logistics and transportation in global supply chains ...jbookbinder/tutorials-2009-cd... ·...
TRANSCRIPT
INFORMS 2009 c© 2009 INFORMS | isbn 978-1-877640-24-7doi 10.1287/educ.1090.0059
Logistics and Transportation in Global SupplyChains: Review, Critique, and Prospects
James H. Bookbinder and Tiffany A. MatukDepartment of Management Sciences, University of Waterloo, Waterloo, Ontario N2L 3G1,Canada {[email protected], [email protected]}
Abstract A supply chain denotes the series of operations by which raw materials are trans-formed into components, subassemblies, and finished goods, and then moved througha distribution system to the ultimate user or consumer of the item produced. Producttransformation and storage occur at nodes of the network. When those are in differentcountries or areas of the world, logistics and transportation (between nodes) becomeeven more important.
In this tutorial, we review and compare a number of research papers. All are inter-national in scope. Each article emphasizes one or more decisions or activities thatdistinguish a global supply chain. Some, such as location decisions and supplier selec-tion, are important for domestic supply chains but are more complicated in the inter-national arena. Major differences include border crossings and exchange rates. Otherdecisions, such as switching production from one country to another, are uniquelyglobal and may depend upon tariffs, quotas, and local content restrictions.
Following our rationale for the subset of literature that we chose, those articles arediscussed, categorized, and compared in several complementary ways. An assessmentis offered of the research conducted to date. Further studies that may be usefullyundertaken are suggested.
Keywords international; operations management; intermodal; distribution; manufacturing;location; trade; NAFTA; EU; supplier selection; transfer pricing; literature review
1. IntroductionGlobalization has become both a buzzword and a truism. Demand for products and servicesoriginating from across the world have made it necessary for companies to reexamine theirsupply chain processes to maintain competitiveness in the international arena. Competitioncan be in the form of new and innovative technology, but perhaps the most easily measuredforms of competition are profits and costs.
Supply chain models attempt to develop optimal facility, supplier, and/or transportationnetworks that are the most profitable (or least costly). Expanding these models to includeinternational locations and suppliers creates additional challenges, because exchange ratefluctuations and transfer prices may affect product costing, and transportation linkages maybecome more complex and expensive. However, labor costs in certain regions of the worldare often so low that they can overcome the large transportation costs from there to markets,or to another location where the more technologically oriented processes will take place tocomplete the production of the end item.
This is the “truism” portion, that the optimal supply chain for such products may procureraw materials on one continent, and complete the manufacturing on another continent, forsale in a third. While “buzzwords” may come and go, globalization is not going anywhereanytime soon. Regional trading blocs exist in North America (North American Free TradeAgreement (NAFTA)), Europe (European Union (EU)), and Asia (Association of Southeast
182
Bookbinder and Matuk: Logistics and Transportation in Global Supply ChainsTutorials in Operations Research, c© 2009 INFORMS 183
Asian Nations). Firms from one region may outsource some production to a firm in anotherregion, or they may purchase components from another region for assembly in their own.
The preceding suggests why we chose “Global Supply Chains” as part of our title. Natu-rally, the international movements of raw materials, components, and finished goods requireexcellent logistics practices. Most authors would include transportation as one of the func-tional areas of logistics, and so do we. But we feel that separate mention of each makes moreobvious the trade-offs of transportation within decisions on inventory or location or man-ufacturing. These trade-offs are fundamental in logistics management. It is for this reasonthat the logistical activities that we discuss in this tutorial will emphasize transportation,or at least place it on a par with the other functions.
In this tutorial, we are not exhibiting a new problem formulation nor teaching a modelingtechnique. Rather, in light of the significance of globalization in the popular press, and ofcourse its actual importance, we intend to put into context the vast number of referenceson global supply chains. We wish to offer guidance to practitioners in companies whosesupply chains have become more international in scope, and to academics, we will suggestresearch opportunities by classifying and critiquing the available models in various waysthat reinforce each other.
Naturally, a model that attempts to cover “everything” must neglect some details.A number of authors have made different choices, even in similar-sounding cases, aboutwhich factors are important. The reasons for those choices, and specific findings in the givensituations, can guide future research and practice. We will show that particular applicationshave so far been little studied or not studied at all. We will offer suggestions for additionalresearch and the underlying rationale.
1.1. Research ApproachOur survey is limited to journal articles pertaining to aspects of global or regional logisticspublished during the past 15 years. To search for articles, certain key terms and variations ofthem were used, such as “global logistics,” “global production distribution,” “internationalsupply chain,” and “international transportation.” The papers were broadly (and subjec-tively) categorized by the following primary topics: previous review papers; facility location;procurement; transportation; production switching, transfer pricing policies, and postpone-ment; overall networks; and other papers. (An additional search for articles was performedusing the references from previous review papers if they did not appear in the initial search.)Because of the nature of some topics, papers may address multiple themes but will be dis-cussed by the main issue or problem that they study. Also, although some articles do notdiscuss logistics on a global scale per se, their subjects and solution methodologies may beeasily transferred to an international supply chain network.
Table 1 summarizes, by primary topic and region of focus, the references that addressinternational logistics. The majority of papers treat global supply chains on a worldwidebasis, whereas the remaining papers are fairly evenly distributed across the continents.The following five articles are not specifically global in nature but discuss topics that areapplicable to international logistics. The models developed by Eskigun et al. [24] (2005)and Grasman [28] (2006) relate to multimodal routing, an aspect that is common to globaltransportation models. Bilgen and Ozkarahan [7] (2004), Gunasekaran and Kobu [30] (2007),and Melo et al. [51] (2009) survey topics in logistics and describe certain features that maybe applicable to an international supply chain.
In the next sections, we will discuss the trends in research appearing in the reviewed liter-ature, providing the reader with a breakdown of papers by category, global region addressedin the research, model objective, and distinguishing features. Publications will be discussedin chronological order.
Bookbinder and Matuk: Logistics and Transportation in Global Supply Chains184 Tutorials in Operations Research, c© 2009 INFORMS
Table 1. Selected papers by primary topic and region of focus.
Procurement Transportation
Worldwide 8 5 4 3 14 4 44
N. America 1 2 1 1 7
S. America 1 1 3 5
Europe 6 1 11
Asia/Oceania 2 1 1 6
Topic total 8 9 4 12 16 10 14 73
PrimaryTopicRegion
Previousreviews
Facilitylocation
Productionswitching, transfer
pricing policies, andpostponement
Overall networkdesign
Regionaltotal
Otherpapers
6
2
4
2
2. Previous Review PapersA number of review articles are concerned with the effect of globalization on supply chains(see Table 2). Bhatnagar et al. [5] (1993) research the importance of multiplant coordi-nation in making optimal production planning decisions for an entire firm. Two varia-tions of coordination are examined: (1) general coordination, referring to the integrationof different functions such as facility location, production, distribution, and marketing; and(2) multiplant coordination, which deals with linking decisions within the same functionaldepartment across different stages of the supply chain. The authors go on to describe howmultiplant coordination is affected by differences in lotsizing at various facilities, nervous-ness in demand, and levels of safety stock. Pontrandolfo and Okogbaa [59] (1999) also reviewliterature pertaining to the configuration and coordination of multinational corporations(MNCs). From their research, the authors develop a framework to evaluate the currentdegree of coordination of a firm and to provide recommendations for improvement.
Boone et al. [10] (1996) explore relevant literature to determine if any regional differencesexist in the approach to international logistics problems. As well, the authors evaluate threepreviously established frameworks for classification of this research. Boone et al. [10] findthat no statistically significant regional differences exist in the chosen topics nor in thesolution methodologies used; however, when a particular framework that is geared towardstrategic classification is applied, regional trends become more apparent.
Vidal and Goetschalckx [73] (1997) conduct a review of strategic production–distributionmodels in global supply chains, focusing particularly on the use of mixed integer program-ming (MIP) as a solution approach. They tabulate the model characteristics of eight selectedMIP papers, as well as international issues present in six selected global supply chain mod-els. Vidal and Goetschalckx [73] (1997) find that most models do not take into account thevolatility of global supply chains, nor do they consider bill of materials (BOM) constraints.
Table 2. Previous review papers.
Authors Year Additional information
Bhatnagar et al. [5] 1993 Multiplant coordination should address nervousness, lot sizing, and safety stock
Boone et al. [10] 1996Determines whether a significant difference exists between regions in terms ofresearch; classifies papers based on 3 established frameworks
Cohen and Mallik [18] 1997 Reviews the use of analytical models in global supply chain management literature
Vidal and Goetschalckx [73] 1997 Reviews production–distribution models with an emphasis on MIP
Pontrandolfo andOkogbaa [59] 1999
Develops framework for global manufacturing planning and uses to analyze anMNC’s current coordination mode
Schmidt and Wilhelm [66] 2000Reviews papers and modeling issues addressing strategic, tactical, andoperational levels
Prasad et al. [60] 2001 Detailed classification of research papers
Meixell and Gargeya [50] 2005 Assesses how well research papers address global logistics decisions
Bookbinder and Matuk: Logistics and Transportation in Global Supply ChainsTutorials in Operations Research, c© 2009 INFORMS 185
Some of their suggestions for future research include the consideration of more stochasticfeatures such as customer service levels and lead times, improved modeling of BOM con-straints, simulation of qualitative factors, and the inclusion of vendor and transportationchannel reliability. Similarly, the review by Cohen and Mallik [18] (1997) proposes that addi-tional focus should be placed on risk management, the benefits of economies of scale versusoperational flexibility, managing the complexity arising from increased product variety, andthe design of incentive systems to facilitate production switching. Certain of these featureshave now appeared in more recent papers.
Schmidt and Wilhelm [66] (2000) examine the strategic, tactical, and operational issuesin supply chain models. For each level of decision making, the authors provide a literaturereview as well as typical solution methodologies employed. The authors find that (1) strategicmodels are normally concerned with facility location, capacity, and production technolo-gies; (2) tactical models seek to develop appropriate material flow management policies;and (3) operational models deal with coordination of activities to meet customer servicerequirements.
Prasad et al. [60] (2001) provide a breakdown of literature on international operationsstrategies from 1986 to 1997, classifying papers by region, solution methodology, author-ship, and cross-country affiliation, to name a few. They recommend that future researchshould focus on how internal changes such as turnover in senior management affect theimplementation of strategic operations, as well as how these firms cope when falling shortof performance targets.
Meixell and Gargeya [50] (2005) classify a collection of global supply chain design papersbased on four dimensions: decision variables (e.g., for facility and/or supply selection), per-formance measurement (nature of the objective function and constraints), supply chainintegration (e.g., number of supply chain tiers, type of coordination, and BOM specifica-tions), and global considerations (such as exchange rates, tariffs, duties, trade barriers, etc.).From their research, the authors recommend that future global supply chain design modelsshould include both the external and internal supplier locations, that more than one tierof the supply chain should be considered, a broadened definition of performance metricsshould be applied, and that case studies should explore different industry settings ratherthan exhausting the automotive, textile, or electrical manufacturing industries.
Each of these previous reviews stresses similar areas for future research, namely, improvinginternal coordination, expanding the unit of analysis to include external players, and estab-lishing appropriate targets with measurable progress. The firm’s goals need not be basedsolely on profit maximization or cost minimization, but rather, firms should consider otherfactors as Vidal and Goetschalckx [73] (1997) and Meixell and Gargeya [50] (2005) suggest.We will see in the following section how qualitative aspects are accounted for in some facilitylocation models.
3. Facility locationFacility location models pertain to the optimal location of a variety of facilities, includingdistribution centers (DCs), manufacturing plants, and finishing plants. A general facilitylocation model addresses open–close decisions for plants or DCs, how to allocate customerdemand to these plants, and the quantity of product to ship to each customer market(Gourdin [27] 2006). A global facility location model is extended by including factors suchas exchange rates, duties, tariffs, and local content rules. Each of these will affect how acompany positions itself internationally to better take advantage of tax incentives, lowerwage costs, and proximity to customers. Table 3 provides a summary of the selected facilitylocation papers. Note that although all papers have been classified according to severalcriteria, only the relevant columns for each topic are shown.
In Badri [3] (1999), the author combines his previous multicriteria integer goal program-ming model (Badri [2] 1996) with the analytic hierarchy process (AHP). Decision makers
Bookbinder and Matuk: Logistics and Transportation in Global Supply Chains186 Tutorials in Operations Research, c© 2009 INFORMS
Tabl
e3.
Loc
atio
nm
odel
s.
