A tax savings model for the emerging global manufacturing network

A tax savings model for the emerging global manufacturing network

Cheng-Min Feng



, Pei-Ju Wu

Institute of Traffic and Transportation, National Chiao Tung University, 4F, 118, Sec. 1, Chung Hsiao W. Rd., Taipei 10012, Taiwan, ROC

a r t i c l e

i n f o

Article history: Received 3 August 2007 Accepted 11 May 2009 Available online 21 June 2009 Keywords:

Global manufacturing network International logistics zone After-tax profit

Tax savings model Logistics behavior

a b s t r a c t

The emerging global manufacturing network involves nodal location features of tax areas and international logistics zones, manufacturing procedures of simple process and deep process, as well as transportation arcs. Since choosing tax savings locations and manufacturing procedures that increase after-tax profit is important to global manufacturers, this study aims to present several tax savings approaches and to develop a tax savings model for maximizing after-tax profit in the emerging network. Numerical illustration demonstrates that the proposed model is an effective approach for global manufacturers to achieve tax savings. The proposed model elucidates the crucial logistics behavior associated with tax savings.

&_{2009 Elsevier B.V. All rights reserved.}

1. Introduction

Global logistics can be conceptualized as the geo-graphic expansion of domestic logistics to markets abroad (Bowersox and Closs, 1996;Sheu, 2004). Fierce competi-tion in a rapidly changing global market has imposed tremendous pressure on manufacturing enterprises to transform and adjust their supply chain operations abroad (Chia et al., 2002). A necessary policy for managing resources is to negotiate at a global level and at a local level supply chain (Albino et al., 2002;Hu¨lsmann et al., 2008). Accordingly, global manufacturers are currently integrating their operations in different countries to achieve manufacturing efficiency across markets and operating units worldwide (Cavusgi et al., 2004).

Manufacturers are increasingly capable of dealing with full system production, and currently seek to add value to existing production systems (Zhai et al., 2007). Minimiz-ing manufacturMinimiz-ing cost and production time combined with increasing quality and shipment reliability are important challenges to all production systems (Mezga´r et al., 2000). Specifically, manufacturers must devise

effective global logistics strategies that maximize profit and fulfill customer orders within manufacturing net-works (Hammami et al., 2003; Jodlbauer, 2008).Hameri and Paatela (2005)also observed that contract manufac-turers focus on integrating value added operations in networks to maintain and recreate profitable business in markets with narrow margins.

Each manufacturing base only produces special ances-tor goods (e.g. components) for the total demand in participating countries, and ships these ancestor goods to other manufacturing bases for further transformation of ancestor goods into descendant goods (e.g. finished products) (Arntzen et al., 1995; Hiraki, 1996). Conse-quently, a distributed product often has its manufacturing activities dispersed throughout many locations (Lakhal et al., 2005). Once goods transfer from one place to another, complicating tax factors arise such as import duties, corporate taxes, value added taxes, sales taxes, etc. (Goetschalckx et al., 2002; Sheu, 2003; Meixell and Gargeya, 2005; Power, 2005; Tsiakis and Papageorgiou, 2008;Das and Sengupta, 2009).

Some of these tax factors have been examined in previous studies.Arntzen et al. (1995)proposed a global supply chain model for minimizing total cost at the Digital Equipment Corporation considering duty drawback and duty relief. Since a differential tax structure contributes to Contents lists available atScienceDirect

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

Int. J. Production Economics

0925-5273/$ - see front matter & 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.ijpe.2009.05.019

_{Corresponding author. Tel.: +886 2 2349 4956; fax: +886 2 2349 4965.} E-mail address:[email protected] (C.-M. Feng).

distribution network decisions that cause logistic ineffi-ciency, Avittathur et al. (2005) developed a model for determining locations of distribution centers (DCs) which considered the impact of differential sales taxes applicable in inter-state trade. Nonetheless, corporate taxes are not easily incorporated into a profit-maximizing model, mainly because some subsidiaries of a global manufac-turer may operate at a loss. Restated, unprofitable subsidiaries are not required to pay corporate taxes, but others are subject to corporate tax. Accordingly,Vidal and Goetschalckx (2001) constructed a global supply chain model to cope with the above problem. Their model maximized the after-tax profit of a multinational enter-prise by considering transfer pricing and transportation cost allocation. Thereafter, Fandel and Stammen (2004)

and Vila et al. (2006) extended previous research to construct an after-tax profit-maximizing model that reflects similar tax factors such as duties and corporate taxes with an emphasis on product life cycles and divergent process industries, respectively. Nonetheless, there has been little research performed to develop a model by simultaneously considering import duty, value added tax and corporate tax.

Governments recognize that most global enterprises pay much attention to the impact of tax factors on their global profit. Therefore, a common governmental strategy is developing ‘‘international logistics zones’’ (Lu and Yang, 2007) offering tax-exemption strategies (e.g. exemptions from corporate tax or import duties) to attract investment and ideally to spark economic growth. Examples of these logistics zones are free trade zones, export processing zones, free port zones, bonded zones and customs-free zones (Prasad and Sounderpandian, 2003;Lee and Yang, 2003; Oum and Park, 2004;Lu and Yang, 2007). For the purposes of this study, ‘‘international logistics zones’’ are defined as zones offering tax-exemptions while ‘‘tax areas’’ are defined as areas which do not offer tax preferences. Taking advantage of preferential taxation is extremely important for global manufacturers to achieve tax savings. Herein, tax savings mean that the amount enterprises save in taxes. Additionally, global enterprises have typically used transfer price to manipulate profit distribution among their subsidiaries. However, enter-prises utilizing transfer price as a means of tax mitigation are easy prey for costly audits and litigation (Lakhal et al., 2005). Many countries now have international logistics zones that reduce taxes for global enterprises. Thus, discussing legitimate tax savings approaches associated with international logistics zones is worthwhile.

Facilities functioning only as manufacturing centers or only as distribution centers are less responsive to rapid changes in global commerce than facilities capable of both. Simchi-Levi et al. (2003)pointed out that intense competitive pressure has forced manufacturers to add manufacturing capability at DCs.Sheu (2004)also noted that manufacturers with combined production and dis-tribution facilities have significant advantages in global logistical management. Furthermore, DCs in international logistics zones can be classified as ‘‘deep process’’ and ‘‘simple process’’ facilities (DHL, 2006). Deep processing DCs have manufacturing functions proceeding with

serious manufacture producing added value, while simple processing DCs cannot manufacture and merely own the functions of simple and convenient processes (e.g. assembling). For clarity, DCs serving the functions of either simple or deep processing, or both are defined as ‘‘processing DCs’’ in our study. In practice, DCs can be divided into three types: deep processing DCs, simple processing DCs and non-bonded DCs, depending on their locations and manufacturing procedures. Deep processing DCs located in international logistics zones have both deep process and simple process functions. Although located in international logistics zones, simple processing DCs only have the simple process function. Non-bonded DCs perform the same functions as deep processing DCs, but they are located in tax areas. Accordingly, the emerging global manufacturing network comprises nodal location characteristics of tax areas and international logistics zones, manufacturing procedures of simple process and deep process in these nodes, as well as transportation arcs. Therefore, manufacturers must re-view their global manufacturing activities.

