ey © mei PI: S1366-5548(98)00012-X STRATEGIC ANALYSIS OF LOGISTICS AND SUPPLY CHAIN MANAGEMENT SYSTEMS USING THE ANALYTICAL NETWORK PROCESSt LAURA MEADE* ‘The University of Tera at Arlingion, Fort Worth, TX, USA. and JOSEPH SARKIS ‘The Graduate School of Management, Crk University, Worcester, USA. (Received 4 July 1996; revised form ! October 1997; aceped 23 April 1998) Abert —This paper explore end ilustats an analytical framework to ates an organization's logitice ‘tratey, The opiom lgateal rye it eveluted based on tree primary levels: the organiaonl/ppiy ‘an reltionsip involved, the principles of logics requled, andthe atsibutes af tase principle. The ncipe of logics sre defined and developed satis or achieving coordination and tegration ofthe {ogistes network and suply chain. The analytic network procs, asjstemic analycal model wil be uid torevlute logics strates for an organization that seks tobe adap to dynamic compete exon: ments. © 1998 Ehevir Scene Lid. All righis reserved -xywords analytical network process (ANP), analytical Neachy process (AHP), logis strategy, supply ‘am management 1. INTRODUCTION Logistics plays an increasingly important strategic role for organizations that strive to Keep pace ‘with market changes and supply chain integration. Traditionally, supply management and logistics Ihave been delegated to operational level personnel in purchasing and distribution departments Logistics and supply management are currently evolving due to external factors such as strategic alliances, technological changes, cycle time compression, and the increasing competitive environ- ment (Bowersox, 1990; Daugherty and Pittman, 1995; LaLonde and Mason, 1993). There have also been changing internal factors such as the implementation of decision support systems, information systems integration, spanning of logistics to impact traditional functional areas, and increasing performance expectations (LaLonde and Mason 1993). Recently, much work has been targeted to the area of logistics strategic development and its categorical modelling. A framework for logistics research includes strategy, structure, and perfor- mance (Chow, et a, 1995). Strategy can be defined as plans to meet relatively long-term organi zational objectives that have broad corporate functional implications. These developments explain hhow a successful logistics strategy has moved from an internal focus emphasizing integration with other enterprise functions, such as production and marketing, with a linkage to the overall cor- porate strategy, to an extemal focus of integrating supply chains and cycle time compression. The ‘complexity of logistics strategic decisions and choices has increased with the number of dimensions that need to be considered. Logistics now includes comprehensive supply chains which require coordination, if not optimization. hath fo corependece Fax: 001 817 272 S801; e-mail: Imeade@arrirs.uta edu SE Sree arated SOF cs i BSE, Fam te Enns ce a —— 20Er 1 Meade and J, Sakis ‘The model presented in this paper helps to structure this complexity by evaluating alternative logistics strategies by utilizing a systemic multiattribute analytical technique, known as the analy- tical network process (ANP) (Saaty 1996). Am illustrative example provides additional insights for research and practical applications. First, a review of some analytical models for strategy devel- ‘opment and decision making in logistics is presented. 2. ANALYTICAL MODELS FOR LOGISTICS STRATEGY ANALYSIS In the strategic management literature, analytical models that incorporate the many dimensions ofa logistics strategy are rare. A number of qualitative models based on strategy development do exist (€.g O'Neil and Tveson, 1992). Limitations of the analytical models that exist for logistics strategy evaluation include a focus on one dimension ofthe logistics strategy or are static in their approaches. ‘One popular model isthe added value strategy modelling (AVSM) system. AVSM measures the relative importance of the main factors that influence a customer's preference. This enables an organization to create a logistics package to suit individual customers at affordable costs which would maximize competitive advantage (U.S. Distribution Journal, 1993). This model is an ‘example ofthe ‘traditional logistics strategy of cost versus service modelling. ‘Hewlett-Packard (HP) (a leader in supply chain management and logistics) has been involved in the development of analytical frameworks and models for managing the supply chain. BP's models have evolved as its focus on the supply chain has evolved (Davis 1994; Lee and Billington, 1994). Yet, most ofthe strategic supply chain management analytic modes used by BP focus on inven- tory control (¢g. WINO) as the only decision variable, this focus is usually associated with more tactical decisions (LaLonde and Mason, 1993). Consultants have devised a number of simple weighted checklists for evaluating company markets, company products, customer service, logis- tics, transportation and product operations to determine whether a logistics strategy needs to be redirected in order to more closely correlate to the busines strategy ofthe company (Johnson and ‘Wood, 1993). The simplicity of these models is very helpful to define gaps, but are static (¢. consideration of a singe point in time) and are not able to directly aid in the selection of alter native strategies. Part ofthe dificulty in analytically inodeling strategic decisions is ther basis on quantitative and qualitative information with multiple dimensions. A quantitative model that can be used to integrate qualitative information and quantitative values and analysis isthe Analytical Hierarchy Process (AHP). This technique has been effectively used for logistics applications such as the analysis of international consolidation terminals (Min, 1994, determining what to benchmark (Partovi, 1994), the allocation of contract incentives based on schedule, quality, radiation expo- sure, and safety (Thompson, 1994), and locating airports (Min, 1994a). AHP is a relatively popu- lar tool for modeling strategic decisions, but a primary limitation is its basic relationships do not allow for an integrated dynamic modelling ofthe environment. ‘AUP assumes the system elements are uncorrelated and are unidrectionally influenced by a hierarchical relationship. A more general evaluation approach defined as the Analytical Network Process (ANP), or systems-with-feedback approach, may be used to assess a dynamic multi- directional relationship among decision attributes (Hamalainen and Seppalainen, 1986; Saaty, 1988; Saaty and Takiawz, 1986). Due to its complex relationships and lack of exemplary model applications, ANP’s application has been very limited. One of the few strategic applications of ANP includes an evaluation of a mult-attribute, multi-year decision process applied to an equip- ‘ment replacement devsion (Azhar and Leting, 1993). The ANP approach has been defined as @ non-linear, network relationship aniong various factors It allows fr the eapabily to model more ‘complex and dynamic environments, environments that are more evident at strategic planning levels which are influenced by ever changing extemal forces. 