Glo
bal f
acto
rs
Aut
hors
Yea
rR
egio
nO
bjec
tive
Tim
epe
riod
Num
ber
ofpr
oduc
tsT
rans
fer
pric
esE
xcha
nge
Rat
esD
utie
s an
dta
riff
sA
dditi
onal
info
rmat
ion
Bad
ri [
2]19
96A
sia/
Oce
ania
Min
cos
tS
SU
ses
qual
itativ
e fa
ctor
s to
aid
inde
cisi
on m
akin
g
Can
el a
nd K
hum
awal
a[1
4]a
1996
Max
pro
fit
MS
��
Dev
elop
s a
0-1
MIP
for
a c
apac
itate
dan
d un
capa
cita
ted
inte
rnat
iona
lfa
cilit
ies
loca
tion
prob
lem
Bad
ri [
3]19
99A
sia/
Oce
ania
Min
cos
tS
SE
xten
ds th
e B
adri
[2]
mod
el to
incl
ude
the
AH
P an
d fo
rm a
mor
e sy
stem
atic
deci
sion
-mak
ing
proc
ess
Can
el a
nd K
hum
awal
a[1
5]a
2001
Max
pro
fit
MS
��
Pres
ents
a h
euri
stic
pro
cedu
re to
solv
e th
e C
anel
and
Khu
maw
ala
[14]
mod
el
Can
el a
nd D
as [
13]a
2002
Max
pro
fit
MS
��
Inte
grat
es m
anuf
actu
ring
and
mar
ketin
g de
cisi
on in
a g
loba
l con
text
Bhu
tta e
t al.
[6]a
2003
Max
pro
fit
MM
��
MIP
mod
el; i
nclu
des
deci
sion
s on
prod
uctio
n, d
istr
ibut
ion,
and
inve
stm
ents
Kou
velis
et a
l. [4
1]20
04M
ax N
PVM
M�
�In
corp
orat
es g
over
nmen
t sub
sidi
esan
d ot
her
ince
ntiv
es
Rob
inso
n an
d B
ookb
inde
r [6
3]a
2007
NA
FTA
Min
cos
tM
SM
IP f
or N
AFT
A s
uppl
y ch
ain;
incl
udes
mod
al d
ecis
ions
Ham
ad a
nd G
uald
a [3
3]a
Not
es. A
ll m
odel
s in
this
sub
set a
ssum
e de
term
inis
tic c
ondi
tions
. S, s
ingl
e; M
, mul
ti.
aPa
per
has
been
app
lied
to in
dust
ry/u
sed
indu
stry
dat
a.
2008
S. A
mer
ica
Min
cos
tS
M�
��
Atte
ntio
n to
val
ue-a
dded
taxe
s in
Bra
zil;
incl
usio
n of
take
-or-
pay
cost
s
b
b NPV
, net
pre
sent
val
ue.
Bookbinder and Matuk: Logistics and Transportation in Global Supply ChainsTutorials in Operations Research, c© 2009 INFORMS 187
assign weightings to qualitative global factors, such as political situations, global competi-tion and survival, economic-related factors, and government regulations, ranking what theydeem most important when considering which facilities to open or close. Badri [3] (1999)finds that this method provides consistency and a systematic methodology to solving thelocation-allocation problem.
Canel and Khumawala [14] (1996) present a thorough description of international factorsand costs to consider for both uncapacitated and capacitated international facility locationproblems. The authors apply their 0-1 MIP to an actual case involving a U.S.-based multi-national chemical company serving markets in North America, Europe, and the Far East. InCanel and Khumawala [15] (2001), a heuristic procedure is used to simplify this multiperiodMIP while treating qualitative international factors. By incorporating multiple periods, themodel is able to take into account price, cost, and demand variations while maximizingprofit.
Bhutta et al. [6] (2003) develop a mixed integer linear program to maximize profit anddetermine optimal capacities, investments, and production allocation for MNCs. They con-sider three scenarios, with varying numbers of facilities and countries, to examine the effectsof exchange rates and tariff rates on locating facilities and quantities to produce at eachplant. Bhutta et al. [6] show that tariff and distribution costs will offset production set upcosts, and that it is advantageous for the MNC to assign some initial capacities and inventoryto each plant to help decrease overall costs. Canel and Das [13] (2002) also consider exchangerates and tariffs in their profit maximization model, while realizing the interdependencythat exists between marketing and manufacturing decisions.
Kouvelis et al. [41] (2004) present a comprehensive model through the consideration ofnumerous realistic global factors. In addition to transfer prices, they examine special cases offacility location models in which government financing, tax incentives, local content require-ments (LCRs), and regional trading agreements are present. The goal of their model is tomaximize after-tax profits through the design of an optimal network of plants and DCs fora firm using a hybrid product-process focus. A numerical example is provided, depicting afirm with expected demand from markets around the world. Six cases are modeled, showingthe effects of financing, tariffs, local content, regional trading zones, and taxation. Withtheir all-encompassing model, Kouvelis et al. [41] provide the reader with many realistic sit-uations to which their model applies, assisting decision makers in choosing optimal facilitylocations and demand allocations under varying circumstances.
Robinson and Bookbinder [63] (2007) examine DC and finishing plant locations for afirm operating in the NAFTA region. Their multiperiod model also determines the flowof material and transportation mode to employ when shipping goods within Canada, theUnited States, and Mexico. Because distribution is contained within the NAFTA region,tariffs are negligible and border-crossing costs are included in transportation costs. Theauthors find that capacity will ultimately determine the degree of centralization of finishingplants and DCs in the system, additional facilities will be added to the network if demandexceeds capacity, and that more than one mode of transportation is necessary when thereis insufficient lead time.
Take-or-pay costs are incurred when buyers must pay for the total contracted volume of ashipment, even though that quantity may exceed the actual demand. Hamad and Gualda [33](2008) incorporate this concept and other traits, such as value-added taxes, in their locationmodel. The authors then apply it to a firm in the chemical industry based in South Americawith worldwide operations.
We note that all mathematical models in this section assume deterministic conditions,perhaps due to the strategic nature of the decisions being made. We would anticipate thatover a long time horizon, parameters that are subject to fluctuations (i.e., exchange rates,tariffs, and duties) will assume their expected values, and that employing random variableswould only complicate the model. If random variables were considered, this would increase
Bookbinder and Matuk: Logistics and Transportation in Global Supply Chains188 Tutorials in Operations Research, c© 2009 INFORMS
the difficulty of solving the problem, and the effort required to obtain representative datamay not contribute any additional insight.
Although many of these models solve location and allocation decisions simultaneouslybased on a number of quantitative and qualitative parameters, region-specific characteristicssuch as LCRs, free trade agreements, and value-added taxes are examined by only a fewarticles (e.g., Kouvelis et al. [41] 2004, Robinson and Bookbinder [63] 2007, Hamad andGualda [33] 2008).
4. ProcurementIn a typical procurement or sourcing problem, a firm wishes to purchase, from one or manysuppliers, components or raw materials and have them delivered to their various facilities.The selection of suppliers will depend on purchasing cost and the ability of the supplier tomeet the firm’s demand. An international sourcing problem tackles additional challengesthat arise due to the risks that may be involved. Diversification of a supplier networkcan result in many advantages, such as competitive prices and higher quality, but a firmmay incur increased costs when transporting goods from farther away when fluctuationsin exchange rates affect purchase prices of goods, or when paying duties and tariffs onimported components. There are few papers that address the supplier selection problem inan international context, as seen in Table 4 and concurrent with the findings of Meixell andGargeya [50] (2005).
Gutierrez and Kouvelis [31] (1995) provide the reader with both deterministic and stochas-tic models whose solutions are robust. The deterministic model generates different scenariosthat reflect fluctuating exchange rates and inflation. These solutions are then used as inputfor the stochastic model, eliminating the need for managers to assign probabilities to multi-ple exchange rate scenarios. The robust nature of the models makes them insensitive to thesefluctuations, and still able to generate a solution that is nearly optimal. In contrast, GonzalezVelarde and Laguna [26] (2004) incorporate exchange rates explicitly in their robust modelby generating a table of scenarios for weak, medium, and strong international economicregions and good, average, and bad economic states. Their mixed integer nonlinear programis solved using a heuristic based on Benders decomposition and tabu search.
Munson and Rosenblatt [55] (1997) examine the effects of LCRs on supplier selectionfrom several countries. They model a single plant situation with one local content rulein a deterministic environment and find that the firm will source at most one componentfrom two different suppliers. This model can accommodate additional LCRs in a multiplantsituation, though the authors note that if some components were produced in-house orobtained through local suppliers, the computational difficulties would be lessened.
Balaji and Viswanadham [4] (2008) develop a tax-integrated model to determine whetherit is in a company’s interest to outsource material or pursue foreign direct investment (FDI).They consider an eight-stage supply chain for the manufacture of computers, with operationsin two countries. To incorporate tax planning in their model, the authors include the taxincurred per tax lot of transferred goods as a parameter. A hub-based model is also proposed,where firms may source their components from a regional hub and benefit from lower taxes,lower costs, and higher-quality freight handling. In their numerical example, Balaji andViswanadham [4] show that it is optimal to outsource production to the country wherecost is lowest, even when demand exists in both countries. It is also in the company’s bestinterest to outsource production when demand is high and to use FDIs for echelons closestto the customer.
Few publications deal with procurement in a global setting. We discussed several of thosearticles and noted that a variety of factors were considered in each of them. Exchange rates,LCRs, and taxes have been incorporated in the development of sourcing policies to meet therequirements set by the decision makers. In contrast to the deterministic facility location
Bookbinder and Matuk: Logistics and Transportation in Global Supply ChainsTutorials in Operations Research, c© 2009 INFORMS 189
Tabl
e4.
Pro
cure
men
tm
odel
s.
Glo
bal f
acto
rs
Aut
hors
Yea
rM
odel
type
Obj
ectiv
eT
ime
peri
odN
umbe
r of
prod
ucts
Tra
nsfe
rpr
ices
Exc
hang
era
tes
Dut
ies
and
tari
ffs
Add
ition
al in
form
atio
n
Gut
ierr
ez a
nd K
ouve
lis [
31]
1995
Min
cos
tS
S*
�M
odel
dea
ls w
ith u
ncer
tain
ty in
cos
ton
ly, n
ot d
eman
d
Mun
son
and
Ros
enbl
att [
55]
1997
DM
in c
ost
SM
Can
inco
rpor
ate
mul
tiple
LC
Rs
Gon
zale
z V
elar
de a
ndL
agun
a [2
6]20
04M
in c
ost
SS
�
Solu
tion
met
hod
is a
heu
rist
ic b
ased
on
Ben
ders
dec
ompo
sitio
n an
d ta
buse
arch
Bal
aji a
nd V
isw
anad
ham
[4]
2008
DM
in c
ost
MM
Mod
el in
tegr
ates
tax
plan
ning
St
Not
es. A
ll m
odel
s in
this
sub
set a
ssum
e de
term
inis
tic c
ondi
tions
.
D
, det
erm
inis
tic; S
t, st
ocha
stic
. S
, sin
gle;
M, m
ulti.
*
Impl
icitl
y co
nsid
ered
.
St
Bookbinder and Matuk: Logistics and Transportation in Global Supply Chains190 Tutorials in Operations Research, c© 2009 INFORMS
models examined in §3, the optimality of a procurement network will depend heavily onthe realized values of the random variables considered. The most prominent of these are theexchange rates. Considerable importance is thus placed on choosing appropriate scenariosfor use in the corresponding stochastic program.
5. Transportation
5.1. IntroductionIn addition to facility location and supplier network decisions, a well-defined transportationplan is necessary to ensure an optimal flow of material throughout the supply chain. Unfor-tunately, transportation is sometimes the forgotten element in a supply chain model. Nodalactivities, such as fabrication, assembly, and inventory, are emphasized when the focus ison conversion of components to subassemblies and finished products. But how do thosematerials actually get from one node to the next? Can the timing or scheduling of thosemovements of goods be improved? Can the flow be accomplished at lower costs?
We feel that transportation should be accorded the same importance as the other oper-ations management activities. The benefits of good production-inventory decisions can beeasily overwhelmed by neglecting the economies of scale available in the various transportmodes.
Two of those modes, air and water, are the only ways to move goods between noncon-tiguous countries. Water transportation and air freight are automatically “global” in thatcontext. Useful references include Groothedde et al. [29] (2005) and Chang [17] (2008) oninternational shipments by water and air, respectively.
In this section, we will limit consideration to papers whose major focus is transportation,but within global supply chains. Morash and Clinton [54] (1997) have studied empirically thegeneral impacts of transportation in those cases. There is of course an extensive literatureon the international transport of goods. The subset of interest to us consists of articles inwhich the shipments are part of an overall supply chain model for the analysis of purchasing,manufacturing, or distribution.
5.2. Transportation by Road or RailOnce a network of facilities and suppliers is established, the flow of goods between theselocations and the mode by which they travel must be determined. A transportation modelseeks to minimize the cost, distance, or time associated with transporting goods from theirorigins to destinations by making appropriate modal choices.