Once both processing DCs and international logistics zones are incorporated in the global manufacturing network, global manufacturers have difficulty in deter-mining the optimal tax savings route and manufacturing procedure for each order.Simchi-Levi et al. (2003)noted that implementing a strategy in which the manufacturing process is completed in a local DC can reduce costs associated with duties as duties are lower for semi-products than for finished products. Accordingly, manufacturers must decide whether to: (1) import semi-products and then convert these semi-products into finished products in tax areas to reduce duties, or (2) manufacture finished products in international logistics zones and then import the finished products to tax areas to reduce corporate tax. Moreover, to identify the best tax savings route and manufacturing procedure, an after-tax model should allow goods free transfer among processing DCs. Restated, a finished product may be processed via simple or deep processing, or both, in various DCs.

Nevertheless, few studies have investigated the choice of tax savings locations and manufacturing procedures to increase after-tax profit for a global manufacturer in the emerging global manufacturing network. The purpose of this study was to represent several tax savings approaches and eventually to develop a tax savings model that maximizes a global manufacturer profit. Furthermore, the proposed model determines the optimal tax savings route and manufacturing procedure for each order.

The rest of this paper is organized as follows. Section 2 presents tax savings approaches concerning the charac-teristics of international logistics zones. Section 3 de-scribes the problem statement to clarify the scope of the study and to facilitate model formulation. Section 4 provides a model with tax savings, incorporating the emerging global manufacturing network, to find after-tax profit-maximizing solutions. Section 5 tests the problem-solving effectiveness of the proposed model and discusses the findings of the numerical results. Finally, Section 6 summarizes the conclusions of the study.

2. Tax savings approaches

According to in-depth interviews with global manu-facturers, the three key tax factors contributing to operating income are the following: import duty, value added tax and corporate tax. Furthermore, considering the characteristics of international logistics zones, the charge condition of import duty, value added tax and corporate tax in the emerging network has become more complex than that in a typical network. The tax savings approaches for these three taxes are outlined below.

2.1. Import duty

Import duties are tariffs paid to the relative govern-ment as goods pass into tax areas. Issues of import duties can be divided into the following three dimensions.

1. Charge condition: As situation depicted in Fig. 1, the charge condition of import duty is that for the same country original flows are in international logistics zones and destination flows are in tax areas, while for different countries destination flows are in tax areas. 2. Import from low duty rate country: Since duty rates may

differ between countries for the same goods, enter-prises can reduce costs by importing goods from countries with lower duties. As Fig. 2shows, import duties from country B ($80 ¼ $800 *10%) are lower than from country A ($400 ¼ $800 *50%) for the same goods ($800). Consequently, assuming all other condi-tions are equal, the enterprise can save import duties by importing via the low duty country.

3. Import duty and product forms: Duty rates change with respect to product form, and manufacturers must then determine the most advantageous trade-off between import duty and processing cost. For instance, assum-ing country B requires the finished products inFig. 3, manufacturers must decide whether to: (1) convert raw materials into finished products in country A and then import the finished products to country B or (2) import raw materials from country A and then convert the raw materials into finished products in country B.

2.2. Value added tax

Assessment of value added tax (VAT) is based on the incremental increase in the value of goods from raw materials to finished products. For each transaction, VAT is levied on the increased value of a product after input from previous chain members. Value added tax is generally formulated as follows:

VATcost ¼ ps_qs_{VAT  p}i_qi_{VAT þ p}o_qo

 ðVAT  DRTÞ (1) where VATcost implies the cost of VAT; ps_{, p}i_{, p}o_represent

the prices associated with sale, input and export, respectively; qs_{, q}i_{, q}o _{denote the quantities associated}

with sale, input and export, respectively; VAT indicates VAT rate (%) on the value of goods; DRT signifies VAT drawback rate (%) on the value of goods.

The first term in Eq. (1) represents sales VAT, and the second term denotes input VAT. Sales VAT can be offset by input VAT. Further, the third term is regarded as export VAT which refers to the VAT imposed on certain exported goods in some countries, e.g. China. Thus, governments adopt strategies for regulating the VAT drawback rate for exports. For example, a country may increase the VAT drawback rate to promote the exporting of certain goods (e.g. mechanical and electrical products) whereas a country may decrease the VAT drawback rate for goods that were restricted to exporting (e.g. natural resources).

AsFig. 4illustrates, according to a DC in tax areas or in international logistics zones, the charge condition of sales VAT is that destination flows are in tax areas for the same

Fig. 1. Charge condition of import duty.

country, while the charge condition of input VAT is that neither original nor destination flows are in international logistics zones. Further, the charge condition of export VAT is that, in the same country, original flows are in tax areas, and destination flows are in international logistics zones; for different countries, both original and destina-tion flows are not in internadestina-tional logistics zones. Consequently, international logistic zones enable enter-prises to avoid government regulation strategies of export VAT.

2.3. Corporate tax

Corporate tax is the tax paid by enterprises on the profit they earn. For tax savings, goods completely manufactured in international logistics zones are exempt from corporate tax. Nevertheless, manufacturers must identify the most advantageous trade-off between corpo-rate tax and other costs (e.g., import duties).

3. Problem statement

To clarify the study scope and facilitate model formulation, the problem statement is postulated as

follows: (1) types of goods, (2) supply chain members and flow of goods, and (3) transactions.

3.1. Types of goods

Goods, ancestral to descendent, are classified in this study as modular components, semi-products and fin-ished products. Modular strategy has been discussed in detail elsewhere (Lamothe et al., 2006).

3.2. Supply chain members and flow of goods

Supply chain members include internal and external supply chain members responsible for different global logistics functions. Internal supply chain members are manufacturing centers and processing DCs, while external supply chain members are vendors and brand companies. Moreover, the number and location of all supply chain members are given. Nevertheless,Kerbache and MacGregor Smith (2004) indicated that manufacturers link their internal processes to external supply chain members and the resulting supply chain often comprises a very large network of activities and resources. Modeling and optimization of such complex systems is very difficult.

Fig. 3. Import duty and product forms.

Manufacturing centers receive modular components from the vendors and then transform modular compo-nents into semi-products or finished products. Further, manufacturing centers send semi-products or finished products to DCs.

The basic functions of DCs are consolidating and distributing finished products from manufacturing cen-ters to brand companies. Nevertheless, processing DCs may have simple or deep processing functions or both depending on their locations in international logistics zones or tax areas. Accordingly, DCs can be divided into three types: deep processing DCs, simple processing DCs and non-bonded DCs. AsFig. 5illustrates, deep processing DCs located in international logistics zones have both deep process and simple process functions. Here, deep process involves transforming modular components into semi-products, transforming modular components into finished products and transforming semi-products into finished products, while simple process involves simple processes of semi-products and finished products (e.g. transfer, assembling, and packaging). Although located in international logistics zones, simple processing DCs only have the functions of a simple process for semi-products and finished semi-products. Non-bonded DCs perform the same functions as deep processing DCs, but they are located in tax areas. Based on the function of DCs mentioned above, deep processing DCs and non-bonded DCs receive modular components from vendors. DCs may also receive semi-products or finished products from manufacturing centers. Further, semi-products or finished products can be transferred between all kinds of DCs.

