Mathematical and Computer Modelling 49 (2009) 12741282Contents lists available at ScienceDirectMathematical and Computer Modellingjournal homepage: www.elsevier.com/locate/mcmSelection of technology acquisition mode using the analytic networkprocessHakyeon Lee1, Sora Lee1, Yongtae ParkDepartment of Industrial Engineering, School of Engineering, Seoul National University, San 56-1, Shillim-Dong, Kwanak-Gu, Seoul 151-742, Republic of Koreaa r t i c l e i n f oArticle history:Received 20 March 2008Received in revised form 3 August 2008Accepted 7 August 2008Keywords:Technology strategyTechnology acquisition modeAnalytic network processa b s t r a c tSelecting the appropriate acquisition mode for a required technology, is one of the criticalstrategic decisions in formulating a technology strategy. Although a number of factors werefound to be influential inthe choice of technology acquisitionmode, it still remains a void inthe literature howto make a strategic decision, based on a huge set of those factors with thehelp of a systematic approach. This study deals with the selection of technology acquisitionmode as a multiple criteria decision making (MCDM) problem. The proposed solution tothe problem in this study, is the analytic network process (ANP) approach. Since the ANPis a MCDM method that can accommodate interdependency among decision attributes, itis capable of providing priorities of alternatives with consideration of interrelationshipsamong strategic factors. The 21 influential factors identified fromthe empirical studies areincludedas sub-criteria inthe ANPmodel, andthey are groupedintofive criteria: capability,strategy, technology, market, andenvironment. The final decisioncanbe made basedontheresulting priorities of the alternative acquisitionmodes. The proposedapproachis expectedto effectively aid decision making on which mode is adopted for acquisition of requiredtechnologies. A case of a software company is presented for the illustration of the proposedapproach.2008 Elsevier Ltd. All rights reserved.1. IntroductionEffective formulation and implementation of technology strategy has been considered as a major driver for competitiveadvantage of a firm. Although much debate is still going on about howto define the scope of technology strategy, fromquitespecifically focusing on technology development, to very broad knowledge-based definitions [1], what the literature has incommonis that technology strategy canbe viewedas a process composedof a series of steps requiring strategic decisions andactions, such as acquisition-management-exploitation [1,2]. One of the critical strategic decisions in formulating technologystrategy is how to acquire the required technology.Technology acquisition concerns whether to acquire technologiesthrough internal development,cooperating with other firms of institutions,or buying the technology [3].A variety oftechnology acquisition strategies (or modes) available and the complexity of modern business environments have led thedecision to be intractably difficult.Several empirical studies have beenconductedto identify key determinants affecting the choice of technology acquisitionmode [47]. However, there is a missing link between influential factors and final decisions. Although a number of factorswere found to be influential in selecting the acquisition mode, it still remains a void in the literature how to make astrategic decision based on a huge set of influential factors with the help of a systematic and quantitative approach. Variousapproaches, basedonmathematical programming, statistical analysis, or multiple criteria decisionmaking (MCDM) methodshave been proposed to aid decisions both prior to and posterior to selection of technology acquisition mode: selectionCorresponding author. Tel.: +82 2 880 8358; fax: +82 2 878 3511.E-mail addresses: yuny29@snu.ac.kr (H. Lee), soralang@snu.ac.kr (S. Lee), parkyt@cybernet.snu.ac.kr (Y. Park).1Tel.: +82 2 878 3511; fax: +82 2 878 3511.0895-7177/$ see front matter 2008 Elsevier Ltd. All rights reserved.doi:10.1016/j.mcm.2008.08.010H. Lee et al. / Mathematical and Computer Modelling 49 (2009) 12741282 1275of technologies to be acquired among identified alternatives, such as technology selection [8], R&D project selection [9],and decisions under the selected acquisition mode such as technology supplier selection [10], go/no-go decision of R&Dprojects [11]. However, very fewsystematic approaches have been proposed to selection of technology acquisition strategy,while there is a growing need of employing sophisticated mathematical modelling for