3. ALLOOISTICS FRAMEWORK FOR SUPPLY CHAIN COORDINATION AND INTEGRATION According to strategic alignment models, the suocess of organizational strategy is dependent upon the alignment of environmental uncertainties, strategy, and adopted technology, practices, or ys- tems (Henderson and Venkatraman, 1993; Luftman; Lewis and Oldach, 1993). For organizations toSteg nays oflopsics and suply chain management ystems 0s bbe competitive and thrive they should accurately measure and evaluate their logistics activities and processes. Understanding and measuring the dynamic nature of logistics will aid an organization's transition to competitiveness. As products advance through the cycle of introduction, rapid ‘growth, maturity, and decline, 2 diferent logistics strategic configuration may be economical (Ballou, 1993). The framework forthe model is shown in Fig. . The overall objective of the model isto determine the optimum strategic logistics system or alternative for a given organization based on a Logistics Strategy. The end result isa weighting of various systems being considered and the Suggested System will be the one which ranks the highest. The logistics system is evaluated on three different levels (or clusters); the Organizational Supply Chain relationship involved, the Principles of Logistics required, and the Attributes of Principles of Logistics. Figure 2 shows the framework in a representation conducive to ANP. As can be seen the current system (if there is one) is compared to two other alternatives (System A and System B). The end result or the model will indicate which system best meets the needs of the decision maker based on interaction between three different levels. The three different levels (organizational/supply chain relationship, principles of logistics and attributes) will be discussed. 4. ORGANIZATIONAL SUPPLY CHAIN RELATIONSHI ‘The organization needs to determine what type of supply chain strategy will help an organiza tion achieve its greatest competitive advantage. Even at the supply chain strategy level a dynamic environment with various choices exist. Supply chain strategies include a continuum extending from commodity providers to virtual enterprise membership. Commodity, partnering, strategic alliances and virtual enterprises form a spectrum of relationships that may exist among enterprises, with the relations of the inter-enterprise business processes becoming more unified and integrated along this spectrum. The supply chain environment in which the enterprise exists will deterniine ‘what type of inter-enterprise relationship is necessary to maintain a competitive position in a particular industry. The strategic alignment of an organization's logistics network needs to be ‘synchronized with the demands of the competitive environment. As a system, an enterprise that Logistics Strategy Grganizational/Supply Chain Relationships Principles of Logistics ‘tributes of Principles of Logistics Fig. Framework for logs srmteic analysis,Fig. 2. Graphical representation of relationships forthe logic strategic analy framework fails to respond to environmental demands is placed a a disadvantage relative to competing rms Some of the characteristics of the supply chain relationship specrum are defined below. This spectrum comprises the strategic dynamic environment elements in the strategic analysis model. Commodity relationships among enterprises focus on customers choosmig suppliers based on price, quality, and reliability. The relevant business processes will be sparsely linked compared ‘withthe inkages tobe found at the virtual relationship end of the spectrum, andthe Snancial legal relationships will be less strongly coupled. Partnering’s goa is to provide benefits to all sides of the relationship. Moody (1994), Schon- berger (1990) and others advocate long term, selective partnerships. Cost and time reductions are tangible benefits, whereas flexibility and customer satisfaction are intangible benefits of partnering. Strategic alliances are more strategically oriented than parinerships. The organization develops a relationship with a specific goal in mind. Retailers, manufacturers and logistics service companies team together for a winning combination (Bowerson, 1950) The supplicr adds value to the cus- tomer’s busines processes as purchases become more unique and customized Virtual relationships focus more on satisfying the customer than on maintaining the relationship among the supply chain entities. The concept of the “Virtual Enterprise’ (Iacocca institute, 1991) has become more evident in recent years. Based on this concept, several independent enterprises join together emphasizing their particular core competencies to form a Virtual Enterprise that is able to compete in a given arena for a specific product or service. Without this merger of resources, the separate companies may be unable to successfully compete in a given market niche. The com- petitive advantage that can be achieved by a virtual enterprise depends on how wel the individual firms complement each other and their ability to integrate with one another ‘Dynwnic environmental characteristics such as produc lifecycle issues and supply chain rela- tionships require a solid base by which to evaluate logistics strategies. Literature shows that cycle time compression, inventory policy, faclity location, transportation selection and routing are additional