Intermodal transportation occurs when a product is shipped using two or more differentmodes. For example, a shipment from a supplier in Hong Kong may travel to California byocean, and then be transloaded from the vessel onto a truck, where the remainder of itsjourney to various DCs will be by road. Intermodal transportation stresses the systematictransfer of such goods between different modes to minimize handling and downtime (Gourdin[27] 2006). In a global context, transportation is inevitably intermodal: products will betransferred at a port from vessel to rail, or at an airport from plane to truck. The landroutes between the NAFTA countries are somewhat unusual in permitting internationalintermodal shipments by truck and rail.
Bookbinder and Fox [9] (1998) analyze the flow of material in the NAFTA region, specif-ically from Canada to Mexico. The authors develop potential intermodal routes from fiveCanadian cities to three Mexican cities and show that shipments from central Canada arethe most direct, regardless of the destination city. Also, they observe that coastal originsrequire more efficient intermodal combinations. Potential future service expansion couldinclude a link by water between Vancouver and Manzanillo, a port on the west coast ofMexico, as well as a direct rail link between Calgary and El Paso, Texas.
Bookbinder and Matuk: Logistics and Transportation in Global Supply ChainsTutorials in Operations Research, c© 2009 INFORMS 191
Most “regional” models concern the logistical activities within a given free trade zone, forexample, flows from Canada to Mexico in a NAFTA supply chain. McCray and Gonzalez[48] (2008), however, have noted that those two countries could play a wider role because ofacute capacity shortages and occasional labor disruptions at American West Coast ports—international shipments, destined to the United States, might take advantage of the NAFTArelationship. Intermodal terminals on the Pacific coast of Canada or Mexico would requirean expanded infrastructure. Only then would transit through those ports enable containershipments to bypass the possible bottlenecks in California or Washington.
European rail service is at a crossroads. Environmental concerns and the fully exploitedcapacity of the road network strongly indicate greater use of rail–truck intermodal trans-portation. The EU thoroughly advocates this. There is now generally also a distinction ineach country between the national ownership of a track and the right to operate rail serviceson that track.
With the EU’s enlarged membership now incorporating 27 countries, including some inEastern Europe, there is potentially a new north–south corridor on which rail intermodalservice could expand. Nair et al. [57] (2008) assess the market potential for this. Kuo et al.[42] (2008) recommend that this expansion be facilitated by the collaboration of multiplerail carriers on what is termed the REORIENT corridor (11 countries, southward fromScandinavia to Greece). The mechanism proposed by those authors is that of “train-slotcooperation.” Furthermore, Janic [36] (2008) suggests that long intermodal freight trains(LIFTS) be operated as a block within Europe, potentially through this same corridor.
5.3. Transportation by AirWarsing et al. [76] (2001) study the economic benefits of using a dedicated multimodalcargo facility (DMCF) as opposed to a traditional passenger-based airport. These authorsseek to minimize delays that result from congestion in the flow of materials. The DMCFacts as a hub, where a firm can locate some of their logistical activities (e.g., warehousing,production/distribution) closer to the cargo handling and transfer facilities, minimizing theamount of disruption in material flow that would normally occur otherwise. The modelconsists of a network of regions, each with demand destinations and supply points. Thelatter act as transshipment points for interregional shipments; each supply point also servescustomers within its own region. The objective is to minimize total transportation costs bydeciding the quantity of products to ship to customers, the aircraft by which products willtravel, and the amount of inventory at each customer’s facility at the end of the period.The authors assume that there is a significant distance between each region that requiresinterregional shipments to go by air, there is a limited number of aircraft available in eachregion, and that the products require “same-day” delivery. To calculate the shipping leadtime, a queueing approach is used to determine the successive waiting times experiencedbetween the manufacturer’s dock and the product’s interregional departure, and then to theproduct’s arrival. Warsing et al. [76] determine that using a DMCF can help reduce delays inmaterial shipment and decrease holding costs compared to those at a nondedicated facility.
Tyan et al. [71] (2003) also take a multimodal approach to global transportation, butfrom the view of a third-party logistics provider (3PL) that must make decisions regardingconsolidation of material and select alternative flights for shipment. In this model, the 3PLmust consolidate the manufacturer’s orders (loose cartons or skids of 40 cartons) in a waythat will minimize costs without compromising customer service. Tyan et al. [71] proposethree consolidation policies, where physical consolidation of material takes place at bothcargo and express terminals. Policy differences relate to degrees of just-in-time (JIT) andcost–service trade-offs. Their findings, consistent with other studies of shipment consoli-dation in a nonglobal context, suggest that collaboration between 3PLs and shippers willbenefit both parties. Customer service performance can improve, and total costs decrease,when the 3PL may exercise some consolidation decisions.
Bookbinder and Matuk: Logistics and Transportation in Global Supply Chains192 Tutorials in Operations Research, c© 2009 INFORMS
Chang [17] (2008) proposes a multiobjective, multicommodity flow problem to determineoptimal routes for an international intermodal carrier. His model comprises three essentialyet complicating characteristics of intermodal routing: (1) the multiobjective nature of theproblem, because shippers may want to minimize costs and/or travel time; (2) scheduledtransportation modes and delivery times to ensure feasible routes are being considered; and(3) the effect of economies of scale on transportation cost. To solve the model, the authorsapply Lagrangian relaxation techniques and decompose the problem into two subproblems:one to minimize travel time and cost of material flow, and the other to minimize travel costfor flow within a specified range. Once a solution is found, the problem is reoptimized toensure the final solution is feasible. Chang [17] provides the reader with a numerical examplefor a distributor of LCD monitors, shipping from Taiwan to the United States.
Additional references include Neiberger [58] (2008) on European air freight shipments,and Huang and Chi [34] (2007). The latter authors approach the consolidation problem facedby air freight forwarders by initially transforming it to a set covering problem: a feasibleconsolidated shipment is treated as a set. Lagrangian relaxation is employed in a heuristicalgorithm.
5.4. Other Transportation ModelsErera et al. [23] (2005) study the movement of intermodal tank containers used in thechemical industry. A tank container operator must manage a fleet of containers as they travelby road, rail, and/or sea, and must take into account that most customers are requestingone-way trips to get their order from point A to point B. To improve utilization of these tankcontainers, the operator should be aware of the imbalance in shipping patterns, where certainregions, such as North America, tend to be net sources of containers, whereas others, such asEast Asia, tend to be net sinks (see McCalla et al. [47] 2004). Empty repositioning helps tocorrect the geographical and temporal imbalances by moving these empty containers betweendepots. The authors develop a deterministic multicommodity network flow model to find theoptimal routing of empty and loaded containers that will minimize total transportation anddepot costs. A computational study is performed to examine the effects of three strategies:(1) weekly repositioning, where depot-to-depot repositioning is allowed only on the first dayof the week; (2) bounded daily repositioning, where depot-to-depot repositioning is allowedevery day, but if this option is exercised a lower bound is imposed to deter numerous smallshipments from being made; and (3) unbounded daily repositioning, where repositioning isallowed every day with no restrictions. The authors find that Option 3 provides the lowestcost, because timing has more of an impact on total cost than does the number of containersto reposition.
Groothedde et al. [29] (2005) propose a hub-based network for intermodal transportation.Trucks deliver products from the manufacturer to the hub, and from the hub to the prod-uct’s final destination; barges are used for interhub transportation in this application in TheNetherlands. The general hub network design problem is relaxed so that not all hubs areinterconnected, direct shipment of material between facilities is allowed, and facilities canbe assigned to more than one hub. In their model, the authors seek to minimize total trans-portation cost by determining the optimal location of hub facilities, the assignment of originand destination facilities to the hubs, linkages between hubs, and the flow of material throughthe network. The hub network allows for economies of scale through the consolidation ofmaterial traveling between hubs, offsetting extra costs incurred from increased shipping dis-tance. Because transportation by sea is slower than by truck, well-forecasted demand shouldbe shipped by barge through the hub network, whereas unpredictable demand should betransported directly by truck from origin to destination. In this way, customer service is notjeopardized, and firms can take advantage of lower shipping costs. Groothedde et al. [29]propose that further collaboration between firms, such as sharing resources and information,can lead to benefits for both parties.
Bookbinder and Matuk: Logistics and Transportation in Global Supply ChainsTutorials in Operations Research, c© 2009 INFORMS 193
6. Production Switching, Transfer Pricing Policies,and Postponement
With the implementation of optimal transportation policies, the nodes in the supply chainare linked by air, water, rail, and road. In §§3–5, we examined the decisions that a companyconsiders when developing its global supply chain. In this section, we focus on the productionof material in an established logistics network. A company with facilities located worldwidewill inevitably encounter fluctuations in demand, exchange rates, and production costs.Depending on the direction of these variations, it may at times be more profitable to producein only one, all, or a combination of some countries. The papers in §6.1 address the issueof production switching, when companies must evaluate whether moving production to alocation with lower wages and a more competitive exchange rate will offset the setup costsassociated with this shift.
Another important issue in international supply chains is transfer pricing. When divisionswithin an MNC transact with one another, a transfer price set by the MNC is charged bythe selling division to the buying division. Because this price will affect profits and expensesreported by each business unit, and with the incorporation of varying exchange rates, it isimportant to set an appropriate transfer price that adheres to government regulation. Thepapers in §6.2 include transfer pricing in their models to evaluate its effect on productionscheduling and company profits.
In §6.3, we address the concept of postponement. Rather than transfer the completeproduction of an item between two or more facilities, a postponement strategy allows thefirm to manufacture a generic product at a number of its facilities. The items are thenshipped to their respective destinations where further value-added processes may take placeto customize the goods.
6.1. Production SwitchingKogut and Kulatilaka [39] (1994) present a model to assess the value of having productionfacilities in two different countries. As exchange rates fluctuate, to which country should acompany shift production to minimize their costs? The authors develop a model to determinethe number and location of facilities while considering variability in real exchange rates.This decision is complicated by a number of costs associated with production shifting, suchas shutdown, start-up, and transportation costs. In general, a plant should be built if thevalue of its flexibility exceeds the initial investment required.
Huchzermeier and Cohen [35] (1996) also examine the value of production flexibility foran international firm by using a hierarchical approach. First, the authors model exchangerate fluctuation, then the firm’s global manufacturing strategy is used to determine poten-tial product and supply chain network designs, and finally, switching costs are studied.Huchzermeier and Cohen [35] develop a stochastic dynamic program that examines differentexchange rate scenarios to determine the number and location of facilities as well as theallocation of demand. A subproblem is also solved to calculate the optimal quantities ofproduct shipped from suppliers to plants and plants to markets to optimize after-tax profits.
Unlike Huchzermeier and Cohen [35] (1996), Dasu and Li [22] (1997) propose a deter-ministic model, as they consider the long-term value of locating facilities internationallyand using excess capacity to hedge against variability. The authors present a model for twofacilities located in two different countries, and determine the optimal amount and time toproduce to minimize costs when the firm is subject to switching costs and exchange rates.This model is then extended to consider n countries, as well as a firm that serves multiplemarkets. Dasu and Li [22] prove that, when switchover costs can be modeled as linear orstepwise functions, regardless of the number of markets that the firm serves, the optimalpolicy is always the barrier policy; i.e., either the upper or lower barrier specified in theavailable options will be the optimal policy.
Bookbinder and Matuk: Logistics and Transportation in Global Supply Chains194 Tutorials in Operations Research, c© 2009 INFORMS
Mohamed [53] (1999) studies the effects of varying inflation and exchange rates on anMNC’s operational decisions. In this model, the author determines the products to produceat each facility, and the market that this facility will serve to minimize total costs. A detaileddescription of relevant costs is provided. In numerical examples, the author finds that thetotal number of units produced and the required capacity for a facility are independent ofthe exchange rate variation, and that the highest profit is made when companies begin withoptimal capacities. Mohamed [53] also examines the sensitivity of profit to the exchangerates when there is a capacity limitation.
Li et al. [43] (2001) propose a stochastic control model to decide which facilities to operateunder varying exchange rates and uncertainties in demand. The authors consider a two-country production network that produces a single product and allows backlogging for thoseorders that are not met. Their results show that for an exchange rate below the intersectionof the two switching cost curves, the primary facility is the overseas plant, and the secondaryfacility is the home plant. When inventory is sufficiently high, neither plant is operated,but as inventory drops, the overseas unit will begin operations, followed by the home plant.When the exchange rate is above the intersection of the two switching curves, the scenariois reversed, as the home plant and the overseas plant become the primary and the secondaryfacilities, respectively. Li et al. [43] extend their model to consider multiple plants in eachof the two countries and in multiple countries.