such strategy selection problems.This studydeals withthe selectionof technologyacquisitionmode as a MCDMproblem. InMCDM, decisionmakers evaluate several alternatives using multiple conflicting criteria. The decision environment of selecting technologyacquisition strategy constitutes a typical form of the MCDM: selecting the appropriate option among several technologyacquisition modes as alternatives by considering various influential factors as criteria. Among a variety of MCDM methods,the analytic network process (ANP) is employed in the proposed approach. The ANP is a generalisation of the analytichierarchy process (AHP),which is one of the most widely used MCDM methods [12]. Since the ANP allows for morecomplex interrelationships among elements, by replacing a hierarchy in the AHP with a network, it is capable of providingpriorities of alternatives that capture interrelationships among strategic factors [13]. In particular, the ANP has been provedto be useful for strategy selection problems, since strategic elements that need to be considered in decision making haveinterdependencytoeachotheratmostcases. TheexampleofusingtheANPforstrategyselectionincludesbusinessstrategy [14], e-business strategy [15], knowledge management strategy [16], and national military strategy [17]. This studyalso employs the ANP for selection of technology acquisition strategy.The remainder of this paper is organized as follows. Section 2 reviews the underlying methodology of the proposedapproach, the ANP. The proposed approach is explained in Section 3 and illustrated with a case study in Section 4. The paperends with conclusions in Section 5.2. Analytic network processThe ANP is a generalisation of the AHP [12]. The AHP, also developed by Saaty [18], is one of the most widely used MCDMmethods. The AHP decomposes a problem into several levels making, up a hierarchy in which each decision element isconsidered to be independent. The ANP extends the AHP to problems with dependence and feedback. The ANP allows formore complex interrelationships among decision elements by replacing the hierarchy in the AHP with a network [19].Due to such advantage, recent years have seen a huge increase in the use of the ANP for various MCDM problems [20]. Inaddition, the ANP has been applied to decision making with the existing frameworks such as quality function deployment(QFD) [21] and balanced scorecard (BSC) [22]. Various attempts have also been made to integrate the ANP with anothertheory or technique such as fuzzy set theory [23,24] and mathematical programming [25,26].The process of the ANP is comprised of the following four major steps [12,19,27]:(i) Step 1 (model construction): A problem is decomposed into a network in which nodes corresponds to components.The elements in a component can interact with some or all of the elements of another component. Also, relationships amongelements in the same component can exist. These relationships are represented by arcs with directions.(ii) Step 2 (pairwise comparisons and local priority vectors): The elements are compared pairwisely with respect to theirimpacts on other elements. The way of conducting pairwise comparisons and obtaining priority vectors is the same as in theAHP. The relative importance values are determined ona scale of 19, where a score of 1 indicates equal importance betweenthe two elements and 9 represents the extreme importance of one element compared with the other one. A reciprocal valueis assigned to the inverse comparison; that is, aji = 1/aij where aij denotes the importance of the ith element compared withthe jth element. Also, aii =1 is preserved in the pairwise comparison matrix. Then, the eigenvector method is employedto obtain the local priority vectors for each pairwise comparison matrix. To test consistency of a pairwise comparison, aconsistency ratio (CR) can be introduced with consistency index (CI) and random index (RI). If the CR is less than 0.1, thepairwise comparison is considered acceptable. For detailed information on how to calculate CR, see the text by Saaty [18].