elements that are to be examined when assessing a logistics network. The principles of logistics, which are described below, form another level necessary forthe framework to evaluate these various logistics strategies. ‘5. THE PRINCIPLES OF LOGISTICS AND THE PRINCIPLES ATTRIBUTES, ‘The principles of logistics, identified by LaLonde and Mason (1993) provide a foundation for consistent evaluation of logistics activities and strategies. Yet, as with any model, limitations to these principles do exist. Even though these principles are comprehensive, they are notStrategic analysis of logistics and supply chain management systems 20s exhaustive (¢.g. component standardization is another principle that is not included among, LaLonde and Mason's principles). In addition, these principles are not necessarily scientifically proven principles and may be quite dynwnic as competitive environments, technology, ete. pressures evolve. The three main areas of logistcs—inbound logistics, materials management, and outbound logistice—are influenced by the principles of logistics. These principles affect and are affected by the logistics environment ofthe frm (eg. supply chain strategies and product lifecycle strategies). The logistics principles will be defined with example attribute for management of these principles recommended for the framework presented inthis paper. These attributes, which form the third level within the analysis framework, are based on expert opinion and literature; they are not exhaustive, Table I summarizes the nine principles of logistics along with the supporting attributes for their effective management. This tabular relationship ean also be viewed 28 a hierarchical linkage, a8 wll be shown in the analysis framework. Initially, we define each of the major principles with example attributes for managing the Principles. Sil. Principle of selective risk ‘The principle of selective risk’s objective is to design logistics systems so that system perfor- ‘mance is directly related to the importance of the product or customer of the enterprise. Logistics strategies are based on the level of service desired for a specific customer. Detailed knowledge about the customer is necessary in order to implement this principle. Attributes of processes and. systems for aiding in the management of selective risk, may include knowledge about customers (KACU) (eg. sales, length of time of customer), knowiedge about competition (KACO) range of service capabilities of logistics system (SRC), and multilevel (capabilities) inventory management system (IMSF). “Table |. Principe of loitice and atibues of systems for management of opis principles Piacpe of opines “Atsbte for management of logit principe Saecive Hak Knowdge about curomen (KACU) ‘Kaowledge abou competion (KACO) ‘Service range capabilites (SRO) Inventory management tem feibity (MSF) Information Selectivity -Fleibity of dat kage (FDL) ‘Accuracy of data (AOD) ‘Accuracy of data bows (ADN) Data seach expebity (DSC) Information Sabato Coverage of information inkages (CIL) ‘Accuracy of deta (AOD) Level of syst ntegraton (LSI) Forecating capitis (FC) “ramscton Simplification serinteface fendines (IF) Data avalabie to wer (DAU) eve of systems integration (LSI) ‘Stplies acest information (SA) ‘Variance Reduction Demand forecasting tools (DFT) Inventory Velocity ‘Elicient third pat relationships (ETPR) ) operations (FMO) Fleble diseibaion options (FDO) Postponement Modular product design (WD) Flexible packaging desig (FPD) Reraljdibaton ste data (RDSD) Shareshied Risk ‘Crenton of sandarés (COS) ‘Outsoureng agreements (OA) Supplier euttombeaton (90)206 1L Meade and J. Sarkis 5.2. Principle of information selectivity ‘With the growth and implementation of information systems, information is abundant. How- ever, the logistics manager needs to determine how the logistics systems are designed, implemented and managed in order to provide the appropriate information. A review of the literature suggests that information technology is a key enabling technology for logistics functions (Forge, 1994; LaLonde and Mason, 1993). Information technology and practices need to include tools that help to filter out unnecessary information, Some of the attributes for effective management of the information selectivity principle are the flexibility of data linkages (FDL), the accuracy of data itself (AOD), the knowledge of what data is needed (AND), and efficient data search capabilities (DSC) (such as data mining tools). 5.3. Principle of information substitution The principle of information substitution is based on the fact thatthe cost of information is ess than other resources. Whenever possible an information resource should be substituted for other production resoures such as labour, capital, and facilities. Some attributes that will id in effec- tively managing information substitution include the amount of coverage of information linkages (CIL) (ie. centralization distribution of data), accuracy of data (AOD), level of systems integra- tion (LSI) and forecasting capabilities (FC). 5.4. Principle of transaction simplification ‘Traditionally logistics has been occupied with numerous operational transactions such as, order entry, order fulfilment, and order delivery. The objective ofthe transaction simplification principle is to improve the efficiency and effectiveness of the transactional processes of the firm. This prin- ciple can be executed from both a technical and managerial perspective. Attributes for effective management of the principle of transaction simplification include user-interface friendliness (VEF), the amount of data available to ¢ user (DAU), level of system integration (LSI), and sup- pliers access to operational information (SA. 5. Principle of variance reduction ‘The principle of variance reduction’s objective is to reduce unplanned variance in the system which can significantly influence productivity. There are several supplier and customer links in 4 logistics system, both externally and internally. Any time there are unplanned surges in demand or requirements, undue stress and expense are created in the customer supplier chain ‘The 1993 Efficient Consumer Response Report estimates that demand variability contributes 5-9 per cent excess warehousing costs, 3-6 per cent excess transportation costs, and 