This analysis assumes venders are below the top upstream suppliers in a typical supply chain. They receive and then process raw materials from upstream suppliers to manufacture modular components, which are then sent to manufacturing centers, deep processing DCs or non-bonded DCs.

Brand companies will request global manufacturers to distribute finished products to assigned locations around

the world. Assigned locations could be DCs or warehouses owned by brand companies.

3.3. Transactions

Since many brand companies often contract with global manufacturers for delivered duty paid (DDP) transactions, transactions in our model are based on the DDP value of the shipment. Herein, DDP means that the seller bears the risks and costs, including taxes, duties and other charges of transporting the goods until they have been delivered.

4. Modeling

Given the problem statement, a tax savings model is formulated to derive after-tax solutions that maximize profit in the emerging global manufacturing network. The proposed model is based on models developed byVidal and Goetschalckx (1998),Vidal and Goetschalckx (2000),

Vidal and Goetschalckx (2001), Fandel and Stammen (2004) and Vila et al. (2006). Nevertheless, once the after-tax model considers the emerging global manufac-turing network, determining the optimal tax savings route and manufacturing procedure for each order is difficult. Furthermore, three principal tax factors—import duty, value added tax and corporate tax—are considered simultaneously in the proposed model. This section is divided into two subsections—(1) objective function and (2) constraints. Appendix A summarizes notations and definitions, and Appendix B presents the equations of the proposed model.

4.1. The objective function

The objective function maximizes global after-tax profit in dollars for the period of analysis. The after-tax profit of internal supply chain members involved in the objective function are expressed in Eq. (2). The operating

income variables oiyx are free variables since operating

income may be positive, zero or negative. Accordingly, each variable is treated as the difference between a plus non-negative variable (operating profit) oiþ

yx¼oi_yx and a

minus non-negative variable (operating loss) oi yx¼ oi_yx

(Vidal and Goetschalckx, 1998, 2001;Fandel and Stammen, 2004;Vila et al., 2006).

Each operating income variable is measured by subtracting the corresponding aggregate costs costyx¼

ðP24_k¼1zk

yxÞ from the respective aggregate revenues

revenueyxð¼

p

1

yxþ

p

2_yxÞ, as Eq. (3) demonstrates.

Trading with internal supply chain members and brand companies produces the corresponding aggregate reven-ue, as expressed in Eqs. (4) and (5), respectively. Here, transfer price TPyx_ly_a is given to avoid costly

auditing and litigation. An effective method for obtaining market-driven transfer prices was proposed in Lakhal et al. (2005).

The aggregate cost is composed of 24 items. They are the corresponding aggregate costs in terms of transform-ing modular components into semi-products (Eq. (6)), transforming modular components into finished products (Eq. (7)), transforming semi-products into finished pro-ducts (Eq. (8)), simple process of semi-propro-ducts (Eq. (9)), simple process of finished products (Eq. (10)), transporta-tion cost of trading with internal supply chain members (Eq. (11)), transportation cost of trading with brand companies (Eq. (12)), inventory cost of trading with internal supply chain members (Eq. (13)), inventory cost of trading with brand companies (Eq. (14)), procurement cost of raw materials (Eq. (15)), procurement cost of semi-products or finished semi-products (Eq. (16)), fixed cost (Eq. (17)), sales VAT trading with internal members (Eq. (18)), sales VAT trading with brand companies (Eq. (19)), input VAT trading with vendors (Eq. (20)), input VAT trading with internal members (Eq. (21)), export VAT trading with internal members in the same country (Eq. (22)), export VAT trading with brand companies in the same country (Eq. (23)), export VAT trading with internal members in different countries (Eq. (24)), export VAT trading with brand companies in different countries (Eq. (25)), import duty trading with internal members in the same country (Eq. (26)), import duty trading with brand companies in the same country (Eq. (27)), import duty trading with internal members in different countries (Eq. (28)), import duty trading with brand companies in different countries (Eq. (29)). Note that Eqs. (20) and (21) are minus items as mentioned in Section 2. Furthermore, the term ½TTyx_ly_mþCSF  FSyx_ly_mþ SSFyx_apffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiTT_yx_ly_m in Eqs. (13) and (14) is the total time

required to calculate inventory costs (Vidal and Goetschalckx, 2000). Herein, the first term is the time required to measure the pipeline inventory; the second term is the time required to measure the cycle inventory; the third term is the time required to measure the safety stock (Vidal and Goetschalckx, 2000). The gamma dis-tribution was adopted in the safety stock for modeling stochastic lead times and inventory problems (Vidal and Goetschalckx, 2000). Additionally,Vidal and Goetschalckx (2000) performed a more exhaustive study of Eqs. (13) and (14).

4.2. Constraints

Given that corresponding logistics conditions are limited by operating requirements, eleven groups of constraints are the following: flow conservation of deep and simple process, inbound flow conservation, outbound flow conservation, identifying goods transformations, maximum goods transformation, assignment of goods, brand company requirements, capacity of chain members, subtour breaking constraints, binary constraints, and non-negative constraints. These constraints are further elabo-rated below.

1. Flow conservation of deep and simple process: AsFig. 5

shows, deep process, including transforming modular components into semi-products, transforming mod-ular components into finished products and trans-forming semi-products into finished products, are expressed as Eqs. (30)–(32), respectively. Simple process involving simple processing of semi-products and finished products are expressed as Eqs. (33) and (34), respectively.

2. Inbound flow conservation:Fig. 5shows three inbound flows: modular components, semi-products, and finished products. Consequently, the corresponding inbound flow constraints are expressed as Eqs. (35)–(37), respectively.

3. Outbound flow conservation: As Fig. 5 shows, two outbound flows are semi-products and finished products. Regarding finished products, manufacturing centers only can convey finished products to DCs, while DCs convey finished products to brand compa-nies or other DCs. Consequently, the corresponding outbound flow constraints are expressed as Eqs. (38)–(40), respectively.

4. Identifying goods transformations: For the sake of rational goods transformations and assignments, the expression gotryxab represents good transformations,

including transformations from modular components into semi-products, from modular components into finished products and from semi-products into fin-ished products. Accordingly, the corresponding con-straints on goods transformations are expressed in Eqs. (41)–(43), respectively.

5. Maximum goods transformation: Eqs. (41)–(43) ensure only that if goods transformation occurs, the sum of gotryx_ab equals or exceeds one. Consequently, it is

necessary to limit the maximum number of goods transformations, including those from modular com-ponents into semi-products, from modular compo-nents into finished products, and from semi-products into finished products. Thus, these constraints are expressed as Eqs. (44)–(46), respectively.

