Chen, Huang - 2007 - International Journal - Bi-Negotiation Integrated AHP in Suppliers Selection

8/6/2019 Chen, Huang - 2007 - International Journal - Bi-Negotiation Integrated AHP in Suppliers Selection

http://slidepdf.com/reader/full/chen-huang-2007-international-journal-bi-negotiation-integrated-ahp 1/19

Bi-negotiation integrated AHPin suppliers selection

Yee-Ming Chen and Pei-Ni Huang Department of Industrial Engineering and Management,

Yuan-Ze University, Tao-Yuan, Taiwan, Republic of China

AbstractPurpose – This paper seeks to propose a new approach for tackling the uncertainty and imprecisionof identifying suitable supplier offers, evaluating these offers and choosing the best alternatives inbi-negotiation. In a build-to-order supply chain, the handling of uncertainties is addressed by a realtime information sharing system and appropriate supplier selection.Design/methodology/approach – A methodology integrated analytic hierarchy process (AHP) withbi-negotiationagentsbased on themulti-criteriadecision-making approach andsoftware agent technique

is then developed to take into account both qualitative and quantitative factors in supplier selection.Findings – During the decision-making between buyer and suppliers, the AHP process matchesproduct characteristics with supplier characteristics. Next, agents assist the user in the debate tonegotiate a joint representation of the supplier chosen and automatically justify proposals with this joint representation.Originality/value – This study focused on a multi-attribute negotiation mechanism includingqualitative conditions, which enables automated negotiation on multiple attributes. Finally, a fuzzymembership function represented the joint representation’s cognition for each condition such asquantity, price, quality, and delivery for the outsourced component. A case study in a high-endcomputer manufacturing company is given to demonstrate the potential of the methodology.Keywords Analytical hierarchy process, Supplier evaluation, Decision making,Uncertainty managementPaper type Research paper

1. IntroductionOne consequence of market globalization has been the growing incidence of collaborative ventures among companies from different countries. Small and large,experienced and novice, companies increasingly are choosing partnerships as a way tocompete in global market place. International joint ventures (IJV) have emerged as thedominant form of partnership in light of intense global competition and the need forstrategic organizational viability. Traditionally, vendors are selected from amongmany suppliers based on their ability to meet the quantity requirements, deliveryschedule, and the price limitation. In this approach, suppliers aggressively competewith each other. The relationship between buyer and supplier is usually adversarial. Inthis global supply chain era, the cooperation between buyer and supplier is the startingpoint to establish a successful supply chain management and a necessary. Therefore,supplier selection and evaluation are very important to the success of the supply chainprocess (Bhutta and Huq, 2002). The success of IJV depends on many factors, but themost critical include recognition of cultural differences, specied work-ow,information-sharing through electronic data interchange and the internet, and jointplanning and other models that facilitate a successful supply chain management. Thesupplier selection negotiation mechanism is often the most complex, since it requires

The current issue and full text archive of this journal is available atwww.emeraldinsight.com/1463-5771.htm

Bi-negotiationintegrated

AHP

575

Benchmarking: An International Journa

Vol. 14 No. 5, 200pp. 575-59

q Emerald Group Publishing Limited1463-577

DOI 10.1108/146357707108192



evaluation and decision-making under uncertainty, based on multiple attributes(criteria) of quantitative and qualitative nature, involving temporal and resourceconstraints, risk and commitment problems, varying tactics and strategies, domainspecic knowledge and information asymmetries, etc. The negotiation cycle typicallyinvolves a sequence of interdependent activities (evaluation and decision-making) -from suppliers’ selection to enter the negotiation, through the negotiationper se to theexecution of the agreed deal. Supplier selection and negotiation then are of a specialimportance for supply chain management. Thus, the objective of this study is todevelop an integrated analytic hierarchy process (AHP) with negotiation mechanismwhich will help to solve the supplier selection problems to obtain the most benecialoffers for the buyer by creating strong competition between suppliers and providing avehicle for negotiating with them. The rest of the paper is organized as follows.Section 2 presents the methodologies of supplier selection process. The details of themethodology we proposed and the reasons that lie behind are given in Section 3. Theapplication is explained through a case study in Section 4. Finally, the last sectioncontains some conclusions and perspectives.

2. Brief review and analysis of supplier selection methodologiesThe problem of supplier selection is notnew. Beforesupplychain management becomes abuzzword,theproblemof supplier selection was called vendor selection. First publicationson vendor selection can be traced back to the early 1960s. These early research activitiesare summarized in a literature review by Weber et al. (1991). The vendor selection is alsocalled supplier selection from now on. In the literature, there are many studies about thesupplier selection process. Traditional methodologies of the supplier selection process inresearch literature include the cost-ratio method, the categorical method, weighted-pointevaluations, mathematical programming models and statistical or probabilisticapproaches (Yan et al., 2003; Oliveira and Lourenco, 2002). Dickson (1966) has identied23 important criteria in the study of supplier decision-making. A study by Vokurkaet al.(1996) looked at the supplier selection decision criteria used in buying different categoriesof products.Themyriad factors weregrouped intoperformancecriteria, economiccriteria,integrative (wiliness to co-operate) criteria and adaptive criteria (the extent to which thebuyingrm mayhave to adapt itsplans to accommodate uncertainty about the capabilityof the suppliers). The advantage of the categorical method is that it helps structure theevaluation process in a clear and systematic way. However, a disadvantage with thisapproach is that typicallyit does not clearly denethe relative importanceofeach criterion(Muralidharan etal.,2002).Weberetal. hascompiled many articles in this area andheuseda linearweightingmodel for supplier selection.Linearweightingmodelsplacea weightoneach criterion and provide a total score for each supplier by summing up the supplier’sperformance on the criteria multiplied by these weights. Mandal and Deshmukh (1994)used an interpretive structural modeling for vendor selection. In this study, Mandaldeveloped an analytical framework, which combines qualitative and quantitative factors.Data envelopmentanalysishasbeenused forthe supplier selection process (Liu etal.,2000;Narasimhan et al., 2001).Another approach for supplier selection is theanalytic hierarchyapproach(Jiang andWicks,1999).In this study,the AHP approachwasemployedto arriveat thesupplier selection decision for theoutsourced component of manufacturing high-endcomputers. It is well-known that problem formulation is critical to the success of optimization.

