Arbitration and bargaining across the pacific.

Deck, Cary ; Farmer, Amy ; Zeng, Dao-Zhi 等

1. Introduction

The increasingly global nature of the economy increases the likelihood of disputes between geographically and culturally separated people from different nations. These international interactions increasingly extend to labor relations, business contracts, and legal conflicts. To the extent that cultural norms and other factors that may vary with nationality affect concepts of fairness, patterns of negotiation, or even beliefs regarding the proper procedure for settling disputes, these differences may increase the degree of conflict or make the probability of self-negotiated settlements decline. This creates an imperative that we understand the role that such differences play in conflict resolution in order to facilitate negotiations and/or choose dispute resolution procedures that can encourage settlements.

A significant literature has addressed international negotiations in a variety of contexts; this literature identifies how individuals in various countries behave differently in dimensions that may matter for resolving conflicts. While this information is critical to understand, the literature has failed to ask whether these preferences or behaviors vary when people interact with a fellow countryman or a foreigner. In other words, do the identified differences change when agents are working with individuals outside of their own nation or culture? Leung (1997) argues that despite a plethora of research on cultural differences in negotiations, there is a lack of research on intercultural negotiations. This paper helps to fill this void through laboratory experiments in which subjects in the United States and Japan interact in real-time final offer arbitration negotiations, not only within their own groups but also across groups. As such, we are not only able to identify differences in within-group behavior; we are also able to identify how subjects alter their behavior when bargaining across international and cultural boundaries. We find that such changes in behavior do occur, suggesting that the vast literature identifying cultural differences is incomplete in helping us understand the interactions of agents. This finding is of importance irrespective of the specific form of negotiations or the countries involved; rather, in documenting behavioral changes when agents bargain across cultures, this research suggests that the quickly expanding literature documenting cultural bargaining differences may not generalize to negotiations between agents of different cultures.

The specific form of dispute resolution addressed in this paper is final-offer arbitration (FOA). In the United States, arbitration is widely applied in settling disputes that fail to be negotiated. (1) FOA requires that each party formally submit a final offer, and an arbitrator then selects one. Contrary to the standard usage, in FOA an "offer" is the amount of money a party is requesting for herself. Proposed by Stevens (1966), FOA has been employed in Major League Baseball since the 1970s to resolve salary disputes for players who are not yet eligible for free agency and can only bargain with their current team. It also has been applied to settle public sector labor contract issues, where the right to strike is not available to laborers in many countries. As we outline below, there has been an inability for researchers to fully explain agents' behavior in FOA. Given that individuals deviate from theoretically predicted behavior, it is possible that beliefs and social norms--factors that could vary by culture or nationality--may influence behavior. To the extent that we can identify such behavior, this work also contributes to the arbitration literature seeking to understand the failure of agents to settle when they are predicted to do so.

Specifically, this paper compares pre-arbitration bargaining and the performance of FOA within and between residents of two countries: the United States and Japan. (2) We have two reasons for this choice. First, Japan and the United States have been major trading partners for several decades (Graham and Sano 1989), and a successful dispute resolution method has implications for both economies. Second, the nations are geographically separated and many papers find that they are quite different in cultural background. For example, Hofstede (1980, 1991) and Schwartz (1994) argue that these cultures differ in two important dimensions: Japan is generally considered collectivist and hierarchical, while the United States is individualistic and egalitarian. (3,4) Buchan, Croson, and Johnson (2004) find that the people in these two countries have different viewpoints of fairness and the role of power that might determine their bargaining behavior. In addition to studies documenting differences in values and bargaining culture, Lituchy (1997), Brett and Okumura (1998), Beard, Beil, and Mataga (2001), and Oosterbeek, Sloof, and van de Kuilen (2004) find that economic behavior differs significantly between these two countries in many situations. However, Slemrod and Katuscak (2005) find that both the United States and Japan are societies that exhibit high levels of trust, potentially making economic transactions across national boundaries less difficult.

It is difficult to investigate the issue using field data. While naturally occurring data from FOA are available from the United States, (5) its application in Japan is rare. (6) Therefore, we rely upon laboratory methods to make this cultural comparison. Laboratory studies also have the advantage that the experimenter can control factors that typically vary across naturally occurring disputes and are often unobservable. For example, in the lab we can control the information each party has about the arbitrator's preferences and the cost of going to arbitration. This could be potentially important in our examination of interactions across the two cultures.

As a prelude to our experimental results, we find that the average settlement rates were lower within Japan at 64% than within the United States at 70%, but the difference is not statistically significant. When agents bargain internationally, we observe a significantly higher settlement rate of 82%. In part, this result may be due to the increased amount that U.S. disputants demand when facing a Japanese rival. Importantly, these results indicate that bargainers alter their behavior when faced with an opponent from the other country. While this study documents such changes in a specific bargaining environment between two specific countries, it suggests that in addition to fully understanding this application, the literature comparing intracultural negotiations across cultures is incomplete and potentially misleading.

This paper is organized as follows: Section 2 provides a more extensive discussion of the literature, section 3 presents a basic model of FOA, and section 4 describes our experimental design. The results are presented in section 5, followed by conclusions in section 6.

2. Literature Review

Relevant Arbitration Literature

Under FOA, an aggressive offer that requests a large share is likely to lose, resulting in the opponent's offer being implemented. Therefore, researchers originally expected disputants to present compromising offers in the arbitration process. However, game-theoretical studies showed that the offers are unlikely to converge when the arbitrator's notion of fairness is not completely known to the disputants (see Farber 1980; Brams and Merrill 1983). Despite this result, rational disputants are still expected to reach an agreement in the pre-arbitration negotiation process because it is rational to save their arbitration costs. In other words, the costs create a contract zone within which disputants prefer to settle rather than pursue arbitration. (7) Despite this prediction, real-world disputes frequently enter the arbitration process.

