آیا توسل به حل اختلاف آنلاین باعث پیشبرد موافقت نامه ها می شود؟ شواهد تجربی

This paper presents an experiment performed to test the properties of an innovative bargaining mechanism (called automated negotiation) used to resolve disputes arising from Internet-based transactions. The main result shows that the settlement rule tends to chill bargaining as it creates incentives for individuals to misrepresent their true valuations, which implies that automated negotiation is not able to promote agreements. However, this perverse effect depends strongly on the conflict situation. When the threat that a disagreement occurs is more credible, the strategic effect is reduced since defendants are more interested in maximizing the efficiency of a settlement than their own expected profit. The implications of these results are then used to discuss the potential role of public regulation and reputation mechanisms in Cyberspace.

By reducing transaction costs, the open structure of the Internet offers businesses and consumers a new and powerful tool for electronic trade (Shapiro and Varian, 1998). For example, Internet technology lowers buyer search costs by providing them a wider array of products and services from different sellers than they would have in geographically defined markets. The Internet reduces seller search costs as well, by allowing them to communicate product information cost effectively to potential buyers, and by offering sellers new ways to reach buyers through targeted advertising and one-to-one marketing (Bakos, 2001; Garicano and Kaplan, 2001). From this point of view, electronic commerce is widely expected to increase social welfare by intensifying competition and helping the consumers to enjoy lower prices and more choices. However, what makes the Internet such an interesting medium for exchange creates also a number of legal obstacles which could hinder the full economic potential of electronic commerce from being reaped. The characteristics of the Internet make traditional dispute resolution through judicial procedures unsatisfactory for many controversies that arise in electronic commerce (Froomkin, 1997). For instance, suppose that a buyer purchases a product from an auction site and something goes wrong with the sale (e.g. the seller may ship a damaged item or the item may have been incorrectly described in the auction). Such a problematic Internet-based transaction raises several issues about how disputes can be resolved in the virtual environment of electronic markets. First, such a transboundary transaction creates legal uncertainty about which jurisdiction is competent and about the applicable law. Second, given that the parties are physically distant, it seems difficult to haul them into court. Third, the low transaction value may simply discourage the parties to resort to a costly legal process. Consumers who participate in this type of commerce expose themselves to a heightened level of risk due to the anonymity and location of the individual making a sale or purchase.1 During the medieval period, such international trade was governed by rules of private international law, the lex mercatoria. 2 Following this idea, many authors have argued that a distinct set of substantive rules should be created in order to regulate the electronic commerce insofar as the application of legal rules which focus on the concept of territory is questionable in the case of ubiquitous computer networks such as the Internet ( Johnson and Post, 1996). The need to regulate the electronic commerce has precipitated the creation of several online dispute resolution companies that offer computer-aided bargaining forums in order to settle conflict situations. These mechanisms consist of proprietary software which utilize the Internet as a means to more efficiently engage parties in automated negotiation of monetary sums. Automated negotiation appears to be an attractive solution to an important part of the jurisdictional challenges presented by the electronic commerce and promotes the idea of lex electronica by providing a self-applying settlement tool in which the legal location and anonymity of the parties do not matter: the resolution is crafted based on the preferences of the parties and does not require the physical presence of them ( Mefford, 1997 and Rule, 2002). In this context, many organizations have called for a variety of Internet companies to integrate online dispute resolution into their practices. Participants to the Hague Conference on Private International Law (11–12 December 2000) explored how online dispute resolution can improve trust for electronic commerce by helping to resolve business-to-consumer disputes. In the same way, the OECD Guidelines for Consumer Protection in the Context of Electronic Commerce, completed in December 1999, encourages the use of online dispute resolution. Let us elaborate the automated negotiation procedure. The resolution process begins when a plaintiff registers with an online dispute resolution service provider, such as “AllSettle” or “SettleItNow”.3 The provider then uses the information provided by the plaintiff to contact the defendant party and invite him/her to participate in online dispute resolution. If the other party accepts the invitation, they will then file a response to the plaintiff's complaint.4 From this point, the software accepts sealed offers from the parties and determine whether a settlement occurs according to the following bargaining rule (Gabuthy, 2004). Acting independently and without prior communication, plaintiff and defendant submit