4-8 per cent excess manufacturing costs (Pratt, 1995). Some attributes for effective management of variance reduction are effective demand forecasting tools (DFT), increased communication with 8 customers/suppliers (M), statistical process control (SPC), and increased internal systems integration (ISD, 5.6. Principle of inventory velocity ‘The logistics role of facilitating the flow of inventory from raw materials to end user serves as the basis of the principle of inventory velocity. Inventory velocity is not new to logistics. Quick Response (QR), a business strategy developed in 1989 to shorten the time for production and distribution and reduce inventory throughout the supply chain, directly impacts the inventory velocity principle (Quick Response'slow but inevitable, 1994) Attributes of processes and systems which aid in managing the inventory velocity principal include third-party (subcontracting) rela- tionships (ETPR), Just-inctime (HT) support mechanisms, flexible manufacturing operations (FMO), and flexible distribution options (FDO) (eg. cross-docking capabilites). 5.7. Principle of postponement The principle of postponement is an attempt to reduce the amount of inventory necessary to ‘eet specific and varied customer service levels. Key benefits of postponement include not having to commit production to specific customers uni alate stage in production (increasing feiblity in demand fulfilment) reduced range of components required for manufacturing, and the simplif- cation of inbound logistics planning. Attributes such as modular product design (MPD), flexibleSategc analysis of logics and supply chain management systems Ee package design (FPD), and retail/distribution site data (RDSD) will aid organizations manage the postponement principles. 58, Principle of shared)shifted risk Due to the increased cost of money a new principle has emerged within logistics which is to share or shift the risk. The application of this principle is the shifting ofthe logistics cost structure from a fixed cost base to a variable cost base, The costs can also be shifted upstream or down- stream in the supply chain, Attributes that aid in managing this principle include the creation of standards (COS) and requiring suppliers and customers to adhere to them, improved outsourcing arrangements and capabilities (OA), and improving the number special requests of suppliers that can be met, supplier customization (SC). 6, ANETWORK FRAMEWORK FOR ASSESSING LOGISTICS STRATEGIES AND SYSTEMS ‘The discussion on logistics strategies, environments, and principles provides the elements for the development ofa strategic assessment framework. This framework is summarized in Fig. 2. The framework is presented through a network of decision model relationships. The levels of the net- work framework include the organizational/supply chain relationships, the principles of logistics level, the attributes level, and the alternative selection level. These levels impact the overall goal of maintaining a competitive logistics/supply chain strategy. This framework is only one general set of relationships that can exist, some variations on this framework are discussed later in the paper. ‘The four components of the organizational/supply chain relationships (we shall defme these as organizational relationships), are commodity, partnership, strategic alliance, and virtual (enter- prise). The principles level contains the eight principles of logistics discussed earlier. The attributes level is composed of the components which help monitor the deployment and management of these principles, For the illustrative example and for maintaining simplicity in its presentation, three logistics systems are considered, the Current (incumbent) System, System A, and System B. The goal of this model is to select the most appropriate logistics system for a given enterprise operating to maintain a competitive logistics strategy. 6.1, The analytical network process ‘ANP is @ more general form of AHP Whereas AHP models a decision making framework using 4 unidirectional hierarchical relationship among decision levels, ANP allows for more complex interrlationships among the decision levels and attributes. Typically, in AHP the top element of the hierarchy isthe overall goal for the decision model. The hierarchy decomposes from the gen- ‘eral to a more specifi attribute until a level of manageable decision criteria is met. ANP does not require this strictly hierarchical structure. Interdependencies may be represented by two way arrows (or arcs) among levels, or if within the same level of analysis, a looped arc. The directions of the ares signify dependence, ares emanate from an attribute to other attributes that may infu- ‘ence it. The relative importance or strength of the impacts on a given element is measured on a ratio scale similar to AHP. A priority vector may be determined by asking the decision maker for their numerical weight directly, but there may be less consistency, since part of the process of decomposing the hierarchy is to provide better definitions of higher level attributes. ‘Similar to the traditional AHP approach, a hierarchical relationship exists within the network ‘model. A major difference between the two techniques isthe existence of a feedback relationship ‘among the levels within ths farnework. In this example, a two way impact relationship exists between the organizational relationships and principles of logistics levels of analysis. The principles of logistics ‘ffect the organizational relationship selected (or in practice) and the organizational strategy selec- ‘ted determines the roles of the principles. For example, in a commodity relationship, the selective risk principle may play a larger role then in a strategic alliance relationship (where shared/shifted tisk would become a more critical principle). In addition, the focus on a certain principle would affect the development ofthese relationships. If there isa higher priority set on shared/shifted risk a ‘commodity relationship may be forced to evolve into a partnership oriented relationship. The ANP approach is capable of handling interdependence among elements by obtaining the composite weights through the development of ‘supermatrix’. Saaty (1988) explains the supermatrix concept a8 & parallel to the Markov chain process. The supermatrix development is208 1 Meade and J. Sarkis described below in the steps for assessing the logistics strategy model. In this example, the only interdependencies that are identified, and will form the supermatrix, are the organizational rela- tionships and principles of logistics components levels. The methodology for building and ana- lysing the logistics strategy model is now detailed in a series of steps with a parallel illustrative ‘example provided. The values used in this example are assumed. In an actual application of this ‘model a complex iterative approach is recommended, one designed to elicit the data from the ‘minds’ of one or more strategic planners who have a stake in the final decision. This may include input from sources outside the immediate enterprise, to include customers and suppliers. 