BIJ14,5

576



Therefore, we should rst answer the following questions:. What supplier selection criteria to use?. How to use them?.

How to automatically trade offers and come to mutually acceptable agreement?The rst question is relatively easy to answer. We should use a set of criteria that arewell accepted. The second question is often ignored by researchers since they usuallyassign xed weights to the criteria. The last question of automating negotiations alsoopens up a number of new possibilities. So, we combined AHP, fuzzy set and softwareagents into multi-criteria decision-making and bi-negotiation mechanism. AHP dealswith the traders’ relative preferences and satisfactions for offers and counter-offers. Inaddition, fuzzy membership functions manipulate theuser’s cognition for each conditionand uncertainty occurring in the agents of bi-negotiation process. Meanwhile, weconcentrate on the software agents for multi-issue (or “attributes”) automatednegotiations. This approach allows us to adapt and change the conditions for a deal

dynamically. Companies rely on strategic alliances based on core competencies andinformation technologies to achieve exibility and responsiveness in their supply chain(Gunasekaran and Ngai, 2005). In this study, we combined AHP approach withbi-negotiationto enable the buyer and seller to cooperate to become the most competitiveweapon in this market. Build-to-order (BTO) supply chain can be dened as the valuechain that manufactures quality products or services based on the requirements of anindividual customer or a group of customers at competitive prices and within in a shortspan of time by leveraging the core competencies of partner rms or suppliers tointegrate such a value chain. Thus, bi-negotiation let seller (supplier) know what buyer(manufacturer) wants and make a common consensus of quality products.

3. Methodology

A typical manufacturing Company A lies in a common manufacturing supply chain,which includes its suppliers, distributors, and nal customers. CompanyA produces ther product. It may consist of n major components, which need to be outsourced (CompanyA might have capacities to produce the other components by itself). For each outsourcedcomponent O s ( s ¼ 1 ,. . . n ), there arekr potential suppliers to choose from them. Eachpotential supplier S p ( p ¼ 1,. . .kr ) has a knownproductioncapacity C p: According to theproduction plan, Company A will purchase q units of component from one or moresuppliers out of the whole set of potential suppliers for outsourced component based oncompany A’s predened supplier selection criteria considering each supplier’sproduction capacity. In summary, Company A will make decision in two phases to:

(1) choose most favorable supplier(s) for various outsourced components to meetits supplier selection criteria; and

(2) order various quantities, prices, etc. from the chosen most favorable supplier tomeet its production plan.

The developed methodology for phases (1) and (2) are based on AHP approach andbi-negotiation mechanism, incorporating both quantitative and qualitative factors.To evaluate the suppliers’ offers of product, we proposed a mechanism for supplierselection. The rst phase was to apply decomposition-synthesis approach using AHPapproach. At the second phase, we rst used linear programming to calculate buyer’s


AHP

577



offer (i.e. company A) and then used the agents of buyer and chosen supplier toautomatically negotiate several times. As a result, buyer and chosen supplier can getthe nal satisfying offers. Steps of the algorithm based on AHP and bi-negotiationagents are briey summarized as follows.

3.1 Phase A: apply deconstruct-synthesis approach using AHP 3.1.1 Step A_1. Deconstruct problem . The underlying multi-criteria decision-makingproblem is decomposed according to its selection issue. The overall goal of supplierselection is to achieve overall efciency of suppliers. The efciency measure consists of fourtop-level attributes,namely,assets,business criteria, cost, anddelivery. Eachattributeconsists of a number of specic performance metrics, which are identied in next step.These attributes were determined by reviewing the literature and using a brainstormingtool with the members of the supplier chain department (Barborosoglu and Yazgac, 1997;Braglia and Petroni, 2000; Tam and Tummala, 2001; Masella and Rangone, 2000).

3.1.2 Step A_2. Dene attributes for supplier selection . Supply Chain Council (SCC,1999) constructed a descriptive framework called SCOR. SCOR is a standard supplychain process reference model designed to embrace all industries. SCOR performancemetrics are used as the second-level attributes for supplier selection. The SCORendorses 16 performance metrics, which fall into four dening top level attributes:

(1) Assets:. cash-to-cash cycle time (a1 );. inventory days of supply (a2 );. order quantity (a3 ); and. visitation to supplier facilities (a4 ).

(2) Business criteria:.

performance history (b1 );. production exibility (b2 );. quality performance (b3 );. position in the industry and reputation (b4 );. EDI capability (b5 ); and. Organization structure (b6 ).

(3) Cost:. price (c1 );. logistics cost (c2 ); and. value-added productivity (c3 ).

(4) Delivery:. supply chain response time (d1 );. delivery lead time (d2 ); and. ll rate (d3 ).

These performance metrics are adopted here as the standard attributes for evaluating asupplier’s performance.

BIJ14,5

578



3.1.3 Step A_3. Design the hierarchy . The hierarchy consists of the overall goal,top-level attributes, second-level attributes (performance metrics), sub-level (couldhave several levels), and the decision alternatives. Figure 1 schematically shows theproposed hierarchy based on SCOR metrics.