(iii) Step 3 (supermatrix formation and transformation): The local priority vectors are entered into the appropriatecolumnsofasupermatrix, whichisapartitionedmatrixwhereeachsegmentrepresentsarelationshipbetweentwocomponents. The supermatrix of a system of N components is denoted as the following:(1)1276 H. Lee et al. / Mathematical and Computer Modelling 49 (2009) 12741282Table 1Factors affecting the selection of technology acquisition modesCriteria Sub-criteria ReferenceCapability Technological position [2,5,2831]R&D resources [2,31]R&D manpower [5]R&D experience [5,29,32,33]Firm size [7,31,34]Complementary asset [3,31]Strategy Fit with business strategy [35]Fit with technology strategy [3]Acquisition urgency [2]Importance to a firm [2,36]Technology Technology life cycle [2,4,5,30,3743]Development cost [3,5,30,42,44,45]Technological relatedness [4]Easiness to imitate [46]Market Commercial uncertainty [5,44,45,4752]Market size [5,53]Competitive intensity [4,5,31,33,35,54,55]Environment Appropriability regime [4,5,30,37,56,57]Availability of external source [3,6]Quality of external technology [6]Dynamism [31,35,46]Ck is the kth component (k = 1, 2, . . . , N), which has nk elements denoted as ek1, ek2, . . . , eknk. A matrix segment, Wij,represents a relationship between the ith component and the jth component. Each column of Wij is the local priority vectorobtained from the corresponding pairwise comparison, representing the importance of the elements in the ith componentto an element in the jth component. When there is no relationship between components, the corresponding matrix segmentis a zero matrix. Then, the supermatrix is transformed into the weighted supermatrix, each of whose columns sums to one.This column stochastic feature of the weighted supermatrix allows convergence to occur in the limit supermatrix. Finally,the weighted supermatrix is transformed into the limit supermatrix by raising it to powers. The reason for multiplying theweighted supermatrix, is to capture the transmission of influence along all possible paths of the supermatrix. Raising theweighted supermatrix allows convergence of the matrix, and the resulting matrix is called the limit supermatrix, whichyields limit priorities capturing all of the direct and indirect influences of each element on every other element.(iv) Step 4 (final priorities): When the supermatrix covers the whole network, the finial priorities of elements are foundin the corresponding columns in the limit supermatrix. If a supermatrix only includes interrelated components, additionalcalculations should be made for obtaining final priorities.3. Proposed approach3.1. Model developmentThissectiondevelopstheANPmodel forselectionoftechnologyacquisitionmode. Thegoal oftheANPmodel isto select the best option for acquiring the required technology among the alternative modes. A number of technologyacquisition modes are available, such as acquisition, merger, licensing, joint venture, joint R&D, R&D contract, alliance,consortium, outsourcing, in-house R&D [3]. Since too many alternatives make the ANP procedure extremely complex andtime-consuming, three broad categories of the technology acquisition modes have been defined as the alternatives of theANP model: Make, Cooperate, and Buy. Make means in-house R&D, and Cooperate includes various forms of cooperation withother firms with or without equity involvement such as joint venture, joint R&D and alliance. Buy constitutes a form of R&Dcontract, acquisition, licensing, and outsourcing.The literature review was conducted to identify factors that need to be considered when evaluating the appropriatenessof the acquisition modes. Total 21 factors were identified and summarised in Table 1 with their references. The factors canalso be classified into five categories: capability, strategy, technology, market, and environment. The five categories and 21factors are employed in the model as criteria and sub-criteria, respectively.Fig. 1 shows the developed ANP model composed of the goal, five criteria, 21 sub-criteria, and three alternatives. Amongthe various form of the network model in the ANP, the proposed model takes the form of a control hierarchy, which issimply hierarchy of criteria and sub-criteria where priorities are derived with respect to the overall goal of the systembeing analysed [12]. It assumes that interdependency occurs at the sub-criteria level; the sub-criteria belonging to the samecriteria have interdependency on each other. For example, technological position of a firm is likely to be influenced byits R&D manpower or R&D experience, and vice versa. Thus, this interrelationships among factors need to be mirrored inevaluation of technology acquisition modes. That is why the ANP is employed in the proposed approach instead of the AHP.This form of the ANP model is similar to the model by Meade and Sarkis [19], Agarwal and Shankar [58], and Jharkharia andShankar [59].H. Lee et al. / Mathematical and Computer Modelling 49 (2009) 12741282 1277Fig. 1. ANP model for selection of technology acquisition mode.3.2. ProcedureThe evaluation