To manage production when facing exchange rate uncertainty, Kazaz et al. [38] (2005)propose the application of two operational hedges. As managers receive new informationthroughout the production period, these hedging strategies allow for adjustments to theproduction plan that either react to (allocation hedging) or act in anticipation of (productionhedging) varying exchange rates. The problem is divided into two stages. The firm mustdecide (1) how much to produce to satisfy demands in two different markets and (2) howmuch of that production to allocate to each market. The objective is to maximize the firm’sexpected profits subject to transportation and demand constraints. Kazaz et al. [38] extendtheir single-period model to a multiperiod scenario and find that, in each of these cases, theoptimal solution applies both allocation and operational hedging. The authors also examinethe effect of uncertain demand, the timing of price-setting decisions, and their combinedeffect.
Wu and Lin [78] (2005) also consider exchange rate uncertainty in their entry–exitdecision-making model. The model consists of a single exporter that must determine theoptimal entry or exit threshold value at which domestic production should be transferredoverseas, subject to real exchange rates and their volatility, tariffs, wages, and prices of rawmaterials. The objective is to maximize the value of the exporter’s products through theoptimization of the quantity of raw materials and labor required.
6.2. Transfer Pricing PoliciesKouvelis and Gutierrez [40] (1997) incorporate transfer prices and exchange rates into theirtwo-market global newsvendor problem, obtaining the optimal quantities of items to shipfrom the primary market to the secondary market and the production quantities to beproduced by the secondary market. The decisions to be made are from the viewpoint of thecorporate planner, who must consider shortage costs attributed to lost sales and overagecosts attributed to selling items at their salvage value at the end of their selling season.
Kouvelis and Gutierrez [40] also examine a decentralized policy, where each division ofthe firm is responsible for making purchasing and production decisions. This arrangementallows the divisions to act independently but requires more coordination in deciding thetransfer prices applied to goods shipped between divisions. In cases where transfer pricesare constant across the firm, this could lead to suboptimal policies, so the authors suggestthat “more sophisticated transfer pricing mechanisms” be used (Kouvelis and Gutierrez[40] 1997). These mechanisms include firm-controlled transfer prices, when a constant and
Bookbinder and Matuk: Logistics and Transportation in Global Supply ChainsTutorials in Operations Research, c© 2009 INFORMS 195
agreed-upon price is applied to shipments between profit centers, and intermediate purchas-ing coordination. This latter mechanism imitates a centralized transfer pricing policy on asmaller scale, because one of the firm’s centers is responsible for coordinating the trade ofgoods between two of the firm’s other profit centers. This allows each of the units to actindependently while maximizing the overall profit of the firm.
Vidal and Goetschalckx [74] (2001) formulate a model to assist MNCs in setting transferprices that will maximize their net income after tax, as well as ensure that the performancetargets of each division are met. In addition, their model decides how transportation costsshould be allocated, the transportation mode employed on each arc, and the quantity ofmaterial to ship from one plant or DC to another. The authors make a distinction betweeninternal and external suppliers: transfer prices do not apply to the latter case, but rathermarket prices will influence how much those suppliers will charge the MNC.
Miller and de Matta [52] (2008) develop a global profit maximization plan that determinesthe optimal production policy for each plant in the firm’s network, as well as a sourcingand distribution plan with associated transfer pricing policies. Their model maximizes profitwhile considering tax and exchange rates in each country and the lower and upper bounds ontransfer prices applied to intramarket and market-to-customer transactions. The nature ofthis model is such that it can be used for tactical or strategic production planning, becausethe firm can evaluate the robustness of the plan and adjust it as necessary in the short runor analyze the effect of exchange rates on the plan over a number of years.
Villegas and Ouenniche’s [75] (2008) model seeks to maximize the profit of an MNCwith a thorough consideration of many global factors, including exchange rates, tariffs, dutydrawbacks, income tax rates and credits, royalties, and transportation costs. The modeldetermines the optimal quantities of the products to produce and sell for each division ofthe firm, the allocation of transportation costs to each division, and the associated transferprices. The results of this model suggest that optimal transfer pricing and transportationcost allocation policies occur at their upper or lower bounds when there is a low level ofexchange rate risk, and that trade quantities and transportation cost allocations do notaffect transfer prices.
6.3. PostponementHadjinicola and Kumar [32] (2002) examine eight different manufacturing and marketingoptions for a firm producing in two countries. The options consider the degree of customiza-tion and centralization of the firm, as well as the prevalent production policy. Core produc-tion options, referring to the manufacture of a core poroduct in one of the two facilities,dominate the other strategies, because the firm achieves economies of scale and still has theflexibility to customize the product further along its supply chain.
Yang and Burns [79] (2003) identify the decoupling point, degree of demand uncertainty,costs, and capacity requirements as the main components of the existing supply chain thatwill be affected by a postponement strategy. In a later article, Yang et al. [80] (2004) proposea framework to determine whether adopting postponement is ideal based on the reliabilityof demand forecasts, nature of the organizational culture and structure, and the degree offlexibility in other areas of the supply chain. Yang et al. [79, 80] have outlined a number ofgeneral principles on postponement that are clearly applicable in the global context.
Employing postponement may influence the proportion of locally and globally sourcedcomponents. Jin [37] (2004) determines that understanding the nature of demand, impor-tance of information and manufacturing technology, vicinity of subcontractors, and relation-ships with suppliers will help to achieve a balance in the sourcing policy that maximizesprofits.
Prasad et al. [61] (2005) conduct a survey, comparing make-to-stock (MTS) and build-to-order (BTO) supply chains in developed and developing countries. The authors concludethat BTO systems are preferable when there is uncertainty in demand, requiring individual
Bookbinder and Matuk: Logistics and Transportation in Global Supply Chains196 Tutorials in Operations Research, c© 2009 INFORMS
items to be tracked along the supply chain to ensure the customers’ requests are satisfied.Developing countries generally have less reliable lead times and tend to employ coordinatedreplenishment policies for cost savings, resulting in an environment more conducive to anMTS system.
The models in §§6.1 and 6.2 are perhaps the most mathematically involved because theyaccount for the variability of global economic factors. As companies strive to meet theirperformance targets, there is a growing importance placed on developing and implementingappropriate transfer pricing and production switching policies that maximize profit for eachdivision of an MNC. Furthermore, the implementation of postponement on a global scalecan influence the policies of the firm due to challenges stemming from demand uncertaintyand differences in organizational processes and practice.
7. Overall Network DesignPrevious sections discussed each functional area of the supply chain on its own. The papersbelow assimilate these functions into one strategic model that can be used to optimize anentire logistics network, deciding optimal facility locations, product-sourcing and manu-facturing plans, and transportation flow. Compared to papers discussed previously, thesemodels have a greater scope, and hence are able to incorporate many global features (as inArntzen et al. [1] 1995) and specific aspects of regional trade (as in Wilhelm et al. [77] 2005).
Arntzen et al. [1] (1995) develop a comprehensive model for the Digital Equipment Cor-poration supply chain that determines the production plan and location of facilities and thequantity of products to produce and ship. Inputs to the model include the global BOM,demand, LCRs, and a variety of costs. Arntzen et al. [1] use an objective weighting factorto integrate both time and costs in the objective function. The scope of this model hasallowed those authors to apply it to a variety of studies involving sourcing and distributiondecisions, revision of manufacturing infrastructure, and repairs and service improvements.
Camm et al. [12] (1997) incorporate a geographic information system (GIS) in their overallsupply chain model for Procter and Gamble. The problem is decomposed into a distribution-location model and a product-sourcing model. The visual interface of GIS allows usersto identify sets of optimal sourcing solutions by inputting product type, potential plantlocations, and transportation modes. The output there can in turn be employed as inputfor the product-sourcing model, which is then used to obtain the optimal manufacturing-distribution plan.
Dasu and de la Torre [21] (1997) examine a network of partially owned and affiliatedtextile facilities in Latin America. The corporate firm faces competition from the externalmarket, but because the affiliates operate in the same region and independently of oneanother, any intrafirm competition could potentially hinder the overall performance of theMNC. Depending on the level of collaboration between the firms, the optimal pricing policiesfor the firms will vary. To capture the dynamics of the MNC’s profits, the authors developa model that fosters intrafirm cooperation when each of the firms faces direct externalcompetition, and another that examines intrafirm rivalry. Each model seeks to maximizeprofits by calculating optimal sales prices and quantities for each firm.
Tsiakis et al. [70] (2001) examine an entire global supply chain network where decisionsneed to be made regarding the number, location, and capacity of warehouses and DCs,the allocation of demand, and the production rate and shipment quantities of materialto minimize cost. The authors apply their model to two case studies, where demand isdeterministic and then uncertain.
Sheu [67] (2004) uses a fuzzy AHP in combination with the fuzzy multiattribute decision-making (MADM) techniques to help identify global logistics strategies for companies. Thismodel is useful in realistic situations where conditions are uncertain or vulnerable to risk,
Bookbinder and Matuk: Logistics and Transportation in Global Supply ChainsTutorials in Operations Research, c© 2009 INFORMS 197
allowing users to subjectively rank decision alternatives according to specific performancecriteria, thus generating an optimal solution that meets the requirements of decision makers.
Contesse et al. [19] (2005) model the purchase and distribution of natural gas in SouthAmerican countries. Their MIP assists a Chilean firm in deciding optimal volumes of naturalgas to purchase from a selection of suppliers and the method of transport. The model isintended for operational planning, allowing the user to update the model daily or as demandinformation becomes known with certainty. The authors extend their deterministic modelby developing a robust optimization approach to consider demand-varying scenarios.
Nagurney and Matsypura [56] (2005) examine the effect of risks and uncertainty on aglobal supply chain network. A three-tier supply chain is considered where electronic ordersare allowed, meaning that retailers have a direct link to their manufacturers, essentiallybypassing the distributors. There are risks associated with each tier, such as political andexchange rate risk for manufacturers and distributors, or demand uncertainty for retailers.In light of this, the authors have developed a multicriteria model such that the objectiveis to minimize risk and maximize profit. Goh et al. [25] (2007) present a similar model,considering the variability in exchange rates, tariff rates, tax rates, and market demands.That model suggests the optimal number of facilities to open or shut and the quantities ofproduct to ship to customers.
Santoso et al. [65] (2005) use the sample average approximation strategy to handle thestochastic nature of particular variables in their supply chain network model. They chooseas random variables the transportation and processing costs, demand, supplies, and capaci-ties. The model determines which processing centers to build, which processing and finishingmachines to procure, and the routing of products from suppliers to customers. The objectiveis to minimize the current investment costs and expected future processing and transporta-tion costs.
Wilhelm et al. [77] (2005) present an intricate model specific to the NAFTA region.The model considers a multitude of variables to capture the reality of international supplychains. To maximize after-tax profits, the firm must decide the amount of material that isbackordered, the number of components that are shipped or held in inventory, by whichmode the products will be transported, and the associated transfer prices. Influencing thedecisions are costs (fixed, transportation, backorder, and material), capacity, tariffs, taxrates, LCRs, BOM restrictions, income taxes, and exchange rates. The model is appliedto 11 different cases, examining the effect of centralized/decentralized decision making,outsourcing versus in-house assembly, flexible versus dedicated technologies, and economiesof scale versus economies of process.
The preceding models consider many global factors and characteristics in the develop-ment of optimal overall supply chain networks. Although LCRs, BOMs, and exchange ratevariations are typical in any international manufacturing setting, we note that only Wilhelmet al. [77] (2005) present a model specific to a free trade zone (NAFTA). Other models con-sider general trade situations between international partners and do not encompass some ofthe particular traits of a given free trade agreement.
8. Other PapersThe literature review thus far has concerned normative models only. In this section, wesummarize several empirical studies that have been done, relating to global logistics andtransportation. These studies employ a variety of data collection methodologies to gainadditional insight on a number of international situations, such as the effects of globalization,reasons for and against expanding sourcing networks, and comparisons of regional logisticspractices. Our discussion is organized by the type of data used to analyze these issues.Although the technicalities of empirical research are not the focus of this study, we do notesome general observations from our own experiences and the articles cited in this section.
Bookbinder and Matuk: Logistics and Transportation in Global Supply Chains198 Tutorials in Operations Research, c© 2009 INFORMS
8.1. Cross-Sectional DataWell-structured surveys, questionnaires, or personal interviews offer informative answers toaid in a researcher’s study. The two former options are less intrusive and usually require lesstime to complete, whereas the latter can be used to elicit higher-quality answers that mayshed light on otherwise ambiguous topics. These methods, however, typically examine issuesat a cross-section in time, and are therefore widely used for comparisons and identificationof areas for further research.
Lynch et al. [45] (1994) conduct a Delphi study to forecast the future of logistics inCanada based on responses from industry experts. Their survey contains seven sections:international, transportation, industry, logistics managers, environment, government, andtechnology, whose conclusions must be forgiven the 15 years’ experience since then.
