Bargaining Power in Electronic Negotiations: A Bilateral Negotiation Mechanism

Bargaining power has a major influence in negotiations. Up to now, a lot of electronic negotiation models have been developed and manifold negotiation challenges have been already addressed, but mainly related to the structure and the process of the negotiation. However, research concerning bargaining power is still inadequate represented. Thus, in order to contribute to the state of the art of electronic negotiations, this paper shows a bilateral automated negotiation mechanism that considers bargaining power. 1 Motivation and Problem Description In spite of a rich developed field of electronic negotiation models, the best initial negotiation offer is still a major challenge. The corresponding problem is the following: Generally, whenever an (initial) offer is made by a negotiation partner, it needs a good reason to vary it. So he is more or less bounded to his offer. If the initial offer is too far away from a potential zone of agreement in order to make a good deal, it will normally take a long time to reach an agreement and some of the negotiation partners will probably exit the negotiation before. However, if the initial offer is too close to the reservation price of the bidder at a risen possibility of a quick agreement, the surplus of the bidder will be smaller. The best initial negotiation offer depends on many factors; for example negotiation deadlines [41], utility functions [17] etc. However, research concerning bargaining power, as one of this major influencing factors [13, 36], is still inadequate represented [11, 19]. Thus, this paper focuses on bargaining power in electronic negotiations. The text of the paper is divided into 4 parts: In section 2 we sketch the role of bargaining power in organizational approaches and related work in multiagent literature. Section 3 presents an overview of electronic negotiations. After that, section 4 shows a bilateral automated negotiation mechanism considering bargaining power. At the end, future research questions are introduced. 2 Power in Organizational Approaches A lot of old and modern scientific theories point out the significant influence of the power distribution in organizations (e. g., Contingency-Theory, BehavioralApproach). 2 Ricardo Buettner and Stefan Kirn 2.1 Contingency-Theory The Contingency approach is rooted in different other theories. For example, research in social sciences pointed out within the scope of the Bureaucracy Approach by M. Weber [56] that organizational structures are not in line with the ideal type of bureaucracy. During the fifties business studies came to the conclusion that a universal organization structure can not exist. One of the first was J. Woodward [57, 58], who argued the importance of taking the individual initial position of the organization into account whenever design recommendations concerning the organizational structure are given. At the beginning of the development of the Contingency-Theory, empirical research studied the impact of the influencing factors to the organization structure [12]. For example, following the popular thesis by T. Burns and G. M. Stalker, organic structures are in line with dynamic environments and mechanistic structures fit to static ones [8]. The Contingency-Theory presumes power in organizations as an important phenomenon. The management tries to make sure its authority, mainly on the basis of organizational rules. 2.2 Behavioral Approach The development of the behavioral approach is marked by the Bayes-Rule [2], the ”St. Petersburg-Game” [4] by D. Bernoulli, the Expected Utility Theory by J. L. von Neumann and O. Morgenstern [55], the concept of Bounded Rationality by H. A. Simon [49], the concept of Satisficing [35, 49], the Subjective Expected Utility Theory by L. J. Savage [45], the Allais-Paradox [1], the Ellsberg-Paradox [16], the Prospect-Theory by D. Kahneman and A. Tversky [24, 25], the RegretTheory by G. Loomes and R. Sugden [33] or rather by D. E. Bell [3] and the concept of Framing by A. Tversky and D. Kahneman [53, 54]. The behavioral approach shows key findings concerning the rationality assumptions in negotiations. Recapitulating, the assumption of perfect rationality (Homo Oeconomicus) was theoretically and empirically disproved. The neoclassic premises were revised by the behavioral approach and replaced by bounded rationality and imperfect information situations [49]. Therewith, bargaining power plays an important role within negotiations [41]. The decision process is significantly influenced by the distribution of power. 2.3 Power in Multi-Agent-Systems M. S. Fox [18], T. W. Malone [34], K. M. Carley and L. Gasser [10], et al. consider Multi-Agent-Systems (MAS) as organizations (in the sense of organized social entities). Availability of different forms of power (e. g., resources, information, coalitions [27]) have been identified since the eighties [42]. It is also well understood that intelligent agents are driven by intentions (e. g., BDI [7]) and that they plan and perform actions in order to meet their objectives. To this purpose, they aim to make optimal use of their resources, which also includes different forms of power. So far, however, there is only little work on integration of power distribution into negotiation protocols. Bargaining Power in Electronic Negotiations 3 3 A Brief Description of Electronic Negotiations Electronic automating of negotiations was forecasted