of the alternatives modes for technology acquisition starts with the proposed ANP model.Pairwisecomparisons are made among the five criteria with respect to the importance to the goal, among sub-criteria with respectthe importance to their criteria. In addition, pairwise comparisons need to be conducted for interdependency among sub-criteria within the same criteria.Then, preference to alternatives in terms of each sub-criterion is obtained through pairwise comparisons. The questionfor comparison is: how much more is a technology acquisition mode appropriate than another mode when considering thegiven sub-criterion? To help answer the question, Table 2 provides general guidelines developed based on the empiricalfindings derived from the previous studies. The number of + indicates the relative preference to each mode when thedegree of the value for each factor is high. If the technological position of a firm is high, for example, the order of preferenceis Make (++), Cooperate (+), and Buy. When a firm has considerable R&D experience, the preferred mode is Make (+). Onething that should be noted here is the + scale is not interval but ordinal; that is, ++ is not twice as good as +. In thepairwise comparison, thus, any scale from 2 to 9 can be used for comparing a mode denoted as ++ with another modedenoted as +. For some factors about which empirical findings are controversial, multiple guidelines are presented withthe previous studies supporting them.After all types of pairwise comparisons are completed, the supermatrix is constructed with priority vectors obtainedfrom pairwise comparisons for interdependency. In the proposed approach, the supermatrix is equivalent to the weightedsupermatrix since the supermatrix is already column stochastic. Therefore, the supermatrix is directly transformed into thelimit supermatrix.The final priorities are calculated by the desirability index approach proposed by Meade and Sarkis [19]. The desirabilityindex (Di) for the alternative i is defined as the following:Di =J

j=1Kj

k=1PjADkjAIkjSikj. (2)Kj is the index set of sub-criteria for criterion j, and J is the index set for criterion j. Pj is the relative importance of criterionj and ADkj is the relative importance of sub-criterion k of criterion j for the dependency (D) relationships. These are derivedfrom the pairwise comparisons among criteria and among sub-criteria, respectively. AIkj is the stabilised importance weightof the sub-criterion k of criterion j for interdependency (I) relationships, which is taken from the limit supermatrix. Sikjis the rating of alternative i on sub-criterion k of criterion j. The appropriateness weighted index (AWI) can be obtained bynormalizing the derived desirability index. The final decision is then made based on the AWIs of the three alternative modes.1278 H. Lee et al. / Mathematical and Computer Modelling 49 (2009) 12741282Table 2General guidelines for comparing acquisition modesCriteria Sub-criteria Preferred mode ReferenceMake Cooperate BuyCapability Technological position ++ +R&D resources ++ +R&D manpower ++ +R&D experience +Firm size ++ +Complementary asset +Strategy Fit with business strategy ++ +Fit with technology strategy ++ +Acquisition urgency + ++Importance to a firm ++ +Technology Stage in technology life cycle + ++ [2,30,42]+ ++ [3741]Development cost +Technological relatedness ++ +Easiness to imitate +Market Commercial uncertainty + ++ [44,45,50,55]+ + [4749]Market size + ++Competitive intensity + [31]+ + [29,52,53]Environment Appropriability regime ++ + [35,54]+ [28,55]Availability of external source +Quality of external technology + ++Dynamism + ++Table 3Pairwise comparison matrix among criteria with respect to the goalCapability Strategy Technology Market Environment PriorityCapability 1 5 2 3 5 0.445Strategy 1 1/3 1 1/3 0.071Technology 1 5 1 0.237Market 1 1/2 0.082Environment 1 0.165CR: 0.064. Illustrative exampleIn this section, the following example is presented for illustration of the proposed approach. The proposed approachwas applied to technology acquisition mode selection in a software company located in Seoul, Korea. Over the last decade,the company has developed and provided a range of advanced IT solutions to clients throughout the world. Although themain product has been middle-ware solutions, the company has a plan to enter the small and medium enterprise (SME)enterprise resource planning (ERP) market by developing an own ERP package. The technologies or products required fordeveloping the designed ERP package are as follows: (1) AJAX, (2) OR mapping, (3) Aspect