Computers ind. Engng, VoL31,No. 3/4,pp. 821 - 826, 1996
Pergamon
Cop~ght© 1995ChinaMachine Publi~edby ElsevierScienceLtd.Printedin GreatBritain S0360-8352(96)00~5-0 0360-8352,96 $15.00+ 0.00
A QUANTITATIVE METHOD USED IN NEGOTIATION SUPPORT SYSTEMS Hnang Jinghna School of Economics and Management, Tsinghna University, Beijing,100084,P.l~China Abstract: It is important for small and medium sized business to enter international market with an electronic international trade network-the Marketplace, which is a large and synthetic compoter-based network to enable an organization to engage in buying and selling products and services across national borders almost as easily as one can conduct such business locally. As one of its necessary components, Negotiation Support Systems(NSS) aid negotiators to bargain and reach agreement in negotiation process. On the basis of discussing the framework and capabilities of the Marketplace and the theory of negotiation, this paper introduces a quantitative method used in negotiation support system, which systematically analyses trade-offs between variables and is an additive scoring method to get to the efficient frontier in order to reconunend high quality alternatives near it. The principle and techniques of this approach are also suggested. Key words: Negotiation Support System, Efficient Frontier, Marketplace, Trade-net 1 INTRODUCTION China's economy has become an important part of the international market along with its gradually deepening reforming and opening. To realize the internationalization, our enterprises need not only the support from fund and policy,but also to make full use of international information and experience, to use methods and techniques of the developed countries' enterprises on international trade. In fact, facing the tendency of integration of the world's economy ,tools and techniques which promote international trade success are studied in the way of information in developed countries. The research about trade network(tradenet)-Marketplace has been noticed widely. Trade-net is a system based on computer network. With its support and help, enterprises can exchange trade information and commodities efficiently among countries. The major users of the trade-net are some middle or smallsized enterprises which are limited by information and are di~colt to do international trade. The successful implementation will convenience and speed up the enterprises to do international trade as in domestic. But the trade-net is still in the stage of theoretical research. It involves many technological interesting problems. Some problems may be solved by present technology. But some problems such as negotiations ,languages and law consulting should be studied further after taking into account integrated framework of this system. This paper will introduce the structure and capabilities of the Marketplace, summarize negotiation theory and study the Negotiation Support Subsystem. A quantitative method will be suggested. This method is useful to the subsystem. 2 THE FRAMEWORK AND CAPABILITY OF THE TRADE-NET 2.1 The frlunework of the Trnde-net The initial consideration about the Trade-net structure is that it consists of six data/model bases and five subsystems. They are Enterprises/Products Data Base which provides information about enterprises and their products/services, Dynamic Market Information Base on which sellers/buyers can post information about products/services one would like to provide or obtain, Model Base which provides models or methods about conducting negotiations, Ancillary Information Base which provides information about customs regulations, shipment of goods, financing, laws and the like, Alternative Cooperators Base which provides possible cooperators about one requirement on the Market Information Base, and Contract Base provides information on all contracts. The five subsystems are Retrieving and Inquiring Subsystem which searches products/services information in Enterprises/Products Data Base , Responding and Posting Requirement Subsystem which responds or posts requirement information from/to Dynamic Market Information Base, Selecting Negotiators Subsystem which selects the most suitable negotiator from Alternative Cooperators Base, Negotiation Support Subsystem which helps negotiators conduct negotiations, Contract Ancillary Management Subsystem which manages all contracts. The framework of the Trade-net is shown in Fig. 1. The whole business process according to this structure is the following: users can inquiry information from DB in the net and find candidates of negotiators,at the same time, users can post or search their requirement in Market on
821
822
18th Imernazional Conference on Computers and Industrial Engineering
~ a ~ r p r i s e s / P r ~ u~~ base
/ Retrieving&Inquiring
~ternative ~---~
LSubsystem
I Cooperators
[ Information Base [
Responding & Posting Requirement Subsystem
/ - -
I
Selecting Negotiators Subsystem [
[ Negotiation Support t Subsystem
B~llary I n f o ~ [
t Contracts Ancillary [ Management Subsystem J
L