6. Assignment of goods: Each modular component can only be used once. Only one of two manufacturing procedures, including from modular components into either semi-products or finished products, can be used. Therefore, the corresponding constraint is given by Eq. (47). Similarly, since each semi-product can only be used once, the corresponding constraint is

given by Eq. (48). Finished products can be transferred among DCs, but one finished product only can be assigned once to a brand company. Restated, one company can only receive one unique finished product during the assignment process. Accordingly, the corresponding constraint is given by Eq. (49). 7. Brand company requirements: To meet brand company

requirements, the corresponding constraint is given by Eq. (50).

8. Capacity of chain members: In addition to vender capacity to supply modular components (Eq. (51)), there are five capacities of internal supply chain members for goods transformation, including from modular components into semi-products, from mod-ular components into finished products, from semi-products into finished semi-products, simple process of semi-products and simple process of finished pro-ducts. Accordingly, the corresponding constraints on five capacities of internal supply chain members are expressed as Eqs. (52)–(56), respectively.

9. Subtour breaking constraints: Since goods can transfer among DCs, Eq. (57) prohibits a formation of any subtour among them.

10. Binary constraints: Constraints denoted by Eqs. (58)–(62) indicate that those variables are binary.

11. Non-negative constraints: Constraints denoted by Eqs. (63) and (64) indicate that operating income variables are non-negative variables.

5. Numerical illustration

To test the applicability and the solvability of the proposed model, a simplified numerical study was conducted by interview. Table 1 outlines the main characteristics of the basic scenario. Moreover, country 1 has a lower logistics cost (such as deep processing costs) and greater processing capacity (such as deep processing capacity) than countries 2 and 3 in the basic scenario.

Fig. 6 displays five main patterns of the numerical results of logistics behavior. First, modular components

were shipped from vendor (no. 2) to deep processing DC (no. 7) or deep processing DC (no. 8). Second, deep processing take place at deep processing DC (no. 7) or deep processing DC (no. 8) to transform modular components into semi-products. Third, semi-products were shipped from deep processing DC (no. 7) to another deep processing DC (no. 8), and simple processing of semi-products then took place at deep processing DC (no. 8). Fourth, semi-products were shipped from deep processing DC (no. 7) to non-bonded DC (no. 9) or from deep processing DC (no. 8) to non-bonded DC (no. 10) for further transformation of semi-products into finished products. Finally, finished products are shipped from non-bonded DC (no. 9) to brand company (no. 11) or from non-bonded DC (no. 10) to brand company (no. 12). Furthermore, some internal supply members operate at a profit (nos. 7, 9, 10), and others operate at a loss (nos. 3–6, 8). More precisely, Table 2 presents an example of the steps in the creation of the finished product (no. 33) to meet the requirement of the brand company (no. 12). Herein, Fig. 7 displays an example of the deep process and the simple process regarding deep processing DC (no. 8).

To further examine logistics behavior, three extended scenarios and their numerical results are briefly narrated as follows. First, if tax areas were exempt from corporate tax as international logistics zones, most finished pro-ducts would be directly manufactured at non-bonded DCs located close to brand companies (extended scenario) rather than at deep processing DCs in international logistics zones (basic scenario). Second, if tax areas were exempted from import duty as international logistics zones, most finished products would be directly manufactured in international logistics zones (extended scenario) rather than in tax areas (basic scenario). Third, if country 2 has the same logistics cost and processing capacity as country 1, then semi-products would not be shipped from country 1 to country 2 (extended scenario). Based on the numerical results of logistics behavior mentioned above, some important findings are summar-ized and discussed as follows. First of all, most semi-products are manufactured in international logistics zones. The main reason for this result may be that goods manufactured in those zones are exempt from corporate tax. Secondly, most domestic (non-bonded) DCs import semi-products from international logistics zones, since import duties are lower for semi-products than for finished products. A similar concept was discussed in

Simchi-Levi et al. (2003). Finally, the model demonstrates that most manufacturing behavior occurs in country 1, and semi-products are then shipped from country 1 to country 2. In reality, this may be owing to that manufacturers relocated their main processing capacity to low-cost zones (e.g. China) and has a lower processing capacity in the proximity of customers or in R&D zones (e.g. Taiwan). Similar situations are apparent elsewhere (Chia et al., 2001, 2002;Sheu, 2003).

Fig. 8 presents the results of sensitivity analysis conducted by varying tax parameters such as corporate tax rate, duty, VAT rate and VAT drawback rate. Expect-edly, lower corporate tax rate, lower VAT rate and lower

Table 1

Main characteristics of the basic scenario. Characteristics Design value Set of supply chain members FAV

¼{1,2}; FAM ¼{3}; FADs ¼{4,5,6}; FADd ¼{7,8}; FADn ¼{9,10}; FAB ¼{11, 12} Set of goods Gr ¼{1,y,20}; Gs ¼{21,y,30}; Gp ¼{31,y,35} Set of countries N ¼ {1,2,3} Set of simple and deep

process product lines

SNrs I ¼ f1g; SNrpI ¼ f2g; SN sp I ¼ f3g; SNss I ¼ f4g; SNppI ¼ f5g; SNrsO¼ f6g; SNrpO¼ f7g; SN sp O¼ f8g; SNssO¼ f9g; SNpp O ¼ f10g

Set of transportation modes O¼{air transportation: 1, sea transportation: 2, truck: 3} Required finished products BR111¼3 (orders); BR₁₂2¼2 (orders)

Equivalent of goods BOMrs¼2; BOMsp¼2; BOMrp¼4 Note: Per order of 100 goods.

duty all tend to increase after-tax profit. Notably, the VAT drawback rate does not affect after-tax profit, since lack of logistics behavior meets the charge condition of VAT drawback rate in the basic scenario. This finding also reveals that manufacturers can avoid government regula-tion strategies of VAT drawback rate through operating in international logistics zones. Overall, the sensitivity analy-sis demonstrates the robustness of the proposed model, and most tax factors are sensitive to after-tax profit. The above tax factors would be of importance to manufacturers seeking to maximize profit by global logistics strategies.

6. Conclusion

The emerging global manufacturing network involves nodal location features of tax areas and international logistics zones, manufacturing procedures of simple process and deep process in these nodes, as well as transportation arcs. This study presented several tax

savings approaches and developed a tax savings model for the emerging global manufacturing network. The numerical illustration demonstrates that the model is valid and viable as an analytical tool for global manu-factures. The major decision-making parameters can be tailored to specific global manufacturers.

The numerical illustration reveals the following crucial findings. First, manufacturers can produce goods in international logistics zones to save corporate tax. Second, manufacturers can import ancestor goods (e.g. semi-products) with lower duty rates and transform them into descendant goods (e.g. finished products) in tax areas to save duty. Third, manufacturers can operate in interna-tional logistics zones to avoid government regulation of VAT drawback rate. Finally, most manufacturing behavior occurs in zones with lower logistics costs and greater processing capacity to maximize their global profit.