oriented programming (AOP),(4) Role based access control (RBAC), and (5) Group ware. The company already possesses the high level of AJAX and groupware technologies, and OR mapping and AOP can be obtained as a freeware. The problem to be faced is how to acquire theRBAC technology.4.1. Pairwise comparisons and priority vectorsFirstly, the relative importance of each criterion with respect to the goal, selection of the most appropriate mode oftechnology acquisition was derived. The pairwise comparison matrix and the resulting priority vectors are shown in Table 3.The priority vectors of the criteria are imported as Pj in Table 9.Secondly, pairwise comparisons among sub-criteria are carried out with respect to their criterion. For example, thepairwise comparison matrix with respect to Capability is shown in Table 4. The additional comparisons were also madefor the other four criteria. The priority vectors obtained here are carried as ADkj in Table 9.Then, pairwise comparisons were conducted to measure interdependency among the sub-criteria. Because the modelincludes 21 sub-criteria, 21 pairwise comparisons were made with respect to the impact on the given sub-criterion amongsub-criteria under the same criteria. The pairwise comparison matrix for sub-criteria with respect to TP under Capability isshown as an example in Table 5. The resulting priority vectors are entered into the supermatrix in Table 7.H. Lee et al. / Mathematical and Computer Modelling 49 (2009) 12741282 1279Table 4Pairwise comparison matrix among sub-criteria with respect to CapabilityTP RR RM RE FS CS PriorityTP 1 3 1 1 5 6 0.277RR 1 1/2 1/3 3 4 0.125RM 1 1 3 5 0.230RE 1 4 6 0.267FS 1 2 0.062CS 1 0.039CR: 0.02Table 5Pairwise comparison matrix among sub-criteria with respect to TP under CapabilityRR RM RE FS CS PriorityRR 1 1/3 1/3 3 2 0.152RM 1 1 3 4 0.330RE 1 4 5 0.362FS 1 1 0.080CS 1 0.076CR: 0.02Table 6Pairwise comparison matrix among alternatives with respect to TPMake Cooperate Buy PriorityMake 1 3 4 0.625Cooperate 1 2 0.238Buy 1 0.136CR: 0.02Finally, thealternativeswerepairwiselycomparedwithrespecttopreference, intermsofeachsub-criterion. Theguidelines in Table 2 were referred to evaluate the relative appropriateness of the three technology acquisition modes.As an example, the pairwise comparison matrix among alternatives with respect to TP is shown in Table 6.4.2. Supermatrix formation and transformationThesupermatrixwasconstructedwithpriorityvectorsobtainedfrompairwisecomparisonsforinterdependenciesamong the sub-criteria, as shown in Table 7. As the supermatrix is already stochastic; it is directly transformed into thelimit supermatrix. In this case, convergence was reached at W77. The limit supermatrix is shown in Table 8. The convergedpriorities for sub-criteria are carried as AIkj in Table 9.4.3. Final prioritiesThe final priorities, the AWIs, were produced by the desirability index approach, as shown in Table 9. The values of thethird column are the priorities of the criteria which come fromTable 4. The values of the fourth column, which are importedfrom Table 5 and the other pairwise comparisons among sub-criteria, are the relative importance of the sub-criteria ininfluencing their criteria. The values of the fifth column represent the priorities of the sub-criteria obtained from the limitsupermatrix in Table 8. The values of the next eight columns correspond to the priorities of the three acquisition modes interms of each sub-criterion, including the priorities with respect to TP in Table 6. The desirability indices obtained by Eq.(1) are presented in the last two rows of Table 9. The final priorities of the three alternative modes, the appropriatenessweighted indices are shown in the last row. The result indicates the AWI of Make (0.465) is about twice higher than those ofCooperate (0.242) and Buy (0.293). Thus, it is recommended that the RBAC technology be acquired through in-house R&D.5. ConclusionsThis study proposed the ANP approach for the selection of a technology acquisition mode.The proposed approachevaluates the appropriateness of alternative modes for technology acquisition, in terms of capability, strategy, technology,market, andenvironment. The case of a software company was presentedfor the illustrationof the proposedapproach. It wasshown that the ANP was successfully employed for producing the priorities of the alternative modes, with a considerationof interdependency among decision elements.This paper contributes to the field, by proposing a method for linking influential factors with a final decision. Most of theprevious