Many authors have addressed the puzzle of why players do not behave as theory suggests. For example, Shavell (1982) and Priest and Klein (1984) argue that settlement failure arises because parties make mistakes in their assessment of the disputes. Bebchuk (1984) and Reinganum and Wilde (1986) also stressed the role of asymmetric information in explaining settlement failure. However, the empirical study of Farmer, Pecorino, and Stango (2004) shows that even in cases where information is widely available, such as baseball labor negotiations, the disputants often fail to settle.

The lack of pre-arbitration settlement has consistently been observed in the laboratory through a variety of experimental designs and arbitration mechanisms. Ashenfelter et al. (1992) compare FOA with conventional arbitration and tri-offer arbitration, all of which are used in practice; Dickinson (2004, 2005) compares FOA and conventional arbitration with combined arbitration of Brams and Merrill (1986) and double-offer arbitration of Zeng, Nakamura, and Ibaraki (1996); and Deck, Farmer, and Zeng (2007) compare FOA with amended final-offer arbitration of Zeng (2003). (8) With the exception of Deck, Farmer, and Zeng (2007), settlement rates for FOA are typically below 50% in the lab.

Relevant Cultural/Nationality Literature

The above arbitration studies draw upon data from the United States. Culture and national character are believed to play an important role in economic decision making, which in part may explain why observed behavior deviates from the predictions of positive game theory. In fact, recent laboratory research has already identified the potential role of cultural norms in influencing economic outcomes. For example, Roth et al. (1991) have investigated the difference of bargaining and market behavior across four countries. Their results are amplified by Henrich (2000). Although cultural factors are not explicitly incorporated in game-theoretical models yet, it is natural to guess that dispute resolution depends on culture, since it is related to egoism, fairness, equality, and reciprocity, which appear to be sensitive to cultural socialization. Given the increase in globalization, it is critical to understand how culture affects behavior and arbitration outcomes both across and between cultures.

International relations scholars have been investigating the role of culture for some time. (9) In an effort to refine our knowledge of the influence of culture, Elgstrom (1999) finds that culture matters more under certain circumstances than it does under others. Following that, Bercovitch and Elgstrom (2001) find that four out of five specific measures of cultural differences produce failed mediation results. Moreover, not only do cultures differ in their approaches to international relations negotiations, there are also significant differences in beliefs regarding the appropriate procedures to utilize in the process of bargaining. Specifically, Fu et al. (2002) find that Chinese and Americans exhibit differences in choosing an appropriate mediator: Chinese agents prefer mutual acquaintances, while Americans prefer a party that is unconnected and disengaged from the situation. As mentioned above, a number of studies document possible differences between U.S. and Japanese negotiators (see Hofstede 1980, 1991; Schwartz 1994; Yoshimura 1997; Brett and Okumura 1998; Beard, Beil, and Mataga 2001; Buchan, Croson, and Johnson 2004; Oosterbeek, Sloof, and van de Kuilen 2004; Slemrod and Katuscak 2005). (10)

Similarly, Lew and Shore (1999) find cultural differences in cross-examination procedures in litigation settings. With regard to arbitration specifically, Gans (1997) advocates for the use of informal procedures prior to invoking formal arbitration procedures, citing that Japanese businessmen expect to solve their problems themselves. Finally, the president of the American Arbitration Association, William Slate II, cites the importance of cultural understanding, noting that international arbitration procedures are increasingly becoming a blend of cultural procedures and practices. (11)

In addition to direct examination of cultural differences in international conflict and preferences for dispute resolution mechanisms, there has been considerable research on how differences in individual behaviors, beliefs, and preferences are expected to impact how agents behave in arbitration. Kilbourne, Grunhagen, and Foley (2005) identify differences in materialism across cultures, while Brandts, Saijo, and Schram (2004) find differences in spite and cooperation behavior in voluntary contribution mechanisms. Buchan, Johnson, and Croson (2006) examine the role that nationality plays in a set of "other regarding preferences" (trust, reciprocity, and altruism), all of which would likely influence behavior in arbitration. Specifically, they find that the U.S. subjects trust in-group members more than out-group members, and they find the opposite pattern from Chinese subjects.

However, most of this literature investigates only the differences on various dimensions within a country, not whether these differences change when interacting with members of other countries. Even Brett and Okumura (1998), who led the way in comparing the outcomes of intercultural negotiations between the United States and Japan with intracultural negotiations, did not examine whether individual subjects alter their behavior when facing an opponent across cultures; rather, they simply show that the interaction across cultures interferes "with the creation of joint gains ..." (p. 507). However, it is unclear exactly what this means, as participants were not compensated based upon the negotiated outcome, a shortcoming of Lituchy (1997) as well. (12) In fact, both of these studies also suffer from confounding factors in their subject pools. Lituchy (1997) draws conclusions regarding Japanese subjects from subjects in the United States for intercultural negotiations, but from subjects in Japan for intranational negotiations. Brett and Okumura (1998) drew participants from three distinct populations, all inside the United States, for each of their three treatment conditions. Further, they actively tried to discourage Japanese participants who could not speak English well.