3.1.4 Step A_4. Construct pair-wise comparison matrix . Once the problem has beendecomposed and the hierarchy constructed, prioritization procedure starts to determinethe relative importance of the elements within each level. Based on productcharacteristics and corresponding supply chain strategies, the relative importance of the top-level attributes and the second-level attributes (performance metrics) isdetermined by experienced managers.

Company A identies its preference to decide the linguistic value of each attribute(Figure 2) to describe how much more important the i th attribute is than the jthattribute. Suppliers also were asked to indicate their preference level as to eachattribute and then to construct a comparison matrix for each attribute.

The comparison matrix of company A for product r determining outsourcedcomponent O s is shown in formula (1) by linguistic attribute values,a ij [ ½1; 3; 5; 7; 9 .a ij represents how much more important the i th attribute is than the jth attribute.

O s ¼ ½a ij m£ m ¼

a11 a12 · · · a1m

a21. .

.a2m

..

. . ..

am1 · · · · · · amm

266666664

377777775ð1Þ

Figure 2.Linguistic representative

preference and theirnumeric values

Numeric values

Equivalent LessImportance

StrongImportance

Very StrongImportance

Absolutely StrongImportance

1 3 5 7 9

Definition

Figure 1.

AHP hierarchy forsuppliers selection

Overall efficiency of suppliers

a1 a2 a3 d3

Supplier 1 Supplier 2 Supplier k r

Objective

Top-LevelAttributes

DecisionAlternatives ……

Second-levelAttributes

Assets DeliveryBusinessCriteria Cost

a3 b1 b2 b3 b3 c1 c2 c3 d1b1 b2 d2


AHP

579



a ij ¼1; i ¼ j

3; 5; 7; 9 otherwise;( a ji ¼1a ij

Where i , j represents each attribute, and m represents the number of attributes.3.1.5 Step A_5. Normalized the comparison matrix . This step was to normalize eachelement of the comparison matrix in the equation (1). Each element in columni isdivided by the sum of the element in columni and form a new matrix N _ O s. The sum of the element in column i of N _ O s is 1.

N _ O s ¼

a11

Pm

i ¼ 1a i 1

a12

Pm

i ¼ 1a i 2

· · · a1m

Pm

i ¼ 1a im

..

. . .. ..

.

am1

Pm

i ¼ 1a i 1

· · · amm

Pm

i ¼ 1a im

2666664

3777775

ð2Þ

3.1.6 Step A_6. Selection of optimal supplier by overall weights . We rst calculated theaverage vector C which is the average of each element in rowi of N _ O s. ci representsthe relative degree of importance for thei th attribute and evaluating score of attributes.

C ¼ ½ci � ¼

a11

Pm

i ¼ 1a i 1

þ a12

Pm

i ¼ 1a i 2

þ · · ·þ a1m

Pm

i ¼ 1a im =m

..

.

a11

Pm

i ¼ 1a i 1

þ a12

Pm

i ¼ 1a i 2

þ · · ·þ a1m

Pm

i ¼ 1a im =m

266666664

377777775

ð3Þ

Then using N _ O s · C to form the overall weight X :

N _ O s · C ¼ X ¼

x1

x2

..

.

xm

26666664

37777775ð4Þ

3.1.7 Step A_7. Consistency check . After step A_5 and A_6, we needed to checkconsistency in the comparison matrix. In AHP, consistency index (CI) and consistency

ratio (CR) are two indexes used to test consistency of the matrix:

r ¼1mX

m

i ¼ 1

xi

ci ð5Þ

CI ðConsistency IndexÞ ¼r 2 mm 2 1

ð6Þ

By test, assume that CI ¼ 0 represents the judgments before and after beingdetermined by the decision maker are completely consistent. CI . 0 represents the

BIJ14,5

580



judgments are not consistent. And CI(0 represents tolerable deviation. So that meansthe smaller the CI value is, the higher the consistency is:

CR ðConsistency RatioÞ ¼CI

RI ðRandom IndexÞ

ð7Þ

CI by random is called random index (RI). And CI/RI is called the consistency ratio (CR). If CR . 0.1, thedegreeofconsistency is satisfactory,but ifCR % 0.1, seriousinconsistenciesmay exist, and the AHP may not yield meaningful results. Then we must go back toconsider another more signicant attribute (Selimet al., 2003; Wanget al., 2004).

3.2 Phase B: Bi-negotiation between company and the chosen most favorable supplier 3.2.1 Step B_1. Fuzzy membership function . The interaction of bi-negotiation agentshappen on two levels: the rst level interaction within user and software agentframework (Chen and Huang) and the second between two agents. The rst level of interaction involves user and software agent communicating in order to come to anagreement on what decision to take as a negotiation agent. A natural way to cope withsuch uncertain communication is to express the interaction as a fuzzy membershipfunction, which incorporates the vagueness of user thinking. The second level of interaction involves bi-negotiation agents interact using offer/counter-offer to reach anagreement. At the rst level, the buyer used his own preference to determine what thesatisfactory degrees of each attribute are. This model assumes that each attribute hadits range that can be changed by the buyer. The buyer can grade from 0 to 1 inaccordance with his satisfaction degree (SD). A triangular fuzzy membership function(Figure 3) with center b can be interpreted as showing attribute quantity “ x isapproximately in the point b.” The others such as a trapezoidal fuzzy number may beseen as an attribute quantity “ x is approximately in the interval.” In order to betterfacilitate interaction between the user and software agent from a practical viewpoint,we then employed triangular membership function to determine the user linguisticvalue of each attribute. When the fuzzy membership function of each attribute isdecided, we can then convert these human linguistic values of attributes into fuzzyvalue (FV). The average of FV of all attributes represents buyer’s and seller’s SDs.