Bolisani and Scarso [8] (1996) carry out a survey of Italian clothing firms to determinewhat factors influence the internationalization of these organizations. The authors find thatmost companies are driven to reduce costs by locating their facilities closer to customer andsupplier markets, and by seeking lower-wage but highly skilled labor. To maintain the samelevel of profitability, these firms generally prefer centralized coordination, where the headoffice is actively involved in ensuring each that division meets the corporatewide goals.
Das and Handfield [20] (1997) investigate the feasibility of using JIT sourcing at a globallevel. The authors survey North American firms that have either domestic or global suppliers,some of which are JIT suppliers. In their research, the authors find that domestic JITsuppliers and global non-JIT suppliers tend to provide superior product quality over theircounterparts, whereas all JIT suppliers provide more reliable and frequent deliveries. Firmsthat transition from domestic to global JIT suppliers will realize a reduction in suppliercosts; however, the reverse is true for non-JIT suppliers.
Morash and Clinton [54] (1997) study the differences in transportation capabilitiesbetween firms in the United States, Korea, Japan, and Australia. Their survey focuses onseven areas related to transportation logistics in industry: time compression, reliability, stan-dardization, JIT delivery, information systems support, flexibility, and customization. Theauthors find that both American and Japanese firms place a higher importance on the reli-ability of transportation logistics than any of the other dimensions. Korean firms are morelikely to improve their information support systems and prefer a centralized managementstyle, whereas Japanese firms will focus their efforts on integrating their external supplychain with the internal activities. Whereas Japanese firms have more formal relationshipswith their partners, Australian and U.S. firms will take a more interactive approach, sharinginformation and resources with their customers and making site visits more frequently.
MacCarthy and Atthirawong [46] (2003), similar to Lynch et al. [45] (1994), use the Del-phi method to survey members from academia, the government, and consultancies aboutfactors affecting location decisions in global supply chains. In the first part of their study,the authors find that access to low labor costs and highly skilled labor are major motivationsfor companies when determining facility locations. In the second part, the influencing fac-tors developed by the panelists are ranked, with costs, infrastructure, labor characteristics,government and political factors, and economic factors being the five most important fordecision makers to consider.
Steenhuis and De Bruijn [68] (2004) develop an assessment tool to support facility locationdecisions. Using the gross domestic product (GDP) per working person and knowing thenature of the industry, companies, as well as regional governments, can assess the feasibilityof having a firm locate one or more of its facilities in their area.
McCalla et al. [47] (2004) examine the differences in containerization between NorthAmerica, Europe, and East Asia. In terms of geography, because North America is muchlarger and continental, rail is the predominant mode of choice for long-distance transporta-tion, whereas there is a tendency in East Asia and Europe to use water transport. Differencesin economic development lead to imbalances in imports and exports: because North America
Bookbinder and Matuk: Logistics and Transportation in Global Supply ChainsTutorials in Operations Research, c© 2009 INFORMS 199
imports more than it exports, a large number of empty containers eventually build up,whereas the opposite is true for East Asia. Institutional constraints also affect the reliabilityand service that companies are able to provide in each area. In North America, containerizedshipments are now subjected to security inspections at the border, following the “10+2”initiative. An importer security filing is sent to U.S. Customs and Border Protection priorto the container’s physical arrival at the border, comprising 10 elements of information per-taining to the importer, seller, buyer, manufacturer, and consolidator of the shipment. Twoadditional pieces of information that can also be provided are the vessel stow plans andcontainer status messages (U.S. Customs and Border Protection [72] 2009). The result willstill be less delay than in Europe, where some delays exist because of mandatory locomotiveswitching and crew changes enforced at the country borders. High levels of bureaucracy inEast Asia make delays unavoidable because intermodal transportation of goods does nothold as much priority as passenger transport.
Carbone and Stone [16] (2005) focus on growth and relational strategies of 3PLs in Europe.Through personal interviews, examination of annual reports, press, and quality reviews,the authors show that mergers and acquisitions are commonly used by 3PLs to be able toserve a larger geographic market and to take advantage of economies of scope and strategicsynergies. Relational strategies are also used to develop vertical or horizontal alliances.A vertical alliance involving 3PLs and their customers can put the 3PL at risk if there istoo much dependence on a given customer for continued business. In contrast, a horizontalalliance involves only 3PLs, allowing costs and risks to be spread out among the firms.
8.2. Case StudiesTypically, a case study examines a problem faced by a particular company or industry overa certain period of time, where some of the relevant data are obtained through secondarysources. We find, however, that many of the case studies we examine have supplemented theirinitial data collection with unstructured interviews or site visits, providing more meaningfulinformation for analysis. These studies, compared to the cross-sectional data studies above,tend to seek answers to more specific problems and are able to draw more definite conclusionspertaining to the effects of certain initiatives on a single firm or an industry.
Robertson et al. [62] (2002) conduct a case study of a steel manufacturer serving firmsin Australia, New Zealand, and Southeast Asia. The authors detail how a “global salesand operations optimizer” is synchronized with the master production schedule, achievingeffective coordination of information, and resources across the tiers of the supply chain.
Meijboom and Voordijk [49] (2003) approach globalization from a different angle, inter-viewing companies in The Netherlands to find out why they have opted to maintain aregional presence despite the growing trend to internationalize. The authors discover that thepolitical and economic environment of the domestic region and proximity to the consumermarket persuade companies to remain in Western Europe.
Liu and Young [44] (2004) develop an information technology model to coordinate manu-facturing decision making. Their model is integrated with two existing information models toallocate production tasks, support producer/consumer relationship coordination, and ensurethat all manufacturing activities are synchronized.
Buxey [11] (2005) interviews senior executives of Australian textile, clothing, andfootwear (TCF) firms to examine how these industries have responded to globalization. Theauthor identifies six different strategies used by the three industries: low cost, market access,global, defender, importer, and exporter.
Stratton and Warburton [69] (2006) examine the implications of outsourcing in three casestudies. The authors find that in each case, as companies transition to offshoring to takeadvantage of reduced production and labor costs, difficulties arise in the reliability of the newsuppliers and increase in lead times. The authors suggest that those trade-offs in outsourcingshould be explicitly considered when evaluating the option to source abroad. This proactive
Bookbinder and Matuk: Logistics and Transportation in Global Supply Chains200 Tutorials in Operations Research, c© 2009 INFORMS
approach may ease the transition to outsourcing and avoid any inconsistencies in expectedsavings and performance.
Rudberg and West [64] (2008) detail a case study for coordinating manufacturing net-works. The “model factory” concept provides standardized guidelines to show the best pos-sible network of factories. Using this network, responsibilities are assigned to each plant, andcompetence groups are established to ensure continued information sharing and knowledgetransfer.
These empirical studies establish useful foundations for normative research, because theknowledge and insight gained from personal interviews and questionnaires can help createmodels more applicable to specific industry problems. From the issues studied, we see thatregional differences account for many of the challenges encountered in global logistics. Lan-guage, regulations, geography, and local customs are all factors that affect the efficiency ofproduction, transportation, and marketing in a given international supply chain. Researchersencourage streamlining processes and sharing meaningful information to improve supplychain coordination.
9. Regional ModelsAlthough the majority of papers in our review apply to global logistics in a general sense,there are a number of models that deal with variables and constraints that are unique tospecific regions. Below we highlight characteristics of four normative regional models. Othersinclude Badri [2, 3] (1996, 1999), Dasu and de la Torre [21] (1997), Dasu and Li [22] (1997),Tsiakis et al. [70] (2001), Tyan et al. [71] (2003), Santoso et al. [65] (2005), Robinson andBookbinder [63] (2007), and Hamad and Gualda [33] (2008).
9.1. North AmericaWilhelm et al. [77] (2005) include a number of NAFTA-specific constraints in their networkmodel. For a laptop manufacturer with suppliers in the United States, Mexico, and China,the authors present 11 cases to model the key logistics decisions. The constraints includeLCRs that are imposed on components moving within the NAFTA region, upper and lowerbounds on transfer prices, border-crossing costs, and transportation cost allocations. Thesafe harbor rule is also modeled to account for the tax credit that will be received by aneligible maquiladora (generally, a labor-intensive plant in northern Mexico).
9.2. South AmericaContesse et al. [19] (2005) include take-or-pay costs in their MIP model, which are also seenin Hamad and Gualda [33] (2008). The buyer is required to pay this amount for a contractedvolume of natural gas, regardless of customer demand. Penalties against the buyer apply ifdaily minimum purchase levels are not met, or maximum levels are exceeded. The contractentered into by the supplier and buyer is such that it persuades the buyer to maintain ahigh purchase volume. The model of Contesse et al. [19] considers the flow of natural gasfrom suppliers in Argentina to a distributor in Chile and accounts for the lack of storagefacilities, a problem that is not normally faced in North America, as the authors mention.
9.3. EuropeGroothedde et al. [29] (2005) develop a hub network based in The Netherlands and connect-ing to Germany and Belgium. Though the authors do not explicitly discuss regional tradingpractices or incorporate international constraints, as seen in Wilhelm et al. [77] (2005), theydo create a model that recognizes common shipping practices in Europe. Transportation byinland waterway is paramount in their model, a feature that is distinctive in Europe butnot as widely available in continental regions, such as North America.
Bookbinder and Matuk: Logistics and Transportation in Global Supply ChainsTutorials in Operations Research, c© 2009 INFORMS 201
9.4. Asia and AustraliaSheu [67] (2004) captures the current trends in Asian logistics, having based his model ona survey of Taiwanese manufacturers. The model compares six global logistics strategiestypically employed in the integrated circuit industry, using a fuzzy AHP and fuzzy MADMto determine which strategy is best for a company to implement. These strategies involve theway products and information are transferred through the entire supply chain. Sheu [67] findsthat Taiwanese firms place a high priority on management control and core competitiveness,and less priority on the firm’s response to its external environment.
10. Critique and ProspectsIn earlier sections of this tutorial, the references that we cited have been discussed underone or more categories. Although the classification of a given article may not be clear cut,an overall assessment of the literature can be seen in Tables 1–8. Here, from those sametables, we will note some opportunities for further research. It should be kept in mind,however, that a particular cell may be empty for technical reasons. That is, perhaps onecould formulate a more complicated model containing additional global factors; solution ofthat model may be inhibited by the absence of appropriate bounds or a good decompositionscheme.
Table 1’s matrix, of primary topic and region of focus, is necessarily incomplete. That tablesimply summarizes those articles that we wished to cite in this tutorial. As just one example,there are clearly many published applications of transportation (such as vehicle routing)in Europe, North America, and elsewhere. Table 1, and our entire tutorial, emphasizestransportation when the materials moved are part of an international supply chain. Similarremarks pertain to most columns in Table 1.
Table 2 did not require as much selective judgment on our part. References such as Vidaland Goetschalckx [73] (1997) and Schmidt and Wilhelm [66] (2000) obviously deal withglobal supply chains. On the other hand, Boone et al. [10] (1996) and Pontrandolfo andOkogbaa [59] (1999) pertain to international operations and global manufacturing; they havewhat we feel is a high degree of overlap with our topic. We know of other review papersthat overlap to a lesser degree, but with further study of a subset of their subject matter,we could find a stronger correspondence with our research.
One new example of the latter is Melo et al. [51] (2009), which reviews location modelsin a supply chain context. Some of these will, in our terminology, be global supply chains.Other review papers may contain similar examples of partial overlap, and these could beworth pursuing.
From Table 3 we see that many factors are considered in location models, making ithard to discern any particular trend. As mentioned in §3, all models in this subset assumedeterministic conditions, but this does not imply that stochastic programming cannot beutilized for this problem; rather, from our observations, incorporating the variable nature ofexchange rates that are typical in global supply chains may only complicate this strategicmodel.
Sourcing from international suppliers would suggest that characteristics of a procurementmodel should include standard global factors, such as exchange rates and tariffs; however,the opposite appears true from the results in Table 4. Only Gutierrez and Kouvelis [31](1995) and Gonzalez Velarde and Laguna [26] (2004) include a combination of exchangerates and duties in their models, whereas Munson and Rosenblatt [55] (1997) and Balaji andViswanadham [4] (2008) consider LCRs and tax planning, respectively. From these articleswe see that there are a number of approaches to determine an optimal procurement strategy,depending on the emphasis a firm places on international factors.
Table 5 shows that transportation models can be applied of course to general internationalsettings (e.g., Chang [17] 2008) or to specific industries and regions. Examples of the latter
Bookbinder and Matuk: Logistics and Transportation in Global Supply Chains202 Tutorials in Operations Research, c© 2009 INFORMS
include a shortest-path model for NAFTA supply chains (Bookbinder and Fox [9] 1998), anevaluation of the benefits of using a DMCF for air transport (Warsing et al. [76] 2001), andcontainer management in the chemical industry (Erera et al. [23] 2005). Regardless of theapplication, these models either seek to minimize cost or travel time, or both.