6.2. ANP analysis and solution methodology ‘Step 1: Model construction and problem sircturing: The frst step isto construct a model to be evaluated. The illustrative example will use the model that was developed earlier in the paper and summarized in Fig. 2. The relevant criteria and alternatives are structured in the form of a hier- archy where the higher the level, the more ‘strategic’ the decision. The topmost clements are decomposed into sub-components and attributes. The model development will require the devel- opment of attributes at each level and a definition of their relationships. In this example, the only interdependence or feedback occurs between the organizational relationships level and the princi- ples of logistics level of attributes. Step 2: Pairwise comparisons matrices of interdependent component levels: Eliiting preferences of various components and attributes will require a series of pairwise comparisons where the decision maker will compare two components at atime with respect to an upper level ‘control criterion In ANP, like AHP, pairwise comparisons ofthe elements in each level are conducted with respect to their relative importance towards their control criterion. Saaty (1988) has suggested a scale of 1 to 9 when comparing two components, with a score of 1 representing indifference between the two components and 9 being overwhelming dominance of the component under consideration (row component) over the comparison component (column component). Ifa component has some level of weaker impact the range of scores willbe from 1 to 119, where 1 represents indifference and 1/9 being an overwhelming dominance by a column ele- rent over the row element. When scoring is conducted for a pair, a reciprocal value is auto- ratically assigned to the reverse comparison within the matrix. That is, if ay is a matrix value assigned to the relationship of component i to components then ay is equal to (or ayay = 1) Since many of these values are strategic, additional strategic group decision a, making tools such as seenario planning or the Delphi approach can be utilized to assign meaningful values to these pairwise comparisons. Especially, when determining the relative impact of various logistics prin- ciples within a given organizational relationship. Within this illustrative example the relative importance of the system attributes with respect to a specific organizational relationship selected (ic. Commodity to Virtual Enterprise) is fist deter- mined. A pairwise comparison matrix will be required for each ofthe four major organizational relationships for calculation of impacts of each of the logistics principles. In addition, eight pair- ‘wise comparison matrices will ned to be determined for calculation of the relative impacts of the ‘organizational relationship on a specific logistics principle. To fully describe these two-way rela tionships, 12 pairwise comparison matrices will be required. Once the pairwise comparisons are completed, the local priority vector w (defined as the eVector in the example figures) is computed asthe unique solution to: = 7 ‘where Amar i the largest eigenvalue of 4. Saaty (1986) provides several algorithms for approx- mating w. In this paper a two-stage algorithm that involved forming a new nxn matrix by dividing each element ina column by the sum of the column elements and then summing the ele- ments in each row of the resultant matrix and dividing by the m elements in the row. This is referred to as the process of averaging over normalized columns. This is represented as: Aw @Siraepc analy of logics and supply chain management stems 209 where: wy = the weighted priority for component i; J = index number of columns (components); 1 = index number of rows (components) In the assessment process there may ovcur a problem in the transitivity or consistency of the pairwise comparisons. For an explanation on inconsistencies in relationships and their calculations see Saaty (1988). It is assumed that the pairwise comparisons are consistent in these examples. ‘An example of the logistics principles pairwise comparison matrix within a commodity organi- zational relationship environment is presented in Fig. 3. In the commodity environment, the selective risk principle is viewed as being slightly more important (a12 = 3) than the information selectivity principle The weighted priorities for this matrix is shown as the last columa in Fig. 3. ‘The weighted priorities for each of the organizational relationships matrices (four in all) are ‘combined to create a matrix A with four columns and eight rows (see Fig. 4) The local priority weights for the relative impact of organizational relationships given a logistics principle are now determined. An example ofthe impact on the selective risk principal by various ‘organizational environments is shown in Fig. 5. Notice that a commodity relationship will infiu- cence the selective risk principle more than each of the other relationships. Whether or not this [COMMODITY [Sel Info Info Tran Var Inv Pp Sah] [Vector IRisk Sel al Risk err 1.000” 3.000 0250 2000 3.000) [0.109 info Sel 333 1.000 3.000 3.000 3.000 | [0.136 '333 0.333 1.000 3.000 0.333 3.000 2.000 3.000 | jo.107 [3.003 1.000 01333 1.000 3.000 $.000 3.000 2.000 | fo228, f2.000 4.000 3.003 0.333 1.000 $.000 3.000 3 fo301 [4.000 0.333 0.333 0.200 0200 1.000 2.000 2.000] Jovoss 1500 0.333 0500 0.333 0.333 0.500 1.000 5.000] Joos 1333 0.333 0333 0.500 0.333 0.500 0200 1.000] lori Fig 3. Logistics