by [14] more than 20 years ago. However, the automation level of present negotiation systems is still different: Full-automated, process support and hybrid negotiation models exist today. Full-automated models work without any human interaction and are strictly structured. Process support models (e. g., INSPIRE [26] or www.ebay.com/) only facilitate the negotiation. Hybrid models are partly-automated, for example [15]. The literature shows many definitions for (electronic) negotiations; e. g. [32, 43]. A pragmatic way to define a negotiation is the following [5, p. 316]: Definition 1. A negotiation is an iterative communication and decision making process between two or more participants who: (1) cannot achieve their objectives through unilateral actions, (2) exchange information comprising offers, counteroffers and arguments, (3) deal with interdependent tasks, and (4) search for a consensus which is a compromise decision. J. S. Rosenschein and G. Zlotkin were the first who analyzed strategic interactions between self-interested agents [42, 43, 60] on the basis of the fundamental game-theoretic work [55] by J. L. von Neumann and O. Morgenstern. Their formal analysis is based on the distributed problem solving approach (see [6]) adapted from the work of J. C. Harsanyi et al. [20, 21] and D. M. Kreps et al. [30]. The underlying game-theoretic approach studies the equilibrium conditions and tries to find out the optimal strategy between identical agents [31, 37–39, 46–48]. Game-theoretic models are deemed to be mathematically elegant, but they are very restricted in use because of their assumptions of perfect rationality, unlimited resources and a perfect information situation [22, 23, 40]. In order to relax this restrictions, heuristic approaches have been adapted for electronic negotiations. Heuristic approaches solve the problematic assumption of unlimited resources by using thumb rules (e. g., [28]). Thus, the assumption of perfect rationality is also rejected. But, electronic negotiation models on the basis of heuristic approaches need an intensive evaluation, regular via simulation or empirical investigations [22, p. 210]. Finally argumentation-based negotiations have been developed. There, the agents have the possibility to reason about their positions. When the negotiation partner is persuaded, it will change its negotiation position. The argumentation-based approach increases the possibility and the quality of an agreement compared to game-theoretic or heuristic solutions and was firstly realized by K. P. Sycara [52] (negotiation support system PERSUADER). Electronic negotiation models can handle many organizational challenges [9]. For example, regarding the negotiation structure, bilateral, one-sided multilateral and double-sided multilateral negotiations are separated (protocol category) [5]. Bilateral negotiations are restricted to two negotiation partners (one buyer and one seller) and were first analyzed by [14, 50]. One-sided multilateral negotiations are deemed to be the standard form of auctions and are either characterized by one seller and many buyers or vice versa. Finally, double-sided multilateral negotiations are characterized by many buyers and many sellers (e. g., 4 Ricardo Buettner and Stefan Kirn [59]). Depending on how many attributes for the negotiation item are taken into account during the evaluation, two attribute types are distinguished: In the single-attribute case the negotiation item is evaluated only by one characteristic, normally the price. All other attributes, for instance quality or warranty, are agreed in advance and will not be negotiated. In contrast, in multi-attribute negotiations more than one characteristic are taken simultaneously into account, e. g., [26]. Beyond, the number of positions describes the quantity of independent items in a single negotiation over there a final decision is made. Electronic negotiation systems with the possibility to contract a high number of positions (e. g., [15]) are of significant practical relevance. Furthermore, negotiations can be separated in non-mediated (e. g., INSPIRE [26]) and mediated negotiations. In addition to the negotiation structure a lot of process-related challenges have been already addressed. Firstly, a negotiation can be separated into public and closed sessions [5, p. 318]. As in public negotiations new participants can take part dynamically, this is not allowed in closed sessions. Furthermore, electronic negotiations can be distinguished by the binding type. Binding negotiations (e. g., www.ebay.com/) ask for an authentication of every participant in advance. Finally, time has an enormous influence to negotiations [41, 51]: At first, S. Kraus et al. [29] took time limits into account in electronic negotiations. 4 A Negotiation Mechanism for Asymmetric Bargaining Power Distribution As shown in the last section, many negotiation challenges have been already addressed. However, works studying behavioral issues, especially concerning bargaining power, are still inadequate represented. Thus, this section presents a negotiation mechanism focussing on bargaining power. 4.1 Initial Assumptions 1. There are two agents, A and B. A is the seller and B the buyer agent. 2. Each agent wants to maximize its own utility (utilityA and utilityB). 3. The distribution of the bargaining power between A and B is different (powerA and powerB). 4. Each agent knows all relevant information, except the reservation price of the other.