This study differs from previous studies addressing profit-maximizing problems in several ways. First, this study examined three primary tax factors associated with operating income—import duty, value added tax and corporate tax—via in-depth interviews with global man-ufacturers. Furthermore, tax savings approaches for the emerging global manufacturing network were also dis-cussed. Second, the tax savings model for the emerging global manufacturing network helps manufacturers iden-tify solutions that maximize after-tax profit. The proposed model can determine the optimal tax savings route and manufacturing procedure for each order. For tax savings, the proposed model allows goods free transfer among processing DCs. Additionally, three principal tax factors are considered simultaneously in the proposed model. Global manufacturers can develop strategies using the proposed model for maximizing preferential tax treat-ment in international logistics zones to achieve tax savings. Moreover, the proposed model identifies the critical logistics behavior associated with tax savings. The proposed model may stimulate further research in the field of global logistics and may help address issues regarding tax savings and international logistics zones.

Fig. 6. Numerical results of logistics behavior.

Table 2

Behavior of P33 for delivering finished products to brand company 12. R6–S23: qu(2, 8, 1, 6)-gt(8, 6, 1)-gotr(8, 6, 23)-gt(8, 23, 6) R10–S23: qu(2, 8, 1, 10)-gt(8, 10, 1)-gotr(8, 10, 23)-gt(8, 23, 6) R7–S24: qu(2, 7, 1, 7)-gt(7, 7, 1)-gotr(7, 7, 24)-gt(7, 24, 6) R13–S24: qu(2, 7, 1, 13)-gt(7, 13, 1)-gotr(7, 13, 24)-gt(7, 24, 6) S24–S24: qu(7, 8, 2, 24)-gt(8, 24, 4)-gt(8, 24, 9) S23–P33: qu(8, 10, 3, 23)-gt(10, 23, 3)-gotr(10, 23, 33)-gt(10, 33, 8) S24–P33: qu(8, 10, 3, 24)-gt(10, 24, 3)-gotr(10, 24, 33)-gt(10, 33, 8) P33: qu(10, 12, 3, 33)

Note: R indicates modular components; S indicates semi-products; P indicates finished products. One dash linking two goods means deep process (e.g., R6–S23) or simple process (e.g., S24–S24) while single goods indicates shipping goods (e.g., P33). Here, qu, gu and gotr are main decision variables. Four terms within the qu bracket represent former member (origin), latter member (destination), transportation mode and goods, respectively. Three terms within the gt bracket denote chain member, goods and product line, respectively. Three terms within the gotr bracket indicate chain member, ancestor goods, and descendant goods, respectively.

Future studies may also incorporate quotas, certificate of origin and local content into the tax savings model. The model may also be extended to a product family and its bill-of-materials (BOM). Moreover, large-scale instances of profit-maximizing problems in a numerical study should be carefully generated to approximate reality as much as possible. The continuing relevance of the proposed model is expected in further studies.

Acknowledgments

This research was supported by Grant NSC 95-2415-H009-003-MY2 from the National Science Council of Taiwan. The valuable suggestions of Professor Mu-Chen Chen to improve this paper are gratefully acknowledged. Also, special thanks to Professor Marc Goetschalckx for his helpful report in conducting this study. Finally, the authors are deeply indebted to numerous personnel in the examined global manufacturers for their cooperation in interviews and scenario design. Any errors or omissions remain the sole responsibility of the authors.

Appendix A. Notations and definitions

All the notations and definitions for sets, decision variables and parameters are summarized as follows.

A.1. Sets

FA Set of internal and external supply chain members. Herein, FAV_{: set of}

vendors (abbreviated as V); FAM_{: set of manufacturing centers}

(abbreviated as M); FADs: set of simple processing DCs (abbreviated as Ds; FADd_{: set of deep processing DCs (abbreviated as Dd); FA}Dn_{: set}

of non-bonded DC (abbreviated as Dn); FAB_{: set of brand companies}

(abbreviated as B); FAI_{: set of all internal supply chain members}

(abbreviated as I); FADC: set of all DCs, including deep processing DCs, simple processing DCs and non-bonded DCs (abbreviated as DC); FAC: set of chain members in international logistics zones

(abbreviated as C); FAT: set of chain members in tax areas

(abbreviated as T)

G Set of types of goods. Herein, Gr_{: set of modular components}

(abbreviated as r); Gs: set of semi-products (abbreviated as s); Gp: set of finished products (abbreviated as p)

N Set of countries

SN Set of simple and deep process product lines. Herein, SNrs_{: set of}

product lines involving transformation of modular components r into semi-products s (abbreviated as rs); SNrp_{: set of product lines}

involving transformation of modular components r into finished products p (abbreviated as rp); SNsp: set of product lines involving transformation of semi-products s into finished products p (abbreviated as sp); SNss_{: set of product lines involving simple}

processing of semi-products s (abbreviated as ss); SNpp_{: set of}

product lines involving simple processing of finished products p (abbreviated as pp); SNI: set of product lines of inbound flow

(abbreviated as I); SNO: set of product lines of outbound flow

(abbreviated as O); SNR: set of product lines of sum of corresponding modular components (abbreviated as R); SNS_{: set of product lines of}

Fig. 7. An illustrative example of the deep process and the simple process.

sum of corresponding semi-products (abbreviated as S); SNP_{: set of}

product lines of sum of corresponding finished products (abbreviated as P)

O Set of transportation modes. Here,Oðyx_;ly

Þis the set of available transportation modes between a given chain memberyin the country x 2 N and another given chain memberlin the country y 2 N

A.2. Decision variables

gotryx_ab Binary decision variable indicates whether goods

transformation occurs at a given chain memberyin country x 2 N when transferring ancestor goodsainto descendant goodsb

gtyx_af Binary decision variable indicates whether goodsais in

progress in a product line f at a given chain memberyin country x 2 N

oiyx Operating income of a given chain memberyin country x 2 N

for the period of analysis (dollar/unit of time)

ordyx_a Number of nodes visited on the transfer path from the origin up

to nodeyx_{for goodsa(i.e., the visit number of they}x_{th node)}

quyx_ly_maBinary decision variable representing whether goodsa2G is shipped from a given chain memberyin country x 2 N to another given chain memberlin country y 2 N, using transportation mode m 2Oðyx;lyÞ.