Perhaps the topic that most epitomizes global supply chains is production switching,transfer pricing policies, and postponement. In Table 6, we see that every analytical modelfeatures exchange rates. About one-half of the models in that table consider duties andtariffs, and one-half consider transfer prices. One notes from the table, however, that thereis not much overlap between these “halves.” Only Vidal and Goetschalckx [74] (2001) andVillegas and Ouenniche [75] (2008) treat both transfer prices and duties and tariffs. In thelatter reference, tariffs were implicitly included in some cost parameters. A generalizationof that idea may permit expansion to the other global factors of the models in Table 6.
We offer an additional comment on production switching. Suppose a product’s manufac-ture is moved from Country A to Country B, with no change in the supplier of raw materials.That vendor, in Country A, had been considered domestic but is now international. Manycases of production switching will thus automatically involve duties and tariffs.
Models for network design (Table 7) are potentially the most complicated. The nodallocations are to be chosen, and decision taken on the activities to be carried out at eachnode. We observe in that table the inclusion of only one, or even zero, global factors for mostof the models. Perhaps such factors are not needed for a regional model whose transactionsare within a free trade zone. But we note that only Wilhelm et al. [77] (2005) and Gohet al. [25] (2007) incorporate all three global factors. Naturally, the complications of anextra subscript or two, and the presence of additional constraints, may seriously impact thesolution’s computation time and its deviation from optimality.
Table 8 summarizes a number of empirical studies. These are in contrast to the normativemodels that are much more prevalent at an INFORMS conference and in its publications.But the papers in that table belong in this tutorial for the following reasons.
Whereas a normative model can show what ought to be done to improve a global supplychain, a good empirical study can present the current state of affairs. This empirical infor-mation can serve a benchmarking function, for that supply chain and its logistics activities,perhaps pointing the way toward a normative model. The articles in Table 8 are thus ofoverall interest. Additional research may be required, however, to uncover a topic there forfurther study.
11. SummaryIn this tutorial, we have tried to balance several issues. We concentrated on the literatureof the past 15 years, the period during which the supply chain concept has emerged andflourished. Even so, various selections needed to be made from our bibliography. Of thoseinteresting references, we were able to discuss only a subset. Those articles are categorizedin various ways in Tables 2–8.
No model can be all things to all people. An excellent treatment of a regional supplychain, i.e., one that happens to involve only a single continent, could be more relevant toa proposed new application than a model that attempts a “whole-world” focus. Such ageographical distinction is one of the six dividing factors that we see in the literature. Andneither that factor, nor any of the others, may be pertinent in judging the value of a modelor application. More is only sometimes better.
Among the excellent regional logistics models are Sheu [67] (2004), Groothedde et al. [29](2005), and Wilhelm et al. [77] (2005). References such as Kouvelis et al. [41] (2004) andVillegas and Ouenniche [75] (2008) are closer to global in scope. The latter citations neededthat focus because of the particular application or problem setting.
Another dividing factor is the use of a deterministic model (Vidal and Goetschalckx [74]2001, Chang [17] 2008) versus one that is stochastic (Wu and Lin [78] 2005, Goh et al. [25]2007). The respective titles indicate the model features treated in each case. In those articles,
Bookbinder and Matuk: Logistics and Transportation in Global Supply ChainsTutorials in Operations Research, c© 2009 INFORMS 203
Tabl
e5.
Tra
nspo
rtat
ion
mod
els.
Aut
hors
Yea
rR
egio
nO
bjec
tive
Tim
epe
riod
Add
ition
al in
form
atio
n
Land
Boo
kbin
der
and
Fox
[9]a
1997
NA
FTA
Min
cos
t/M
in tr
avel
tim
eN
/AN
/ASe
lect
ion
of r
oute
s fo
r in
term
odal
net
wor
ks
McC
ray
and
Gon
zale
z[4
8]a
2008
N. A
mer
ica
Min
con
tain
erpr
oces
sing
tim
eN
/AN
/A
Nai
r et
al.
[57]
a20
08E
urop
eM
in c
ost/
Min
trav
el ti
me
N/A
N/A
Inve
stig
ates
pot
entia
l for
new
nor
th–s
outh
rai
l cor
rido
r in
Eur
ope
(RE
OR
IEN
T)
and
the
degr
ee to
whi
ch th
ese
rout
es a
reco
mpe
titiv
e
Kuo
et a
l. [4
2]a
2008
Eur
ope
Min
pro
cess
ing
time
N/A
N/A
Prop
oses
thre
e st
rate
gies
for
col
labo
rativ
e de
cisi
on m
akin
g to
stre
amlin
e tr
ansp
ort a
long
RE
OR
IEN
T c
orri
dor
Jani
c [3
6]a
2008
Eur
ope
N/A
MC
ompa
res
perf
orm
ance
of
usin
g L
IFT
S ve
rsus
con
vent
iona
lin
term
odal
fre
ight
trai
ns o
r ro
ad
Air
War
sing
et a
l. [7
6]20
01M
MB
enef
its o
f us
ing
a D
MC
F ve
rsus
a tr
aditi
onal
pas
seng
er-b
ased
airp
ort
Tya
n et
al.
[71]
a20
03A
sia
(Tai
wan
)M
SE
xam
ines
con
solid
atio
n po
licie
s fo
r a
3PL
Hua
ng a
nd C
hi [
34]
2007
Eur
ope
Eur
ope
Min
cos
t
Min
cos
t
Min
cos
t
Min
cos
t
SM
Dev
elop
s a
Lag
rang
ian-
base
d he
uris
tic a
lgor
ithm
for
an
air
frei
ght f
orw
arde
r to
det
erm
ine
optim
al c
onso
lidat
ion
polic
ies
Cha
ng [
17]
2008
Min
cos
t/M
in tr
avel
tim
eM
MPr
oble
m is
dec
ompo
sed
into
two
min
imiz
atio
n pr
oble
ms
usin
gL
agra
ngia
n re
laxa
tion
Nei
berg
er [
58]a
2008
N/A
N/A
N/A
Stud
ies
the
effe
ct o
f su
pply
cha
in c
onfi
gura
tion
on th
e ai
rfr
eigh
t sec
tor
base
d on
inte
rvie
ws
with
60
com
pani
es
Water
Ere
ra e
t al.
[23]
2005
Min
cos
tM
SD
eter
min
es o
ptim
al p
olic
y fo
r co
ntai
ner
repo
sitio
ning
Gro
othe
dde
et a
l. [2
9]a
2005
Eur
ope
(Net
herl
ands
)M
in c
ost
N/A
N/A
Des
igns
a c
olla
bora
tive
inte
rmod
al h
ub n
etw
ork
Exa
min
es th
e fe
asib
ility
of
usin
g C
anad
ian
or M
exic
an te
rmin
als
to m
itiga
te c
onge
stio
n at
U.S
. Wes
t Coa
st
Num
ber
of
prod
ucts
Not
es. A
ll m
odel
s in
this
sub
set a
ssum
e de
term
inis
tic c
ondi
tions
. S, s
ingl
e; M
, mul
ti.
aPa
per
has
been
app
lied
to in
dust
ry/u
sed
indu
stry
dat
a.
Bookbinder and Matuk: Logistics and Transportation in Global Supply Chains204 Tutorials in Operations Research, c© 2009 INFORMS
Tabl
e6.
Mod
els
for
prod
ucti
onsw
itch
ing,
tran
sfer
pric
ing
polic
ies,
and
post
pone
men
t.
Glo
bal f
acto
rs
Aut
hors
Yea
rM
odel
type
Obj
ectiv
eT
ime
peri
odN
umbe
r of
prod
ucts
Tra
nsfe
rpr
ices
Exc
hang
era
tes
Dut
ies
and
tari
ffs
Add
ition
al in
form
atio
n
Production switching
Kog
ut a
nd K
ulat
ilaka
[39
]19
94St
/DM
in c
ost
MS
��
Stoc
hast
ic d
ynam
ic p
rogr
amm
ing
mod
elH
uchz
erm
eier
and
Coh
en [
35]
1996
Max
pro
fit
MS
*�
Mod
els
exch
ange
rat
es a
s di
ffus
ion
proc
esse
sw
ith in
terc
ount
ry c
orre
latio
n
Das
u an
d L
i [22
]19
97St
/DM
in c
ost
MS
��
Det
erm
ines
by
how
muc
h an
d w
hen
to a
lter
prod
uctio
n qu
antit
ies
for
a L
atin
Am
eric
an M
NC
Moh
amed
[53
]19
9SSt
Min
cos
tM
MIn
tegr
ated
pro
duct
ion
and
dist
ribu
tion
mod
elun
der
vary
ing
exch
ange
rat
e an
d in
flat
ion
rate
s
Li e
t al.
[43]
2001
Min
cos
tM
S�
�U
ncer
tain
ties
in d
eman
d, e
xcha
nge
rate
s, a
ndpr
oces
sing
tim
es
Kaz
az e
t al.
[38]
2005
St/D
Max
pro
fit
Max
pro
fit
Max
pro
fit
Max
pro
fit
Max
pro
fit
Max
pro
fit
S/M
S�
Tw
o-st
age
mod
el th
at o
ffer
s th
e fl
exib
ility
of
usin
g pr
oduc
tion
and
allo
catio
n he
dges
Wu
and
Lin
[78
]20
05M
S�
�R
eal o
ptio
ns a
naly
sis;
con
tinuo
us-t
ime
mod
el
Transfer pricingpolicies
Kou
velis
and
Gut
ierr
ez[4
0]19
97M
in c
ost
SS
�“S
tyle
goo
ds”
sold
in tw
o in
tern
atio
nal m
arke
tsw
ith n
onov
erla
ppin
g se
lling
sea
sons
Vid
al a
nd G
oets
chal
ckx
[74]
2001
DM
S�
��
Tra
nsfe
r pr
ices
and
allo
catio
n of
tran
spor
t cos
tsar
eex
plic
itde
cisi
ons
Mill
er a
nd d
e M
atta
[52
]20
08D
SM
��
Max
pro
fits
on
a gl
obal
bas
is; b
oth
stra
tegi
c an
dta
ctic
al le
vels
Vill
egas
and
Oue
nnic
he[7
5]20
08D
SM
��
*D
eter
min
es h
ow d
ecis
ions
on
trad
e qu
antit
ies
affe
ct a
fter
-tax
rep
atri
ated
ear
ning
s
Postponement
Had
jinic
ola
and
Kum
ar[3
2]20
02D
SM
�E
xam
ines
eig
ht m
anuf
actu
ring
-mar
ketin
g op
tions
for
a fi
rm w
ith p
lant
s in
two
coun
trie
s
Yan
g an
d B
urns
[79
]20
03D
iscu
ssio
n of
cur
rent
tren
ds in
pra
ctic
e an
d lit
erat
ure
Iden
tifie
s fo
ur s
uppl
y ch
ain
char
acte
rist
ics
affe
cted
by
post
pone
men
t
Jin
[37]
2004
Find
ings
bas
ed o
n a
revi
ew o
f U
.S. a
ppar
el m
anuf
actu
ring
fir
ms
Yan
g et
al.
[80]
2004
Lite
ratu
re r
evie
wD
evel
ops
a fr
amew
ork
to a
sses
s fe
asib
ility
of
adop
ting
a po
stpo
nem
ent s
trat
egy
Pras
ad e
t al.
[61]
2005
Find
ings
bas
ed o
n st
ruct
ured
inte
rvie
ws
with
30
com
pani
es
aN
otes
.D, d
eter
min
istic
; St,
stoc
hast
ic.
b
S, s
ingl
e; M
, mul
ti .
Lat
in A
mer
ican
reg
ion .
Nor
th A
mer
ican
reg
ion.
*
Impl
icitl
y co
nsid
ered
.
Com
pare
s an
d co
ntra
sts
BT
O a
nd M
TS
syst
ems
inde
velo
ping
and
dev
elop
ed c
ount
ries
Prop
oses
fou
r in
dica
tors
to a
chie
ve a
bal
ance
betw
een
glob
ally
and
loca
lly s
ourc
ed g
oods
a
b
St St StSt
Bookbinder and Matuk: Logistics and Transportation in Global Supply ChainsTutorials in Operations Research, c© 2009 INFORMS 205
Tabl
e7.
Ove
rall
netw
ork
desi
gnm
odel
s.