principles pairwise comparison matrix for commodity organizational relationship environment and ‘genet (relative inpotancelmpac weigh). [AMATRIX [COMMODITY PARTNERSHIP STRAT ALL VIRTUAL, Sel Risk 0.109 015 oz nfo Sel 0136 0135 o1ss ote] listo Sub 0107 0135 0100 ou [Trans Simp 0228 02% 014s o.125 ari Red 0301 0.40 oist 0s lav vel 0055 0078 00s 08s Pp oo 0.028 002 028s] [si Risk oat 0018 aio 0.147 Fig 4 Toe A mati formed fm eigenvector (tv importance weighs) fr orion each impliatons ‘ogc pins GREECHVE [Comedy Parmcmtip Seat Viewal ] fer I-00 900 3.000 3.000 am pears 500 1.000 »—«-2.000 2.000 265, All 1333 S00 «1.002.000 18s frees 13330500 0.500__1.000 140 Fig, 5. Organizational rlaonshipe psirwne compariton matin for wletve stk loptics principle environment and ‘dbmvectr (elatieimportace/ipect weigh).210 1, Meade and J. Sakis principle is practiced and at what level of consideration is dependent on the organizational rela- tionship thatthe organization is practising with a certain group of customers. Each of the eight principles will have asa result an eigenvector of priority weights, altogether these vectors form an ‘c8 matrix B) shown in Fig. 6 Step 3: Supermairix formation: The supermatrx allows a resolution of the effects of inter- dependence that exists between the elements of the system. The supermatrix is a partitioned matrix, where each submatrix is composed ofa st of relationships between two levels in the gra- phical model. Three types of relationships may be encountered in this model: (1) 14 independence from succeeding components, (2) interdependence among components, and (3) interdependence between levels of components. In Fig. 6 we show comparable graphical and matrix relationships among two levels in the model. The A and B matrix relationships currently exist in the model. A and B in Fig. 7 (a) represent the locations in the supermatrix of the matrices A and B calculated in the previous steps and represent interdependence between two levels of components. Relationships Cand D in Fig. 72) represent the interdependence of level of components on ite. For example, in an organization there may be an influence on various orgeniational relationships (eg. com- ‘modity) and their relative importances if they are in a specific environment (partnership). In this illustrative example the swne level impacts are not deemed to be significant. Thus, inthe ill ‘tative example supermatriy, all the values in submatrices C and D will be assigned a value of "0 IFC and D were non-zero matroes, a normalization step will be needed for making sure that the matrix is ‘column-stochastc’ « necessary requirement for achieving convergence to a set of weights, MATRIX JSe1Risk Taf Sel Taf Sub Tran Var Tov Pp Sa Rink _Simp Red_Vel sy Tail Gas 04a 0436 0305 01920100 137 Partnership 0265 0261 0245 0275 016002120225 027] strategie Altance| 0185 0.198 0.162 0.177 025802550311 0332 0140 0085 _o.108 0.112 0279 0341 0362 _ 0259 Fig. 6, The B matrix formed from eigenvector (relative importance weights) for logistics prises implications on or ‘istionalfelstonstipe ‘Organizational Relationships Logistics Principles (A) Network Model of Supermatrix Relationships Emery Pe 7 (B)Matrix Model of Supermatrix Relationships Fi 7 Two viewpoint of sperma aosSratepc analy of logistics and supply chain management systems a ‘The two compiled matrices A and B, are now combined to form the supermatrix M shown in Fig. 8. Raising the supermatrix to the power 2k + /, where k is an arbitrarily large number, allows ‘convergence of the interdependent relationships between enterprise strategies and principles of logistics. In this example, convergence is reached at M."° The ‘long term’ stable weighted values 10 bbe used in the analysis are shown in Fig. 9. ‘Step 4: Analyze principles of logistics attributes: In this illustration no interdependence between the principles level and the attributes level is assumed to exist. A similar pairwise comparison that ‘was made in Step 2 is made for the attributes level for relative importance weight calculation (or eigenvector determination). There are eight separate pairwise comparison matrices that have to be developed for this step in the analysis. ‘Step 5: Alternative evaluations: Each alternative will need to be evaluated on each of the prin- ciple attributes or management decision categories. This is completed by making a pairwise com- parison of the performance of each alternative on each attributes. Since there are 30 atributes, an ‘additional 30 3x3 pairwise comparison matrices will be needed for evaluation. The size of the ‘pairwise matrices is dependent on the number of systems alternatives that are to be evaluated. This illustration includes three alternatives, a current logistics (incumbent) system that is to be eval- uated against two new alternatives. The pairwise comparisons are completed by asking the relative impact of one system on a logistics principle attribute. For example in Fig. 10, the fist attribute “knowledge about customer’ (KACU), is compared between the current system and alternatives ‘A’ and ‘B’, The current system is assumed to perform better on the KAW attribute than system ‘a’ and °B, since it already tracks sales and length of time company has been a customer. ‘Step 6: Selection of best alternative: The selection of the best alternative depends on the calcu ie index’ for an alternative i(D). The equation for D; is defined by: Tnkages.22 1, Meade nd 5, Sarkis D, =SE rysy @ where: 2, is the relative importance weight of principles j ‘Avis the relative importance weight for attribute k of principles j, and ‘Say is the relative impact of alternative ‘on attribute k of principle j AK; isthe index set of attributes for principle j is the index set of principles. ‘The alternative with the largest desirability index should be the one selected. In the illustrative example the results of the logistics strategic analysis (see Fig. 11) point to selection of Alternative System A, which has the largest desirability index of 0.438, 17. MODEL EXTENSIONS ‘The current model that is presented may still require additional extensions for modelling prac- tical situations. For example we have not introduced the true dynamics of long and short term planning horizons and their influences on the principles and the model. Nor have we