3.2.2 Step B_2.Determine initial offers of buyer . In Phase A, we already knew theoffers of chosen supplier. Because the buyer wanted the average SD to be higher, wenext used linear programming to calculate a hypothetical counter-offer from the seller

Figure 3.Triangular membership

function

1

0AttributeCba

FuzzyValue(FV)


AHP

581



based on seller’s initial requirements. Table I shows some notations about thedenitions. We assumed the SD is the average of all attributes’ FVs:

SDB ¼ average ðsum of all FVBÞ ð8Þ

SDS ¼ average ðsum of all FVSÞ ð9Þ

3.2.3 Step B_3. Determine negotiation bargain strategy of buyer and seller . Havingalready determined the most suitable supplier, the buyer still hoped that thesatisfaction level could be improved. So the offers between the chosen supplier and thebuyer now needed to be negotiated again. For this step, a bi-negotiation mechanismusing agent-based bargain strategies was used to let both sides make concessions oneach attribute.

The bargain strategies were decided by the buyer and seller. The objective of bargain strategy was to nd out mutually satisfying compromise between buyer andseller (In this study, the terms “seller/buyer” and “supplier/Company A” are usedinterchangeably).

Figure 4 shows bargain strategy of each condition. (1), (2), and (3) are seller’sstrategies and (4), (5), and (6) represent buyer’s strategies. Thex-axis stands for theseries of negotiation offers and counter-offers, andy-axis is the changing value of eachattribute from the initial offer/counteroffer represented by IOS (seller agent) and IOB

(buyer agent):

IOB Initial offers of buyerIOS Initial offers of sellerFVB Fuzzy value of buyerFVS Fuzzy value of sellerSDB Satisfaction degree of buyerSDS Satisfaction degree of seller

Table I.Notation

Figure 4.Bargain strategies of buyer and seller agents

(a) (b) (c)

(f)(e)(d)

Seller’s bargain strategy

Buyer’s bargain strategy

IOS IO S IOS

IOB IO B IOB

Negotiationtimes

Negotiationtimes

Negotiationtimes

Negotiationtimes

Negotiationtimes

Negotiationtimes

BIJ14,5

582



(1) The seller makes an initial offer, and then compromises. The seller then addscompromise to a larger degree as negotiations continue.

(2) The seller presents an initial offer and negotiates with a series of incrementallydecreasing offers.

(3) The seller tries advantages an initial offer then they compromise substantiallyin order to let the buyer know their sincerity but stick at this lower offer.

(4) The buyer tries advantages an initial offer then they compromise substantiallyin order to let the seller know their sincerity but stick at this lower offer.

(5) The buyer presents an initial offer and negotiates with a series of incrementallydecreasing offers.

(6) The buyer makes an initial offer, and then compromises. The buyer then addscompromise to a larger degree as negotiations continue.

As shown in Figure 4, buyer and seller predetermine their patterns of compromise andthese strategies can be inputted into their agents so that the negotiation process itself can be automated via the action of the agents.

3.2.4 Step B_4. Bi-negotiation process . According to step B_2 in this phase, whenthe initial offers of both sides were determined, we chose the biggest difference of attributes m i between buyer and seller as our main factor to negotiate rst. Thenusing the bargain strategies of the buyer and seller, the buyer and seller agents beganto negotiate the values of each attribute. This continues until the buyer and selleragents were achieved a common consensus for all attributes at which point thenegotiation stops with a mutually satisfactory offer for both sides. Figure 5 shows theintegrated AHP with bi-negotiation algorithm for supplier selection and negotiationprocess.

4. Case implementationTo demonstrate this proposed mechanism for supplier selection, we used a BTOcomputer manufacturer as our sample case. In order to maintain the condentiality of the rm utilized in the case illustration, the high-end computer manufacturingcompany is referred to as Company A. It produces various functional components,such as mother boards, interface cards, and peripheral components connectivityhardware, etc. The specic component to be outsourced in our hypothetical sample wasthe power supply unit.

It is assumed that only three potential sellers are qualied to supply the outsourcedcomponent. So, in this study three international suppliers of Company A will beevaluated and named as Supplier 1 (Taiwan), Supplier 2 (China), and Supplier 3(Malaysia). Supplier 1 is a famous outsourcing rm which is good at ODM. Supplier 2emphasizes quality and makes the best products. And Supplier 3 is well-known for itsproduction rate. With this basic background about the suppliers, we started oursupplier selection mechanism process step by step.

In Phase A, we create the comparison matrix of priorities attributes desired byCompany A and each supplier relative ability to meet these requirements. To beginPhase A, we met with Company A to determine which attribute was most important tothem in choosing suppliers. We arrived at four top-level quantitative and qualitativeattributes which were assets, business criteria, cost, and delivery. These four were


AHP

583



subdivide further into 16 performance metrics with respective to top level attributes.In this sample case, there were three suppliers to choose from, and we used fourtop-level attributes and 16 performance metrics to compare the offers of each supplier.In Table II, the suppliers’ information in relation to these criteria is listed for theoutsourced power supply.

At this point in Phase A, managers of Company A were asked to prioritize the fourattributes in view of the suppliers’ offer to generate the accompanies matrix. This isdone by using pair-wise comparison with Saaty’s 1-9 scales (Figure 2). To computeeach supplier’s weights, one need is to calculate the overall priority. Overall, priority iscalculated by multiplying the respective terms in priority of the top-level by the priorityof the second-level and the priority of decision alternatives. And Tables III-VIII showthe process of constructing the pair-wise comparison matrices and their normalizedprocess with the second-level attributes “Order quantity” in the top level attribute“Assets” as an example, and the same steps used to compute the other attributes. Tofacilitate Phase A calculation, Matlabw programming utilities were used.