Fig. 1 The Framework of the Trade-net which users present their cooperative idca;through this two kinds of search , users can get the list of candidates of cooperation; after the final cooperator is selected, the two sides conduct a negotiation and sign a contract ; the two sides carry out the contract. 2.2 The capabilities of the Trade-net .Dynamic Information Retrieval and Publication The objective of retrieving and posting information in Enterprises/Products Data Base and Dynamic Market Information Base is to ensure users may find some possible candidates of negotiators to be selected. If there are too many retrieved candidates, it may be very difficult for users to select satisfied negotiators.If it happens, the whole system will lose its value. So in order to approach this objective a set of effective retrieval and filtering software are needed. On the other hand, specification of enterprises/products must be effective. But how to define is also a problem to be studied. The "market" is defined as a set of electronic notice board. The difference between the "market" and the general database is that the "market" will provide dynamic information to all of users to let them know who is seeking for cooperators and which type of cooperators he is seeking for; responders who intent to cooperate with him can respond actively, while the retrieval of the general database is static and is operated passively. The method used to retrieve "market" may be the same with that used to general database. It is noticeable that, as a dynamic system, the "market" should have the function of automotive searching to let all enterprises know chances that exist in the market, have opportunity to get information in time and become cooperators. .Select a Suitable Negotiator When responders/cooperators receive information, they respond at once. In order to continue cooperation, cooperators should suggest feasible alternatives in the first respondeney, which needs searchers ( who post requirement ) put forward perfect specification of products/service requirement at the beginning. Responders should use flexible language and format to make searchers know them deeply. After responding,a searcher should select one or two negotiators from those responders to conduct negotiations. Because initially the searcher understand responders only from their respondency, so it is needed for the system to supply a function by which the searcher can deeply communicate with responders in order to select a suitable negotiator. Under ideal conditions, the information communication should be as open as possible. But, during international trade, because of language limitation, it needs a translation system, so that the two sides can communicate smoothly. .Support Negotiations Conducting a negotiation and signing a contract are complex. They are more difficult when they are done on computer network in long distance. In order to decrease difference of negotiators' culture and habits, their
18th International Conference on Computers and Industrial Engineering
823
mmmunigaUo"nshouldbe limited. But the communication shoeld not be little to mti~y the needs of negotiatiou. The suitable relation is to use the lest conummication to include all ef mgotiatm"n problems. It should involve two contents: one b the m n degree of two side~tbe ether is entries or vambles of m n t r a ~ . .lVhm~ CoaUat~ C~tragtsJotddbecdear and definite so that two ~ caaumkrslaadtheirconnotation. The genernl way to ~rtify that two sides are in agreement on a o~ntract is to sign on a oontract. And this way can be done by computer technology. But signing a contract does not mean finishing a trade. In a practical trado,after that, there are several activities. These activities will be controlled by Contract Ancillary Management Subsystem. It will supply capabilities like quality test ,payment and law consulting. $ SUMMARY OF NEGOTIATION THEORY 3.1 Negotiation models .Game Tbeony Game models are fundamentally static models in that they focus on outcomes rather than on processes. Consequently, these models are relevant for prescriptive purposes, but are inadequate for descriptive purposes. Game models assume that the number and identity of players are fixed and known,that players are completely rational, that alternatives and utility functions are fixed and known, and that communication takes place only within the model and does not affect either the form or the content of a game's payoff matrix.J2][3] .Economic Model Economic models treat negotiations as a process of convergence over time involving a sequence of offers and counteroffers. They are dynamic models that focus on the bargaining process as well as the ultimate outcome of bargaining.But they assume that the utility functions of the participants are fixed and known from the outset, that a compromise zone exists, is identifiable, and remains stable over time. They tend to emphasize the formation of expectations about the behavior of the relevant others. [2][3] .Multiple.Criteria Model Multiple-criteria mod.els recognize the multiplicity of goals and objectives underlying negotiators' behavior and aim at developing a decision rude. These models assume that the utility function of each negotiator is stationary, and may be assessed separately and then aggregated. Following the utility independence assumption, the negotiations process is reduced to specifying preferences, which are combined for each participant and then for the group. The group utility function is used to generate compromises. But the models that do not require the definition of a group utility use decision rules derived from available alternatives.J3] .Experience Model Experience models focus on the recording agenda, process and the final outcome in each instance of negotiations. Such recordings allow the development of a database of negotiaticm histories that can be used to study negotiation strategies adopted by different individuals and simulate consequences of adopting a different strategy from the one in current use. Expert system techniques, like case-based reasoning, can be used to provide expertise to one or more players in a particular problem domain.