3. The Model

The game-theoretical model of FOA is simple. Consider two risk-neutral disputants deciding how to allocate a known fixed sum of money, [PI]. The game consists of two stages. In the first stage, two disputants (players) bargain with each other and try to reach a mutually agreeable resolution. If they fail, then they proceed to the second stage, in which arbitration occurs with cost c to each disputant. In arbitration, the arbitrator is assumed to have a belief regarding the fair share for disputant 1, which we denote by z. Thus the fair share for disputant 2 is [PI] - z. Neither disputant knows z exactly. For the sake of the experiments, we assume that z is distributed as a random variable on [0, [PI]] with uniform distribution. (13) The randomness regarding the fairness might result from the fact that different arbitrators have different notions of fairness and the indeterminacy in selecting arbitrators (Bloom and Cavanagh 1986). For example, on October 25, 2005, Tokyo District Court handed down contradictory rulings in two separate compensation suits filed by former Taiwanese and South Korean leprosy patients who were locked up in special isolation facilities during Japanese colonial rule. One judge ruled that the Taiwanese plaintiffs were entitled to compensation, but the other judge rejected the case for South Koreans (Asahi News, October 26, 2005). Alternatively, arbitrators may intentionally maintain randomness to avoid being excluded from the pool of potential arbitrators for future disputes (Ashenfelter 1987). (14)

The disputants provide final offers representing the share of disputant 1. Let x and y be the offers of disputants 1 and 2, respectively. If the offers are compatible (that is, y [greater than or equal to] x), then the final result is the average of the offers. That is, disputant 1 receives (x + y)/2 - c, and disputant 2 receives the remaining [PI] - (x + y)/2 - c. If the offers are not compatible, then the arbitrator must choose one of the offers as the final result. Specifically, if |x - z| < |y - z|, then disputant 1 receives x - c, and disputant 2 receives [PI] - x - c. If |x - z| > |y - z|, then disputant 1 receives y - c, and disputant 2 receives [PI] - y - c. If both disputants are equally distant from the arbitrator's z, then the arbitrator randomly picks one of the offers as the final result, each with probability 0.5.

According to Brams and Merrill (1983), offers x* = [PI], y* = 0 constitute the unique Nash equilibrium in this zero-sum game. The expected payoffs of the two players are both [PI]/2 - c. The contract zone is defined as the set of settlements preferred by both disputants to reaching arbitration. If the arbitration cost c is positive, then two disputants are expected to reach a pre-arbitration agreement. The range of possible agreements is [[PI]/2 - c, [PI]/2 + c] and is referred to as the contract zone. (15)

4. Experimental Design and Procedures

We conducted a series of four experimental sessions for each of the three treatments determined by the location of the subjects. In the UU treatment, both parties were students at a state university in the United States. Disputants in the JJ treatment were both students at a Japanese university. Lastly, the UJ treatment matched students at the two universities. Each subject participated in only one session. (16)

A session began with the subjects reading written directions and completing a comprehension quiz. (17) For the UU and JJ treatments, the subjects knew that their randomly selected counterparts were in the same room. However, for the UJ treatment, the subjects were told that their counterparts were being randomly selected from a group of people on the other side of the world. To help ensure the existence of the counterparts, web cameras were used to broadcast live images of each lab. The subjects at each location could see the people at the other lab. Further, one subject in the United States was randomly selected to draw a pattern on a sheet of paper and show it to the camera. Then a randomly selected subject in Japan copied that pattern on a sheet of paper and showed it to the camera. The procedure was then reversed, allowing the Japanese subject to draw some new symbol, which was redrawn by the American subject. This process is similar to the one developed by Eckel and Wilson (2006).

After all questions had been answered, the computerized experiment began. The experiment consisted of two phases. In the first 15 periods, subjects were randomly and anonymously assigned a counterpart. (18) The two parties both made final offers for the computerized arbitrator to consider. As was common information for the subjects, the arbitrator's preferred outcome was randomly selected from the interval [0,100]. (19) For these first 15 periods, c = 0, as the subjects had no choice but to be in arbitration. A one-page sheet of direction was distributed after the end of the first phase, and this was the first time the subjects learned of a second phase. The second phase of the experiment allowed the randomly selected counterparts to bargain for one minute prior to going to arbitration. During the minute, either party could make a proposal or accept the counterpart's standing proposal. Any pair that reached a self-negotiated agreement did not go to arbitration, and the two parties were not charged the cost of c = 15. (20) The second phase consisted of 10 periods. (21)

Following Roth et al. (1991), we were careful to minimize the effects of uncontrolled procedural differences across locations. The same experimenters conducted all of the sessions. (22) Unfortunately, this dictated that the treatments were not all conducted at the same time and hence the exchange rate varied somewhat. The payoffs in the experiment were denominated in Experiment Dollars. A subject's payoff was converted into cash at the rate of Experiment Dollars 100 = $1 = [yen]100. At the time of the JJ sessions the exchange rate was approximately $1 [approximately equal to] [yen]106 and for the UJ sessions it was $1 [approximately equal to] [yen]116. Hence the real payoffs are similar but not identical. Of course, studies done in different cities in the same country or over an extended period of time have different real incentives given variation in the cost of living. For each laboratory session, subjects were recruited to the lab for one hour and received a participation payment of $5 or [yen]500.

To keep the designs parallel across treatments, the number of subjects differed across treatments. In the JJ and UU treatments, where everyone was in the same room, four subjects participated in each session. Hence, for any subject there were three other people who could be randomly selected as a counterpart in a given period. In the UJ treatment there were six subjects, three at each laboratory, so that again there were three possible counterparts each period. Given the repeated nature of the experiment, it is important that the likelihood of interacting with a person again or one's ability to discipline the group be held constant across treatments.