After repeating this calculation for each of the 16 second-level attributes, all of theoverall weights were computed (Table IX).

Several implications for company manager are evident in Table IX. For example,“Price” (c1 ), as expected, played a key role in the supplier selection process (indicatedby the overall weight of 0.18169). In buying and selling situations, price is typically themost important concern for buyers. Next important to the company was “Qualityperformance” (b3 ) (the overall weight of 0.14635). Therefore, suppliers emphasizingstrong quality control would be more successful in this competition.

Figure 5.An integrated AHP withbi-negotiation algorithm

Deconstruct problem

Define attributes for supplier selection

Design the hierarchy

Construct pair-wise comparison matrix

Normalized the comparison matrix

Selection of optimal supplier by overall weights

Consistency check

CR<0.1

Determine initial offers of buyer

Bargain strategy of buyer and seller

Bi-negotiation process

Deal achieved?

Optimal offers of buyer and seller

Yes

YesNo

No

Fuzzy membership function

Choose the most favorable supplier

Phase A

Phase B

BIJ14,5

584



Additionally, Table IX reveals that “Supply chain response time” (d1 ) and “Position inthe industry and reputation” (b 4 ) were also important factors in supplier selection. So,the managers of suppliers can refer to this information sharing in Table IX as a guideto what to prioritize in managing a company.

Assets Business criteria Cost Delivery

Os ¼ Assets 1 1/3 1/5 1/3Business criteria 3 1 2 2Cost 5 1/2 1 3Delivery 3 1/2 1/3 1

Table III.Initial comparison matrix

of top-level attributesamong the suppliers

a1 a2 a3 a4

OA ¼ a1 1 2 5 3a2 1/2 1 5 3a3 1/5 1/5 1 1/3a4 1/3 1/3 3 1

Table IV.Initial comparison matrixof second-level attributes

“order quantity” in thetop-level attribute

“assets”

SellersAttributes Supplier 1 Supplier 2 Supplier 3

Assets Cash-to-cash cycle time (a1 ) (days) 5 7 8Inventory days of supply (a2 ) (days) 15 11 16Order quantity (a3 ) (units) 400 380 375Visitation to supplier facilities (a4 ) (percent) 90 93 95

Business Performance history (b1 ) (percent) 85 88 80criteria Production exibility (b2 ) (percent) 78 85 75

Quality performance (b3 ) (grades) 5 2 4Position in the industry and reputation (b 4 ) (grades) 4 3 1EDI capability (b5 ) (percent) 4 3 5Organization structure (b 6 ) (grades) 3 2 1

Cost Price (c1 ) ($) 13 12 10Logistics cost (c2 ) ($) 30 32 40Value-added productivity (c3 ) (grades) 4 5 3

Delivery Supply chain response time (d1 ) (days) 7 3 4

Delivery lead time (d2 ) (days) 22 28 25Fill rate (d3 ) (percent) 95 96 92

Table II.The initial offers of each

attribute between threesuppliers

Supplier 1 Supplier 2 Supplier 3

O a3 ¼ Supplier 1 1 2 3Supplier 2 1/2 1 2Supplier 3 1/3 1/2 1

Table V.Initial comparison matrix

the attribute “orderquantity” among thesuppliers


AHP

585



A s s e t s

B u s i n e s s c r i t e r i a

C o s t

D e l i v e r y

A s s e t s


C o s t

D e l i v e r y

O s

¼

A s s e t s

1

1 / 3

1 / 5

1 / 3

0 . 0 8 3 3 3

0 . 1 4 2 8 6

0 . 0 5 6 6 0

0 . 0 5 2 6 3


3

1

2

2

N o r m a l i z e d

ÿ ÿ ÿ ÿ ÿ ÿ !

0 . 2 5 0 0 0

0 . 4 2 8 5 7

0 . 5 6 6 0 4

0 . 3 1 5 7 9

C o s t

5

1 / 2

1

3

0 . 4 1 6 6 7

0 . 2 1 4 2 9

0 . 2 8 3 0 2

0 . 4 7 3 6 8

D e l i v e r y

3

1 / 2

1 / 3

1

0 . 2 5 0 0 0

0 . 2 1 4 2 9

0 . 0 9 4 3 4

0 . 1 5 7 8 9

Table VI.Normalized comparisonmatrix of each attributeamong the suppliers

BIJ14,5

586



The last step in the selection process is to calculate the nal rank of potential suppliersfor the outsourced component by using the Equation (10) based on Table IX. Thesupplier selection results are shown in Table X.

Total weights of supplier i ðTW i Þ ¼ C ACa1Csi þ C ACa2Csi þ C ACa3Csi þ · · ·þ CC Cc2Csi þ CC Cc3Csi

ð10Þ

Based on the results for overall weights in Phase A, we recommended Supplier 1 as themost favorable supplier for the Company A. Therefore, the buyer would select Supplier1 as their partner. Then, Company A and Supplier 1 will negotiate with each other inthe next phase.

4.1 Phase B: Bi-negotiation between company and the chosen most favorable supplier 4.1.1 Step B_1. Fuzzy membership function . In this phase, the buyer and seller agentsoffer their negotiating positions reecting their relative priorities for a deal.

Negotiating the deal involves quantitative and qualitative attributes. Each attributehas a predetermined acceptable range for the buyer and seller. The buyer agentsgenerate a starting value for each attribute and form within this range as the rstnegotiation position. They then proceed to use alternate FV in response to on another’scounter-offers until equitable deal is reached (Table XI).