price offers bPbP and bDbD, respectively. If the offers converge or crisscross (i.e. bD⩾bPbD⩾bP), then the case is settled and the defendant has to pay the price asked by the plaintiff: b=bPb=bP. If the offers diverge but are within a specified range (i.e. bD(1+δ)⩾bP>bDbD(1+δ)⩾bP>bD), then the settlement price is determined by splitting the difference between the parties’ offers: b=(bP+bD)/2b=(bP+bD)/2.5 Compared to traditional bargaining, it seems that the automated negotiation procedure would be able to help the disputants to reach an agreement by providing them an additional possibility to settle their dispute (i.e. when bD-bP<0bD-bP<0), through an enlargement of the settlement zone proportional to the compatibility factor, δ∈[0,1)δ∈[0,1) (i.e. provided that bD(1+δ)-bP⩾0bD(1+δ)-bP⩾0). Our main concern is to investigate this issue by evaluating whether automated negotiation is effectively able to generate efficiency and help the parties to resolve their conflict. In order to do so, we formulate a simple model of bargaining under incomplete information that captures many of the important elements of the automated negotiation process, and then test it by conducting an experiment where we compare the individuals’ behavior to the derived theoretical predictions. Laboratory experiments serve as a powerful tool for investigating many kinds of economic phenomena because they provide the means to fully control the economic environment and simulate the basic assumptions of the model under consideration ( Smith, 1982). Furthermore, the use of experiments to generate original data on automated negotiation is necessary for an even practical reason: the confidentiality which characterizes the online dispute resolution procedures creates important limitations to get field data. The experimental methodology offers the only way to obtain initial data on automated negotiation and therefore to shed some empirical light on how disputants respond to the incentives of this innovative settlement mechanism. In literature, one mechanism that has been proposed to structure two-person bargaining under conditions of two-sided incomplete information is the well-known sealed bid k-double auction. The sealed bid k-double auction is a one-parameter family of bargaining rules for determining the terms of trade when a single seller and a single buyer voluntarily negotiate the transfer of an indivisible item. Under this mechanism, buyer and seller simultaneously choose bids pbpb and psps, respectively. Trade occurs if and only if pb⩾pspb⩾ps; in this case, the buyer pays the seller p=kpb+(1-k)psp=kpb+(1-k)ps, where k∈[0,1]k∈[0,1]. In other words, when the compatibility factor is set equal to 0, the automated negotiation procedure investigated in our paper reduces to the sealed bid k-double auction mechanism where k=0k=0 (i.e. where the seller—or the plaintiff—sets the price unilaterally if an agreement is reached with the buyer—or the defendant). In order to clearly understand the analogy between our bargaining situation and the sealed bid k-double auction, it is helpful to think of the seller (S) as a plaintiff (P) and the buyer (B) as a defendant (D) who bargain over the price at which the plaintiff will sell his claim to the lawsuit. Starting with the seminal paper of Chatterjee and Samuelson (1983), considerable theoretical attention has been given to the sealed-bid mechanism ( Myerson and Satterthwaite, 1983 and Leininger et al., 1989; Satterthwaite and Williams, 1989 and Satterthwaite and Williams, 1993; Brams and Kilgour, 1996 and Ausubel et al., 2002) and, recently, a number of authors have experimentally investigated its empirical properties ( Daniel et al., 1998, Rapoport et al., 1998, Seale et al., 2001, Parco et al., 2004 and Parco, 2006 and many others). 6 We depart from these previous studies precisely by focusing the analysis on the role that the compatibility factor may have on the individuals’ bargaining behavior. Indeed, our main insightful result shows that, contrary to what may appear to be intuitive on an a priori basis, an increase in the parameter δδ does not enhance the extent to which agreement is struck. As δδ increases, the disputants are discouraged to converge on their own which induce that the automated negotiation procedure does not significantly increase the range of possible settlements: each party has a strong individual incentive to exploit strategically the compatibility factor and to adopt aggressive positions, which leads to a collective inefficient result. The results of the experiment state that the compatibility factor plagues human interaction and show that the ability of the procedure to generate efficiency increases only when the threat that a disagreement occurs becomes more credible. When the threat that a disagreement occurs is more credible, the strategic effect due to δδ is reduced since defendants are more interested in maximizing the efficiency of a settlement than their own expected profit. The remainder of the paper is organized as follows. Section 2 presents the game theoretical analysis of automated negotiation which is based on Gabuthy (2004). Section 3 then describes the experiment designed to examine the strategic behavior of subjects and presents the theoretical predictions. The results of the experiment are analyzed in Section 4, and conclusions are drawn with respect to the observed behavior and the factors contributing to it. The implications of these results are used finally in Section 5 to discuss the potential role of public regulation and reputation mechanisms in Cyberspace.