The instructions for the experiment and the computer interface were initially written in English for the U.S. subjects. The directions were subsequently translated into Japanese by Zeng and one of his Japanese students for the Japanese subjects. Unfortunately, translating the subject's computer interface was not practical, as Japanese characters require more bytes than English characters. To make the software informative to the Japanese subjects, symbols replaced text on the screen. For example, the symbol "[??]" replaced the word "counterpart." The English directions and software for the UJ treatment also used the pictorial representation.

5. Results

In total, 56 unique subjects participated in the experiment. The data consist of 420 pairs of disputants in arbitration during the first phase of the experiment and 266 pairs of disputants who had the opportunity to settle pre-arbitration in the second phase. (23) As pairs were randomly selected each period from the subjects in a given session, the observations are not independent within sessions but are independent across sessions. Therefore we rely upon nonparametric tests of session-level data and use econometric models that handle repeated measures for decision-level data. To control for possible learning effects we omit data from the five initial periods of the experiment.

We begin our analysis by focusing on the offers that are made in arbitration during the first phase of the experiment. Figure 1 shows the distribution of offers by location of the counterpart. The left-hand panel is for subjects in the United States, and the fight-hand panel is for subjects in Japan.

Evident from this figure is that U.S. subjects want a greater share of the surplus when paired with a subject in Japan than when paired with another subject in the United States. However, Japanese subjects appear to be nearly equally demanding against counterparts from either country. This pattern is borne out statistically. We estimate a linear mixed effects model for the average offer. This model estimates a fixed effect for the treatments while allowing for each subject and each session to have a random effect. (24) Specifically, we estimate

[FIGURE 1 OMITTED]

[Offer.sub.ijt] = [[beta].sub.0] + [e.sub.i] + [zeta].sub.ij] + [[beta].sub.1][JapSub.sub.i] + [[beta].sub.2][JapCount.sub.i] + [[beta].sub.3][JapSub.sub.i] x [JapCount.sub.i] + [[epsilon].sub.ijt],

where [e.sub.i] ~ N(0, [[sigma].sup.2.sub.1]), [zeta].sub.ij] ~ N(0, [[sigma].sup.2.sub.2]), [[epsilon].sub.ijt] = [rho][[epsilon].sub.ij(t-1)] + [u.sub.ijt], and [u.sub.ijt] ~ N(O, [[sigma].sup.2.sub.3]). [Offer.sub.ijt], is the offer made by subject j in session i during period t. [JapSub.sub.ij] and [JapCount.sub.ij] are dummy variables that take on the value 1 if subject j in session i was Japanese and had a Japanese counterpart, respectively. Table 1 gives the results of the estimation for the treatment effects.

While the theoretical prediction for risk-neutral agents with common beliefs of arbitrator behavior is an offer of 100, consistent with many previous studies of FOA, what we observe is substantially more moderate. On average, U.S. subjects with counterparts in the United States make final offers of 56.4. U.S. subjects make offers of 56.4 + 10.5 >> 66.9 when facing Japanese counterparts, a statically significant difference (p-value = 0.0186). Japanese subjects make nominally higher offers when facing U.S. counterparts (56.4 + 4.1 >> 60.5) than U.S. subjects do when facing U.S. counterparts (56.4), but the difference is not significant (p-value = 0.3424). Further, Japanese subjects ask for slightly less money from Japanese counterparts (56.4 + 4.1 + 10.5 - 12.6 >> 58.4) than from U.S. counterparts (60.5). This difference is not significant based upon a likelihood ratio test of [[beta].sub.2] + [[beta].sub.3] = 0 at any standard significance level. Ultimately, there is no statistical difference between offers in the UU and JJ treatments (56.4 vs. 58.4), based upon a likelihood ratio test that [[beta].sub.1] + [[beta].sub.2] + [[beta].sub.3] = 0.

The results, reported in Table 1, allow us to formally look at the cultural hypotheses laid out by Lituchy (1997). Specifically, she asserts that Japanese subjects, coming from a collectivist society, will seek less money when facing a Japanese counterpart than when facing an American. But Americans, being individualistic, will not alter their behavior based upon the culture of their counterpart. We do not find statistical support for either of the two cultural hypotheses. (25) The first hypothesis is in the form of a one-sided alternative that [[beta].sub.2] + [[beta].sub.3] < 0, for which we fail to reject the null. The second hypothesis is in the form of a null that [[beta].sub.2] = 0, which we reject in favor of the two-sided alternative.

We now focus on the settlement rate, a typical measure of an arbitration mechanism's success. Figure 2 plots the self negotiated pre-arbitration settlement rates by session. (26) From this figure it appears that FOA is more effective in resolving international disputes than disputes in which both parties are in Japan (Wilcoxon Rank Sum test statistic W = 25, p-value = 0.058). However, there is considerable variation in the United States, so that while the average agreement rate of 70% is nominally between the Japanese rate of 64% and the international rate of 82%, it is not statistically different from either (W = 19.5, 0.886 < p-value < 0.686 and W = 20, p-value = 0.686, respectively). It is worth noting that while these agreement rates are higher than those typically reported in the laboratory, they are more in line with empirical estimates. For example, Farmer, Pecorino, and Stango (2004) report that 85% of baseball salary disputes are settled prior to arbitration.

A random effects probit model also leads to the same conclusion--that Japanese subjects are more likely to settle with U.S. counterparts than with Japanese counterparts. Table 2 provides the estimation results for a random effects probit model where the estimating equation for the probability a pair p in session j will reach arbitration is given by

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

where [PHI] denotes the cumulative distribution function of the standard normal distribution. JJ and UJ are dummy variables for the treatments. The baseline case is the UU treatment. That agreements are less likely in JJ than in UJ is evidenced by the fact that we reject the null hypothesis that [[beta].sub.1] = [[beta].sub.2] (likelihood ratio test statistic = 6.151, p-value = 0.0131).