The workload involved in this process of negotiation and recalculation would beheavy if all possible combinations needed to be tried. However, this is rarely necessarybecause in practical negotiations, only specied attributes are negotiated and these arenegotiated within a limited range of mutually desirable values (Steel and Beasor, 1999).In this negotiation, the important issues are quantity, price, quality and delivery.Figure 6 shows results of the fuzzy membership functions for each attribute.

4.1.2 Step B_2.Determine initial offers of buyer . After Supplier 1 was chosen, theinitial offers still needed to be ne-tuned between Supplier 1 and Company A. In orderto manufacture high-quality end product, Company A would like to improve thesatisfaction in the supplier’s offer from 78.75 to 95 percent. So in this step, we usedlinear programming to achieve this objective (Table XII). In Table XIII, we can see that

a1 a2 a3 a4 a1 a2 a3 a4

OA ¼ a1 1 2 5 3 0.49180 0.56604 0.35714 0.40909a2 1/2 1 5 3 Normalized

ÿÿÿÿÿÿÿ !0.24590 0.28302 0.35714 0.40909

a3 1/5 1/5 1 1/3 0.09836 0.05660 0.07143 0.04545a4 1/3 1/3 3 1 0.16393 0.09434 0.21429 0.13636

Table VII.Normalized comparison

matrix of the second-leveattribute “order quantity”

in attribute “assets”

Supplier1

Supplier2

Supplier3

Supplier1

Supplier2

Supplier3

O a3 ¼ Supplier 1 1 2 3 0.54545 0.57143 0.50000Supplier 2 1/2 1 2 Normalized

ÿÿÿÿÿÿÿ !0.27272 0.28571 0.33333

Supplier 3 1/3 1/2 1 0.18181 0.14286 0.16667

Table VIII.Normalized comparison

matrix the attribute“order quantity” among

the suppliers


AHP

587



Top levelattributes

AHPweights

( C i ) Second level attributes

AHPweights

(Cai )C.R.

valueDecisionalternatives

AHPweights

(Csi )

Assets 0.08386 Cash-to-cash cycle time (a1 ) 0.03824 2 1.7844 Supplier 1 0.02384Supplier 2 0.00526Supplier 3 0.00914

Inventorydaysof supply(a 2 ) 0.02715 2 1.9152 Supplier 1 0.00455Supplier 2 0.02004Supplier 3 0.00256

Order quantity (a3 ) 0.005702 1.9195 Supplier 1 0.00307Supplier 2 0.00169Supplier 3 0.00093

Visitation to supplier 0.01277 2 1.8234 Supplier 1 0.00220facilities (a4 ) Supplier 2 0.00669

Supplier 3 0.00388Business 0.39010 Performance history (b1 ) 0.03584 2 1.7844 Supplier 1 0.00857

criteria Supplier 2 0.02235Supplier 3 0.00493

Production exibility (b2 ) 0.08030 2 1.9213 Supplier 1 0.01846Supplier 2 0.05203Supplier 3 0.00981

Quality performance (b3 ) 0.14635 2 1.9004 Supplier 1 0.08136Supplier 2 0.01322Supplier 3 0.05177

Position in the industry andreputation (b4 )

0.08197 2 1.9168 Supplier 1Supplier 2Supplier 3

0.048420.027360.00619

EDI capability (b5 ) 0.027822 1.8949 Supplier 1 0.00507Supplier 2 0.00320

Supplier 3 0.01955Organization structure (b 6 ) 0.01782 2 1.9005 Supplier 1 0.01049Supplier 2 0.00449Supplier 3 0.00284

Cost 0.34691 Price (c1 ) 0.181692 1.9053 Supplier 1 0.01929Supplier 2 0.04733Supplier 3 0.11506

Logistics cost (c2 ) 0.105532 1.9215 Supplier 1 0.06134Supplier 2 0.03262Supplier 3 0.01157

Value 2 added 0.05971 2 1.9053 Supplier 1 0.01555productivity (c3 ) Supplier 2 0.03782

Supplier 3 0.00634Delivery 0.17913 Supply chain response 0.11345 2 1.8455 Supplier 1 0.07266

time (d1 ) Supplier 2 0.02336Supplier 3 0.01743

Delivery lead time (d2 ) 0.046662 1.9161 Supplier 1 0.03428Supplier 2 0.00396Supplier 3 0.00843

Fill rate (d3 ) 0.0190122 1.8527 Supplier 1 0.00402Supplier 2 0.00194Supplier 3 0.01305

Table IX.All of the overall ratingweights in phase A

BIJ14,5

588



the FV for the attribute “Quality” between both seller and buyer was 1 which means thiscomponent’s quality wasassured to be high. Therefore, we did notneed to negotiate thisitem. However, the difference between the respective FVs for the attribute “Quantity”(0.4 and 1) was the greatest so we regarded this attribute as our main factor to negotiate.

4.1.3 StepB_3. Determine negotiation bargain strategy of buyer and seller . In thisstep, we rst set the bargain strategy of seller and buyer then we set the agents for each

Figure 6.Fuzzy membershipfunctions for each

attribute from userinteraction

Price (Dollars)

0

0.2

0.4

0.6

0.8

1

1.2

1 0 1 1 1 2 1 3 1 4 1 5 1 6 1 7 1 8 1 9 2 0

quantity (unit)

0

0.2

0.4

0.6

0.8

1

1.2

3 0 0

3 5 0

4 0 0

4 5 0

5 0 0

5 5 0

6 0 0

6 5 0

7 0 0

7 5 0

8 0 0

Delivery (Days)

0

0.2

0.4

0.6

0.81

1.2

2 0 2 1 2 2 2 3 2 4 2 5 2 6 2 7 2 8 2 9 3 0

Quality (grades)

