGRICULTURAL
PRICE VOLATILITY UNDER CLIMATE CHANGE: The Impact of Multiple Objectives
on Commodity Prices (123rd Seminar, February 23–24, 2012, Dublin, Ireland
No. 122539). European Association of Agricultural Economists.
Gorman, W.D., Mori, H., 1988. Relative prices of imported grain fed beef in Japan:
the case of air-freighted chilled carcasses. Agribusiness 4, 535–548.
Hsu, J.L., Wann, J.J.W., 2004. Competitiveness and consumer preferences of US fruits
in Taiwan. Agribusiness 20, 433–448.
Iran Ministry of Agriculture, 2011. <www.maj.ir>, Tehran, Iran.
IRI Central Bank, 2012. Economic Statistics. Central Bank of the Islamic Republic of
Iran, <www.cbi.ir>.
Karimi-Nasab, M., Ghomi, S.M.T.F., 2012. Multi-objective production scheduling
with controllable processing times and sequence-dependent setups for
deteriorating items. International Journal of Production Research 50, 7378–
7400.
Karimi-Nasab, M., Konstantaras, I., 2012. A random search heuristic for a multi-
objective production planning. Computers & Industrial Engineering 62, 479–
490.
Kishimoto, Y., Yamauchi, H., 1987. A citrus system dynamic model of Japan.
Agribusiness 3, 43–50.
Kumar, S., Nigmatullin, A., 2011. A system dynamics analysis of food supply chains –
case study with non-perishable products. Simulation Modelling Practice and
Theory 19, 2151–2168.
Lambert, D.K., Miljkovic, D., 2010. The sources of variability in US food prices.
Journal of Policy Modeling 32, 210–222.
Mergenthaler, M., Weinberger, K., Qaim, M., 2009. The food system transformation
in developing countries: a disaggregate demand analysis for fruits and
vegetables in Vietnam. Food Policy 34, 426–436.
Messac, A., Ismail-Yahaya, A., Mattson, C.A., 2003. The normalized normal
constraint method for generating the Pareto frontier. Structural and
Multidisciplinary Optimization 25, 86–98.
Özbayrak, M., Papadopoulou, T.C., Akgun, M., 2007. Systems dynamics modelling of
a manufacturing supply chain system. Simulation Modelling Practice and
Theory 15, 1338–1355.
Reinert, K.A., Rajan, R.S., Glass, A.J., 2008. The Princeton Encyclopedia of the World
Economy. Princeton University Press.
Rickard, B.J., Lei, L., 2011. How important are tariffs and nontariff barriers in
international markets for fresh fruit? Agricultural Economics 42, 19–32.
Romero, C., 2000. Risk programming for agricultural resource allocation: a
multidimensional risk approach. Annals of Operations Research 94, 57–68.
Rong, A., Akkerman, R., Grunow, M., 2011. An optimization approach for managing
fresh food quality throughout the supply chain. International Journal of
Production Economics 131, 421–429.
Sarker, R., Ray, T., 2009. An improved evolutionary algorithm for solving multi-
objective crop planning models. Computers and Electronics in Agriculture 68,
191–199.
Sterman, J., 2000. Business dynamics: systems thinking and modeling for a complex
world. Irwin/McGraw-Hill.
Sushil, 1993. System dynamics: a practical approach for managerial problems.
Wiley Eastern Ltd.
Tako, A.A., Robinson, S., 2012. The application of discrete event simulation and
system dynamics in the logistics and supply chain context. Decision Support
Systems 52, 802–815.
The Islamic Republic of Iran Customs Administration (IRICA), 2011.
<www.irica.gov.ir>, Tehran, Iran.
Wilson, W.W., Preszler, T., 1993. Quality and price competition in international
wheat trade: a case study of the United Kingdom wheat import market.
Agribusiness 9, 377–389.