Theory predicts not only that parties should settle, but that they should settle in the contract zone, which is the interval [35, 65] given the parameters. Overall, the vast majority of disputants settled inside the contract zone conditioned on having reached an agreement; 88% in UU, 93% in J J, and 97% in UJ. The attraction of the contract zone is evident from Figure 3, which plots the frequency of a given split by treatment. Since any agreement allocates the entire 100 experimental dollars, the pairs are identified by the size of the lion's share. The column NA denotes pairs that did not reach an agreement.

Clearly the agreements are overwhelmingly for near-equal shares of the surplus, a result that is not too surprising given previous experimental studies of bargaining; see Kagel and Roth (1995). But one may wonder why there is greater agreement typically with an equal split in UJ even though the subjects ask for more money during arbitration in this treatment.

[FIGURE 2 OMITTED]

In fact, both the Japanese subjects and the U.S. subjects take softer bargaining stances in the pre-arbitration phase of the international disputes. Looking at the final proposals made pre-arbitration for pairs that did not reach agreement, Japanese subjects were demanding 65.6 on average from their countrymen, as compared to 61.6 from Americans. Similarly, U.S. subjects demanded more of their own countrymen, asking for 63.8 on average, compared to the 58.2 requested from Japanese subjects. This pattern is consistent with the early theory regarding FOA. The more money people request once they reach arbitration, which is the case here with the international disputes, the riskier arbitration is and the greater is the incentive to settle ex ante. As demonstrated above, the U.S. subjects want more money in international arbitration, and in fact 62% of the agreements in UJ arise from a Japanese subject accepting a proposal made by their U.S. counterpart prior to reaching arbitration. For pairs that reached agreement, the Americans received 50.2 on average in the UJ treatment. Given the constant sum nature of the problem, this average was necessarily 50 for agreements in JJ and UU.

Thus, not only do we find evidence that U.S. agents alter their behavior when paired with a Japanese subject, we also find that as arbitration becomes riskier settlement increases. The increased demands of U.S. subjects have the socially positive effect of increasing settlement rates, but the distribution of the surplus is shifted in their favor as well.

[FIGURE 3 OMITTED]

6. Conclusions

This paper examines behavioral differences in arbitration. If we simply compare behavior within the United States to behavior within Japan, we do not observe significant differences. That is, disputants from both countries place similar offers in arbitration and face the same probability of reaching arbitration in disputes between countrymen. The examination of behavior across nationalities is where most previous studies have stopped; however, it is incomplete. When we directly evaluate international pairs, we see that the behavior of U.S. subjects changes as they become more aggressive, seeking a larger share of the surplus. Meanwhile, Japanese subjects become more likely to settle ex ante in international disputes, behavior that is consistent with facing counterparts that are demanding more money. In both cases we find the same nominal pattern for the other country as well, but the effect is not significant. It is important to note that while the specifics of the behavioral changes are an important part of our analysis, the primary contribution of this paper is the finding that behavior changes when agents interact interculturally when compared with their behaviors intraculturally. This finding questions the usefulness of an expanding literature documenting behavioral differences across cultures in helping us to understand intercultural bargaining.

Such intercultural effects are important in an ever more global economy. Some preliminary research using auction and public goods experiments suggests that the shift is not simply a function of varied attitudes of risk, cooperation, or optimism, but further work is needed to fully control for these possibilities. (27) It is possible that the behavioral changes are due to the two groups failing to anticipate the similar predispositions along these lines. An alternative explanation is the degree of separation in the experiments. Previous research has shown that social distance can greatly impact behavior. Specifically, studies have shown that behavior differs when agents play within their group rather than against an external party (for example, Tajfel 1979; Frey and Bohnet 1997). Glaeser et al. (2000) find that people are less trustworthy when facing someone from a different racial or national heritage. International disputes inherently involve parties from different groups to a greater degree than do disputes involving parties from the same country. When measuring social distance in another way, Cox and Deck (2005) find that U.S. subjects behave in a more monetarily self-interested way as social distance increases. This could explain in part why we observe U.S. subjects in international disputes requesting more money than they do in domestic ones. (28) In terms of pre-arbitration negotiations, there may be several possible explanations for the increased willingness of subjects to reach a settlement when in an international dispute. First, if the agents are risk-averse, then settlement is more attractive when arbitration becomes riskier, an outcome that arises from the increased demands of the U.S. subjects' behavior in arbitration. Alternatively, one could argue that subjects in the intercultural experiments were more aware that they were representing their culture and thus wanted to present a good image. While this may be the case, much like the increased social distance, such a cultural representation is an inherent part of naturally occurring international disputes as well. Although in this research we cannot separate the effects of geographic or cultural distance, this paper takes a critical first step in identifying the changes in behavior that result when individuals bargain intra- versus internationally.

Our findings suggest that future research is warranted to determine the precise factors behind the behavioral changes in these international negotiations and to determine if they exist in other contexts. (29) One thing is clear, however: it is not at all sufficient to simply document the differences in subjects' behaviors across cultures; it is critical to identify and understand the motivations behind the fact that subjects alter their behaviors when facing a member of another country or culture.

Received February 2008; accepted September 2008.

References

Asahi News. 1991. Dream Baseball Market, Ochiai Receives a Salary of 220 Million Yen (in Japanese). Asahi News 9 March, p. 3.