0

0.2

1 2 3 4 5

0.4

0.6

0.81

1.2

(a) Quantity (b) Price

(c) Quality (d) Delivery

FV FV

FV FV

Final rank

Supplier 1 0.41316Supplier 2 0.30335Supplier 3 0.28349

Table X.Supplier selection results

Attributes Acceptable range

Quantity 300-800 (units)Price 10-20 ($)Quality 1-5 (grades)Delivery 20-30 (days)

Table XI.Acceptable ranges of

each attribute


AHP

589



side to negotiate. Figure 7 shows the strategy of seller agent for “Quantity”. We can seethat the seller will increase the quantity value steeply in the rst two rounds.After that, the seller compromises less and less. The buyer agent bargain strategy asshown in Figure 8. The buyer will decrease the quantity value only slightly in the rstve rounds. After that, the buyerdecreases the quantity value to a larger degree to reachan agreement.

4.1.4 Step B_4.Bi-negotiation process . With the respective bargain strategies nowset, the buyer and seller’s agents negotiate automatically on behalf of each side toachieve common consensus with the highest possible level of satisfaction for both.

Objective SDB ¼ 0.95

Constraint FVB $ FVS

FVB $ 1

Table XII.The objective and theconstraints of linearprogramming

IOS FVS IOB FVB

Seller (Supplier 1) Buyer (Company A)Quantity 400 0.4 Quantity 600 0.8Price 13 0.95 Price 12 1Quality 5 1 Quality 5 1Delivery 22 0.8 Delivery 20 1SDS 78.75 percent SDB 95 percent

Table XIII.The initial offers of theseller and the buyer

Figure 7.Bargain strategy of selleragent

Negotiation times of Quantity of Suppleir B

0.000 1 2 3 4 5 6 7 8 9 10 11

100.00200.00300.00400.00500.00600.00700.00800.00

Figure 8.Bargain strategy of buyeragent 0.00

100.00

200.00

300.00

400.00

500.00

600.00

700.00

0 1 2 3 4 5 6 7 8 9 10 11

Negotiation times of Quantity of Buyer

BIJ14,5

590



In Table XIII, we saw that the attribute “Quantity” marked the largest gap between theneeds of Company A and the supplier’s offer. The initial offers of Supplier 1 onlyprovided for 400 units, but the manufacturer wanted 600 units. Thus, this is wheretheir agents began the automatic bi-negotiate process.

Figure 9 shows the bi-negotiation results each time. As we see in Figure 9, at thethird negotiation, buyer and seller reach a common consensus for each attribute.With the exception of “Quality,” all attributes adjusted their values during thisnegotiation. The negotiations continued until a mutually satisfactory compromisevalue was reached for quality, price, and delivery at which point the negotiationprocess ended. The nal offers of both sides are shown in Table XIV. As a result of thebi-negotiation the satisfaction level approaching the target of 95 percent, reaching90.8 percent, an increasing of 12.05 percent over the starting level of 78.75 percent.

From the moment, the supplier was chosen as the partner of Company A, theinteraction between both sides was crucial. Integrated AHP with bi-negotiation agentsprovided a vehicle to negotiate and achieve a win-win situation.

5. Conclusions and future workSupplier selection and evaluation is very important to the success of a manufacturingrm. This is because of the cost and quality of goods and services sold is directlyrelated to the cost and quality of goods and services purchased. Therefore, purchasingand supplier selection play an important role in supply chain management. Sellers andbuyers in internet-based supply chain through negotiations have signicant impact onsupplier selection and partners’ prot. In this study, the integrated analytichierarchical process approach with bi-negotiation agents have been proposed as apotential tool for analyzing and evaluating suppliers in the electronic supply chain.Using the main and sub-attributes for supplier selection in AHP were clearly identiedand the problem solved was structured systematically. This enables decision-makersto examine the strengths and weaknesses of vendor systems by comparing them with

Figure 9.Bi-negotiation agents

process

Quantity

(3.591)

0.00100.00200.00300.00400.00

0 1 2 3 4 0 1 2 3 4 0 1 2 3 4

500.00600.00700.00

Negotiation times

Quantity

Seller Buyer

Price

(3.13)

0.002.004.006.008.00

10.0012.0014.00

Negotiation times

Quantity

Seller Buyer

Delivery

(3.21)

0.004.00

8.0012.0016.00

20.00

Negotiation times

Quantity

Seller Buyer

(a) Quantity (b) Price (c) Delivery

Final offers

Quantity 591 0.782Price 13 0.95Quality 5 1Delivery 21 0.9SD 90.8per cent

Table XIV.The nal result between

buyer and seller


AHP

591



respect to appropriate top and second-level attributes. During the decision-makingbetween buyer and seller, it is crucial to negotiate on multiple attributes for a deal suchas price, quantity, quality, delivery, and relative preferences. Therefore, this paper isalso attempt to develop how to elicit the user’s bargain strategies in order for hisautonomous agent to negotiate on their behalf. Moreover, the approach is recognized asone of the best bargain strategies for automated negotiation. Based on this work, ourfuture extension is to investigate other decision phases in supplier selection andprovide similar approaches to enrich the available literature. We will evaluate a moredetailed form, the inuence of other methodologies on the nal quality and accuracy of decisions. We will also try to enhance our decision support system with softwareagents’ techniques to enable managers comparing different solutions and making morerigorous and practical decisions.

ReferencesBarborosoglu, G. and Yazgac, T. (1997), “An application of the analytic hierarchy process to

the supplier selection problem”, Production & Inventory Management Journal , Firstquarter, pp. 14-21.Bhutta, K.S. and Huq, F. (2002), “Supplier selection problem: a comparison of the total cost of

ownership and analytic hierarchy process approaches”, Supply Chain Management: An International Journal , Vol. 7 No. 3, pp. 126-35.