explicitly included risks, costs and benefits as specific attributes that need to be considered. To be able to ‘handle these two circumstances we shall provide a general description of the necessary alterations to the basic model we have thus far presented. First the external competitive environment and its pressures could be integrated into the model by the inclusion of an additional ‘cluster in the network model. This cluster can be defined as, competitive pressures over a short time period horizon (within a 1 year time period), a medium time period horizon (1-5 years), and a long time period horizon (over 5 years). The categorization of competitive time horizons will be dependent on the type of organization and industry market. ‘his three level categorization for the competitive environment can influence and be influenced by ‘both the type of inter-organizational relationships and by the principles of logistics. This set of relationships will not be detailed, but specific questions can be developed to aid in their evaluation. AA general analytical network hierarchy framework that can model the evaluation of these types of relationships is shown in Fig. 12. Notice the two way arrows showing an interdependence of the principles, horizon, and relationships clusters. The supermatrix to help evaluate the relationships reeesy a pedgageaagen Fig 11, Desirability index cleltion for lopntcs tems alternative,ee ' a er > Fig (2. Genera graphical representation of reltloaships forthe extended logistics strategie analy framework TH OR oP mem |B | FIG at omnes | H | C | B Prete (2) | 1 Fig. 13. Genera upermatix structure forthe extended lgitice wage analy framework ‘within this extended model can be formed from a set of nine submatrices as shown in Fig. 13. The characteristics ofthe submatrices, A,B, C, and D, are similar to those described in step 3 (above). ‘The submatrices F, G, H, and Tin Fig 13, will be non-zero matrices. While submatrix E is assigned a z2r0 value. The evaluation of this network can be completed with the same steps as described in the previous section. ‘There are two ways for integrating costs, benefits, and risks into the analysis. One way isto have ‘another cluster (or set of clusters, levels) that will consider various costs, benefits and risks and hhow they interact with other clusters and levels of the network hierarchy. Alternatively, a techni- ‘que recommended by Saaty (1996) isto have a separate network hierarchy for each category (cost, ‘benefit or risk) and get a score for each alternative in this network by emphasizing contribution to cach ofthese categories when evaluating the network. Then each alternative can be measured on a simple ratio of Bille The networks used to evaluate each of the ratios may have exactly the same structure. 8. DISCUSSION AND CONCLUSIONS ‘The framework that was used in this example could serve as one of the tools for making a strategic decision. The criteria and attributes that were used in the model focused on logistics strategy and requirements. The model is capable of taking into consideration both qualitative, quantitative, and multiple dimensions of information into the analysis, a powerful and necessary ‘characteristic for any strategic evaluation,ae 1. Meade and J. Sarkis ‘Since any logistics system or strategy that is selected or recommended by the initial model will impact other functional strategies, this framework requires integration with other models for strategic decision making, For example, in the model consideration of costs, revenues and profits associated with the system do not have their own level of analysis but are taken into consideration indirectly when determining which organizational relationship to pursue and what principles of logistics are deemed important. These financial measures can be explicitly considered in a more ‘complete analysis or through reevaluation based on the contributions to a measure, as described in the model extensions section. The final values of any models that are determined should be crti- cally analyzed. The use of auditing and iterative approaches with this model should be pursued in ‘an actual presentation, The issues relevant to the ANP process and eliciting weights and the pre- ference structure are cognitive decision making issues that can be studied with actual model implementation. ‘The framework represents only one set of possible relationships. A variation in the attributes or organizational strategies can also be made to this model. The attributes selected for this model were not justified, primarily due to space requirements, In an actual implementation of ths fra- mework, the number of attributes will be specific to the organization that applies this model. Additionally, the organizational strategies may not be limited to whether the organization prac tices one of the supply chain relationships mentioned earlier, a cross-sectional look at supply chain strategies. In addition, a mult-organizational framework could be developed fl-om the basic fra ‘mework. This can be accomplished by changing the focus of the decision from a single organiza- tion to the whole supply chain. The enhanced multi-organizational framework will have to address issues relevant to the whole supply chain where organizational relationships among supply chain ‘members can vary from commodity to virtual enterprises. Another enhancement to the model may bee the inclusion of produet life eycle strategies. The logistics strategies and relationships can be driven by the (actual or expected) portfolio of product families. Some product types may be in introductory phases, while others may be in decline. Since a strategic logistics strategy should impact all product families this set of components, a its own level, may be added to the model ‘As we have shown, the model may be enhanced through the inclusion of additional relation- ships. In maintaining some parsimony for ease of exposition, a number of additional relationships Were not considered. For example, a feedback or systems relationship may be extended to include the effect of logistics attributes on organizational relationships and the interdependence of same level components on each other. ‘Typically, each of the enhancements to the model win increase the amount of analysis that needs to be completed. From