Asahi News. 2005. Taiwan is Entitled but Korea is Rejected (in Japanese). Asahi News 26 October, p. 1.

Ashenfelter, Orley. 1987. Arbitrator behavior. American Economic Review 77:342-46.

Ashenfelter, Orley, and David Bloom. 1984. Models of arbitrator behavior: Theory and evidence. American Economic Review 74:111-24.

Ashenfelter, Orley, Janet Currie, Henry Farber, and Matthew Spiegel. 1992. An experimental comparison of dispute rates in alternative arbitration systems. Econometrica 60:1407-33.

Avrock, Kevin. 1998. Culture and conflict resolution. Washington, D.C.: U.S. Institute of Peace Press.

Beard, T. Randolph, Richard Beil, and Yoshiharu Mataga. 2001. Reliant behavior in the United States and Japan. Economic Inquiry 39:270-9.

Bebchuk, Lucian. 1984. Litigation and settlement under imperfect information. RAND Journal of Economics 15:404-15.

Bernstein, Lisa. 1993. Understanding the limits of court-connected ADR: A critique of federal court-annexed arbitration programs. University of Pennsylvania Law Review 141:2169-259.

Bercovitch, Jacob, and Ole Elgstrom. 2001. Culture and international mediation: Exploring theoretical and empirical linkages. International Negotiation 6:3-23.

Bloom, David, and Christopher Cavanagh. 1986. An analysis of the selection of arbitrators. American Economic Review 76:408-22.

Brams, Steven, and Samuel Merrill. 1983. Equilibrium strategies for final-offer arbitration: There is no median convergence. Management Science 29:927-41.

Brams, Steven, and Samuel Merrill. 1986. Binding versus final-offer arbitration: A combination is best. Management Science 32:1346-55.

Brandts, Jordi, Tatsuyoshi Saijo, and Arthur Schram. 2004. How universal is behavior? A four country comparison of spite and cooperation in voluntary contribution mechanisms. Public Choice 119:381-424.

Brett, Jeanne, and Tetsushi Okumura. 1998. Inter- and intracultural negotiation: U.S. and Japanese negotiators. Academy of Management Journal 41:495-510.

Buchan, Nancy, and Rachel Croson. 2004. The boundaries of trust: Own and others' actions in the U.S. and China. Journal of Economic Behavior and Organization 55:484-504.

Buchan, Nancy, Rachel Croson, and Eric Johnson. 2004. When do fair beliefs influence bargaining behavior? Experimental bargaining in Japan and the United States. Journal of Consumer Research 31:181-90.

Buchan, Nancy, Eric Johnson, and Rachel Croson. 2006. Let's get personal: An international examination of the influence of communication, culture and social distance on other regarding preferences. Journal of Economic Behavior and Organization 60:373-98.

Cox, James, and Cary Deck. 2005. On the nature of reciprocal motives. Economic Inquiry 43:623-35.

Cox, James, Bruce Roberson, and Vernon Smith. 1982. Theory and behavior of single price auctions. In Research in experimental economics, edited by V. L. Smith, Volume 2. Greenwich, CT: JAI Press, pp. 1-43.

Deck, Cary, and Amy Farmer. 2007a. Bargaining over an uncertain value: Arbitration mechanisms compared. Journal of Law. Economics and Organization 23:547-79.

Deck, Cary, and Amy Farmer. 2007b. Bargaining with asymmetric uncertainty: Arbitration mechanisms compared. Research in Labor Economics 26:415-45.

Deck, Cary, Amy Farmer, and Dao-Zhi Zeng. 2007. Amended final-offer arbitration outperforms final-offer arbitration. American Law and Economics Review 9:384-407.

Dickinson, David. 2004. A comparison of conventional, final-offer, and 'combined' arbitration for dispute resolution. Industrial and Labor Relations Review 57:288 301.

Dickinson, David. 2005. Bargaining outcomes with double-offer arbitration. Experimental Economics 8:145-66.

Eckel, Catherine, and Rick Wilson. 2006. Internet cautions: Experimental games with Internet partners. Experimental Economics 9:53-66.

Elgstrom, Ole. 1999. The role of culture: A discussion and a case study. In Eclectics of Negotiation, edited by T. Hill. Cambridge, MA: The Program of Negotiation Press, pp. 31-49.

Farber, Henry. 1980. An analysis of final-offer arbitration. Journal of Conflict Resolution 24:683-705.

Farmer, Amy, Paul Pecorino, and Victor Stango. 2004. The causes of bargaining failure: Evidence from Major League Baseball. Journal of Law and Economics 47:543-68.

Frey, Bruno, and Iris Bohnet. 1997. Identification in Democratic society. Journal of Socio-Economics 26:25-38.

Fu, Ho-Ying, Michael Morris, Sau-Lai Lee, and Chi-Yue Chiu. 2002. Why do individuals follow cultural scripts? A dynamic constructivist account of American-Chinese differences in choice of mediators to resolve conflicts. Academy of Management Proceedings CM:DI D6.

Fuller, James. 1993. U.S. Tax Review. Tax notes international 7:1041-2.

Gans, Walter. 1997. Saving time and money in cross-border commercial disputes. Dispute Resolution Journal 52:50-7.

Gibbons. Robert. 1988. Learning in equilibrium models of arbitration. American Economic Review 78:896-912.

Glaeser, Edward, David Laibson, Jose Scheinkman, and Christine Soutter. 2000. Measuring trust. Quarterly Journal of Economics 115:811-46.