studies were limited to identifying factors affecting the choice of a technology acquisition mode; it has not dealt1280 H. Lee et al. / Mathematical and Computer Modelling 49 (2009) 12741282Table 7SupermatrixCapability Strategy Technology Market EnvironmentTP RR RM RE FS CS BS TS AU IF TL DC TR IM CS MS CI AR AE QE DYCapability TP 0.000 0.421 0.169 0.107 0.287 0.287RR 0.152 0.000 0.169 0.207 0.142 0.142RM0.330 0.190 0.000 0.504 0.142 0.142RE 0.362 0.190 0.449 0.000 0.142 0.142FS 0.080 0.096 0.123 0.134 0.000 0.287CS 0.076 0.103 0.090 0.048 0.287 0.000Strategy BS 0.000 0.747 0.200 0.637TS 0.481 0.000 0.200 0.258AU 0.114 0.119 0.000 0.105IF 0.405 0.134 0.600 0.000Technology TL 0.000 0.405 0.413 0.714DC 0.200 0.000 0.260 0.143TR 0.600 0.481 0.000 0.143IM 0.200 0.114 0.327 0.000Market CS 0.000 0.333 0.667MS 0.500 0.000 0.333CI 0.500 0.667 0.000EnvironmentAR 0.000 0.200 0.163 0.600AE 0.250 0.000 0.540 0.200QE 0.250 0.600 0.000 0.200DY 0.500 0.200 0.297 0.000Table 8Limit supermatrixCapability Strategy Technology Market EnvironmentTP RR RM RE FS CS BS TS AU IF TL DC TR IM CS MS CI AR AE QE DYCapability TP 0.185 0.185 0.185 0.185 0.185 0.185RR 0.144 0.144 0.144 0.144 0.144 0.144RM0.233 0.233 0.233 0.233 0.233 0.233RE 0.229 0.229 0.229 0.229 0.229 0.229FS 0.115 0.115 0.115 0.115 0.115 0.115CS 0.094 0.094 0.094 0.094 0.094 0.094Strategy BS 0.380 0.380 0.380 0.380TS 0.268 0.268 0.268 0.268AU 0.102 0.102 0.102 0.102IF 0.251 0.251 0.251 0.251Technology TL 0.331 0.331 0.331 0.331DC 0.173 0.173 0.173 0.173TR 0.309 0.309 0.309 0.309IM 0.187 0.187 0.187 0.187Market CS 0.341 0.341 0.341MS 0.293 0.293 0.293CI 0.366 0.366 0.366EnvironmentAR 0.241 0.241 0.241 0.241AE 0.250 0.250 0.250 0.250QE 0.260 0.260 0.260 0.260DY 0.248 0.248 0.248 0.248with how to make a strategic decision based on a huge set of influential factors. The proposed approach incorporates theinfluential factors identified in the previous studies in the ANP model and helps come to a final decision with those factors,using the ANP procedure.However, the criteria or the alternatives included in the ANP model are by no means exhaustive or fixed. The ANP modelcan be customised depending on the context. The proposed ANP model in Fig. 1 only includes the three alternative modesfor technology acquisition at the high level of aggregation, but they can be divided into more specific forms. If the acquisitionmode at the high level is already selected (e.g. Make, Cooperate, or Buy), only the specific modes for the selected mode needto be included (e.g. joint venture, joint R&D, alliance for Cooperate). The criteria can also be added to or removed from themodel upon judgment of a firm.H. Lee et al. / Mathematical and Computer Modelling 49 (2009) 12741282 1281Table 9Calculation of desirability indices and final prioritiesCriteria Sub-criteria PiADkjAIkjS1kjS2kjS3kjMake Cooperate BuyCapability TP 0.445 0.277 0.185 0.625 0.238 0.137 0.014 0.005 0.003RR 0.445 0.125 0.144 0.540 0.297 0.163 0.004 0.002 0.001RM 0.445 0.230 0.233 0.547 0.263 0.190 0.013 0.006 0.005RE 0.445 0.267 0.229 0.674 0.225 0.101 0.018 0.006 0.003FS 0.445 0.062 0.115 0.200 0.600 0.200 0.001 0.002 0.001CS 0.445 0.039 0.094 0.600 0.200 0.200 0.001 0.000 0.000Strategy BS 0.071 0.486 0.380 0.637 0.258 0.105 0.008 0.003 0.001TS 0.071 0.207 0.268 0.637 0.258 0.105 0.003 0.001 0.000AU 0.071 0.080 0.102 0.192 0.634 0.174 0.000 0.000 0.000IF 0.071 0.227 0.251 0.571 0.286 0.143 0.002 0.001 0.001Technology TL 0.237 0.241 0.286 0.188 0.081 0.731 0.003 0.001 0.012DC 0.237 0.570 0.145 0.429 0.142 0.429 0.008 0.003 0.008TR 0.237 0.124 0.188 0.637 0.258 0.105 0.004 0.001 0.001IM 0.237 0.065 0.143 0.625 0.238 0.137 0.001 0.001 0.000Market CS 0.082 0.455 0.341 0.444 0.444 0.112 0.006 0.006 0.001MS 0.082 0.090 0.293 0.648 0.230 0.122 0.001 0.000 0.000CI 0.082 0.455 0.366 0.500 0.250 0.250 0.007 0.003 0.003Environment AR 0.165 0.169 0.241 0.105 0.258 0.637 0.001 0.002 0.004AE 0.165 0.368 0.250 0.143 0.143 0.714 0.002 0.002 0.011QE 0.165 0.368 0.260 0.221 0.319 0.460 0.003 0.005 0.007DY 0.165 0.096 0.248 0.200 0.200 0.600 0.101 0.053 0.064Desirability indices (Di) 0.203 0.106 0.128Appropriateness weighted indices (AWIi) 0.465 0.242 0.293The refinement of the proposed approach for more sophisticated modelling will be a fruitful area for future research.The proposed ANP model only mirrors the interdependency among sub-criteria under the same criteria, but there can beinterrelationships between sub-criteria in different criteria. Incorporating those relationships in the ANP model is expectedto produce more accurate and realistic results. Fuzzy numbers can also be introduced in the pairwise comparison matrices,to more effectively measure the appropriateness in terms of sub-criteria having great uncertainty.References[1] S. Davenport, C. Campbell-Hunt, J. Solomon, The dynamics of technology strategy: An exploratory study, R&D Manage. 