A.3. Parameters

BRly Required finished products for a given chain memberlin

country y 2 N (units of p/unit of time)

BOMab Units of ancestor goodsa2G required to make one unit of

descendant goods B 2 G (a-units/b-unit)

BN A big number

CSF Cycle stock factor (%)

COTyx Corporate tax rate (%) of country x 2 N of a supply chain

membery

CPRICEa International contract price of goodsa2G (dollar/unit of goodsa)

DRTyx_a Value added tax drawback rate (%) on the value of goodsa2G of country x 2 N of supply chain membery

DUTYyx_ly_aImport duty rate (%) on the value of goodsa2G shipped from a given chain memberyin country x 2 N to another given chain memberlin country y 2 N

Eyx Exchange rate of country x 2 N of supply chain membery

(monetary units of the respective country/dollar)

FIXyx Fixed cost associated with a given chain member_yin country

x 2 N (monetary units of country of memberyper unit of time)

FSyx_ly_{m Frequency of goods shipments from a given chain membery}

in country x 2 N to another given chain memberlin country y 2 N, using transportation mode m 2Oðyx;lyÞ(units of time ) H Holding cost ($/($. unit of time))

IVyx_a Inventory value of goodsa2G, given in monetary units of a given chain memberyin country x 2 N per unit of goodsa

NODE Number of DC nodes

PROCly_yx_a Procurement cost (including total cost and taxes) of goods

a2G shipped from a given chain memberlin country y 2 N to another given chain memberyin country x 2 N (monetary units of country of memberl/unit of goodsa)

PPAyx Simple processing capacity of finished products in a given

chain memberyin country x 2 N (finished product units/unit of time)

PPCyx Simple processing cost of finished products in a given chain

memberyin country x 2 N (monetary units of country of membery/unit of finished product)

RSAyx Capacity to transform goods associated with a given chain

memberyin country x 2 N for transferring modular components into semi-products (semi-product units/unit of time)

RSCyx Cost of transforming goods associated with a given chain

memberyin country x 2 N for transferring modular

components into semi-products (monetary units of country of membery/unit of semi-products)

RPAyx Capacity to transform goods associated with a given chain

memberyin country x 2 N for transferring modular components into finished products (finished product units/ unit of time)

RPCyx Cost of transforming goods associated with a given chain

memberyin country x 2 N for transferring modular components into finished products in country x 2 N (monetary units of country of membery/unit of finished products)

SPAyx Capacity to transform goods associated with a given chain

memberyin country x 2 N for transferring semi-products into finished products (finished product units/unit of time) SPCyx Cost of transforming goods associated with a given chain

memberyin country x 2 N for transferring semi-products into finished products (monetary units of country of member

y/unit of finished products)

SSAyx Simple processing capacity of semi-product in a given chain

memberyin country x 2 N (semi-product units/unit of time) SSCyx Simple processing cost of semi-product in a given chain

memberyin country x 2 N (monetary units of country of membery/unit of s)

SSFyx_a Safety stock factor of goodsa2G at a given chain membery

in country x 2 N

TPyx_a Transfer price of goodsa2G shipped from a given chain memberyin country x 2 N (monetary units of country of membery/unit of goodsa)

TRCyx_ly_m Transportation cost per weight unit of goods shipped from a

given chain memberyin country x 2 N to another given chain memberlin country y 2 N, using transportation mode m 2

Oðyx;lyÞ(monetary units of country of membery/weight unit)

TTyx_ly_m Average transportation time from a given chain memberyin

country x 2 N to another given chain memberlin country y 2 N, using transportation mode m 2Oðyx_;ly_{Þ(units of time)}

VCyx Capacity of a given chain memberyin country x 2 N for

supplying modular components (modular component units/ unit of time)

VATyx_a Value added tax rate (%) on the value of goodsa2G of country x 2 N of supply chain membery

Wa Weight of a unit of goodsa2G (weight units/unit of goods)

Appendix B. The equations of the proposed model

The objective function and constraints of the proposed model are presented as follows:

Maximize X yx_2FAI ½ð1  COTyxÞoiþ yxoi  yx (2) oiþ yxoi  yx¼revenue_yxcost_yx 8

y

x2FAI (3)

p

1 yx¼ X ly_ðyx_aly_Þ2FAI X m2Oðyx;lyÞ X a2Gsp 1 Eyx TPyx_aqu_yx_ly_ma 8

y

x2FAI (4)

p

2 yx¼ X ly_2FAB X m2Oðyx ;ly_Þ X a2Gp CPRICEaquyx_ly_ma 8

y

x2FADC (5) z1 yx¼ X a2Gs X f 2SNrs O 1 Eyx RSC_yxgt_yx_a f 8

y

x2 fFAM;FADd;FADng (6)

z2 yx¼ X a2Gp X f 2SNrp O 1 Eyx RPC_yxgt_yx_a f 8

y

x2 fFAM;FADd;FADng (7) z3 yx¼ X a2Gp X f 2SNsp O 1 Eyx SPCyxgtyxaf 8

y

x 2 fFADd;FADng (8) z4 yx¼ X a2Gs X f 2SNss O 1 Eyx SSCyxgt_yx_af 8

y

x2FADC (9) z5_yx¼ X a2Gp X f 2SNpp O 1 Eyx PPCyxgtyxaf 8

y

x 2FADC (10) z6 yx¼ X ly_ðyx_aly_Þ2FAI X m2Oðyx ;ly_Þ X a2GSP 1 Eyx TRCyx_ly_m Waquyx_ly_ma 8

y

x2FAI (11) z7 yx¼ X ly_2FAB X m2Oðyx ;ly_Þ X a2GP 1 Eyx TRCyx_ly_mW_aqu_yx_ly_ma 8

y

x2FADC (12) z8 yx¼ X ly_ðyx_aly_Þ2FAI X m2Oðyx;lyÞ X a2GSP IVyx_aH Eyx ½TTy x_ly_mþCSF FSyx_ly_mþSSFyx_apffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiTT_yx_ly_mqu_yx_ly_ma 8

y

x2FAI (13) z9 yx¼ X ly_2FAB X m2Oðyx;lyÞ X a2GP IVyx_aH Eyx ½TTy x_ly_mþCSF  FS_yx_ly_m þSSFyx_apffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiTT_yx_ly_mqu_yx_ly_ma 8

y

x2FADC (14) z10 yx ¼ X ly2FAV X m2Oðly_;yx_Þ X a2Gr 1 E_lyPROCl y_yx_aquly_yx_ma 8

y

x2 fFAM;FADd;FADng (15) z11 yx ¼ X ly_ðyx_aly_Þ2FAI X m2Oðly ;yx_Þ X a2Gsp 1 Ely TPly_aqu_ly_yx_ma 8

y

x2FAI (16) z12 yx ¼ 1 Eyx FIXyx 8

y

x2FAI (17) z13 yx ¼ X lyðyxaly&x¼yÞ2FAI T X m2Oðyx_;ly_Þ X a2Gsp 1 Eyx TPyx_aquyx_ly_ma VATyx_a 8

y

x2FAI (18) z14 yx ¼ X lyðx¼yÞ2FAB T X m2Oðyx_;ly_Þ X a2GP CPRICEaquyxlymaVATyxa 8

y

x2FADC (19) z15 yx ¼ X ly2FAV_;yx &lyeFAC X m2Oðly;yxÞ X a2Gr 1 ElyPROCl y_yx_a qulyyxmaVATlya 8

y

x 2 fFAM;FADd;FADng (20) z16 yx ¼ X lyðyxalyÞ2FAI_;yx &lyeFAC X m2Oðly ;yx_Þ X a2Gsp 1 Ely TP_ly_a qu_ly yxmaVATlya 8

y

x 2FAI ₍₂₁₎ z17 yx ¼ X lyðyxaly&x¼yÞ2FAI C X m2Oðyx ;ly_Þ X a2Gsp 1 Eyx TPyx_aquyx_ly_ma  ðVATyx_aDRTyx_aÞ 8