Glo
bal f
acto
rs
Aut
hors
Yea
rR
egio
nM
odel
type
Obj
ectiv
eT
ime
peri
od
a
Num
ber
of
prod
ucts
Tra
nsfe
rpr
ices
Exc
hang
era
tes
Dut
ies
and
&ta
riff
sA
dditi
onal
info
rmat
ion
Arn
tzen
et a
l. [1
]19
95D
Min
cos
tM
M�
Con
side
rs g
loba
l BO
M, o
ffse
t tra
de, a
ndL
CR
s; c
an h
andl
e m
ultip
le m
easu
res
oftim
e
Cam
m e
t al.
[12]
1997
DM
in c
ost
MM
Use
s G
IS to
sol
ve d
istr
ibut
ion
loca
tion
and
prod
uct s
ourc
ing
mod
els
Das
u an
d de
la T
orre
[21]
1997
Lat
inA
mer
ica
DM
ax P
rofi
tS
S�
Con
side
rs th
e co
ordi
natin
g ac
tiviti
esan
d al
loca
tion
of g
ains
for
an
MN
C’s
subs
idia
ries
Tsi
akis
et a
l. [7
0]20
01E
urop
eSt
/DM
in c
ost
SM
*E
cono
mie
s of
sca
le f
or tr
ansp
orta
tion
cost
s; d
ecid
es lo
catio
ns f
or f
acili
ties
at
two
eche
lons
(in
a f
our-
eche
lon
syst
em)
Sheu
[67
]20
04A
sia
(Tai
wan
)
Fuzz
yM
AD
M/
Fuzz
y A
HP
N/A
N/A
N/A
*D
eter
min
es b
est g
loba
l log
istic
s st
rate
gyfo
r an
inte
grat
ed c
ircu
it m
anuf
actu
rer
usin
g fu
zzy
AH
P an
d fu
zzy
MA
DM
Con
tess
e et
al.
[19]
2005
S. A
mer
ica
(Chi
le)
DM
in c
ost
MS
MIP
mod
el f
or a
nat
ural
gas
supp
ly c
hain
Nag
urne
y an
d M
atsy
pura
[56]
2005
StM
ax p
rofi
t/M
in r
isk
SS
�
Man
ufac
ture
rs, d
istr
ibut
ors,
and
reta
ilers
; bot
h su
pply
-sid
e an
d de
man
d-si
de r
isks
Sant
oso
et a
l. [6
5]20
05U
.S. a
ndL
atin
Am
eric
aM
in c
ost
SM
*L
arge
-sca
le s
toch
astic
pro
gram
min
g;sa
mpl
e-av
erag
e-ap
prox
imat
ion
sche
me
with
acc
eler
ated
Ben
ders
dec
ompo
sitio
n
Wilh
elm
et a
l. [7
7]20
05N
AFT
AD
Max
pro
fit
MM
��
�In
clud
es m
odal
dec
isio
ns, B
OM
rest
rict
ions
, and
inco
me
taxe
s
Goh
et a
l. [2
5]20
07M
ax p
rofi
tM
S�
��
Ris
ks in
sup
ply,
dem
and,
exc
hang
era
tes,
and
dis
rupt
ion
D, d
eter
min
istic
; St,
stoc
hast
ic.
S, s
ingl
e; M
, mul
ti.
Pape
r ha
s be
en a
pplie
d to
indu
stry
/use
d in
dust
ry d
ata .
*
Impl
icitl
y co
nsid
ered
.
a
a
a
a
a
St
aN
otes
.
St
Bookbinder and Matuk: Logistics and Transportation in Global Supply Chains206 Tutorials in Operations Research, c© 2009 INFORMS
Tabl
e8.
Oth
erpa
pers
(em
piri
calst
udie
s).
Aut
hors
Yea
rR
egio
nD
ata
type
Add
ition
al in
form
atio
n
Cross-sectional data
Lyn
ch e
t al.
[45]
a19
94N
. Am
eric
a(C
anad
a)Pr
imar
yFo
reca
sts
the
futu
re o
f lo
gist
ics
in C
anad
a us
ing
a D
elph
i stu
dy
Bol
isan
i and
Sca
rso
[8]a
1996
Eur
ope
(Ita
ly)
Seco
ndar
y +
Int
ervi
ews
Det
erm
ines
why
, whe
re, a
nd h
ow I
talia
n cl
othi
ng f
irm
s in
tern
atio
naliz
ed
Das
and
Han
dfie
ld [
20]a
1997
N. A
mer
ica
Prim
ary
Prim
ary
Prim
ary
Prim
ary
Use
s a
surv
ey to
com
pare
JIT
and
glo
bal s
ourc
ing
Mor
ash
and
Clin
ton
[54]
a19
97U
ses
a su
rvey
to c
ompa
re tr
ansp
orta
tion
in K
orea
, Jap
an, A
ustr
alia
, and
U.S
.
Mac
Car
thy
and
Atth
iraw
ong
[46]
a20
03D
eter
min
es f
acto
rs in
flue
ncin
g in
tern
atio
nal f
acili
ty lo
catio
n de
cisi
ons
usin
g a
Del
phi s
tudy
McC
alla
et a
l. [4
7]a
2004
Com
pari
son
of c
onta
iner
izat
ion
in E
. Asi
a, N
. Am
eric
a, a
nd N
W E
urop
e
Car
bone
and
Sto
ne [
16]a
2005
Eur
ope
Seco
ndar
y +
Int
ervi
ews
Exa
min
es g
row
th o
f 3P
Ls
in E
urop
e
Stee
nhui
s an
d D
e B
ruijn
[68
]20
04N
/AU
ses
GD
P to
ass
ess
feas
ibili
ty o
f lo
catin
g a
faci
lity
in a
cer
tain
cou
ntry
;ba
sed
on th
e ai
rcra
ft in
dust
ry
Case studies
Rob
erts
on e
t al.
[62]
a20
02O
cean
iaPr
imar
yD
escr
ibes
how
the
deve
lopm
ent o
f a
sale
s an
d op
erat
ions
pla
n an
dm
aste
r pr
oduc
tion
sche
dule
for
a p
artic
ular
fir
m h
as im
prov
ed it
spr
oduc
tion
and
soci
al p
roce
sses
Mei
jboo
m a
nd V
oord
ijk [
49]a
2003
Eur
ope
(Net
herl
ands
)Pr
imar
yD
eter
min
es w
hy D
utch
com
pani
es r
emai
n re
gion
al in
spi
te o
f w
ides
prea
dgl
obal
izat
ion
Liu
and
You
ng [
44]
2004
N/A
Coo
rdin
atio
n of
glo
bal m
anuf
actu
ring
with
em
phas
is o
nin
form
atio
n an
d kn
owle
dge
Bux
ey [
11]a
2005
Aus
tral
iaPr
imar
yD
ocum
ents
how
Aus
tral
ian
TC
F in
dust
ry h
as r
espo
nded
to g
loba
lizat
ion
Stra
tton
and
War
burt
on [
69]a
2006
Prim
ary
Eva
luat
es h
ow o
utso
urci
ng a
ffec
ts c
ompa
ny p
erfo
rman
ce
Rud
berg
and
Wes
t [64
]a20
08E
urop
e(S
wed
en)
Seco
ndar
yC
once
pt f
or c
oord
inat
ing
man
ufac
turi
ng n
etw
orks
aPa
per
has
been
app
lied
to in
dust
ry/u
sed
indu
stry
dat
a.
Bookbinder and Matuk: Logistics and Transportation in Global Supply ChainsTutorials in Operations Research, c© 2009 INFORMS 207
the choice of model type therefore seems logical. Further research would be desirable toelaborate the trade-offs in such a choice. In the case of international supply chains, a greaternumber of factors may impact the selection of model type here than in some other operationsresearch applications.
Manufacturing is included in some models (e.g., Arntzen et al. [1] 1995, Wilhelm et al.[77] 2005) but not in others. Even when a supply chain model involves manufacturing, thatmay not be an essential component; Wilhelm et al. [77] (2005) is one exception.
Similar remarks pertain to location decisions. When the nodes for production and dis-tribution are already fixed, as they are in Gonzalez Velarde and Laguna [26] (2004) andin Balaji and Viswanadham [4] (2008), the model will often emphasize decisions on whichproducts to produce in a particular plant, and the choice of DC through which they shouldflow to satisfy demand of specific customers. On the other hand, references such as Caneland Das [13] (2002), Kouvelis et al. [41] (2004), and Hamad and Gualda [33] (2008) areessentially location models for global logistics and manufacturing.
Questions of transportation are an obvious feature of international supply chains; rawmaterials and end items need to get from here to there. What is not so obvious is that themodel’s time scale can influence whether decisions on mode can be included in such a model.Let us give several examples.
The location of long-term facilities, as a strategic choice, may be based on an annual model(e.g., Canel and Khumawala [15] 2001). This might involve a weighted average over transportmodes, effectively concentrating on decisions other than selection of mode of transportation.Even a “domestic” model, where given arcs may be traversed by rail or truck, requires aweekly time scale, say, for the model to aid in this choice. (Those are the relevant modes andtime scales in the NAFTA location model of Robinson and Bookbinder [63] (2007).) Thepossible modes are more likely air and water in a global model. But the time scale must stillbe fine enough to account for differences in transportation lead times, and to incorporatepipeline inventories.
Finally, we distinguish between the “business features” of a problem or case study, andthe “modeling features” that the analysis must therefore comprise. Issues such as tariffs,duties, and quotas (e.g., Gutierrez and Kouvelis [31] 1995, Bhutta et al. [6] 2003, Goh et al.[25] 2007) are business features that can be difficult to model. Models of supplier selectionin an international setting, such as in Munson and Rosenblatt [55] (1997) and GonzalezVelarde and Laguna [26] (2004), are challenging for business reasons, but perhaps not formathematical ones. Models that incorporate multiple stages of a global supply chain willpose mathematical challenges. References such as Kazaz et al. [38] (2005) on implementingproduction and allocation hedging when faced with exchange rate uncertainty, and Santosoet al. [65] (2005) on designing a supply chain network in a stochastic environment, have thusstressed the solution methodologies involved.
AcknowledgmentsThe research leading to this tutorial was supported by NSERC, the Natural Sciences and Engineer-ing Research Council of Canada. The authors are grateful to Professor Hau Lee for his insightfulcomments on an earlier draft of this manuscript.
References[1] B. C. Arntzen, G. G. Brown, T. P. Harrison, and L. L. Trafton. Global supply chain manage-
ment at Digital Equipment Corporation. Interfaces 25(1):69–93, 1995.
[2] M. A. Badri. A multicriteria approach to global facility location-allocation problem. Interna-tional Journal on Information and Management Sciences 7(3):1–19, 1996.
[3] M. A. Badri. Combining the analytic hierarchy process and goal programming for global facilitylocation-allocation problem. International Journal of Production Economics 62(3):237–248,1999.
Bookbinder and Matuk: Logistics and Transportation in Global Supply Chains208 Tutorials in Operations Research, c© 2009 INFORMS
[4] K. Balaji and N. Viswanadham. A tax integrated approach for global supply chain networkplanning. IEEE Transactions on Automation Science and Engineering 5(4):587–596, 2008.
[5] R. Bhatnagar, P. Chandra, and S. K. Goyal. Models for multi-plant coordination. EuropeanJournal of Operational Research 67(2):141–160, 1993.
[6] K. S. Bhutta, F. Huq, G. Frazier, and Z. Mohamed. An integrated location, production,distribution and investment model for a multinational corporation. International Journal ofProduction Economics 86(3):201–216, 2003.
[7] B. Bilgen, I. Ozkarahan. Strategic tactical and operational production-distribution models:A review. International Journal of Technology Management 28(2):151–171, 2004.
[8] E. Bolisani and E. Scarso. International manufacturing strategies: Experiences from the cloth-ing industry. International Journal of Operations and Production Management 16(11):71–84,1996.
[9] J. H. Bookbinder and N. S. Fox. Intermodal routing of Canada-Mexico shipmentsunder NAFTA. Transportation Research Part E: Logistics and Transportation Review34E(4):289–303, 1998.
[10] T. Boone, S. P. Johnson, M. Sisk, and D. C. Whybark. An analysis of research on internationaloperations networks. International Journal of Production Economics 46:477–488, 1996.
[11] G. Buxey. Globalisation and manufacturing strategy in the TCF industry. International Jour-nal of Operations and Production Management 25(2):100–113, 2005.
[12] J. D. Camm, T. E. Chorman, F. A. Dill, J. R. Evans, D. J. Sweeney, and G. W. Wegryn. Blend-ing OR/MS, judgment, and GIS: Restructuring P&G’s supply chain. Interfaces 27(1):128–142,1997.
[13] C. Canel and S. R. Das. Modeling global facility location decisions: Integrating marketing andmanufacturing decisions. Industrial Management and Data Systems 102(2):110–118, 2002.