33 (5) (2003) 481499.[2] D. Ford, Develop your technology strategy, Long. Range. Plann. 21 (5) (1988) 8595.[3] V. Chiesa, R&D Strategy and Organization: Managing Technical Change in Dynamic Contexts, Imperial College Press, London, 2001.[4] S. Kurokawa, Make-or-buy decisions in R&D: Small technology based firms in the United States and Japan, IEEE Trans. Eng. Manage. 44 (2) (1997)124134.[5] D.H. Cho, P.I. Yu, Influential factors in the choice of technology acquisition mode: An empirical analysis of small and medium size firms in Koreantelecommunication industry, Technovation 20 (12) (2000) 691704.[6] M. Hemmert, The influence of institutional factors on the technology acquisition performance of high-tech firms: Survey results from Germany andJapan, Res. Policy 33 (67) (2004) 10191039.[7] S.W. Hung, R.H. Tang, Factors affecting the choice of technology acquisition mode: An empirical analysis of the electronic firms of Japan, Korea andTaiwan, Technovation (2007) doi:10.1016/j.technovation.2007.10.005.[8] N. Gerdsri, D.F. Kocaoglu, Applying the Analytic Hierarchy Process (AHP) to build a strategic framework for technology roadmapping, Math. Comput.Modelling 46 (7-8) (2007) 10711080.[9] L.M. Meade, A. Presley, R&D project selection using the analytic network process, IEEE Trans. Eng. Manage. 49 (1) (2002) 5966.[10] R.F. Saen, A decision model for selecting technology suppliers in the presence of nondiscretionary factors, Appl. Math. Comput. 181 (2) (2006)16091615.[11] V. Kumar, A.N.S. Persaud, U. Kumar, To terminate or not an ongoing R&D project: A managerial dilemma, IEEE Trans. Eng. Manage. 43 (3) (1996)273284.[12] T.L. Saaty, Decision Making with Dependence and Feedback: The Analytic Network Process, RWS Publications, Pittsburgh, 1996.[13] H.J. Shyur, H.S. Shih, A hybrid MCDM model for strategic vendor selection, Math. Comput. Modelling 44 (78) (2006) 749761.[14] I. Yuksel, M. Dagdeviren, Using the analytic network process (ANP) in a SWOT analysis: A case study for a textile firm, Inform. Sci. 177 (16) (2007)33643382.[15] M.S. Raisinghani, L. Meade, L.L. Schkade, Strategic e-business decision analysis using the analytic network process, IEEE Trans. Eng. Manage. 54 (4)(2007) 673686.[16] W.W. Wu, Y.T. Lee, Selecting knowledge management strategies by using the analytic network process, Expert Syst. Appl. 32 (3) (2007) 841847.[17] B. Simunich, The ANP had shown a better way to deal with Iraq, Math. Comput. Modelling 46 (7-8) (2007) 11301143.[18] T.L. Saaty, The Analytic Hierarchy Process, McGraw-Hill, New York, 1980.[19] L.M. Meade, J. Sarkis, Analyzing organizational project alternatives for agile manufacturing processes: An analytic network approach, Int. J. Prod. Res.37 (2) (1999) 241261.[20] H. Lee, H. Seol, N. Sung, Y.S. Hong, Y. Park, An analytic network process approach to measuring design change impacts in modular products. J. Eng.Des., doi:10.1080/09544820802232517.[21] C. Kahraman, T. Ertay, G. Bykzkan, A fuzzy optimization model for QFD planning process using analytic network approach, Eur. J. Oper. Res. 171(2) (2006) 390411.[22] L.C. Leung, K.C. Lam, D. Cao, Implementing the balanced scorecard using the analytic hierarchy process & the analytic network process, J. Oper. Res.Soc. 57 (6) (2006) 682691.[23] M. Dadeviren, Yksel, M. Kurt, A fuzzy analytic network process (ANP) model to identify faulty behaviors risk (FBR) in work systems, Safety Sci. 46(5) (2008) 771783.1282 H. Lee et al. / Mathematical and Computer Modelling 49 (2009) 12741282[24] M.A.B. Promentilla, T. Furuichi, K. Ishii, N. Tanikawa, A fuzzy analytic network process for multi-criteria evaluation of contaminated site remedialcountermeasures, J. Environ. Manage. 88 (3) (2008) 479495.[25] J.W. Lee, S.H. Kim, Using analytic network process and goal programming for interdependent information system project selection, Comput. Oper.Res. 27 (4) (2000) 367382.[26] O. Ustun, E.A. Demirtas, An integrated multi-objective decision-making process for multi-period lot-sizing with supplier selection, Omega 36 (4)(2008) 509521.[27] H. Lee, C. Kim, H. Cho, Y. Park, An ANP-based technology network for identification of core technologies: A case of telecommunication technologies,Expert Syst. Appl. (2007) doi:10.1016/j.eswa.2007.10.026.