y

x2FAIT (22) z18 yx ¼ X ly_ðyx_aly_&x¼yÞ2FAB C X m2Oðyx ;ly_Þ X a2Gp CPRICEaquyx_ly_ma  ðVATyx_aDRTyx_aÞ 8

y

x2FADCT (23) z19 yx ¼ X lyðyxaly&xayÞ2FAI ;yx&lyeFAC X m2Oðyx;lyÞ X a2Gsp 1 Eyx TPyx_a quyx_ly_ma ðVAT_yx_aDRTyx_aÞ 8

y

x2FAI (24) z20 yx ¼ X

ly_ðyx_aly_&xayÞ2FAB_;yx_&ly_eFA C X m2Oðyx ;ly_Þ X a2Gp CPRICEa quyxlyma ðVATyxaDRTyxaÞ 8

y

x 2FADC (25) z21 yx ¼ X ly_ðyx_aly &x¼yÞ2FAIT X m2Oðyx ;ly_Þ X a2Gsp 1 Eyx  ðTP_yx_aþTRC_yx lym WaÞ quyx_ly_maDUTYyx_ly_a 8

y

x2FAIC (26) z22 yx ¼ X lyðyxaly&x¼yÞ2FAB T X m2Oðyx ;ly_Þ X a2Gp CPRICEaþ 1 Eyx TRCyxlymWa   quyx_ly_maDUTY_yx_ly_a 8

y

x2FADC_C (27) z23 yx ¼ X ly_ðyx_aly_&xayÞ2FAI T X m2Oðyx ;ly_Þ X a2GSP 1 Eyx ðTPy x_aþTRC_yx_ly_m WaÞ quyxlymaDUTYyxlya 8

y

x 2FAI (28) z24 yx ¼ X ly_ðyx_aly_&xayÞ2FAB T X m2Oðyx ;ly_Þ X a2GP CPRICEaþ 1 Eyx TRCyx_ly_mW_a   qu_yx lymaDUTYyxlya 8

y

x 2FADC ₍₂₉₎ X a2Gr X f 2SNrs I gtyxaf¼ X b2Gs X f 2SNrs O gtyxbfBOMab 8

y

x2 fFAM;FADd;FADng (30) X a2Gr X f 2SNrp I gtyxaf¼ X b2Gp X f 2SNrp O gtyxbfBOMab 8

y

x2 fFAM;FADd;FADng (31) X a2Gs X f 2SNsp_I gtyx_af¼X b2Gp X f 2SNsp_O gtyx_bfBOMab 8

y

x2 fFADd;FADng (32)

X f 2SNss I gtyxaf¼ X f 2SNss O gtyxaf 8

y

x 2 fFADd;FADs;FADng 8

a

2Gs (33) X f 2SNpp I gtyx_af¼ X f 2SNpp O gtyx_af 8

y

x2 fFADd;FADs;FADng 8

a

2Gp (34) X yx2FAV X m2Oðyx_;ly_Þ quyx_ly_ma¼ X f 2SNR I gtly_af 8

l

y2 fFAM_;FADd_;FADn_g;

_a

2Gr ₍₃₅₎ X yxðyxalyÞ2FAI X m2Oðyx;lyÞ quyx_ly_ma¼ X f 2SNS I gtly_af 8

l

y2 fFADCg;

a

2Gs (36) X yx_ðyx_aly_Þ2FAI X m2Oðyx ;ly_Þ quyx_ly_ma¼ X f 2SNP I gtly_af 8

l

y2 fFADC_g;

_a

2Gp ₍₃₇₎ X f 2SNS O gtyx_af¼ X lyðyxalyÞ2FADC X m2Oðyx;lyÞ quyx_ly_ma 8

y

x2FAI;

a

2Gs (38) X f 2SNP O gtyx_af¼ X ly_2FADC X m2Oðyx ;ly_Þ quyx_ly_ma 8

y

x2FAM_;

_a

2Gp ₍₃₉₎ X f 2SNP O gtyx_af¼ X lyðyxalyÞ2fFADC_;FAB_g X m2Oðyx_;ly_Þ quyx_ly_ma 8

y

x2FADC;

a

2Gp (40) P a2Gr gotr_yx_ab P f 2SNrs O gt_yx bfnBN P f 2SNrs O gt_yx bf P a2Gr gotr_yx_abnBN 8 > > < > > : 8

y

x2 fFAM;FADd;FADng;

b

2Gs (41) P a2Gr gotryx_ab P f 2SNrp O gt_yx bfnBN P f 2SNrp_O gt_yx_bf P a2Gr gotr_yx_abnBN 8 > > > < > > > : 8

y

x2 fFAM;FADd;FADng;

b

2Gp (42) P a2Gs gotr_yx_ab P f 2SNsp_O gt_yx bfnBN P f 2SNsp O gtyxbf P a2Gs gotryxabnBN 8 > > > < > > > : 8

y

x2 fFADd;FADng;

b

2Gp (43) X

yx2fFAM_;FADd_;FADn_g

X a2Gr gotryx_ab X a2Gr BOMab 8

b

2Gs (44) X

yx_2fFAM_;FADd_;FADn_g

X a2Gr gotryx_abX a2Gr BOMab 8

b

2Gp (45) X yx2fFADd_;FADn_g X a2Gs gotryx_ab X a2Gs BOMab 8

b

2Gp (46) X

yx_2fFAM_;FADd_;FADn_g

X

b2Gs

gotryxabþ

X

yx_2fFAM_;FADd_;FADn_g

X o2Gp gotryxao 1 8

a

2Gr (47) X yx2fFADd_;FADn_g X b2Gp gotryx_ab1 8

a

2Gs (48) X yx_2FADC X ly_2FAB X m2Oðyx ;ly_Þ quyx_ly_ma¼1 8

a

2Gp (49) X yx_2FADC X m2ðyx ;ly_Þ X a2Gp quyx_ly_ma¼BRly 8

l

y2FAB (50) X

ly2fFAM_;FADd_;FADn_g

X m2Oðyx;lyÞ X a2Gr quyx_ly_maVC_yx 8

y

x2FAV (51) X a2Gs X f 2SNrs O gtyxafRSAyx 8

y

x 2 fFAM;FADd;FADng (52) X a2Gp X f 2SNrp_O gtyx_afRPA_yx 8

y

x2 fFAM;FADd;FADng (53) X a2Gp X f 2SNsp O gtyxafSPAyx 8

y

x 2 fFADd;FADng (54) X a2Gs X f 2SNss O gtyx_afSSA_yx 8

y

x2 fFADCg (55) X a2Gp X f 2SNpp O gtyx_afPPA_yx 8

y

x2 fFADCg (56)

ordyxaordlyaþNODE

X m2Oðyx ;ly_Þ quyxlymaNODE  1 8

y

xa

l

y2FADC;

a

2 fGs;Gpg (57) gotryxab2 f0; 1g 8

y

x 2 fFAM;FADd;FADng;

a

2Gr;

b

2Gs (58) gotryx_ab2 f0; 1g 8

y

x2 fFAM;FADd;FADng;

a

2Gr;

b

2Gp (59) gotryx_ab2 f0; 1g 8

y

x2 fFADd;FADng;

a

2Gs;

b

2Gp (60) gtyx_af2 f0; 1g 8

y

x2FAI;

a

2G; f 2 SN (61) quyx_ly_ma2 f0; 1g 8

y

x2FA;

l

y2FA; m 2

O

;

a

2G (62) oiþ yx0 8

y

x 2FAI (63) oi yx0 8

y

x2FAI (64)

References

Albino, V., Izzo, C., Ku¨htz, S., 2002. Input–output models for the analysis of a local/global supply chain. International Journal of Production Economics 78 (2), 119–131.