3.2 Negotiation Types .Distributive bargaining It's a competitive or win-lose bargaining. The objectives of two parties are in fundamental conflia. Resources are constant and limited and each party wants to maximize its share. .Integrative Bargaining It's bargaining in which all parties share objectives and outcomes and they gain benefits simultaneously. Commercial bargaining belongs to the second. 4 NEGOTIATION SUPPORT SYSTEM There is growing interest in Negotiation Support Systems as a branch of GDSS , which combines communication,computer and decision/negotiation technologies to support negotiators to arrive at an agreement.Technological advancements such as electronic boardrooms, local- and wide- area networks, telccoofcrencing, and decision support sollware have spurred research in this area. A complete NSS consists of two parts: individual decision/negotiation support systems and communication support systems. Individual decision/negotiation support systems not only have characteristics of general DSS but also are equipped with domain models,such as game, economic, psychological ones of negotiation.The complete NSS should have the following capabilities:[2] a)The system should support requirements analysis.
824
18th International Conference on Computers and Industrial Engineering
b)lt should support strategic analysis,which includes analysis of the other party's needs(and thus may involve some degree of guessing to begin with) as well as the team's joint position with respect to the efficiency frontier.This new functionality could even be utilized to locate the fairest solution(for example,the Nash solution) for the team. c)It should support the interaction per se.The system should provide a communication channel through which the negotiators are able to refer to some comnmn referents for facilitating the interaetion.The common referents may appear in the form of reasoning and justifications for supporting one's arguments. This paper is Uying to examine an approach which has the second functionality.
5 ADDITIVE SCORING METHOD FOR COMMERCIAL NEGOTIATION 5.1 The principle of additive scoring method This method combines the utility idea of game theory and economic model,methods of multiple-criteria and the experience of some experts. Negotiation variables or issues are those which negotiators use to make trade-offs. They can be quantitative or qualitative. For example, in commercial negotiations, the main variables are price, quantity, quality, delivery ,etc. Values refer to the particular endowed quantity on variables. There are two assumptions of this approach. a)negotiation variables(issnes) are independent. b)all possible values on each variable are given(two parties have agreed upon it). The process of this method is as following: a)For each side ,rank all variables according to their importance and one's preference. b)For each side ,rank all possible values on each variable according to one's preference. c)For each side, give each variable a score according to the result of a) and the addition of each score equals to I. d)For each side,use a scoring system to give all values on each variable scores according to the result of b). The highest score of the value on each variable equals to 100. e)For each side, score the best alternative to a negotiated agreement(a cutoff score for acceptance of agrcement).(BATNA) t)Compote the part of the efficient frontier which dominates two sides joint agreement point and improve a noneflicient negotiated outcome. The last two steps work are done by mediator. In the process of nsing this approach to assist in negotiation,we have to solve the following problems: .How to get all values on each variable? .How to rank and score the importance of all variables? .How to rank and score values of each variable? .When is it appropriate to use an additive scoring system? .How to cope with nonadditivity? .How to identify and score a BATNA? .How to get efficient frontier? 5.2 Some techniques used to those problems There are many variables in international trade commercial negotiations.But some are focused on trade-offs.They are the following: produet/service specification, price, quantity, quality, delivery, terms of payment, delay penalty, warranty. Firstly, negotiators agree on the range of each variable. Second, within the range they discuss and agree with all possible values of each variable. The process of getting all values on each variable can be supported by a communication system. The second key problem is how to rank and score the importance of all variables. In theory we exam two approaches. One way is to compare all variables pairwise. For each pair a negotiator determines which of the two options is preferred. And at last, not only the order of all the preferences but also the numerical representation of his preferences are got. We can use AHP software to solve this problem. The second way is given by Keeney, Raiffa, van Winteffeldt and Edwards.(See 141,151) The third key problem is how to rank and score the values of each variable. The approach is just the second way of the second problem. It points out that quantification of preference can be characterized by five activities. The first two are qualitative and consist of identifying a complete set of issues and specifying possible agreement values on each variable. The third activity is to determine a reasonable value model, like the additive scoring system for combining the multiple variables. The fourth and fifth activities are quantitative in nature. The fourth is to assign relative numerical preferences to the values of each variables. The fd~h activity is to address the preference trade-offs and state exactly how much achievement on variable M that one will just give up to increase achievement by a specific amount on variable N. The forth problem is the condition of additive scoring system. There are two assumptions of appropriateness of the additive scoring system. First all possible values on each variable do not depend on the particular value of another