[28] E.B. Roberts, C.A. Berry, Entering new business: Selecting strategies for success, Sloan Manage. Rev. 26 (3) (1985) 317.[29] R.K. Moenaert, D. Desschoolmeester, A. de Meyer, J. Barbe, Organizational strategy and resource allocation for technological turnaround, R&DManage.20 (4) (1990) 291303.[30] B.B. Tyler, H.K. Steensma, Evaluating technological collaborative opportunities: Acognitive modeling perspective, Strategic Manage. J. 16(1995) 4370.[31] J. Lowe, P. Taylor, R&D and technology purchase through license agreements: Complementary strategies and complementary assets, R&D Manage. 28(4) (1998) 263278.[32] R.R. Nelson, S.G. Winter, An Evolutionary Theory of Economic Change, Harvard University Press, Cambridge, 1982.[33] G.P. Pisano, The R&D boundaries of the firm: An empirical analysis, Admin. Sci. Quart. 35 (1) (1990) 153176.[34] J. Poon, A. MacPherson, Asian firms technology acquisition strategies in the United States, J. Eng. Technolog. Manage. 42 (2005) 321342.[35] B.B. Allred, K.S. Swan, Contextual influences on international subsidiaries product technology strategy, J. Int. Manage. 10 (2) (2004) 259286.[36] T. Baines, An integrated process for forming manufacturing technology acquisition decisions, Int. J. Oper. Prod. Man. 24 (5) (2004) 447467.[37] D.J. Teece, Profiting from technological innovation: Implications for integration, collaboration, licensing and public policy, Res. Policy 15 (6) (1986)285305.[38] B. Kogut, Joint ventures: Theoretical and empirical perspectives, Strategic Manage. J. 9 (4) (1988) 319332.[39] E.R. Auster, The relationship of industry evolution to patterns of technological linkages, joint ventures, and direct investment between U.S. and Japan,Manage. Sci. 38 (6) (1992) 778792.[40] G.C. Cainarca, M.G. Colombo, S. Mariotti, Agreements between firms and the technological life cycle model: Evidence from information technologies,Res. Policy 21 (1) (1992) 4562.[41] A. Madhok, Cost, value and foreign market entry mode: The transaction and the firm, Strategic Manage. J. 18 (1) (1997) 3961.[42] B. Croisier, The governance of external research: Empirical test of some transaction-cost related factors, R&D Manage. 28 (4) (1998) 289298.[43] L. Canez, D. Probert, Technologysourcing:Thelinktomake-or-buy, in:ProceedingsofPortlandInternational ConferenceonManagementofEngineering and Technology, 1999 pp. 4752.[44] G. Hamel, Y.L. Doz, C.K. Prahalad, Collaborate with your competitors and win, Harvard Bus. Rev. 89 (1) (1989) 133139.[45] M. Dodgson, The strategic management of R&D collaboration, Technol. Anal. Strateg. 3 (1992) 227244.[46] H.K. Steensma, K.G. Corley, Ontheperformanceof technology-sourcingpartnerships: Theinteractionbetweenpartnerinterdependenceandtechnology attributes, Acad. Manage. J 43 (6) (2000) 10451067.[47] G. Walker, D. Webber, Supplier competition, uncertainty, and Make-or-Buy decisions, Acad. Manage. J 30 (3) (1987) 589596.[48] C.C. Baughn, R.N. Osborn, The role of technology in the formation and form of multinational cooperative arrangements, J. High Technol. Manage. Res.1 (2) (1990) 181192.[49] J.T. Mahoney, The choice of organizational form: Vertical financial ownership versus other methods of vertical integration, Strategic Manage. J. 13 (8)(1992) 559584.[50] P. Llerena, S. Wolf, Inter-firmagreements in telecommunications: Elements of an analytical framework, in: G. Pogoral (Ed.), GlobalTelecommunications Strategies and Technical Changes, Elsevier Science, Amsterdam, 1994, pp. 257276.[51] R. Veugelers, Internal R&D expenditures and external technology sourcing, Res. Policy 26 (3) (1997) 303315.[52] V. Chiesa, R. Manzini, Organizing for technological collaborations: A managerial perspective, R&D Manage. 28 (3) (1998) 199212.[53] R.S. Rosenbloom, M.A. Cusumano, Technological pioneering and competitive advantage: The birth of VCR industry, Calif. Manage. Rev. 29 (4) (1987)5176.[54] A.C. Perrino, J.W. Tipping, Global management of technology, Res. Technol. Mange. 32 (3) (1989) 1219.[55] W. Shan, An empirical analysis of organizational strategies by entrepreneurial high technology firms, Strategic Manage. J. 11 (2) (1990) 129139.[56] M. Spence, Cost reduction, competition and industry performance, Econometrica 52 (1) (1984) 101122.[57] R. Veugelers, B. Cassiman, Make and buy in innovation strategies: Evidence from Belgian manufacturing firms, Res. Policy 28 (1) (1999) 6380.[58] A. Agarwal, R. Shankar, On-line trust building in e-enabled supply chain, Supply Chain Manage. 8 (4) (2003) 324334.[59] S. Jharkharia, R. Shankar, Selection of logistics service provider: An analytic network process (ANP), Omega 35 (3) (2007) 274289.