Braglia, M. andPetroni,A. (2000), “A qualityassurance-oriented methodology forhandling trade – offs in supplier selection”, International Journal of Physical Distribution & Logistics Management , Vol. 30 No. 2, pp. 96-111.

Dickson, G. (1966), “An analysis of vendor selection systems and decisions”,Journal of Purchasing , Vol. 2 No. 1, pp. 5-17.

Gunasekaran, A. and Ngai, E.W.T. (2005), “Build-to-order supply chain management: a literaturereview and framework for development”, Journal of Operations Management , Vol. 23,

pp. 423-51. Jiang, K. and Wicks, E.M. (1999), “Integrated investment justication approach for cellular

manufacturing systems using activity-based costing and the analytic hierarchy process”, Journal of Engineering Valuation and Cost Analysis , Vol. 2 No. 4, pp. 271-84.

Liu, J., Ding, F.Y. and Lall, V. (2000), “Using data envelopment analysis to compare suppliers forsupplier selection and performance improvement”, Supply Chain Management: An International Journal , Vol. 5 No. 3, pp. 143-50.

Mandal, A. andDeshmukh, S.G. (1994), “Vendor selection using interpretive structural modeling”, International Journal of Operations & Production Management , Vol. 14 No. 6, pp. 52-9.

Masella, C. and Rangone, A. (2000), “A contingent approach to the design of vendor selectionsystems for different types of co-operative customer/supplier relationship”, International Journal of Operations & Production Management , Vol. 20 No. 1, pp. 70-84.

Muralidharan, C., Anantharaman, N. and Deshmukh, S.G. (2002), “A multi-criteria groupdecisionmaking model for supplier rating”, The Journal of Supply Chain Management ,Fall, pp. 22-33.

Narasimhan, R., Talluri, S. and Mendez, D. (2001), “Supplier evaluation and rationalization viadata envelopment analysis: an empirical examination”, The Journal of Supply Chain Management , Summer, pp. 28-35.

Oliveira, R.C. and Lourenco, J.C. (2002), “A multicriteria model for assigning new orders toservice suppliers”, European Journal of Operational Research , Vol. 139, pp. 390-9.

BIJ14,5

592



SCC (1999),Overview of the SCOR Model V2.0 , Supply Chain Council, Inc., Washington, DC,available at: www.supplychain.org

Selim, Z., Mehmet, S. and Mehves, T. (2003), “Fuzzy analytic hierarchy based approach forsupplier selection”, Journal of Euromarketing , Vol. 12 Nos 3/4, pp. 147-76.

Steel, P.T. and Beasor, T. (1999),Business Negotiation: A Practical Workbook , Gower PublishingLimited, Aldershot.

Tam, M.C.Y. and Tummala, V.M.R. (2001), “An application of the AHP in vendor selection of atelecommunication system”, Omega , Vol. 29, pp. 171-82.

Vokurka, J.R., Choobineh, J. and Vadi, L. (1996), “A prototype expert system for the evaluationand selection of potential suppliers”, International Journal of Operations & Production Management , Vol. 16 No. 12, pp. 106-27.

Wang, G., Huang, S.H. and Dismukes, J.P. (2004), “Product-driven supply chain selection usingintegrated multi-criteria decision-making methodology”, International Journal of Production Economics , Vol. 91, pp. 1-15.

Weber, C.A., Current, J.R. and Benton, W.C. (1991), “Vendor selection criteria and methods”, European Journal of Operational Research , Vol. 50, pp. 2-18.

Yan, H., Yu, Z. and Cheng, T.C.E. (2003), “A strategic model for supply chain design with logicalconstraints:formulation andsolution”, Computers& Operations Research , September,pp.1-21.

Further readingBanuelas, R. and Antony, J. (2004), “Modied analytic hierarchy process to incorporate

uncertainty and managerial aspects”, International Journal of Production Research , Vol. 42No. 18, pp. 3851-72.

Buyukozkan, G. and Feyzioglu, O. (2004), “A new approach based on soft computing toaccelerate the selection of new product ideas”,Computers in Industry , Vol. 54, pp. 151-67.

Chen, Y.M. and Huang, P.N. (n.d.), “Bilateral multi-issue negotiations in multiagent framework”, Articial Intelligence (in press).

Dubois, D. and Prade, H. (1980),Fuzzy Sets and Systems , Academic Press, New York, NY.Kim, J.S. (2003), “Negotiation support in electronic commerce using fuzzy membership functions

and AHP”, Proceedings of the International Conference of Korea Intelligent InformationSystems Society (ICKIISS) 2003, Seoul (Korea) , pp. 99-104.

Xudong, L., Nicholas, R.J. and Nigel, S. (2003),Acquiring Tradeoff Preferences for Automated Negotiations: A Case Study , AMEC, Southampton.

About the authorsYee-Ming Chen is an Associate Professor in the Department of Industrial Engineering andManagement at Yuan Ze University, where he carries out basic and applied research inagent-based computing. His current research interests include soft computing, supply chainmanagement, and system diagnosis/prognosis. Yee-Ming Chen is the corresponding author andcan be contacted at: [email protected]

Pei-Ni Huang is a graduated student in the Department of Industrial Engineering andManagement at Yuan Ze University, where she is studying basic and applied research inagent-based negotiation. Her current research interests include negotiation and supply chainmanagement. E-mail: [email protected]


AHP

593

To purchase reprints of this article please e-mail: [email protected] visit our web site for further details: www.emeraldinsight.com/reprints

Chen, Huang - 2007 - International Journal - Bi-Negotiation Integrated AHP in Suppliers Selection

Documents