an application perspective this may require significant time resources from ‘managers and decision makers. Yet, when seeking to invest in systems that can potentially each in the millions of dollars, a structured analysis, which is provided by this model, can help to reduce the risk of poor investment decisions. Eliciting values for these models can be difficult and may require various levels of management and staff to be involved in the value elicitation. Preference value elicitation is deseribed and researched in behavioral and cognitive sciences research and is beyond the scope ofthis paper. The use of software and group decision support systems may lessen the barriers to implementing this technique. The use of ANP and AHP should also be critically observed from a mathematical ‘operations perspective since there are problems with ‘rank reversal’. Currently, techniques exist to help reduce the occurrence of this event (Saaty and Takiawz, 1986). In addition, since the appli- cation of AHP requires qualitative assessment, fiery set theory has been looked at to improve analysis of the problem. The use of fuzzy set approaches in the ANP approach require develop- ‘ment and research. In conctusion, this paper has set the foundation for a systemic framework that can be used for selection or justification of various logistics strategies and systems. A major contribution of the paper lis in its linkage of disparate strategic logistics and systems issues in a single systemic fra- ‘mework. Even though the technique, as models become more complex, may become time con- suming, the benefits of risk reduction and improving organizational ‘fi of the system hopefully will outweigh the costs. The technique not only provides solutions to strategic logistics alternative selections, it also provides a structure for an organization to develop and enhance @ logistics strategy. The time resources required by an organization to implement this type of model will decrease as the additional learning occurs along with reuse of static data portions of the model.‘Strategic analysis of logistics and supply chain management systems as Finally, this paper provides an exploratory evaluation of an analytical approach for managerial decision making through a modelling technique that has not been fully explored by researchers or practitioners in the logistics field REFERENCES ‘Azhar, T. M. and Leung, L.C. (1993) A Mult-Atibute Product Lie-Cjle Approuch to Replacement Decisis: An “Apolaton of Sauls System With-Feodback Method. The Engineering Economist 28, 321-343, Dalioo, RH. (192) Bune Logics Management, Prentice Hal, Englewood Cli, NI. ‘Ballou, RH. (1993) Reformultng Logics Srey: A concern forte as, present and future. Intemational Journal ofPhyial Disributon & Lapses Management 23, 30-38, ‘owersor, DJ (1990) The strategie benefits of logata sliances, Harvard Busines Review, July/August, 36-48, (Chow, Gy, Hoar, T. D- and Henritson, L.E. (199) Strategy, structure aod performance: framework for logistics serch. The Logics and Tranporation Review, 3, 288-308 Daaghery, PL and Pan, P. H. (1998) Usaason cf tne based Strate: creating 2 dissibution Seiity eos ‘tren Jtematonal Jounal of Operations and Production Managemen 15, 4-0. avis. RV (1994) The Dinibaion Revolution, Plawrng Review, 4623 Form, 8. (1994) High-Power Computing abd the Value Chats, Pures, 26, 430-452 arsine, RP and Spal, T- 01986) Te Anuye Network rows a aryy Paley Making Selo Sele 3, 399-405. Haring, A095) Log, Ant for Chang: Shaping he etd Supt Chin, report and Dire Ton 3630-34 ‘Henderson, J.C. and Venkatraman, N. (1983) Strategic Aliment: Leveraging Information Technology for Transforming TBM Systeme Joura 3,416 ncoat Insite (1991) 210 Cennry Manifactring Enterpris. Lehigh University, Harold S. Mohler Laboratory #200, Bethlehem, PA 18015. Johnson, LC. and Wood, DE. (193) Contemporary Logistics, Sh Edn, MaeMilan, New York. [aLonde, Band Mason, KE (1992) Some Thoughts On Lopes Policy and Statepe. International Journal of Physical ‘Dutribwtion nd Loos Management 23, 39-4. LaLonde, Band Mates, J. M. (1994) Emerging Logistics Stratepes: Blueprints forthe Next Century, Iiematinal Inurl of Phsleal Dstribution and Logistics Management 24, 38-47 Les, H. Land Bling, C-(1996) Evolution of supply chain management models and practice at Hewlett-Packard Working Paper, ay Luftman, 1 X- Levis . R, and Oldach, SH. (1993) Transforming the Enterprise: The Strategic Alignment of Busnes and Information Technology Strategie 2M Syste Journal 32, 198-221 Min, H (1990 Ineratonal Supper Slaion A Muli-Atbute UUliy Approsch.Iernaional Joumal of Physical ‘Busrbuion nd Loisics Management 24, 24-33. ‘Mio, H. (1954) Location Analy of Interoatonal Consolidation Tenmnals Using the Ansiytic Hierarchy Process, Jounal (Business Logics 15, 25-4 ‘Moody, P.E (1996 Breakthrough Parering: Creating a Colecive Enterprise Advanage. Olver Weigh, Esex Junction, ve (O'Neil BF. and Iveson, J. (1992) Strategically Managing the Logistvs Fonction. Logistics nd Tramportation Review 77, 4389-37. ure (0 Deering Mat To Reha: An Ani Mercy Pre Appreciation! nl of Operations & Production Sanagement 14, 25-39. a (95) ae ans, tein he Fourth Ana Ay rn Contes, At, TX, Mac 1999) (Quick Response: slow bat nevible, (1994) Material Handing Engineering. ‘Richards, HL 1999) Flex Your Labour Muscle. Transportation and Distton 35, 4-8 Saat, TL (198 Decson Making: The Analyte Hierarchy Proce Pitsburg, PA ‘Samy, T. Land Takizawz M. 1986) DependeceIndependence: From Linear Hierarchies to Nonlinear Networks. Ew pean Joumal of Operational Research 26, 229-237 ‘Sang. TL (1996) Decuon Making with Dependence and Feedback: The Analytic Neowork Proces, RWS Publications, Praaburgh, PA. Salo, A Aad Hamasinen, RP (1992) Rank retrial in the feedback technique ofthe analytic hierarchy process, Hel- Shaki Uaiversity of Technology Systems Analss Laboratory Research Repors, Aust Schonberg, RJ (1990) Bulga Chat of Cutters: Linking Business Funcion To Create The Word Class Company. ree Press, New York ‘Thompeon, B, Me (1994) Using AHP to allocate contract incentives, Transaction ofthe American Associaton of Cost ‘Engine, DCLT.1-DCLT3, US. Dstttion Journal, (983) New Directions in Logistics. 18 March ‘Wilbon, C192) Tae Logatics Trade-Ott How Modeling Caz Help. Business Marketing Digee 17, 33-40,