18th International Conference on Computers and Industrial Engineering
825
variable. Second the relative importance of two variables does not depend on the value of another variable. Or in another way to express these two conditions is if and only if the value trade-otis between any two variables do not del~ad on the particular value chosen of the remaining variables. The fifth problem is how to cope with nonadditivity. In real world there is more nonadditivity than additivity.How to cope with nonadditivity7 The answer is to convert no_ta~litivity to additivity. First, if dependence is simple, we can reorganize the set of variables. For example, value trade-offs between variable 1 and 2 depend on the value chosen of variable 3, but the value trade-offs among variables 1,2 and 3 do not depend on remaining variables. We would say that the set of variables{l,2,3} is preferentially independent of remaining variables.Becuuse of this preferential dependence within the set of variables{ 1,2,3 }, however, we will not be able to achieve an additive scoring system over the whole variable set.But we may be able to define a new composite variable X that comprises
variables{ 1,2,3} and achieve additivity across the redefined set of variables{x,4,5....... }. Secondly, it is to redefine the set of variables/issues m order to accommodate variables to additivity. The sixth problem is how to identify and score a BATNA. On the basis of a scoring system, a negotiator might establish a cutoff score for acceptance of an agreement: below the cutoff means no agreement; above the cutoff means an acceptable agreement. The BATNA is scored at this cutoff point~ It's important for a negotiator to identify BATNA. Raiffa[5] give a methodology to identify and evaluate BATNA. The seventh problem is how to get efficient frontier. The efficient frontier is defined as the locus of achievable joint evaluations from which no joint gains are possible. It's the set of pairs of scores for two parties such that neither score can be improved without lowering the other score. Overall, the results of negotiations are quite spread out and usually fall below the efficient frontier. Howard Raiffa[7] gives out a method to compute the efficient frontier. Figure 2 expresses efficient frontier. Let i (for i=1,2) designate a negotiator; let j (for j=l ...... J) designate a variable;let X/be a value of the jth variable with domain [aj,bj] (for j=l,...,J),iet ~.(xj) be the component score of negotiator i on j at xy ;let co# be its important weight on variable j where~"~ co# =1 (for i=l,2),let i's total score for contract x=(x]..... ,X j) be given by J
¢,(x)=~ c%¢,j(xj) J If we weight VI by 1 and V2 by 2 (which we will vary to get different points on the efficient frontier),we observe that
¢~(x)+X¢:(x)--~ c% ¢,~(x~)+x~ c% ¢~j(xj) J
J
--~[c% v,j(xj)+x c% ¢:~(xj)l J Hence, if we want to choose x to maximize Vl ( x ) + 2 V: (x),then we merely have to choose Xj in i a~, bjl to maximize •
co~j V~i(xj)+A co2j V:/(x~ ) forj=l ...... J.Letxj be an optimum for that 2 and let X c~t) =(x,6t),....,X~'t),.....,X~x)) .Tbejoint evaluation (Vl(x(x)), Vz(x(x))) whose supporting tangent line is the locus of (V~, V~) points for which
V~+2~ =V,(x~))+2 ~( x ~))
will be a point on the efficient frontier
826
18th International Conference on Computers and Industrial Engineering Part A's Score
/\ lO0 100
p~rt B's BATNA score R •
.T
80
35
-
J is BATNA point I*
30
60
90 100
>
Part B's Score Fig.2 Efficient Frontier
6 Conclusien Because the purpose of our research is to support commercial negotiation ,the question is how to accommodate commercial negotiation to the Additive Scoring System.Some variables in commercial negotiation are interdependent. So we have to convert original variables to new variables in order to accommodate them to additive scoring systems.Generaly,there is relationship like this: within some certain range of variable 1 ,the values of the other variable 2 does not depend on variable 1. But ,when the value of variable 1 is given by value q ,the value function of variable 2 is changed compared with the first part. Hence, we can change the set of variables {1,2,3........ J} to {1',3........ J}.The possible values of 1' are the values of pairs of {1,2}. The advantage of this method is interactive between user/negotiator and computer because user can input his preference in the process of negotiation. In order to verify this method,we'll develop a software through a color graphics, mouse-driven windows environment. REFERENCES [l]David W. Conrath. Some ideas about an Electronic International Trade Network-the Marketplace.working paper, 1994. [2]Lai-hnatlim. A Theoretical Perspective of Negotiation Support Systems Journal of Management Information Systems, $1ol.9,No.3,pp.27-44. [3]Gregory E.Kersten.An Analytic Basis for Decision Support in Negotiations.Naval Research Logistics, Vol.38,1991. [4]Keeney,RalphL.,andHowardRaiffa. Decisions with Multiple Objectives. New York:Wiley, 1976. [5]Von Winteffeidt,Delof and Ward Edwards.Decisian Analysis and Behavioral Research. 1986,New York:Cambridge University Press. [6]I~Ftffit,Howard.The Art and Science of Negotiations. Cmbridge,Mass.: Hard University Press,1982. [7]Young,H.P. Negotiation Analysis . The University of Michigan Press,1991.