Graham, John, and Yoshihiro Sano. 1989. Smart bargaining: Doing business with the Japanese. Cambridge, MA: Ballinger.

Henrich, Joseph. 2000. Does culture matter in economic behavior? Ultimatum game bargaining among the Machiguenga of the Peruvian Amazon. American Economic Review 90:973-9.

Hofstede, Geert. 1980. Culture's consequences." International differences in work-related values. Newbury Park, CA: Sage.

Hofstede, Geert. 1991. Cultures and organizations: Software of the mind." Intercultural cooperation and its importance for survival. New York: McGraw-Hill.

Isaac, R. Mark, and James Walker. 1988. Group size effects in public goods provision: The voluntary contributions mechanism. Quarterly Journal of Economics 103:179-99.

Kagel, John, and Alvin Roth. 1995. The handbook of experimental economics. Princeton, NJ: Princeton University Press.

Kilbourne, William, Marko Grunhagen, and Janice Foley. 2005. A cross-cultural examination of the relationship between materialism and individual values. Journal of Economic Psychology 26:624-41.

Leung, Kwok. 1997. Negotiation and reward associations across cultures. In New Perspectives on International Industrial/Organizational Psychology, edited by P. C. Earley and M. Erez. San Francisco: Jossey-Bass, pp. 640-75.

Lew, Julian, and Laurence Shore. 1999. International commercial arbitration: Harmonizing cultural differences. Dispute Resolution Journal 54:32-8.

Lituchy, Terry. 1997. Negotiations between Japanese and Americans: The effects of collectivism on integrative outcomes. Canadian Journal of Administrative Sciences 14:386-95.

Oosterbeek, Hessel, Randolph Sloof, and Gijs van de Kuilen. 2004. Cultural differences in ultimatum game experiments: Evidence from a meta-analysis. Experimental Economics 7:171-88.

Oyserman, Daphna, Heather Coon, and Markus Kemmelmeier. 2002. Rethinking individualism and collectivism: Evaluation of theoretical assumptions and meta-analyses. Psychological Bulletin 128:3-72.

Pecorino, Paul, and Mark Van Boening. 2001. Bargaining and information: An empirical analysis of a multistage arbitration game. Journal of Labor Economics 19:922-48.

Pecorino, Paul, and Mark Van Boening. 2004. An empirical analysis of bargaining with voluntary transmission of private information. The Journal of Legal Studies 33:131-56.

Priest, George, and Benjamin Klein. 1984. The selection of disputes for litigation. The Journal of Legal Studies 13:1-55.

Reinganum, Jennifer, and Louise Wilde. 1986. Settlement, litigation and the allocation of litigation costs. RAND Journal of Economics 17:557-66.

Roth, Alvin, Vesna Prasnikar, Masahiro Okuno-Fujiwara, and Shmuel Zamir. 1991. Bargaining and market behavior in Jerusalem, Ljubljana, Pittsburgh, and Tokyo: An experimental study. American Economic Review 81:1068-95.

Schwartz, Shalom. 1994. Beyond individualism/collectivism: New cultural dimensions of values. In Individualism and Collectivism: Theory, Method, and Application, edited by U. Kim, H. C. Triandis, C. Kagitcibasi, S.-C. Chol, and G. Yoon. London: Sage, pp. 85-117.

Shavell, Steven. 1982. Suit, settlement and trial: A theoretical analysis under alternative methods for the allocation of legal costs. The Journal of Legal Studies 11:55-81.

Slate, William K. II. 2004. The culture connection in international commercial arbitration. Dispute Resolution Journal. August 1, 2004: Editorial Address.

Slemrod, Joel, and Peter Katuscak. 2005. Do trust and trustworthiness pay off? Journal of Human Resources 40:621-46.

Smith, Vernon, and James Walker. 1993. Rewards, experience and decision costs in first price auctions. Economic Inquiry 31:237-44.

Stevens, Carl. 1966. Is compulsory arbitration compatible with bargaining? Industrial Relations 5:38-52.

Tajfel, Henry. 1979. Individuals and groups in social psychology. British Journal of Social and Clinical Psychology 18:183-90.

Zeng, Dao-Zhi. 2003. An amendment to final-offer arbitration. Mathematical Social Sciences 46:9-19.

Zeng, Dao-Zhi, Shinya Nakamura, and Toshihide Ibaraki. 1996. Double-offer arbitration. Mathematical Social Sciences 31:147-70.

(1) Arbitration is typically less expensive and does not require the same level of disclosure as traditional litigation (Bernstein 1993; Fuller 1993).

(2) The goal of the current paper is to identify if and how behavior varies between domestic and international disputes. Thus the distinction between nationality, culture, etc. is not critical to our purposes.

(3) Of course this does not mean that everyone from a particular nation is identical along these dimensions, but rather suggests that, on average, people from these two countries differ in this way.

(4) There is some contradictory evidence regarding the collectivist/individualistic differences between the two peoples. Oyserman, Coon, and Kemmelmeier (2002) conducted a meta-analysis and found that Japanese people in aggregate are no more or less collectivist than Americans.

(5) Data are available from Canada as well, but of course the United States and Canada are close geographically and culturally.

(6) To the authors' knowledge, the only baseball player's salary in Japan determined by FOA was Hiromitsu Ochiai of Team Chunichi-Dragon in 1991 by FOA (see Asahi News of March 9, 1991).

(7) This result assumes symmetric information on the part of the players. Deck and Farmer (2007b) show that a contract zone may not exist when there is asymmetric information about surplus to be divided.

(8) Other experimental studies comparing FOA and conventional arbitration include Pecorino and Van Boening (2001, 2004) and Deck and Farmer (2007a, b).