In this paper, we analyze the theoretical properties of the automated negotiation procedure and derive equilibrium strategies for the plaintiff and the defendant. The empirical properties of this innovative bargaining mechanism are also tested by performing a set of experiments. In particular, we consider the factors that appear to determine whether a subject places bids that are close to, or exaggerated from, his reservation value. Following the experimental results, we can state that the value of the compatibility factor and the extent of the conflict are such factors: the compatibility factor creates a chilling effect insofar as the settlement rule deliberately splits the difference between the disputants’ proposals and gives them incentives to adopt aggressive bargaining positions, while this effect is reduced when the extent of the conflict is higher. The intuition behind these results is consistent with a basic finding of studies on arbitration which shows that arbitration procedures, by lowering the overall cost of disagreement, increase the incidence of disagreement: bargaining with arbitration lessens the likelihood that bargainers will reach a settlement on their own (Currie and McConnell, 1991; Ashenfelter et al., 1992 and Dickinson, 2004). In other words, despite the significant evidence that arbitrators do not simply split the difference (Bloom, 1986 and Farber and Bazerman, 1986), there does appear to be empirical evidence of a chilling effect to arbitration.19 Another source of evidence is the narcotic effect of arbitration: going to arbitration engenders “dependence” on the procedure ( Currie, 1989; Bolton and Katok, 1998 and Bolton and Katok, 2004). More precisely, a dispute decreases the probability a dispute will happen in subsequent rounds; however, this learning effect with arbitration tends to be lower than it is without. Furthermore, our experimental results raise the crucial question of how to enforce agreements reached via automated negotiation and give some elements of thinking about the potential role of public regulation and reputation mechanisms in Cyberspace. Indeed, such automated negotiation systems are offered by private companies on the electronic justice market and are, by definition, contractual. Therefore, the problem is to know how a private electronic constraint can ensure that the disputants will enter in this type of procedure ex ante and will accept not to renegotiate the settlement ex post (given that nothing other than public justice can force an agent to settle a conflict and/or execute a settlement decision). In this context, we could argue that the reputation mechanisms existing on the Internet would be a powerful way to enforce such contracts. As mentioned in the Introduction, many of the online market sites (e.g. eBay, Amazon) offer reputation management systems that allow the trading parties to submit a rating of the counterpart's performance. Therefore, we could conjecture that if one of the disputants does not respect the settlement stated by the automated algorithm, then a naming and shaming strategy would occur and allow to enforce it. Furthermore, the question concerning the acceptance (or not) of the settlement by the parties arises obviously only if the latter managed to reach an agreement during the automated negotiation process. In other words, what happens if no agreement is reached at the end of the negotiation? This question is not trivial given the poor economic performance of the mechanism and we could think intuitively that the parties will recourse to an alternative dispute resolution system, such as arbitration or mediation (which are also available online). In summary, this paper may be considered as a first step in the empirical investigation of online dispute resolution. Indeed, following the above arguments, it is obvious that further experiments will have to be done before a clear picture of how the type of mechanisms studied here perform well. Such experiments would take into account, for example, the impact of reputation and the role of alternative dispute resolution mechanisms. In this context, we feel confident that the types of question raised by our experiment will be central to the final unraveling of the puzzles presented by the computer-aided bargaining systems available in the online environment.