[14] C. Canel and B. M. Khumawala. A mixed-integer programming approach for the internationalfacilities location problem. International Journal of Operations and Production Management16(4):49–68, 1996.
[15] C. Canel and B. M. Khumawala. International facilities location: A heuristic procedurefor the dynamic uncapacitated problem. International Journal of Production Research39(17):3975–4000, 2001.
[16] V. Carbone and M. A. Stone. Growth and relational strategies used by the European logis-tics service providers: Rationale and outcomes. Transportation Research Part E: Logistics andTransportation Review 41(6):495–510, 2005.
[17] T. S. Chang. Best routes selection in international intermodal networks. Computers and Oper-ations Research 35(9):2877–2891, 2008.
[18] M. A. Cohen and S. Mallik. Global supply chains: Research and applications. Production andOperations Management 6(3):193–210, 1997.
[19] L. Contesse, J. C. Ferrer, and S. Maturana. A mixed-integer programming model for gaspurchase and transportation. Annals of Operations Research 139(1):39–63, 2005.
[20] A. Das and R. B. Handfield. Just-in-time and logistics in global sourcing: An empirical study.International Journal of Physical Distribution and Logistics Management 27(3/4):244–259,1997.
[21] S. Dasu and J. de la Torre. Optimizing an international network of partially owned plantsunder conditions of trade liberalization. Management Science 43(3):313–333, 1997.
[22] S. Dasu and L. Li. Optimal operating policies in the presence of exchange rate variability.Management Science 43(5):705–722, 1997.
[23] A. L. Erera, J. C. Morales, and M. Savelsbergh. Global intermodal tank container managementfor the chemical industry. Transportation Research Part E: Logistics and Transportation Review41(6):551–566, 2005.
[24] E. Eskigun, R. Uzsoy, P. V. Preckel, G. Beaujon, S. Krishnan, and J. D. Tew. Outbound supplychain network design with mode selection, lead times and capacitated vehicle distributioncenters. European Journal of Operational Research 165(1):182–206, 2005.
[25] M. Goh, J. Y. S. Lim, and F. Meng. A stochastic model for risk management in global supplychain networks. European Journal of Operational Research 182(1):164–173, 2007.
Bookbinder and Matuk: Logistics and Transportation in Global Supply ChainsTutorials in Operations Research, c© 2009 INFORMS 209
[26] J. L. Gonzalez Velarde and M. Laguna. A Benders-based heuristic for the robust capacitatedinternational sourcing problem. IIE Transactions 36(11):1125–1133, 2004.
[27] K. N. Gourdin. Global Logistics Management: A Competitive Advantage for the 21st Century.Blackwell Publishing, Malden, MA, 2006.
[28] S. E. Grasman. Dynamic approach to strategic and operational multimodal routing decisions.International Journal of Logistics Systems and Management 2(1):96–106, 2006.
[29] B. Groothedde, C. Ruijgrok, and L. Tavasszy. Towards collaborative, intermodal hub networks.Transportation Research Part E: Logistics and Transportation Review 41(6):567–583, 2005.
[30] A. Gunasekaran and B. Kobu. Performance measures and metrics in logistics and supplychain management: A review of recent literature (1995–2004) for research and applications.International Journal of Production Research 45(12):2819–2840, 2007.
[31] G. J. Gutierrez and P. Kouvelis. A robustness approach to international sourcing. Annals ofOperations Research 59:165–193, 1995.
[32] G. C. Hadjinicola and K. R. Kumar. Modeling manufacturing and marketing options in inter-national operations. International Journal of Production Economics 75(3):287–304, 2002.
[33] R. Hamad and N. D. F. Gualda. Model for facilities or vendors location in a global scaleconsidering several echelons in the chain. Networks and Spatial Economics 8(2–3):297–307,2008.
[34] K. Huang and W. Chi. A Lagrangian relaxation based heuristic for the consolidation problem ofairfreight forwarders. Transportation Research Part C: Emerging Technologies 15(4):235–245,2007.
[35] A. Huchzermeier and M. A. Cohen. Valuing operational flexibility under exchange rate risk.Operations Research 44(1):100–113, 1996.
[36] M. Janic. An assessment of the performance of the European long intermodal freight trains(LIFTS). Transportation Research Part A: Policy and Practice 42(10):1326–1339, 2008.
[37] B. H. Jin. Achieving an optimal global versus domestic sourcing balance under demand uncer-tainty. International Journal of Operations and Production Management 24(11–12):1292–1305,2004.
[38] B. Kazaz, M. Dada, and H. Moskowitz. Global production planning under exchange-rate uncer-tainty. Management Science 51(7):1101–1119, 2005.
[39] B. Kogut and N. Kulatilaka. Operating flexibility, global manufacturing, and the option valueof a multinational network. Management Science 40(1):123–139, 1994.
[40] P. Kouvelis and G. J. Gutierrez. The newsvendor problem in a global market: Optimal cen-tralized and decentralized control policies for a two-market stochastic inventory system. Man-agement Science 43(5):571–585, 1997.
[41] P. Kouvelis, M. J. Rosenblatt, and C. L. Munson. A mathematical programming model forglobal plant location problems: Analysis and insights. IIE Transactions 36(2):127–144, 2004.
[42] A. Kuo, E. Miller-Hooks, K. Zhang, and H. Mahmassani. Train slot cooperation in multi-carrier, international rail-based intermodal freight transport. Transportation Research Record2043:31–40, 2008.
[43] L. Li, E. L. Porteus, and H. Zhang. Optimal operating policies for multiplant stochastic man-ufacturing systems in a changing environment. Management Science 47(11):1539–1551, 2001.
[44] S. Liu and R. I. M. Young. Utilizing information and knowledge models to support globalmanufacturing co-ordination decisions. International Journal of Computer Integrated Manu-facturing 17(6):479–492, 2004.
[45] M. E. Lynch, S. J. Imada, and J. H. Bookbinder. The future of logistics in Canada: A Delphi-based forecast. Logistics and Transportation Review 30(1):95–112, 1994.
[46] B. L. MacCarthy and W. Atthirawong. Factors affecting location decisions in internationaloperations—A Delphi study. International Journal of Operations and Production Management23(7–8):794–818, 2003.
[47] R. J. McCalla, B. Slack, and C. Comtois. Dealing with globalisation at the regional and locallevel: The case of contemporary containerization. Canadian Geographer-Geographe Canadien48(4):473–487, 2004.
Bookbinder and Matuk: Logistics and Transportation in Global Supply Chains210 Tutorials in Operations Research, c© 2009 INFORMS
[48] J. P. McCray and J. J. Gonzalez. Critical infrastructure at US west coast intermodal ter-minals: Are Canadian or Mexican terminals an alternative? Transportation Research Record2062:25–31, 2008.
[49] B. Meijboom and H. Voordijk. International operations and location decisions: A firm levelapproach. Tijdschrift Voor Economische En Sociale Geografie 94(4):463–476, 2003.
[50] M. J. Meixell and V. B. Gargeya. Global supply chain design: A literature review and critique.Transportation Research Part E: Logistics and Transportation Review 41(6):531–550, 2005.
[51] M. T. Melo, S. Nickel, and F. Saldanha-da-Gama. Facility location and supply chainmanagement—A review. European Journal of Operational Research 196(2):401–412, 2009.
[52] T. Miller and R. de Matta. A global supply chain profit maximization and transfer pricingmodel. Journal of Business Logistics 29(1):175–199, 2008.
[53] Z. M. Mohamed. An integrated production-distribution model for a multi-national com-pany operating under varying exchange rates. International Journal of Production Economics58(1):81–92, 1999.
[54] E. A. Morash and S. R. Clinton. The role of transportation capabilities in international supplychain management. Transportation Journal 36(3):5–17, 1997.
[55] C. L. Munson and M. J. Rosenblatt. The impact of local content rules on global sourcingdecisions. Production and Operations Management 6(3):277–290, 1997.
[56] A. Nagurney and D. Matsypura. Global supply chain network dynamics with multicriteriadecision-making under risk and uncertainty. Transportation Research Part E: Logistics andTransportation Review 41(6):585–612, 2005.
[57] R. Nair, E. D. Miller-Hooks, H. S. Mahmassani, V. C. Arcot, A. Kuo, K. Zhang, A. Kozuki,and J. Ludvigser. Market potential for international rail-based intermodal services in Europefrom sea to shining sea. Transportation Research Record 2066:21–30, 2008.
[58] C. Neiberger. The effects of deregulation, changed customer requirements and new technologyon the organisation and spatial patterns of the air freight sector in Europe. Journal of TransportGeography 16(4):247–256, 2008.
[59] P. Pontrandolfo and O. G. Okogbaa. Global manufacturing: A review and a framework forplanning in a global corporation. International Journal of Production Research 37(1):1–19,1999.
[60] S. Prasad, S. Babbar, and J. Motwani. International operations strategy: Current effortsand future directions. International Journal of Operations and Production Management21(5–6):645–665, 2001.
[61] S. Prasad, J. Tata, and M. Madan. Build to order supply chains in developed and developingcountries. Journal of Operations Management 23(5):551–568, 2005.
[62] P. W. Robertson, P. R. Gibson, and J. T. Flanagan. Strategic supply chain developmentby integration of key global logistical process linkages. International Journal of ProductionResearch 40(16):4021–4040, 2002.
[63] A. G. Robinson and J. H. Bookbinder. NAFTA supply chains: Facilities location and logistics.International Transactions in Operational Research 14(2):179–199, 2007.
[64] M. Rudberg and B. M. West. Global operations strategy: Coordinating manufacturing net-works. Omega-International Journal of Management Science 36(1):91–106, 2008.
[65] T. Santoso, S. Ahmed, M. Goetschalckx, and A. Shapiro. A stochastic programming approachfor supply chain network design under uncertainty. European Journal of Operational Research167(1):96–115, 2005.
[66] G. Schmidt and W. E. Wilhelm. Strategic, tactical and operational decisions in multi-nationallogistics networks: A review and discussion of modelling issues. International Journal of Pro-duction Research 38(7):1501–1523, 2000.
[67] J. Sheu. A hybrid fuzzy-based approach for identifying global logistics strategies. Transporta-tion Research Part E: Logistics and Transportation Review 40(1):39–61, 2004.
[68] H. J. Steenhuis and E. J. De Bruijn. Assessing manufacturing location. Production Planningand Control 15(8):786–795, 2004.
[69] R. Stratton and R. D. H. Warburton. Managing the trade-off implications of global supply.International Journal of Production Economics 103(2):667–679, 2006.
Bookbinder and Matuk: Logistics and Transportation in Global Supply ChainsTutorials in Operations Research, c© 2009 INFORMS 211
[70] P. Tsiakis, N. Shah, and C. C. Pantelides. Design of multi-echelon supply chain networks underdemand uncertainty. Industrial and Engineering Chemistry Research 40(16):3585–3604, 2001.
[71] J. C. Tyan, F.-K. Wang, and T. C. Du. An evaluation of freight consolidation policies in globalthird party logistics. Omega 31(1):55–62, 2003.
[72] U.S. Customs and Border Protection. Security Filing “10+2.” Accessed April 6, 2009 from U.S.Customs and Border Protection: http://www.cbp.gov/xp/cgov/trade/cargo security/carriers/security filing/, 2009.
[73] C. J. Vidal and M. Goetschalckx. Strategic production-distribution models: A critical reviewwith emphasis on global supply chain models. European Journal of Operational Research98(1):1–18, 1997.
[74] C. J. Vidal and M. Goetschalckx. A global supply chain model with transfer pricing andtransportation cost allocation. European Journal of Operational Research 129(1):134–158, 2001.
[75] F. Villegas and J. Ouenniche. A general unconstrained model for transfer pricing in multina-tional supply chains. European Journal of Operational Research 187(3):829–856, 2008.
[76] D. P. Warsing, G. C. Souza, and N. P. Greis. Determining the value of dedicated multi-modal cargo facilities in a multi-region distribution network. European Journal of OperationalResearch 133(1):81–93, 2001.
[77] W. Wilhelm, D. Liang, B. Rao, D. Warrier, X. Zhu, and S. Bulusu. Design of internationalassembly systems and their supply chains under NAFTA. Transportation Research Part E:Logistics and Transportation Review 41(6):467–493, 2005.
[78] C. Wu and C. Lin. Hedging entry and exit decisions: Optimizing location under exchange rateuncertainty. IMA Journal of Management Mathematics 16(4):355–367, 2005.
[79] B. Yang and N. Burns. Implications of postponement for the supply chain. International Jour-nal of Production Research 41(9):2075–2090, 2003.
[80] B. Yang, N. D. Burns, and C. J. Backhouse. Postponement: A review and an integrated frame-work. International Journal of Operations and Production Management 24(5–6):468–487, 2004.