Arntzen, B., Brown, G., Harrison, T., Trafton, L., 1995. Global supply chain management at digital equipment corporation. Interfaces 21 (1), 69–93.

Avittathur, B., Shah, J., Gupta, O.K., 2005. Distribution centre location modelling for differential sales tax structure. European Journal of Operational Research 162 (1), 191–205.

Bowersox, D.J., Closs, D.J., 1996. Global logistics. In: Bowersonx, D.J., Closs, D.J. (Eds.), Logistical Management: The Integrated Supply Chain Process. McGraw-Hill, Singapore, pp. 129–135.

Cavusgi, S.T., Yeniyurt, S., Townsend, J.D., 2004. The framework of a global company: a conceptualization and preliminary validation. Industrial Marketing Management 33 (8), 711–716.

Chia, K.-C., Feng, C.-M., Chang, C.-J., 2001. The effects of global logistics strategies on international freight transportation demand. Journal of Eastern Asia Society for Transportation Studies 6 (6), 343–358.

Chia, K.-C., Feng, C.-M., Chang, C.-J., 2002. Survey analysis of supply chain adjustment for Taiwanese information technology firms. Transport Reviews 22 (4), 473–497.

DHL, 2006. China logistics service overview. Greater China Cross Trade Customer Seminar, Taipei.

Das, K., Sengupta, S., 2009. A hierarchical process industry production–distribution planning model. International Journal of Production Economics 117 (2), 402–419.

Fandel, G., Stammen, M., 2004. A general model for extended strategic supply chain management with emphasis on product life cycles including development and recycling. International Journal of Production Economics 89 (3), 293–308.

Goetschalckx, M., Vidal, C.J., Dogan, K., 2002. Modeling and design of global logistics systems: a review of integrated strategic and tactical models and design algorithms. European Journal of Operational Research 143 (1), 1–18.

Hameri, A.-P., Paatela, A., 2005. Supply network dynamics as a source of new business. International Journal of Production Economics 98 (1), 41–55.

Hammami, A., Burlat, P., Campagne, J.P., 2003. Evaluating orders allocation within networks of firms. International Journal of Produc-tion Economics 86 (3), 233–249.

Hiraki, S., 1996. Comparative analysis of ordering models for an international co-operative global complementary production sys-tem. International Journal of Production Economics 44 (1-2), 105–117.

Hu¨lsmann, M., Grapp, J., Li, Y., 2008. Strategic adaptivity in global supply chains—competitive advantage by autonomous cooperation. Inter-national Journal of Production Economics 114 (1), 14–26.

Jodlbauer, H., 2008. Customer driven production planning. International Journal of Production Economics 111 (2), 793–801.

Kerbache, L., MacGregor Smith, J., 2004. Queueing networks and the topological design of supply chain systems. International Journal of Production Economics 91 (3), 251–272.

Lakhal, S.Y., H’Mida, S., Venkatadri, U., 2005. A market-driven transfer price for distributed products using mathematical programming. European Journal of Operational Research 162 (3), 690–699. Lamothe, J., Hadj-Hamou, K., Aldanondo, M., 2006. An optimization

model for selecting a product family and designing its supply chain. European Journal of Operational Research 169 (3), 1030–1047. Lee, H., Yang, H.M., 2003. Strategies for a global logistics and economic

hub: Incheon International Airport. Journal of Air Transport Manage-ment 9 (2), 113–121.

Lu, C.-S., Yang, C.-C., 2007. An evaluation of the investment environment in international logistics zones: a Taiwanese manufacturer’s per-spective. International Journal of Production Economics 107 (1), 279–300.

Meixell, M.J., Gargeya, V.B., 2005. Global supply chain design: a literature review and critique. Transportation Research Part E 41 (6), 531–550. Mezga´r, I., Kova´cs, G.L., Paganelli, P., 2000. Co-operative production planning for small- and medium-sized enterprises. International Journal of Production Economics 64 (1-3), 37–48.

Oum, T.H., Park, J.H., 2004. Multinational firms’ location preference for regional distribution centers: focus on the Northeast Asian region. Transportation Research Part E 40 (2), 101–121.

Power, D., 2005. Supply chain management integration and implemen-tation: a literature review. Supply Chain Management: An Interna-tional Journal 10 (4), 252–263.

Prasad, S., Sounderpandian, J., 2003. Factors influencing global supply chain efficiency: implications for information systems. Supply Chain Management: An International Journal 8 (3), 241–250.

Sheu, J.-B., 2003. Locating manufacturing and distribution centers: an integrated supply chain-based spatial interaction approach. Trans-portation Research Part E 39 (5), 381–397.

Sheu, J.-B., 2004. A hybrid fuzzy-based approach for identifying global logistics strategies. Transportation Research Part E 40 (1), 39–61. Simchi-Levi, D., Kaminsky, P., Simchi-Levi, E., 2003. Coordinated product

and supply chain design. In: Levi, D., Kaminsky, P., Simchi-Levi, E. (Eds.), Designing and Managing the Supply Chain: Concepts, Strategies, and Case Studies. McGraw-Hill, Singapore, pp. 207–235. Tsiakis, P., Papageorgiou, L.G., 2008. Optimal production allocation and

distribution supply chain networks. International Journal of Produc-tion Economics 111 (2), 468–483.

Vidal, C.J., Goetschalckx, M., 1998. A heuristic for the design of global supply chains with transfer pricing and transportation cost alloca-tion. Research Report, School of Industrial and Systems Engineering, Georgia Institute of Technology, Atlanta, Georgia.

Vidal, C.J., Goetschalckx, M., 2000. Modeling the effect of uncertainties on global logistics systems. Journal of Business Logistics 21 (1), 95–120.

Vidal, C.J., Goetschalckx, M., 2001. A global supply chain model with transfer pricing and transportation cost allocation. European Journal of Operational Research 129 (1), 134–158.

Vila, D., Martel, A., Beauregard, R., 2006. Designing logistics networks in divergent process industries: a methodology and its application to the lumber industry. International Journal of Production Economics 102 (2), 358–378.

Zhai, E., Shi, Y., Gregory, M., 2007. The growth and capability develop-ment of electronics manufacturing service (EMS) companies. Inter-national Journal of Production Economics 107 (1), 1–19.