(9) See Avrock (1998) for a thorough discussion of culture and conflict.

(10) We are not aware of any studies identifying the percentage of parties engaging in international negotiations that receive specific intercultural training or any studies identifying performance changes due to such training. However, with continued economic globalization and an increased reliance on arbitration, we anticipate that more and more novice individuals will be dealing with such situations.

(11) This is based upon comments in the editorial address of Dispute Resolution Journal, February 2004.

(12) The effect of rewards in an experimental study is known to be important, according to Smith and Walker (1993).

(13) A uniform distribution is easier to describe to subjects than other distributions.

(14) While most papers in the literature assume that the arbitrator's sense of fairness is independent of the offers, Ashenfelter and Bloom (1984), Ashenfelter (1987), and Gibbons (1988) consider situations in which the arbitrator learns from disputant behavior.

(15) Risk aversion (risk seeking) preferences would increase (decrease) the width of the contract zone. Disputant optimism, that is, the parties holding different beliefs as to the distribution of z, would also impact the size and location of the contract zone.

(16) While some of the subjects had participated in unrelated experiments, none had participated in previous experiments involving bargaining or arbitration.

(17) Copies of the directions and the handouts are available from the authors upon request.

(18) Throughout the experiment and the directions neutral terms were used; "counterpart" for opponent, "random number" for arbitrator's preferred outcome, etc.

(19) In reality, it is unclear which distribution best mimics any particular real situation, and as a result, any choice we make suffers from that reality. However, the uniform distribution is most easily described to subjects, thereby avoiding the difficulty that subjects don't fully understand the arbitrator's decision process. Further, the theoretical solutions are easy to derive using the uniform distribution.

(20) To the extent that the two-stage decision generates a bias, this bias generates bargainers who are more experienced with the arbitration process by the time they are placed in pre-arbitration bargaining. This then produces subjects more similar to those who would truly engage in real arbitration bargaining.

(21) Two of the JJ sessions had fewer second phase periods (six and seven periods, respectively) due to time constraints.

(22) Due to logistical constraints, the sessions were conducted over the course of a year. All of the American sessions were conducted first, followed by the Japanese sessions and then the international sessions. An additional function of the web cameras is to let all the participants in the international sessions see that the experiments are organized by an international team.

(23) Due to time constraints, some sessions did not complete 10 periods in the bargaining (second) phase of the experiment.

(24) The use of a fixed effect for each subject controls for individual specific variation. This includes a person's degree of collectivism relative to their countrymen, which Buchan, Johnson, and Croson (2006) find to be significant when controlled for based upon survey responses.

(25) Lituchy (1997) claims to find support for these hypotheses, but as described in section 2, there are some concerning aspects to her experimental design. It also bears repeating that Oyserman, Coon, and Kemmelmeier (2002) call into question the difference between the United States and Japan along this dimension.

(26) The figure and associated statistics are essentially unchanged if attention is restricted to the first six bargaining periods, the minimum number of periods in any session.

(27) In particular, we conducted a series of public goods experiments similar to the small groups treatments of Isaac and Walker (1988) and auction experiments similar to treatments dfd9 and dfd9' of Cox, Roberson, and Smith (1982). Nominally, the results are similar to previously published work.

(28) Buchan and Croson (2004) find that American subjects, as well as subjects from China, a country considered to be collectivist, are influenced by a natural manipulation of social distance in a hypothetical investment game.

(29) Two methodology improvements over the current study that should be incorporated in similar future work include the back translation of directions and adjustments for subject costs of living. The former refers to translating directions back into the original language and ensuring the two versions convey the same meaning. The latter addresses the point that exchange rates do not necessarily reflect purchasing power parity.

Cary Deck, University of Arkansas, Walton College of Business, Department of Economics, 402 WCOB, Fayetteville, AR 72701, USA; E-mail edeck@walton.uark.edu.

Amy Farmer, University of Arkansas, Walton College of Business, Department of Economics, 402 WCOB, Fayetteville, AR 72701, USA; E-mail afarmer@walton.uark.edu; corresponding author.

Dao-Zhi Zeng, Tohoku University, Graduate School of Information Sciences, Aoba 6-3-09, Aramaki, Aoba-ku, Sendai 980-8579, Japan; E-mail zeng@se.is.tohoku.ac.jp.

The authors would like to thank two anonymous referees and the editor for their valuable comments, which served to improve this paper. Deck and Farmer gratefully acknowledge support from the National Science Foundation (SES 0350709). Zeng acknowledges financial support from the Japanese Ministry of Education, Culture, Sports, Science, and Technology (Grant-in-Aid for Science Research 18530179 and 19203013).

Table 1. Mixed Effect Estimation for Final Offers

Variable          Parameter        Value      t-value   df    p-value

Constant          [[beta].sub.0]    56.3895   19.9404   504   <.0001
JapSub            [[beta].sub.1]     4.1366    0.9604    42   0.3424
JapCount          [[beta].sub.2]    10.5445    2.4480    42   0.0186
JapSub'JapCount   [[beta].sub.3]   -12.5770   -2.0472    10   0.0678

Table 2. Random Effects Probit Model for the Probability
of Reaching Arbitration

Variable   Parameter        Estimates

Constant   [[beta].sub.0]   -0.9076 (0.4082)
UJ         [[beta].sub.1]   -0.5864 (0.4813)
JJ         [[beta].sub.2]    0.2819 (0.4790)

Listed are the Generalized Estimating Equation parameter
estimates. The standard errors in parentheses are the
empirical estimates.